Mercurial > repos > guerler > hhblits

annotate lib/python3.8/site-packages/pip/_vendor/html5lib/html5parser.py @ 0:9e54283cc701 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
author guerler
date Mon, 27 Jul 2020 03:47:31 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1 from __future__ import absolute_import, division, unicode_literals
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2 from pip._vendor.six import with_metaclass, viewkeys
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
3
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
4 import types
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
5 from collections import OrderedDict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
6
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
7 from . import _inputstream
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
8 from . import _tokenizer
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
9
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
10 from . import treebuilders
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
11 from .treebuilders.base import Marker
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
12
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
13 from . import _utils
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
14 from .constants import (
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
15 spaceCharacters, asciiUpper2Lower,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
16 specialElements, headingElements, cdataElements, rcdataElements,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
17 tokenTypes, tagTokenTypes,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
18 namespaces,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
19 htmlIntegrationPointElements, mathmlTextIntegrationPointElements,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
20 adjustForeignAttributes as adjustForeignAttributesMap,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
21 adjustMathMLAttributes, adjustSVGAttributes,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
22 E,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
23 _ReparseException
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
24 )
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
25
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
26
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
27 def parse(doc, treebuilder="etree", namespaceHTMLElements=True, **kwargs):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
28 """Parse an HTML document as a string or file-like object into a tree
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
29
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
30 :arg doc: the document to parse as a string or file-like object
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
31
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
32 :arg treebuilder: the treebuilder to use when parsing
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
33
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
34 :arg namespaceHTMLElements: whether or not to namespace HTML elements
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
35
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
36 :returns: parsed tree
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
37
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
38 Example:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
39
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
40 >>> from html5lib.html5parser import parse
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
41 >>> parse('<html><body><p>This is a doc</p></body></html>')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
42 <Element u'{http://www.w3.org/1999/xhtml}html' at 0x7feac4909db0>
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
43
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
44 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
45 tb = treebuilders.getTreeBuilder(treebuilder)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
46 p = HTMLParser(tb, namespaceHTMLElements=namespaceHTMLElements)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
47 return p.parse(doc, **kwargs)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
48
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
49
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
50 def parseFragment(doc, container="div", treebuilder="etree", namespaceHTMLElements=True, **kwargs):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
51 """Parse an HTML fragment as a string or file-like object into a tree
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
52
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
53 :arg doc: the fragment to parse as a string or file-like object
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
54
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
55 :arg container: the container context to parse the fragment in
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
56
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
57 :arg treebuilder: the treebuilder to use when parsing
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
58
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
59 :arg namespaceHTMLElements: whether or not to namespace HTML elements
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
60
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
61 :returns: parsed tree
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
62
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
63 Example:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
64
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
65 >>> from html5lib.html5libparser import parseFragment
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
66 >>> parseFragment('<b>this is a fragment</b>')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
67 <Element u'DOCUMENT_FRAGMENT' at 0x7feac484b090>
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
68
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
69 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
70 tb = treebuilders.getTreeBuilder(treebuilder)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
71 p = HTMLParser(tb, namespaceHTMLElements=namespaceHTMLElements)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
72 return p.parseFragment(doc, container=container, **kwargs)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
73
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
74
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
75 def method_decorator_metaclass(function):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
76 class Decorated(type):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
77 def __new__(meta, classname, bases, classDict):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
78 for attributeName, attribute in classDict.items():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
79 if isinstance(attribute, types.FunctionType):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
80 attribute = function(attribute)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
81
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
82 classDict[attributeName] = attribute
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
83 return type.__new__(meta, classname, bases, classDict)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
84 return Decorated
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
85
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
86
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
87 class HTMLParser(object):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
88 """HTML parser
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
89
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
90 Generates a tree structure from a stream of (possibly malformed) HTML.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
91
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
92 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
93
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
94 def __init__(self, tree=None, strict=False, namespaceHTMLElements=True, debug=False):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
95 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
96 :arg tree: a treebuilder class controlling the type of tree that will be
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
97 returned. Built in treebuilders can be accessed through
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
98 html5lib.treebuilders.getTreeBuilder(treeType)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
99
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
100 :arg strict: raise an exception when a parse error is encountered
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
101
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
102 :arg namespaceHTMLElements: whether or not to namespace HTML elements
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
103
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
104 :arg debug: whether or not to enable debug mode which logs things
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
105
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
106 Example:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
107
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
108 >>> from html5lib.html5parser import HTMLParser
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
109 >>> parser = HTMLParser() # generates parser with etree builder
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
110 >>> parser = HTMLParser('lxml', strict=True) # generates parser with lxml builder which is strict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
111
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
112 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
113
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
114 # Raise an exception on the first error encountered
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
115 self.strict = strict
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
116
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
117 if tree is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
118 tree = treebuilders.getTreeBuilder("etree")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
119 self.tree = tree(namespaceHTMLElements)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
120 self.errors = []
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
121
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
122 self.phases = dict([(name, cls(self, self.tree)) for name, cls in
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
123 getPhases(debug).items()])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
124
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
125 def _parse(self, stream, innerHTML=False, container="div", scripting=False, **kwargs):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
126
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
127 self.innerHTMLMode = innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
128 self.container = container
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
129 self.scripting = scripting
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
130 self.tokenizer = _tokenizer.HTMLTokenizer(stream, parser=self, **kwargs)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
131 self.reset()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
132
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
133 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
134 self.mainLoop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
135 except _ReparseException:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
136 self.reset()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
137 self.mainLoop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
138
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
139 def reset(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
140 self.tree.reset()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
141 self.firstStartTag = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
142 self.errors = []
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
143 self.log = [] # only used with debug mode
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
144 # "quirks" / "limited quirks" / "no quirks"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
145 self.compatMode = "no quirks"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
146
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
147 if self.innerHTMLMode:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
148 self.innerHTML = self.container.lower()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
149
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
150 if self.innerHTML in cdataElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
151 self.tokenizer.state = self.tokenizer.rcdataState
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
152 elif self.innerHTML in rcdataElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
153 self.tokenizer.state = self.tokenizer.rawtextState
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
154 elif self.innerHTML == 'plaintext':
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
155 self.tokenizer.state = self.tokenizer.plaintextState
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
156 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
157 # state already is data state
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
158 # self.tokenizer.state = self.tokenizer.dataState
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
159 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
160 self.phase = self.phases["beforeHtml"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
161 self.phase.insertHtmlElement()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
162 self.resetInsertionMode()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
163 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
164 self.innerHTML = False # pylint:disable=redefined-variable-type
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
165 self.phase = self.phases["initial"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
166
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
167 self.lastPhase = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
168
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
169 self.beforeRCDataPhase = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
170
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
171 self.framesetOK = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
172
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
173 @property
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
174 def documentEncoding(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
175 """Name of the character encoding that was used to decode the input stream, or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
176 :obj:`None` if that is not determined yet
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
177
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
178 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
179 if not hasattr(self, 'tokenizer'):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
180 return None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
181 return self.tokenizer.stream.charEncoding[0].name
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
182
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
183 def isHTMLIntegrationPoint(self, element):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
184 if (element.name == "annotation-xml" and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
185 element.namespace == namespaces["mathml"]):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
186 return ("encoding" in element.attributes and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
187 element.attributes["encoding"].translate(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
188 asciiUpper2Lower) in
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
189 ("text/html", "application/xhtml+xml"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
190 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
191 return (element.namespace, element.name) in htmlIntegrationPointElements
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
192
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
193 def isMathMLTextIntegrationPoint(self, element):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
194 return (element.namespace, element.name) in mathmlTextIntegrationPointElements
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
195
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
196 def mainLoop(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
197 CharactersToken = tokenTypes["Characters"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
198 SpaceCharactersToken = tokenTypes["SpaceCharacters"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
199 StartTagToken = tokenTypes["StartTag"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
200 EndTagToken = tokenTypes["EndTag"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
201 CommentToken = tokenTypes["Comment"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
202 DoctypeToken = tokenTypes["Doctype"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
203 ParseErrorToken = tokenTypes["ParseError"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
204
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
205 for token in self.normalizedTokens():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
206 prev_token = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
207 new_token = token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
208 while new_token is not None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
209 prev_token = new_token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
210 currentNode = self.tree.openElements[-1] if self.tree.openElements else None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
211 currentNodeNamespace = currentNode.namespace if currentNode else None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
212 currentNodeName = currentNode.name if currentNode else None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
213
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
214 type = new_token["type"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
215
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
216 if type == ParseErrorToken:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
217 self.parseError(new_token["data"], new_token.get("datavars", {}))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
218 new_token = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
219 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
220 if (len(self.tree.openElements) == 0 or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
221 currentNodeNamespace == self.tree.defaultNamespace or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
222 (self.isMathMLTextIntegrationPoint(currentNode) and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
223 ((type == StartTagToken and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
224 token["name"] not in frozenset(["mglyph", "malignmark"])) or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
225 type in (CharactersToken, SpaceCharactersToken))) or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
226 (currentNodeNamespace == namespaces["mathml"] and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
227 currentNodeName == "annotation-xml" and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
228 type == StartTagToken and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
229 token["name"] == "svg") or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
230 (self.isHTMLIntegrationPoint(currentNode) and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
231 type in (StartTagToken, CharactersToken, SpaceCharactersToken))):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
232 phase = self.phase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
233 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
234 phase = self.phases["inForeignContent"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
235
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
236 if type == CharactersToken:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
237 new_token = phase.processCharacters(new_token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
238 elif type == SpaceCharactersToken:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
239 new_token = phase.processSpaceCharacters(new_token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
240 elif type == StartTagToken:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
241 new_token = phase.processStartTag(new_token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
242 elif type == EndTagToken:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
243 new_token = phase.processEndTag(new_token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
244 elif type == CommentToken:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
245 new_token = phase.processComment(new_token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
246 elif type == DoctypeToken:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
247 new_token = phase.processDoctype(new_token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
248
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
249 if (type == StartTagToken and prev_token["selfClosing"] and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
250 not prev_token["selfClosingAcknowledged"]):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
251 self.parseError("non-void-element-with-trailing-solidus",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
252 {"name": prev_token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
253
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
254 # When the loop finishes it's EOF
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
255 reprocess = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
256 phases = []
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
257 while reprocess:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
258 phases.append(self.phase)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
259 reprocess = self.phase.processEOF()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
260 if reprocess:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
261 assert self.phase not in phases
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
262
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
263 def normalizedTokens(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
264 for token in self.tokenizer:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
265 yield self.normalizeToken(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
266
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
267 def parse(self, stream, *args, **kwargs):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
268 """Parse a HTML document into a well-formed tree
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
269
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
270 :arg stream: a file-like object or string containing the HTML to be parsed
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
271
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
272 The optional encoding parameter must be a string that indicates
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
273 the encoding. If specified, that encoding will be used,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
274 regardless of any BOM or later declaration (such as in a meta
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
275 element).
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
276
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
277 :arg scripting: treat noscript elements as if JavaScript was turned on
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
278
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
279 :returns: parsed tree
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
280
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
281 Example:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
282
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
283 >>> from html5lib.html5parser import HTMLParser
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
284 >>> parser = HTMLParser()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
285 >>> parser.parse('<html><body><p>This is a doc</p></body></html>')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
286 <Element u'{http://www.w3.org/1999/xhtml}html' at 0x7feac4909db0>
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
287
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
288 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
289 self._parse(stream, False, None, *args, **kwargs)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
290 return self.tree.getDocument()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
291
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
292 def parseFragment(self, stream, *args, **kwargs):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
293 """Parse a HTML fragment into a well-formed tree fragment
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
294
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
295 :arg container: name of the element we're setting the innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
296 property if set to None, default to 'div'
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
297
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
298 :arg stream: a file-like object or string containing the HTML to be parsed
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
299
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
300 The optional encoding parameter must be a string that indicates
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
301 the encoding. If specified, that encoding will be used,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
302 regardless of any BOM or later declaration (such as in a meta
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
303 element)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
304
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
305 :arg scripting: treat noscript elements as if JavaScript was turned on
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
306
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
307 :returns: parsed tree
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
308
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
309 Example:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
310
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
311 >>> from html5lib.html5libparser import HTMLParser
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
312 >>> parser = HTMLParser()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
313 >>> parser.parseFragment('<b>this is a fragment</b>')
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
314 <Element u'DOCUMENT_FRAGMENT' at 0x7feac484b090>
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
315
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
316 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
317 self._parse(stream, True, *args, **kwargs)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
318 return self.tree.getFragment()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
319
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
320 def parseError(self, errorcode="XXX-undefined-error", datavars=None):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
321 # XXX The idea is to make errorcode mandatory.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
322 if datavars is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
323 datavars = {}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
324 self.errors.append((self.tokenizer.stream.position(), errorcode, datavars))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
325 if self.strict:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
326 raise ParseError(E[errorcode] % datavars)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
327
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
328 def normalizeToken(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
329 # HTML5 specific normalizations to the token stream
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
330 if token["type"] == tokenTypes["StartTag"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
331 raw = token["data"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
332 token["data"] = OrderedDict(raw)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
333 if len(raw) > len(token["data"]):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
334 # we had some duplicated attribute, fix so first wins
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
335 token["data"].update(raw[::-1])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
336
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
337 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
338
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
339 def adjustMathMLAttributes(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
340 adjust_attributes(token, adjustMathMLAttributes)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
341
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
342 def adjustSVGAttributes(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
343 adjust_attributes(token, adjustSVGAttributes)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
344
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
345 def adjustForeignAttributes(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
346 adjust_attributes(token, adjustForeignAttributesMap)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
347
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
348 def reparseTokenNormal(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
349 # pylint:disable=unused-argument
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
350 self.parser.phase()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
351
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
352 def resetInsertionMode(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
353 # The name of this method is mostly historical. (It's also used in the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
354 # specification.)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
355 last = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
356 newModes = {
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
357 "select": "inSelect",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
358 "td": "inCell",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
359 "th": "inCell",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
360 "tr": "inRow",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
361 "tbody": "inTableBody",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
362 "thead": "inTableBody",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
363 "tfoot": "inTableBody",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
364 "caption": "inCaption",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
365 "colgroup": "inColumnGroup",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
366 "table": "inTable",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
367 "head": "inBody",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
368 "body": "inBody",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
369 "frameset": "inFrameset",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
370 "html": "beforeHead"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
371 }
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
372 for node in self.tree.openElements[::-1]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
373 nodeName = node.name
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
374 new_phase = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
375 if node == self.tree.openElements[0]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
376 assert self.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
377 last = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
378 nodeName = self.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
379 # Check for conditions that should only happen in the innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
380 # case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
381 if nodeName in ("select", "colgroup", "head", "html"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
382 assert self.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
383
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
384 if not last and node.namespace != self.tree.defaultNamespace:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
385 continue
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
386
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
387 if nodeName in newModes:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
388 new_phase = self.phases[newModes[nodeName]]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
389 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
390 elif last:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
391 new_phase = self.phases["inBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
392 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
393
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
394 self.phase = new_phase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
395
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
396 def parseRCDataRawtext(self, token, contentType):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
397 # Generic RCDATA/RAWTEXT Parsing algorithm
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
398 assert contentType in ("RAWTEXT", "RCDATA")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
399
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
400 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
401
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
402 if contentType == "RAWTEXT":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
403 self.tokenizer.state = self.tokenizer.rawtextState
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
404 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
405 self.tokenizer.state = self.tokenizer.rcdataState
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
406
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
407 self.originalPhase = self.phase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
408
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
409 self.phase = self.phases["text"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
410
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
411
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
412 @_utils.memoize
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
413 def getPhases(debug):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
414 def log(function):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
415 """Logger that records which phase processes each token"""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
416 type_names = dict((value, key) for key, value in
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
417 tokenTypes.items())
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
418
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
419 def wrapped(self, *args, **kwargs):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
420 if function.__name__.startswith("process") and len(args) > 0:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
421 token = args[0]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
422 try:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
423 info = {"type": type_names[token['type']]}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
424 except:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
425 raise
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
426 if token['type'] in tagTokenTypes:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
427 info["name"] = token['name']
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
428
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
429 self.parser.log.append((self.parser.tokenizer.state.__name__,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
430 self.parser.phase.__class__.__name__,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
431 self.__class__.__name__,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
432 function.__name__,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
433 info))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
434 return function(self, *args, **kwargs)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
435 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
436 return function(self, *args, **kwargs)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
437 return wrapped
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
438
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
439 def getMetaclass(use_metaclass, metaclass_func):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
440 if use_metaclass:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
441 return method_decorator_metaclass(metaclass_func)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
442 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
443 return type
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
444
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
445 # pylint:disable=unused-argument
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
446 class Phase(with_metaclass(getMetaclass(debug, log))):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
447 """Base class for helper object that implements each phase of processing
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
448 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
449
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
450 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
451 self.parser = parser
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
452 self.tree = tree
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
453
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
454 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
455 raise NotImplementedError
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
456
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
457 def processComment(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
458 # For most phases the following is correct. Where it's not it will be
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
459 # overridden.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
460 self.tree.insertComment(token, self.tree.openElements[-1])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
461
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
462 def processDoctype(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
463 self.parser.parseError("unexpected-doctype")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
464
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
465 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
466 self.tree.insertText(token["data"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
467
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
468 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
469 self.tree.insertText(token["data"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
470
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
471 def processStartTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
472 return self.startTagHandler[token["name"]](token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
473
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
474 def startTagHtml(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
475 if not self.parser.firstStartTag and token["name"] == "html":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
476 self.parser.parseError("non-html-root")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
477 # XXX Need a check here to see if the first start tag token emitted is
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
478 # this token... If it's not, invoke self.parser.parseError().
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
479 for attr, value in token["data"].items():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
480 if attr not in self.tree.openElements[0].attributes:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
481 self.tree.openElements[0].attributes[attr] = value
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
482 self.parser.firstStartTag = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
483
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
484 def processEndTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
485 return self.endTagHandler[token["name"]](token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
486
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
487 class InitialPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
488 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
489 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
490
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
491 def processComment(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
492 self.tree.insertComment(token, self.tree.document)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
493
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
494 def processDoctype(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
495 name = token["name"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
496 publicId = token["publicId"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
497 systemId = token["systemId"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
498 correct = token["correct"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
499
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
500 if (name != "html" or publicId is not None or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
501 systemId is not None and systemId != "about:legacy-compat"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
502 self.parser.parseError("unknown-doctype")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
503
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
504 if publicId is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
505 publicId = ""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
506
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
507 self.tree.insertDoctype(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
508
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
509 if publicId != "":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
510 publicId = publicId.translate(asciiUpper2Lower)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
511
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
512 if (not correct or token["name"] != "html" or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
513 publicId.startswith(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
514 ("+//silmaril//dtd html pro v0r11 19970101//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
515 "-//advasoft ltd//dtd html 3.0 aswedit + extensions//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
516 "-//as//dtd html 3.0 aswedit + extensions//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
517 "-//ietf//dtd html 2.0 level 1//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
518 "-//ietf//dtd html 2.0 level 2//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
519 "-//ietf//dtd html 2.0 strict level 1//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
520 "-//ietf//dtd html 2.0 strict level 2//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
521 "-//ietf//dtd html 2.0 strict//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
522 "-//ietf//dtd html 2.0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
523 "-//ietf//dtd html 2.1e//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
524 "-//ietf//dtd html 3.0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
525 "-//ietf//dtd html 3.2 final//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
526 "-//ietf//dtd html 3.2//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
527 "-//ietf//dtd html 3//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
528 "-//ietf//dtd html level 0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
529 "-//ietf//dtd html level 1//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
530 "-//ietf//dtd html level 2//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
531 "-//ietf//dtd html level 3//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
532 "-//ietf//dtd html strict level 0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
533 "-//ietf//dtd html strict level 1//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
534 "-//ietf//dtd html strict level 2//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
535 "-//ietf//dtd html strict level 3//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
536 "-//ietf//dtd html strict//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
537 "-//ietf//dtd html//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
538 "-//metrius//dtd metrius presentational//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
539 "-//microsoft//dtd internet explorer 2.0 html strict//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
540 "-//microsoft//dtd internet explorer 2.0 html//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
541 "-//microsoft//dtd internet explorer 2.0 tables//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
542 "-//microsoft//dtd internet explorer 3.0 html strict//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
543 "-//microsoft//dtd internet explorer 3.0 html//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
544 "-//microsoft//dtd internet explorer 3.0 tables//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
545 "-//netscape comm. corp.//dtd html//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
546 "-//netscape comm. corp.//dtd strict html//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
547 "-//o'reilly and associates//dtd html 2.0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
548 "-//o'reilly and associates//dtd html extended 1.0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
549 "-//o'reilly and associates//dtd html extended relaxed 1.0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
550 "-//softquad software//dtd hotmetal pro 6.0::19990601::extensions to html 4.0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
551 "-//softquad//dtd hotmetal pro 4.0::19971010::extensions to html 4.0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
552 "-//spyglass//dtd html 2.0 extended//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
553 "-//sq//dtd html 2.0 hotmetal + extensions//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
554 "-//sun microsystems corp.//dtd hotjava html//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
555 "-//sun microsystems corp.//dtd hotjava strict html//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
556 "-//w3c//dtd html 3 1995-03-24//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
557 "-//w3c//dtd html 3.2 draft//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
558 "-//w3c//dtd html 3.2 final//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
559 "-//w3c//dtd html 3.2//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
560 "-//w3c//dtd html 3.2s draft//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
561 "-//w3c//dtd html 4.0 frameset//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
562 "-//w3c//dtd html 4.0 transitional//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
563 "-//w3c//dtd html experimental 19960712//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
564 "-//w3c//dtd html experimental 970421//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
565 "-//w3c//dtd w3 html//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
566 "-//w3o//dtd w3 html 3.0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
567 "-//webtechs//dtd mozilla html 2.0//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
568 "-//webtechs//dtd mozilla html//")) or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
569 publicId in ("-//w3o//dtd w3 html strict 3.0//en//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
570 "-/w3c/dtd html 4.0 transitional/en",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
571 "html") or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
572 publicId.startswith(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
573 ("-//w3c//dtd html 4.01 frameset//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
574 "-//w3c//dtd html 4.01 transitional//")) and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
575 systemId is None or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
576 systemId and systemId.lower() == "http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
577 self.parser.compatMode = "quirks"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
578 elif (publicId.startswith(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
579 ("-//w3c//dtd xhtml 1.0 frameset//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
580 "-//w3c//dtd xhtml 1.0 transitional//")) or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
581 publicId.startswith(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
582 ("-//w3c//dtd html 4.01 frameset//",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
583 "-//w3c//dtd html 4.01 transitional//")) and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
584 systemId is not None):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
585 self.parser.compatMode = "limited quirks"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
586
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
587 self.parser.phase = self.parser.phases["beforeHtml"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
588
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
589 def anythingElse(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
590 self.parser.compatMode = "quirks"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
591 self.parser.phase = self.parser.phases["beforeHtml"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
592
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
593 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
594 self.parser.parseError("expected-doctype-but-got-chars")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
595 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
596 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
597
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
598 def processStartTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
599 self.parser.parseError("expected-doctype-but-got-start-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
600 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
601 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
602 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
603
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
604 def processEndTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
605 self.parser.parseError("expected-doctype-but-got-end-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
606 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
607 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
608 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
609
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
610 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
611 self.parser.parseError("expected-doctype-but-got-eof")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
612 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
613 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
614
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
615 class BeforeHtmlPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
616 # helper methods
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
617 def insertHtmlElement(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
618 self.tree.insertRoot(impliedTagToken("html", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
619 self.parser.phase = self.parser.phases["beforeHead"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
620
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
621 # other
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
622 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
623 self.insertHtmlElement()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
624 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
625
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
626 def processComment(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
627 self.tree.insertComment(token, self.tree.document)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
628
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
629 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
630 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
631
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
632 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
633 self.insertHtmlElement()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
634 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
635
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
636 def processStartTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
637 if token["name"] == "html":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
638 self.parser.firstStartTag = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
639 self.insertHtmlElement()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
640 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
641
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
642 def processEndTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
643 if token["name"] not in ("head", "body", "html", "br"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
644 self.parser.parseError("unexpected-end-tag-before-html",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
645 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
646 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
647 self.insertHtmlElement()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
648 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
649
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
650 class BeforeHeadPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
651 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
652 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
653
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
654 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
655 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
656 ("head", self.startTagHead)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
657 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
658 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
659
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
660 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
661 (("head", "body", "html", "br"), self.endTagImplyHead)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
662 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
663 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
664
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
665 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
666 self.startTagHead(impliedTagToken("head", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
667 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
668
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
669 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
670 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
671
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
672 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
673 self.startTagHead(impliedTagToken("head", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
674 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
675
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
676 def startTagHtml(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
677 return self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
678
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
679 def startTagHead(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
680 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
681 self.tree.headPointer = self.tree.openElements[-1]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
682 self.parser.phase = self.parser.phases["inHead"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
683
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
684 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
685 self.startTagHead(impliedTagToken("head", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
686 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
687
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
688 def endTagImplyHead(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
689 self.startTagHead(impliedTagToken("head", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
690 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
691
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
692 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
693 self.parser.parseError("end-tag-after-implied-root",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
694 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
695
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
696 class InHeadPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
697 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
698 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
699
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
700 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
701 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
702 ("title", self.startTagTitle),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
703 (("noframes", "style"), self.startTagNoFramesStyle),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
704 ("noscript", self.startTagNoscript),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
705 ("script", self.startTagScript),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
706 (("base", "basefont", "bgsound", "command", "link"),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
707 self.startTagBaseLinkCommand),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
708 ("meta", self.startTagMeta),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
709 ("head", self.startTagHead)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
710 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
711 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
712
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
713 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
714 ("head", self.endTagHead),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
715 (("br", "html", "body"), self.endTagHtmlBodyBr)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
716 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
717 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
718
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
719 # the real thing
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
720 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
721 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
722 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
723
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
724 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
725 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
726 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
727
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
728 def startTagHtml(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
729 return self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
730
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
731 def startTagHead(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
732 self.parser.parseError("two-heads-are-not-better-than-one")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
733
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
734 def startTagBaseLinkCommand(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
735 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
736 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
737 token["selfClosingAcknowledged"] = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
738
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
739 def startTagMeta(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
740 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
741 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
742 token["selfClosingAcknowledged"] = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
743
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
744 attributes = token["data"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
745 if self.parser.tokenizer.stream.charEncoding[1] == "tentative":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
746 if "charset" in attributes:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
747 self.parser.tokenizer.stream.changeEncoding(attributes["charset"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
748 elif ("content" in attributes and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
749 "http-equiv" in attributes and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
750 attributes["http-equiv"].lower() == "content-type"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
751 # Encoding it as UTF-8 here is a hack, as really we should pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
752 # the abstract Unicode string, and just use the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
753 # ContentAttrParser on that, but using UTF-8 allows all chars
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
754 # to be encoded and as a ASCII-superset works.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
755 data = _inputstream.EncodingBytes(attributes["content"].encode("utf-8"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
756 parser = _inputstream.ContentAttrParser(data)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
757 codec = parser.parse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
758 self.parser.tokenizer.stream.changeEncoding(codec)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
759
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
760 def startTagTitle(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
761 self.parser.parseRCDataRawtext(token, "RCDATA")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
762
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
763 def startTagNoFramesStyle(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
764 # Need to decide whether to implement the scripting-disabled case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
765 self.parser.parseRCDataRawtext(token, "RAWTEXT")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
766
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
767 def startTagNoscript(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
768 if self.parser.scripting:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
769 self.parser.parseRCDataRawtext(token, "RAWTEXT")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
770 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
771 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
772 self.parser.phase = self.parser.phases["inHeadNoscript"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
773
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
774 def startTagScript(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
775 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
776 self.parser.tokenizer.state = self.parser.tokenizer.scriptDataState
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
777 self.parser.originalPhase = self.parser.phase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
778 self.parser.phase = self.parser.phases["text"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
779
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
780 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
781 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
782 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
783
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
784 def endTagHead(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
785 node = self.parser.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
786 assert node.name == "head", "Expected head got %s" % node.name
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
787 self.parser.phase = self.parser.phases["afterHead"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
788
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
789 def endTagHtmlBodyBr(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
790 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
791 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
792
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
793 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
794 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
795
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
796 def anythingElse(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
797 self.endTagHead(impliedTagToken("head"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
798
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
799 class InHeadNoscriptPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
800 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
801 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
802
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
803 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
804 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
805 (("basefont", "bgsound", "link", "meta", "noframes", "style"), self.startTagBaseLinkCommand),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
806 (("head", "noscript"), self.startTagHeadNoscript),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
807 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
808 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
809
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
810 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
811 ("noscript", self.endTagNoscript),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
812 ("br", self.endTagBr),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
813 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
814 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
815
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
816 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
817 self.parser.parseError("eof-in-head-noscript")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
818 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
819 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
820
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
821 def processComment(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
822 return self.parser.phases["inHead"].processComment(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
823
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
824 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
825 self.parser.parseError("char-in-head-noscript")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
826 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
827 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
828
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
829 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
830 return self.parser.phases["inHead"].processSpaceCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
831
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
832 def startTagHtml(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
833 return self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
834
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
835 def startTagBaseLinkCommand(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
836 return self.parser.phases["inHead"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
837
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
838 def startTagHeadNoscript(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
839 self.parser.parseError("unexpected-start-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
840
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
841 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
842 self.parser.parseError("unexpected-inhead-noscript-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
843 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
844 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
845
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
846 def endTagNoscript(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
847 node = self.parser.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
848 assert node.name == "noscript", "Expected noscript got %s" % node.name
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
849 self.parser.phase = self.parser.phases["inHead"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
850
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
851 def endTagBr(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
852 self.parser.parseError("unexpected-inhead-noscript-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
853 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
854 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
855
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
856 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
857 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
858
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
859 def anythingElse(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
860 # Caller must raise parse error first!
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
861 self.endTagNoscript(impliedTagToken("noscript"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
862
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
863 class AfterHeadPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
864 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
865 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
866
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
867 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
868 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
869 ("body", self.startTagBody),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
870 ("frameset", self.startTagFrameset),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
871 (("base", "basefont", "bgsound", "link", "meta", "noframes", "script",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
872 "style", "title"),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
873 self.startTagFromHead),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
874 ("head", self.startTagHead)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
875 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
876 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
877 self.endTagHandler = _utils.MethodDispatcher([(("body", "html", "br"),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
878 self.endTagHtmlBodyBr)])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
879 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
880
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
881 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
882 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
883 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
884
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
885 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
886 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
887 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
888
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
889 def startTagHtml(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
890 return self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
891
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
892 def startTagBody(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
893 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
894 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
895 self.parser.phase = self.parser.phases["inBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
896
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
897 def startTagFrameset(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
898 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
899 self.parser.phase = self.parser.phases["inFrameset"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
900
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
901 def startTagFromHead(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
902 self.parser.parseError("unexpected-start-tag-out-of-my-head",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
903 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
904 self.tree.openElements.append(self.tree.headPointer)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
905 self.parser.phases["inHead"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
906 for node in self.tree.openElements[::-1]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
907 if node.name == "head":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
908 self.tree.openElements.remove(node)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
909 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
910
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
911 def startTagHead(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
912 self.parser.parseError("unexpected-start-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
913
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
914 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
915 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
916 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
917
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
918 def endTagHtmlBodyBr(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
919 self.anythingElse()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
920 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
921
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
922 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
923 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
924
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
925 def anythingElse(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
926 self.tree.insertElement(impliedTagToken("body", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
927 self.parser.phase = self.parser.phases["inBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
928 self.parser.framesetOK = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
929
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
930 class InBodyPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
931 # http://www.whatwg.org/specs/web-apps/current-work/#parsing-main-inbody
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
932 # the really-really-really-very crazy mode
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
933 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
934 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
935
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
936 # Set this to the default handler
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
937 self.processSpaceCharacters = self.processSpaceCharactersNonPre
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
938
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
939 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
940 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
941 (("base", "basefont", "bgsound", "command", "link", "meta",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
942 "script", "style", "title"),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
943 self.startTagProcessInHead),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
944 ("body", self.startTagBody),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
945 ("frameset", self.startTagFrameset),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
946 (("address", "article", "aside", "blockquote", "center", "details",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
947 "dir", "div", "dl", "fieldset", "figcaption", "figure",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
948 "footer", "header", "hgroup", "main", "menu", "nav", "ol", "p",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
949 "section", "summary", "ul"),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
950 self.startTagCloseP),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
951 (headingElements, self.startTagHeading),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
952 (("pre", "listing"), self.startTagPreListing),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
953 ("form", self.startTagForm),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
954 (("li", "dd", "dt"), self.startTagListItem),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
955 ("plaintext", self.startTagPlaintext),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
956 ("a", self.startTagA),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
957 (("b", "big", "code", "em", "font", "i", "s", "small", "strike",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
958 "strong", "tt", "u"), self.startTagFormatting),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
959 ("nobr", self.startTagNobr),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
960 ("button", self.startTagButton),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
961 (("applet", "marquee", "object"), self.startTagAppletMarqueeObject),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
962 ("xmp", self.startTagXmp),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
963 ("table", self.startTagTable),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
964 (("area", "br", "embed", "img", "keygen", "wbr"),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
965 self.startTagVoidFormatting),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
966 (("param", "source", "track"), self.startTagParamSource),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
967 ("input", self.startTagInput),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
968 ("hr", self.startTagHr),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
969 ("image", self.startTagImage),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
970 ("isindex", self.startTagIsIndex),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
971 ("textarea", self.startTagTextarea),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
972 ("iframe", self.startTagIFrame),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
973 ("noscript", self.startTagNoscript),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
974 (("noembed", "noframes"), self.startTagRawtext),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
975 ("select", self.startTagSelect),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
976 (("rp", "rt"), self.startTagRpRt),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
977 (("option", "optgroup"), self.startTagOpt),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
978 (("math"), self.startTagMath),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
979 (("svg"), self.startTagSvg),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
980 (("caption", "col", "colgroup", "frame", "head",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
981 "tbody", "td", "tfoot", "th", "thead",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
982 "tr"), self.startTagMisplaced)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
983 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
984 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
985
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
986 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
987 ("body", self.endTagBody),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
988 ("html", self.endTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
989 (("address", "article", "aside", "blockquote", "button", "center",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
990 "details", "dialog", "dir", "div", "dl", "fieldset", "figcaption", "figure",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
991 "footer", "header", "hgroup", "listing", "main", "menu", "nav", "ol", "pre",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
992 "section", "summary", "ul"), self.endTagBlock),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
993 ("form", self.endTagForm),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
994 ("p", self.endTagP),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
995 (("dd", "dt", "li"), self.endTagListItem),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
996 (headingElements, self.endTagHeading),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
997 (("a", "b", "big", "code", "em", "font", "i", "nobr", "s", "small",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
998 "strike", "strong", "tt", "u"), self.endTagFormatting),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
999 (("applet", "marquee", "object"), self.endTagAppletMarqueeObject),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1000 ("br", self.endTagBr),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1001 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1002 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1003
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1004 def isMatchingFormattingElement(self, node1, node2):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1005 return (node1.name == node2.name and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1006 node1.namespace == node2.namespace and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1007 node1.attributes == node2.attributes)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1008
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1009 # helper
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1010 def addFormattingElement(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1011 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1012 element = self.tree.openElements[-1]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1013
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1014 matchingElements = []
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1015 for node in self.tree.activeFormattingElements[::-1]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1016 if node is Marker:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1017 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1018 elif self.isMatchingFormattingElement(node, element):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1019 matchingElements.append(node)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1020
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1021 assert len(matchingElements) <= 3
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1022 if len(matchingElements) == 3:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1023 self.tree.activeFormattingElements.remove(matchingElements[-1])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1024 self.tree.activeFormattingElements.append(element)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1025
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1026 # the real deal
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1027 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1028 allowed_elements = frozenset(("dd", "dt", "li", "p", "tbody", "td",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1029 "tfoot", "th", "thead", "tr", "body",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1030 "html"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1031 for node in self.tree.openElements[::-1]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1032 if node.name not in allowed_elements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1033 self.parser.parseError("expected-closing-tag-but-got-eof")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1034 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1035 # Stop parsing
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1036
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1037 def processSpaceCharactersDropNewline(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1038 # Sometimes (start of <pre>, <listing>, and <textarea> blocks) we
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1039 # want to drop leading newlines
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1040 data = token["data"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1041 self.processSpaceCharacters = self.processSpaceCharactersNonPre
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1042 if (data.startswith("\n") and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1043 self.tree.openElements[-1].name in ("pre", "listing", "textarea") and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1044 not self.tree.openElements[-1].hasContent()):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1045 data = data[1:]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1046 if data:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1047 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1048 self.tree.insertText(data)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1049
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1050 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1051 if token["data"] == "\u0000":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1052 # The tokenizer should always emit null on its own
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1053 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1054 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1055 self.tree.insertText(token["data"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1056 # This must be bad for performance
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1057 if (self.parser.framesetOK and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1058 any([char not in spaceCharacters
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1059 for char in token["data"]])):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1060 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1061
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1062 def processSpaceCharactersNonPre(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1063 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1064 self.tree.insertText(token["data"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1065
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1066 def startTagProcessInHead(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1067 return self.parser.phases["inHead"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1068
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1069 def startTagBody(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1070 self.parser.parseError("unexpected-start-tag", {"name": "body"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1071 if (len(self.tree.openElements) == 1 or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1072 self.tree.openElements[1].name != "body"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1073 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1074 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1075 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1076 for attr, value in token["data"].items():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1077 if attr not in self.tree.openElements[1].attributes:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1078 self.tree.openElements[1].attributes[attr] = value
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1079
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1080 def startTagFrameset(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1081 self.parser.parseError("unexpected-start-tag", {"name": "frameset"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1082 if (len(self.tree.openElements) == 1 or self.tree.openElements[1].name != "body"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1083 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1084 elif not self.parser.framesetOK:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1085 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1086 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1087 if self.tree.openElements[1].parent:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1088 self.tree.openElements[1].parent.removeChild(self.tree.openElements[1])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1089 while self.tree.openElements[-1].name != "html":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1090 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1091 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1092 self.parser.phase = self.parser.phases["inFrameset"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1093
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1094 def startTagCloseP(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1095 if self.tree.elementInScope("p", variant="button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1096 self.endTagP(impliedTagToken("p"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1097 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1098
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1099 def startTagPreListing(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1100 if self.tree.elementInScope("p", variant="button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1101 self.endTagP(impliedTagToken("p"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1102 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1103 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1104 self.processSpaceCharacters = self.processSpaceCharactersDropNewline
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1105
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1106 def startTagForm(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1107 if self.tree.formPointer:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1108 self.parser.parseError("unexpected-start-tag", {"name": "form"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1109 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1110 if self.tree.elementInScope("p", variant="button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1111 self.endTagP(impliedTagToken("p"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1112 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1113 self.tree.formPointer = self.tree.openElements[-1]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1114
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1115 def startTagListItem(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1116 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1117
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1118 stopNamesMap = {"li": ["li"],
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1119 "dt": ["dt", "dd"],
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1120 "dd": ["dt", "dd"]}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1121 stopNames = stopNamesMap[token["name"]]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1122 for node in reversed(self.tree.openElements):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1123 if node.name in stopNames:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1124 self.parser.phase.processEndTag(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1125 impliedTagToken(node.name, "EndTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1126 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1127 if (node.nameTuple in specialElements and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1128 node.name not in ("address", "div", "p")):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1129 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1130
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1131 if self.tree.elementInScope("p", variant="button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1132 self.parser.phase.processEndTag(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1133 impliedTagToken("p", "EndTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1134
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1135 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1136
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1137 def startTagPlaintext(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1138 if self.tree.elementInScope("p", variant="button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1139 self.endTagP(impliedTagToken("p"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1140 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1141 self.parser.tokenizer.state = self.parser.tokenizer.plaintextState
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1142
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1143 def startTagHeading(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1144 if self.tree.elementInScope("p", variant="button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1145 self.endTagP(impliedTagToken("p"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1146 if self.tree.openElements[-1].name in headingElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1147 self.parser.parseError("unexpected-start-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1148 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1149 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1150
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1151 def startTagA(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1152 afeAElement = self.tree.elementInActiveFormattingElements("a")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1153 if afeAElement:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1154 self.parser.parseError("unexpected-start-tag-implies-end-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1155 {"startName": "a", "endName": "a"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1156 self.endTagFormatting(impliedTagToken("a"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1157 if afeAElement in self.tree.openElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1158 self.tree.openElements.remove(afeAElement)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1159 if afeAElement in self.tree.activeFormattingElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1160 self.tree.activeFormattingElements.remove(afeAElement)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1161 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1162 self.addFormattingElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1163
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1164 def startTagFormatting(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1165 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1166 self.addFormattingElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1167
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1168 def startTagNobr(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1169 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1170 if self.tree.elementInScope("nobr"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1171 self.parser.parseError("unexpected-start-tag-implies-end-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1172 {"startName": "nobr", "endName": "nobr"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1173 self.processEndTag(impliedTagToken("nobr"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1174 # XXX Need tests that trigger the following
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1175 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1176 self.addFormattingElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1177
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1178 def startTagButton(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1179 if self.tree.elementInScope("button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1180 self.parser.parseError("unexpected-start-tag-implies-end-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1181 {"startName": "button", "endName": "button"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1182 self.processEndTag(impliedTagToken("button"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1183 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1184 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1185 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1186 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1187 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1188
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1189 def startTagAppletMarqueeObject(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1190 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1191 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1192 self.tree.activeFormattingElements.append(Marker)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1193 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1194
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1195 def startTagXmp(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1196 if self.tree.elementInScope("p", variant="button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1197 self.endTagP(impliedTagToken("p"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1198 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1199 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1200 self.parser.parseRCDataRawtext(token, "RAWTEXT")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1201
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1202 def startTagTable(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1203 if self.parser.compatMode != "quirks":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1204 if self.tree.elementInScope("p", variant="button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1205 self.processEndTag(impliedTagToken("p"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1206 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1207 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1208 self.parser.phase = self.parser.phases["inTable"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1209
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1210 def startTagVoidFormatting(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1211 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1212 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1213 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1214 token["selfClosingAcknowledged"] = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1215 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1216
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1217 def startTagInput(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1218 framesetOK = self.parser.framesetOK
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1219 self.startTagVoidFormatting(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1220 if ("type" in token["data"] and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1221 token["data"]["type"].translate(asciiUpper2Lower) == "hidden"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1222 # input type=hidden doesn't change framesetOK
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1223 self.parser.framesetOK = framesetOK
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1224
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1225 def startTagParamSource(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1226 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1227 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1228 token["selfClosingAcknowledged"] = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1229
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1230 def startTagHr(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1231 if self.tree.elementInScope("p", variant="button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1232 self.endTagP(impliedTagToken("p"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1233 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1234 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1235 token["selfClosingAcknowledged"] = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1236 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1237
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1238 def startTagImage(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1239 # No really...
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1240 self.parser.parseError("unexpected-start-tag-treated-as",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1241 {"originalName": "image", "newName": "img"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1242 self.processStartTag(impliedTagToken("img", "StartTag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1243 attributes=token["data"],
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1244 selfClosing=token["selfClosing"]))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1245
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1246 def startTagIsIndex(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1247 self.parser.parseError("deprecated-tag", {"name": "isindex"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1248 if self.tree.formPointer:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1249 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1250 form_attrs = {}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1251 if "action" in token["data"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1252 form_attrs["action"] = token["data"]["action"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1253 self.processStartTag(impliedTagToken("form", "StartTag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1254 attributes=form_attrs))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1255 self.processStartTag(impliedTagToken("hr", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1256 self.processStartTag(impliedTagToken("label", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1257 # XXX Localization ...
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1258 if "prompt" in token["data"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1259 prompt = token["data"]["prompt"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1260 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1261 prompt = "This is a searchable index. Enter search keywords: "
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1262 self.processCharacters(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1263 {"type": tokenTypes["Characters"], "data": prompt})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1264 attributes = token["data"].copy()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1265 if "action" in attributes:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1266 del attributes["action"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1267 if "prompt" in attributes:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1268 del attributes["prompt"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1269 attributes["name"] = "isindex"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1270 self.processStartTag(impliedTagToken("input", "StartTag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1271 attributes=attributes,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1272 selfClosing=token["selfClosing"]))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1273 self.processEndTag(impliedTagToken("label"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1274 self.processStartTag(impliedTagToken("hr", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1275 self.processEndTag(impliedTagToken("form"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1276
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1277 def startTagTextarea(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1278 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1279 self.parser.tokenizer.state = self.parser.tokenizer.rcdataState
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1280 self.processSpaceCharacters = self.processSpaceCharactersDropNewline
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1281 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1282
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1283 def startTagIFrame(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1284 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1285 self.startTagRawtext(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1286
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1287 def startTagNoscript(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1288 if self.parser.scripting:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1289 self.startTagRawtext(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1290 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1291 self.startTagOther(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1292
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1293 def startTagRawtext(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1294 """iframe, noembed noframes, noscript(if scripting enabled)"""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1295 self.parser.parseRCDataRawtext(token, "RAWTEXT")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1296
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1297 def startTagOpt(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1298 if self.tree.openElements[-1].name == "option":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1299 self.parser.phase.processEndTag(impliedTagToken("option"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1300 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1301 self.parser.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1302
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1303 def startTagSelect(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1304 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1305 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1306 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1307 if self.parser.phase in (self.parser.phases["inTable"],
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1308 self.parser.phases["inCaption"],
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1309 self.parser.phases["inColumnGroup"],
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1310 self.parser.phases["inTableBody"],
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1311 self.parser.phases["inRow"],
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1312 self.parser.phases["inCell"]):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1313 self.parser.phase = self.parser.phases["inSelectInTable"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1314 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1315 self.parser.phase = self.parser.phases["inSelect"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1316
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1317 def startTagRpRt(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1318 if self.tree.elementInScope("ruby"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1319 self.tree.generateImpliedEndTags()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1320 if self.tree.openElements[-1].name != "ruby":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1321 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1322 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1323
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1324 def startTagMath(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1325 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1326 self.parser.adjustMathMLAttributes(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1327 self.parser.adjustForeignAttributes(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1328 token["namespace"] = namespaces["mathml"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1329 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1330 # Need to get the parse error right for the case where the token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1331 # has a namespace not equal to the xmlns attribute
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1332 if token["selfClosing"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1333 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1334 token["selfClosingAcknowledged"] = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1335
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1336 def startTagSvg(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1337 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1338 self.parser.adjustSVGAttributes(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1339 self.parser.adjustForeignAttributes(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1340 token["namespace"] = namespaces["svg"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1341 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1342 # Need to get the parse error right for the case where the token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1343 # has a namespace not equal to the xmlns attribute
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1344 if token["selfClosing"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1345 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1346 token["selfClosingAcknowledged"] = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1347
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1348 def startTagMisplaced(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1349 """ Elements that should be children of other elements that have a
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1350 different insertion mode; here they are ignored
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1351 "caption", "col", "colgroup", "frame", "frameset", "head",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1352 "option", "optgroup", "tbody", "td", "tfoot", "th", "thead",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1353 "tr", "noscript"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1354 """
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1355 self.parser.parseError("unexpected-start-tag-ignored", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1356
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1357 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1358 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1359 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1360
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1361 def endTagP(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1362 if not self.tree.elementInScope("p", variant="button"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1363 self.startTagCloseP(impliedTagToken("p", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1364 self.parser.parseError("unexpected-end-tag", {"name": "p"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1365 self.endTagP(impliedTagToken("p", "EndTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1366 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1367 self.tree.generateImpliedEndTags("p")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1368 if self.tree.openElements[-1].name != "p":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1369 self.parser.parseError("unexpected-end-tag", {"name": "p"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1370 node = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1371 while node.name != "p":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1372 node = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1373
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1374 def endTagBody(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1375 if not self.tree.elementInScope("body"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1376 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1377 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1378 elif self.tree.openElements[-1].name != "body":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1379 for node in self.tree.openElements[2:]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1380 if node.name not in frozenset(("dd", "dt", "li", "optgroup",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1381 "option", "p", "rp", "rt",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1382 "tbody", "td", "tfoot",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1383 "th", "thead", "tr", "body",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1384 "html")):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1385 # Not sure this is the correct name for the parse error
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1386 self.parser.parseError(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1387 "expected-one-end-tag-but-got-another",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1388 {"gotName": "body", "expectedName": node.name})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1389 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1390 self.parser.phase = self.parser.phases["afterBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1391
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1392 def endTagHtml(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1393 # We repeat the test for the body end tag token being ignored here
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1394 if self.tree.elementInScope("body"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1395 self.endTagBody(impliedTagToken("body"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1396 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1397
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1398 def endTagBlock(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1399 # Put us back in the right whitespace handling mode
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1400 if token["name"] == "pre":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1401 self.processSpaceCharacters = self.processSpaceCharactersNonPre
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1402 inScope = self.tree.elementInScope(token["name"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1403 if inScope:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1404 self.tree.generateImpliedEndTags()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1405 if self.tree.openElements[-1].name != token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1406 self.parser.parseError("end-tag-too-early", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1407 if inScope:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1408 node = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1409 while node.name != token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1410 node = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1411
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1412 def endTagForm(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1413 node = self.tree.formPointer
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1414 self.tree.formPointer = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1415 if node is None or not self.tree.elementInScope(node):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1416 self.parser.parseError("unexpected-end-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1417 {"name": "form"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1418 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1419 self.tree.generateImpliedEndTags()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1420 if self.tree.openElements[-1] != node:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1421 self.parser.parseError("end-tag-too-early-ignored",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1422 {"name": "form"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1423 self.tree.openElements.remove(node)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1424
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1425 def endTagListItem(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1426 if token["name"] == "li":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1427 variant = "list"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1428 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1429 variant = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1430 if not self.tree.elementInScope(token["name"], variant=variant):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1431 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1432 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1433 self.tree.generateImpliedEndTags(exclude=token["name"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1434 if self.tree.openElements[-1].name != token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1435 self.parser.parseError(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1436 "end-tag-too-early",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1437 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1438 node = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1439 while node.name != token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1440 node = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1441
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1442 def endTagHeading(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1443 for item in headingElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1444 if self.tree.elementInScope(item):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1445 self.tree.generateImpliedEndTags()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1446 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1447 if self.tree.openElements[-1].name != token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1448 self.parser.parseError("end-tag-too-early", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1449
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1450 for item in headingElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1451 if self.tree.elementInScope(item):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1452 item = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1453 while item.name not in headingElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1454 item = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1455 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1456
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1457 def endTagFormatting(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1458 """The much-feared adoption agency algorithm"""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1459 # http://svn.whatwg.org/webapps/complete.html#adoptionAgency revision 7867
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1460 # XXX Better parseError messages appreciated.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1461
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1462 # Step 1
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1463 outerLoopCounter = 0
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1464
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1465 # Step 2
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1466 while outerLoopCounter < 8:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1467
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1468 # Step 3
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1469 outerLoopCounter += 1
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1470
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1471 # Step 4:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1472
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1473 # Let the formatting element be the last element in
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1474 # the list of active formatting elements that:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1475 # - is between the end of the list and the last scope
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1476 # marker in the list, if any, or the start of the list
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1477 # otherwise, and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1478 # - has the same tag name as the token.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1479 formattingElement = self.tree.elementInActiveFormattingElements(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1480 token["name"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1481 if (not formattingElement or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1482 (formattingElement in self.tree.openElements and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1483 not self.tree.elementInScope(formattingElement.name))):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1484 # If there is no such node, then abort these steps
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1485 # and instead act as described in the "any other
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1486 # end tag" entry below.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1487 self.endTagOther(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1488 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1489
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1490 # Otherwise, if there is such a node, but that node is
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1491 # not in the stack of open elements, then this is a
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1492 # parse error; remove the element from the list, and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1493 # abort these steps.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1494 elif formattingElement not in self.tree.openElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1495 self.parser.parseError("adoption-agency-1.2", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1496 self.tree.activeFormattingElements.remove(formattingElement)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1497 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1498
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1499 # Otherwise, if there is such a node, and that node is
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1500 # also in the stack of open elements, but the element
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1501 # is not in scope, then this is a parse error; ignore
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1502 # the token, and abort these steps.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1503 elif not self.tree.elementInScope(formattingElement.name):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1504 self.parser.parseError("adoption-agency-4.4", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1505 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1506
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1507 # Otherwise, there is a formatting element and that
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1508 # element is in the stack and is in scope. If the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1509 # element is not the current node, this is a parse
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1510 # error. In any case, proceed with the algorithm as
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1511 # written in the following steps.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1512 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1513 if formattingElement != self.tree.openElements[-1]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1514 self.parser.parseError("adoption-agency-1.3", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1515
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1516 # Step 5:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1517
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1518 # Let the furthest block be the topmost node in the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1519 # stack of open elements that is lower in the stack
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1520 # than the formatting element, and is an element in
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1521 # the special category. There might not be one.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1522 afeIndex = self.tree.openElements.index(formattingElement)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1523 furthestBlock = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1524 for element in self.tree.openElements[afeIndex:]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1525 if element.nameTuple in specialElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1526 furthestBlock = element
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1527 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1528
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1529 # Step 6:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1530
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1531 # If there is no furthest block, then the UA must
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1532 # first pop all the nodes from the bottom of the stack
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1533 # of open elements, from the current node up to and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1534 # including the formatting element, then remove the
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1535 # formatting element from the list of active
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1536 # formatting elements, and finally abort these steps.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1537 if furthestBlock is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1538 element = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1539 while element != formattingElement:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1540 element = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1541 self.tree.activeFormattingElements.remove(element)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1542 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1543
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1544 # Step 7
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1545 commonAncestor = self.tree.openElements[afeIndex - 1]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1546
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1547 # Step 8:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1548 # The bookmark is supposed to help us identify where to reinsert
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1549 # nodes in step 15. We have to ensure that we reinsert nodes after
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1550 # the node before the active formatting element. Note the bookmark
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1551 # can move in step 9.7
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1552 bookmark = self.tree.activeFormattingElements.index(formattingElement)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1553
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1554 # Step 9
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1555 lastNode = node = furthestBlock
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1556 innerLoopCounter = 0
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1557
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1558 index = self.tree.openElements.index(node)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1559 while innerLoopCounter < 3:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1560 innerLoopCounter += 1
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1561 # Node is element before node in open elements
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1562 index -= 1
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1563 node = self.tree.openElements[index]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1564 if node not in self.tree.activeFormattingElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1565 self.tree.openElements.remove(node)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1566 continue
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1567 # Step 9.6
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1568 if node == formattingElement:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1569 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1570 # Step 9.7
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1571 if lastNode == furthestBlock:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1572 bookmark = self.tree.activeFormattingElements.index(node) + 1
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1573 # Step 9.8
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1574 clone = node.cloneNode()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1575 # Replace node with clone
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1576 self.tree.activeFormattingElements[
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1577 self.tree.activeFormattingElements.index(node)] = clone
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1578 self.tree.openElements[
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1579 self.tree.openElements.index(node)] = clone
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1580 node = clone
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1581 # Step 9.9
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1582 # Remove lastNode from its parents, if any
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1583 if lastNode.parent:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1584 lastNode.parent.removeChild(lastNode)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1585 node.appendChild(lastNode)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1586 # Step 9.10
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1587 lastNode = node
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1588
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1589 # Step 10
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1590 # Foster parent lastNode if commonAncestor is a
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1591 # table, tbody, tfoot, thead, or tr we need to foster
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1592 # parent the lastNode
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1593 if lastNode.parent:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1594 lastNode.parent.removeChild(lastNode)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1595
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1596 if commonAncestor.name in frozenset(("table", "tbody", "tfoot", "thead", "tr")):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1597 parent, insertBefore = self.tree.getTableMisnestedNodePosition()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1598 parent.insertBefore(lastNode, insertBefore)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1599 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1600 commonAncestor.appendChild(lastNode)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1601
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1602 # Step 11
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1603 clone = formattingElement.cloneNode()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1604
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1605 # Step 12
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1606 furthestBlock.reparentChildren(clone)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1607
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1608 # Step 13
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1609 furthestBlock.appendChild(clone)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1610
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1611 # Step 14
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1612 self.tree.activeFormattingElements.remove(formattingElement)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1613 self.tree.activeFormattingElements.insert(bookmark, clone)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1614
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1615 # Step 15
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1616 self.tree.openElements.remove(formattingElement)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1617 self.tree.openElements.insert(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1618 self.tree.openElements.index(furthestBlock) + 1, clone)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1619
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1620 def endTagAppletMarqueeObject(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1621 if self.tree.elementInScope(token["name"]):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1622 self.tree.generateImpliedEndTags()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1623 if self.tree.openElements[-1].name != token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1624 self.parser.parseError("end-tag-too-early", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1625
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1626 if self.tree.elementInScope(token["name"]):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1627 element = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1628 while element.name != token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1629 element = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1630 self.tree.clearActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1631
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1632 def endTagBr(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1633 self.parser.parseError("unexpected-end-tag-treated-as",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1634 {"originalName": "br", "newName": "br element"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1635 self.tree.reconstructActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1636 self.tree.insertElement(impliedTagToken("br", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1637 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1638
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1639 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1640 for node in self.tree.openElements[::-1]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1641 if node.name == token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1642 self.tree.generateImpliedEndTags(exclude=token["name"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1643 if self.tree.openElements[-1].name != token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1644 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1645 while self.tree.openElements.pop() != node:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1646 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1647 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1648 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1649 if node.nameTuple in specialElements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1650 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1651 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1652
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1653 class TextPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1654 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1655 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1656 self.startTagHandler = _utils.MethodDispatcher([])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1657 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1658 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1659 ("script", self.endTagScript)])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1660 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1661
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1662 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1663 self.tree.insertText(token["data"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1664
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1665 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1666 self.parser.parseError("expected-named-closing-tag-but-got-eof",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1667 {"name": self.tree.openElements[-1].name})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1668 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1669 self.parser.phase = self.parser.originalPhase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1670 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1671
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1672 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1673 assert False, "Tried to process start tag %s in RCDATA/RAWTEXT mode" % token['name']
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1674
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1675 def endTagScript(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1676 node = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1677 assert node.name == "script"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1678 self.parser.phase = self.parser.originalPhase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1679 # The rest of this method is all stuff that only happens if
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1680 # document.write works
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1681
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1682 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1683 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1684 self.parser.phase = self.parser.originalPhase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1685
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1686 class InTablePhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1687 # http://www.whatwg.org/specs/web-apps/current-work/#in-table
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1688 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1689 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1690 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1691 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1692 ("caption", self.startTagCaption),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1693 ("colgroup", self.startTagColgroup),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1694 ("col", self.startTagCol),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1695 (("tbody", "tfoot", "thead"), self.startTagRowGroup),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1696 (("td", "th", "tr"), self.startTagImplyTbody),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1697 ("table", self.startTagTable),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1698 (("style", "script"), self.startTagStyleScript),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1699 ("input", self.startTagInput),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1700 ("form", self.startTagForm)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1701 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1702 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1703
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1704 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1705 ("table", self.endTagTable),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1706 (("body", "caption", "col", "colgroup", "html", "tbody", "td",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1707 "tfoot", "th", "thead", "tr"), self.endTagIgnore)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1708 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1709 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1710
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1711 # helper methods
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1712 def clearStackToTableContext(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1713 # "clear the stack back to a table context"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1714 while self.tree.openElements[-1].name not in ("table", "html"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1715 # self.parser.parseError("unexpected-implied-end-tag-in-table",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1716 # {"name": self.tree.openElements[-1].name})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1717 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1718 # When the current node is <html> it's an innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1719
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1720 # processing methods
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1721 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1722 if self.tree.openElements[-1].name != "html":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1723 self.parser.parseError("eof-in-table")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1724 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1725 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1726 # Stop parsing
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1727
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1728 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1729 originalPhase = self.parser.phase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1730 self.parser.phase = self.parser.phases["inTableText"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1731 self.parser.phase.originalPhase = originalPhase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1732 self.parser.phase.processSpaceCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1733
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1734 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1735 originalPhase = self.parser.phase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1736 self.parser.phase = self.parser.phases["inTableText"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1737 self.parser.phase.originalPhase = originalPhase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1738 self.parser.phase.processCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1739
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1740 def insertText(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1741 # If we get here there must be at least one non-whitespace character
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1742 # Do the table magic!
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1743 self.tree.insertFromTable = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1744 self.parser.phases["inBody"].processCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1745 self.tree.insertFromTable = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1746
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1747 def startTagCaption(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1748 self.clearStackToTableContext()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1749 self.tree.activeFormattingElements.append(Marker)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1750 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1751 self.parser.phase = self.parser.phases["inCaption"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1752
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1753 def startTagColgroup(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1754 self.clearStackToTableContext()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1755 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1756 self.parser.phase = self.parser.phases["inColumnGroup"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1757
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1758 def startTagCol(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1759 self.startTagColgroup(impliedTagToken("colgroup", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1760 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1761
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1762 def startTagRowGroup(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1763 self.clearStackToTableContext()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1764 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1765 self.parser.phase = self.parser.phases["inTableBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1766
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1767 def startTagImplyTbody(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1768 self.startTagRowGroup(impliedTagToken("tbody", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1769 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1770
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1771 def startTagTable(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1772 self.parser.parseError("unexpected-start-tag-implies-end-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1773 {"startName": "table", "endName": "table"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1774 self.parser.phase.processEndTag(impliedTagToken("table"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1775 if not self.parser.innerHTML:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1776 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1777
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1778 def startTagStyleScript(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1779 return self.parser.phases["inHead"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1780
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1781 def startTagInput(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1782 if ("type" in token["data"] and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1783 token["data"]["type"].translate(asciiUpper2Lower) == "hidden"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1784 self.parser.parseError("unexpected-hidden-input-in-table")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1785 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1786 # XXX associate with form
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1787 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1788 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1789 self.startTagOther(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1790
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1791 def startTagForm(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1792 self.parser.parseError("unexpected-form-in-table")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1793 if self.tree.formPointer is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1794 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1795 self.tree.formPointer = self.tree.openElements[-1]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1796 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1797
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1798 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1799 self.parser.parseError("unexpected-start-tag-implies-table-voodoo", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1800 # Do the table magic!
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1801 self.tree.insertFromTable = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1802 self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1803 self.tree.insertFromTable = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1804
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1805 def endTagTable(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1806 if self.tree.elementInScope("table", variant="table"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1807 self.tree.generateImpliedEndTags()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1808 if self.tree.openElements[-1].name != "table":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1809 self.parser.parseError("end-tag-too-early-named",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1810 {"gotName": "table",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1811 "expectedName": self.tree.openElements[-1].name})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1812 while self.tree.openElements[-1].name != "table":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1813 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1814 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1815 self.parser.resetInsertionMode()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1816 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1817 # innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1818 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1819 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1820
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1821 def endTagIgnore(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1822 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1823
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1824 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1825 self.parser.parseError("unexpected-end-tag-implies-table-voodoo", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1826 # Do the table magic!
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1827 self.tree.insertFromTable = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1828 self.parser.phases["inBody"].processEndTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1829 self.tree.insertFromTable = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1830
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1831 class InTableTextPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1832 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1833 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1834 self.originalPhase = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1835 self.characterTokens = []
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1836
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1837 def flushCharacters(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1838 data = "".join([item["data"] for item in self.characterTokens])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1839 if any([item not in spaceCharacters for item in data]):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1840 token = {"type": tokenTypes["Characters"], "data": data}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1841 self.parser.phases["inTable"].insertText(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1842 elif data:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1843 self.tree.insertText(data)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1844 self.characterTokens = []
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1845
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1846 def processComment(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1847 self.flushCharacters()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1848 self.parser.phase = self.originalPhase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1849 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1850
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1851 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1852 self.flushCharacters()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1853 self.parser.phase = self.originalPhase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1854 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1855
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1856 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1857 if token["data"] == "\u0000":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1858 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1859 self.characterTokens.append(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1860
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1861 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1862 # pretty sure we should never reach here
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1863 self.characterTokens.append(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1864 # assert False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1865
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1866 def processStartTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1867 self.flushCharacters()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1868 self.parser.phase = self.originalPhase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1869 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1870
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1871 def processEndTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1872 self.flushCharacters()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1873 self.parser.phase = self.originalPhase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1874 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1875
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1876 class InCaptionPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1877 # http://www.whatwg.org/specs/web-apps/current-work/#in-caption
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1878 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1879 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1880
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1881 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1882 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1883 (("caption", "col", "colgroup", "tbody", "td", "tfoot", "th",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1884 "thead", "tr"), self.startTagTableElement)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1885 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1886 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1887
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1888 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1889 ("caption", self.endTagCaption),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1890 ("table", self.endTagTable),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1891 (("body", "col", "colgroup", "html", "tbody", "td", "tfoot", "th",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1892 "thead", "tr"), self.endTagIgnore)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1893 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1894 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1895
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1896 def ignoreEndTagCaption(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1897 return not self.tree.elementInScope("caption", variant="table")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1898
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1899 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1900 self.parser.phases["inBody"].processEOF()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1901
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1902 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1903 return self.parser.phases["inBody"].processCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1904
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1905 def startTagTableElement(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1906 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1907 # XXX Have to duplicate logic here to find out if the tag is ignored
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1908 ignoreEndTag = self.ignoreEndTagCaption()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1909 self.parser.phase.processEndTag(impliedTagToken("caption"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1910 if not ignoreEndTag:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1911 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1912
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1913 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1914 return self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1915
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1916 def endTagCaption(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1917 if not self.ignoreEndTagCaption():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1918 # AT this code is quite similar to endTagTable in "InTable"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1919 self.tree.generateImpliedEndTags()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1920 if self.tree.openElements[-1].name != "caption":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1921 self.parser.parseError("expected-one-end-tag-but-got-another",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1922 {"gotName": "caption",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1923 "expectedName": self.tree.openElements[-1].name})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1924 while self.tree.openElements[-1].name != "caption":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1925 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1926 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1927 self.tree.clearActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1928 self.parser.phase = self.parser.phases["inTable"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1929 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1930 # innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1931 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1932 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1933
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1934 def endTagTable(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1935 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1936 ignoreEndTag = self.ignoreEndTagCaption()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1937 self.parser.phase.processEndTag(impliedTagToken("caption"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1938 if not ignoreEndTag:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1939 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1940
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1941 def endTagIgnore(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1942 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1943
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1944 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1945 return self.parser.phases["inBody"].processEndTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1946
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1947 class InColumnGroupPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1948 # http://www.whatwg.org/specs/web-apps/current-work/#in-column
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1949
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1950 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1951 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1952
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1953 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1954 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1955 ("col", self.startTagCol)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1956 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1957 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1958
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1959 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1960 ("colgroup", self.endTagColgroup),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1961 ("col", self.endTagCol)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1962 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1963 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1964
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1965 def ignoreEndTagColgroup(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1966 return self.tree.openElements[-1].name == "html"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1967
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1968 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1969 if self.tree.openElements[-1].name == "html":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1970 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1971 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1972 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1973 ignoreEndTag = self.ignoreEndTagColgroup()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1974 self.endTagColgroup(impliedTagToken("colgroup"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1975 if not ignoreEndTag:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1976 return True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1977
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1978 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1979 ignoreEndTag = self.ignoreEndTagColgroup()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1980 self.endTagColgroup(impliedTagToken("colgroup"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1981 if not ignoreEndTag:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1982 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1983
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1984 def startTagCol(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1985 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1986 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1987 token["selfClosingAcknowledged"] = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1988
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1989 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1990 ignoreEndTag = self.ignoreEndTagColgroup()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1991 self.endTagColgroup(impliedTagToken("colgroup"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1992 if not ignoreEndTag:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1993 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1994
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1995 def endTagColgroup(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1996 if self.ignoreEndTagColgroup():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1997 # innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1998 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
1999 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2000 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2001 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2002 self.parser.phase = self.parser.phases["inTable"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2003
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2004 def endTagCol(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2005 self.parser.parseError("no-end-tag", {"name": "col"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2006
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2007 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2008 ignoreEndTag = self.ignoreEndTagColgroup()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2009 self.endTagColgroup(impliedTagToken("colgroup"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2010 if not ignoreEndTag:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2011 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2012
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2013 class InTableBodyPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2014 # http://www.whatwg.org/specs/web-apps/current-work/#in-table0
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2015 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2016 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2017 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2018 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2019 ("tr", self.startTagTr),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2020 (("td", "th"), self.startTagTableCell),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2021 (("caption", "col", "colgroup", "tbody", "tfoot", "thead"),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2022 self.startTagTableOther)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2023 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2024 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2025
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2026 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2027 (("tbody", "tfoot", "thead"), self.endTagTableRowGroup),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2028 ("table", self.endTagTable),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2029 (("body", "caption", "col", "colgroup", "html", "td", "th",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2030 "tr"), self.endTagIgnore)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2031 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2032 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2033
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2034 # helper methods
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2035 def clearStackToTableBodyContext(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2036 while self.tree.openElements[-1].name not in ("tbody", "tfoot",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2037 "thead", "html"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2038 # self.parser.parseError("unexpected-implied-end-tag-in-table",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2039 # {"name": self.tree.openElements[-1].name})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2040 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2041 if self.tree.openElements[-1].name == "html":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2042 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2043
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2044 # the rest
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2045 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2046 self.parser.phases["inTable"].processEOF()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2047
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2048 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2049 return self.parser.phases["inTable"].processSpaceCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2050
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2051 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2052 return self.parser.phases["inTable"].processCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2053
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2054 def startTagTr(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2055 self.clearStackToTableBodyContext()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2056 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2057 self.parser.phase = self.parser.phases["inRow"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2058
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2059 def startTagTableCell(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2060 self.parser.parseError("unexpected-cell-in-table-body",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2061 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2062 self.startTagTr(impliedTagToken("tr", "StartTag"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2063 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2064
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2065 def startTagTableOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2066 # XXX AT Any ideas on how to share this with endTagTable?
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2067 if (self.tree.elementInScope("tbody", variant="table") or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2068 self.tree.elementInScope("thead", variant="table") or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2069 self.tree.elementInScope("tfoot", variant="table")):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2070 self.clearStackToTableBodyContext()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2071 self.endTagTableRowGroup(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2072 impliedTagToken(self.tree.openElements[-1].name))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2073 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2074 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2075 # innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2076 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2077 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2078
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2079 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2080 return self.parser.phases["inTable"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2081
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2082 def endTagTableRowGroup(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2083 if self.tree.elementInScope(token["name"], variant="table"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2084 self.clearStackToTableBodyContext()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2085 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2086 self.parser.phase = self.parser.phases["inTable"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2087 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2088 self.parser.parseError("unexpected-end-tag-in-table-body",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2089 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2090
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2091 def endTagTable(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2092 if (self.tree.elementInScope("tbody", variant="table") or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2093 self.tree.elementInScope("thead", variant="table") or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2094 self.tree.elementInScope("tfoot", variant="table")):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2095 self.clearStackToTableBodyContext()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2096 self.endTagTableRowGroup(
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2097 impliedTagToken(self.tree.openElements[-1].name))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2098 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2099 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2100 # innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2101 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2102 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2103
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2104 def endTagIgnore(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2105 self.parser.parseError("unexpected-end-tag-in-table-body",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2106 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2107
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2108 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2109 return self.parser.phases["inTable"].processEndTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2110
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2111 class InRowPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2112 # http://www.whatwg.org/specs/web-apps/current-work/#in-row
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2113 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2114 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2115 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2116 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2117 (("td", "th"), self.startTagTableCell),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2118 (("caption", "col", "colgroup", "tbody", "tfoot", "thead",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2119 "tr"), self.startTagTableOther)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2120 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2121 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2122
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2123 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2124 ("tr", self.endTagTr),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2125 ("table", self.endTagTable),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2126 (("tbody", "tfoot", "thead"), self.endTagTableRowGroup),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2127 (("body", "caption", "col", "colgroup", "html", "td", "th"),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2128 self.endTagIgnore)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2129 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2130 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2131
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2132 # helper methods (XXX unify this with other table helper methods)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2133 def clearStackToTableRowContext(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2134 while self.tree.openElements[-1].name not in ("tr", "html"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2135 self.parser.parseError("unexpected-implied-end-tag-in-table-row",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2136 {"name": self.tree.openElements[-1].name})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2137 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2138
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2139 def ignoreEndTagTr(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2140 return not self.tree.elementInScope("tr", variant="table")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2141
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2142 # the rest
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2143 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2144 self.parser.phases["inTable"].processEOF()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2145
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2146 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2147 return self.parser.phases["inTable"].processSpaceCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2148
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2149 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2150 return self.parser.phases["inTable"].processCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2151
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2152 def startTagTableCell(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2153 self.clearStackToTableRowContext()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2154 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2155 self.parser.phase = self.parser.phases["inCell"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2156 self.tree.activeFormattingElements.append(Marker)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2157
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2158 def startTagTableOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2159 ignoreEndTag = self.ignoreEndTagTr()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2160 self.endTagTr(impliedTagToken("tr"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2161 # XXX how are we sure it's always ignored in the innerHTML case?
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2162 if not ignoreEndTag:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2163 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2164
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2165 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2166 return self.parser.phases["inTable"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2167
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2168 def endTagTr(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2169 if not self.ignoreEndTagTr():
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2170 self.clearStackToTableRowContext()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2171 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2172 self.parser.phase = self.parser.phases["inTableBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2173 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2174 # innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2175 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2176 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2177
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2178 def endTagTable(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2179 ignoreEndTag = self.ignoreEndTagTr()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2180 self.endTagTr(impliedTagToken("tr"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2181 # Reprocess the current tag if the tr end tag was not ignored
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2182 # XXX how are we sure it's always ignored in the innerHTML case?
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2183 if not ignoreEndTag:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2184 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2185
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2186 def endTagTableRowGroup(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2187 if self.tree.elementInScope(token["name"], variant="table"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2188 self.endTagTr(impliedTagToken("tr"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2189 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2190 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2191 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2192
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2193 def endTagIgnore(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2194 self.parser.parseError("unexpected-end-tag-in-table-row",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2195 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2196
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2197 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2198 return self.parser.phases["inTable"].processEndTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2199
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2200 class InCellPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2201 # http://www.whatwg.org/specs/web-apps/current-work/#in-cell
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2202 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2203 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2204 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2205 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2206 (("caption", "col", "colgroup", "tbody", "td", "tfoot", "th",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2207 "thead", "tr"), self.startTagTableOther)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2208 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2209 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2210
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2211 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2212 (("td", "th"), self.endTagTableCell),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2213 (("body", "caption", "col", "colgroup", "html"), self.endTagIgnore),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2214 (("table", "tbody", "tfoot", "thead", "tr"), self.endTagImply)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2215 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2216 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2217
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2218 # helper
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2219 def closeCell(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2220 if self.tree.elementInScope("td", variant="table"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2221 self.endTagTableCell(impliedTagToken("td"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2222 elif self.tree.elementInScope("th", variant="table"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2223 self.endTagTableCell(impliedTagToken("th"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2224
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2225 # the rest
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2226 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2227 self.parser.phases["inBody"].processEOF()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2228
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2229 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2230 return self.parser.phases["inBody"].processCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2231
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2232 def startTagTableOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2233 if (self.tree.elementInScope("td", variant="table") or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2234 self.tree.elementInScope("th", variant="table")):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2235 self.closeCell()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2236 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2237 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2238 # innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2239 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2240 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2241
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2242 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2243 return self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2244
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2245 def endTagTableCell(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2246 if self.tree.elementInScope(token["name"], variant="table"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2247 self.tree.generateImpliedEndTags(token["name"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2248 if self.tree.openElements[-1].name != token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2249 self.parser.parseError("unexpected-cell-end-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2250 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2251 while True:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2252 node = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2253 if node.name == token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2254 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2255 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2256 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2257 self.tree.clearActiveFormattingElements()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2258 self.parser.phase = self.parser.phases["inRow"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2259 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2260 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2261
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2262 def endTagIgnore(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2263 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2264
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2265 def endTagImply(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2266 if self.tree.elementInScope(token["name"], variant="table"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2267 self.closeCell()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2268 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2269 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2270 # sometimes innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2271 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2272
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2273 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2274 return self.parser.phases["inBody"].processEndTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2275
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2276 class InSelectPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2277 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2278 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2279
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2280 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2281 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2282 ("option", self.startTagOption),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2283 ("optgroup", self.startTagOptgroup),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2284 ("select", self.startTagSelect),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2285 (("input", "keygen", "textarea"), self.startTagInput),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2286 ("script", self.startTagScript)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2287 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2288 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2289
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2290 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2291 ("option", self.endTagOption),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2292 ("optgroup", self.endTagOptgroup),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2293 ("select", self.endTagSelect)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2294 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2295 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2296
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2297 # http://www.whatwg.org/specs/web-apps/current-work/#in-select
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2298 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2299 if self.tree.openElements[-1].name != "html":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2300 self.parser.parseError("eof-in-select")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2301 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2302 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2303
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2304 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2305 if token["data"] == "\u0000":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2306 return
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2307 self.tree.insertText(token["data"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2308
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2309 def startTagOption(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2310 # We need to imply </option> if <option> is the current node.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2311 if self.tree.openElements[-1].name == "option":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2312 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2313 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2314
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2315 def startTagOptgroup(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2316 if self.tree.openElements[-1].name == "option":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2317 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2318 if self.tree.openElements[-1].name == "optgroup":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2319 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2320 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2321
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2322 def startTagSelect(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2323 self.parser.parseError("unexpected-select-in-select")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2324 self.endTagSelect(impliedTagToken("select"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2325
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2326 def startTagInput(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2327 self.parser.parseError("unexpected-input-in-select")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2328 if self.tree.elementInScope("select", variant="select"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2329 self.endTagSelect(impliedTagToken("select"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2330 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2331 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2332 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2333
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2334 def startTagScript(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2335 return self.parser.phases["inHead"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2336
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2337 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2338 self.parser.parseError("unexpected-start-tag-in-select",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2339 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2340
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2341 def endTagOption(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2342 if self.tree.openElements[-1].name == "option":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2343 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2344 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2345 self.parser.parseError("unexpected-end-tag-in-select",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2346 {"name": "option"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2347
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2348 def endTagOptgroup(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2349 # </optgroup> implicitly closes <option>
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2350 if (self.tree.openElements[-1].name == "option" and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2351 self.tree.openElements[-2].name == "optgroup"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2352 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2353 # It also closes </optgroup>
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2354 if self.tree.openElements[-1].name == "optgroup":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2355 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2356 # But nothing else
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2357 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2358 self.parser.parseError("unexpected-end-tag-in-select",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2359 {"name": "optgroup"})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2360
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2361 def endTagSelect(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2362 if self.tree.elementInScope("select", variant="select"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2363 node = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2364 while node.name != "select":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2365 node = self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2366 self.parser.resetInsertionMode()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2367 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2368 # innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2369 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2370 self.parser.parseError()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2371
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2372 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2373 self.parser.parseError("unexpected-end-tag-in-select",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2374 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2375
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2376 class InSelectInTablePhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2377 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2378 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2379
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2380 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2381 (("caption", "table", "tbody", "tfoot", "thead", "tr", "td", "th"),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2382 self.startTagTable)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2383 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2384 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2385
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2386 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2387 (("caption", "table", "tbody", "tfoot", "thead", "tr", "td", "th"),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2388 self.endTagTable)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2389 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2390 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2391
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2392 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2393 self.parser.phases["inSelect"].processEOF()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2394
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2395 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2396 return self.parser.phases["inSelect"].processCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2397
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2398 def startTagTable(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2399 self.parser.parseError("unexpected-table-element-start-tag-in-select-in-table", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2400 self.endTagOther(impliedTagToken("select"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2401 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2402
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2403 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2404 return self.parser.phases["inSelect"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2405
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2406 def endTagTable(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2407 self.parser.parseError("unexpected-table-element-end-tag-in-select-in-table", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2408 if self.tree.elementInScope(token["name"], variant="table"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2409 self.endTagOther(impliedTagToken("select"))
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2410 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2411
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2412 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2413 return self.parser.phases["inSelect"].processEndTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2414
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2415 class InForeignContentPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2416 breakoutElements = frozenset(["b", "big", "blockquote", "body", "br",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2417 "center", "code", "dd", "div", "dl", "dt",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2418 "em", "embed", "h1", "h2", "h3",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2419 "h4", "h5", "h6", "head", "hr", "i", "img",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2420 "li", "listing", "menu", "meta", "nobr",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2421 "ol", "p", "pre", "ruby", "s", "small",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2422 "span", "strong", "strike", "sub", "sup",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2423 "table", "tt", "u", "ul", "var"])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2424
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2425 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2426 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2427
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2428 def adjustSVGTagNames(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2429 replacements = {"altglyph": "altGlyph",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2430 "altglyphdef": "altGlyphDef",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2431 "altglyphitem": "altGlyphItem",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2432 "animatecolor": "animateColor",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2433 "animatemotion": "animateMotion",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2434 "animatetransform": "animateTransform",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2435 "clippath": "clipPath",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2436 "feblend": "feBlend",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2437 "fecolormatrix": "feColorMatrix",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2438 "fecomponenttransfer": "feComponentTransfer",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2439 "fecomposite": "feComposite",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2440 "feconvolvematrix": "feConvolveMatrix",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2441 "fediffuselighting": "feDiffuseLighting",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2442 "fedisplacementmap": "feDisplacementMap",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2443 "fedistantlight": "feDistantLight",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2444 "feflood": "feFlood",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2445 "fefunca": "feFuncA",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2446 "fefuncb": "feFuncB",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2447 "fefuncg": "feFuncG",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2448 "fefuncr": "feFuncR",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2449 "fegaussianblur": "feGaussianBlur",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2450 "feimage": "feImage",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2451 "femerge": "feMerge",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2452 "femergenode": "feMergeNode",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2453 "femorphology": "feMorphology",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2454 "feoffset": "feOffset",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2455 "fepointlight": "fePointLight",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2456 "fespecularlighting": "feSpecularLighting",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2457 "fespotlight": "feSpotLight",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2458 "fetile": "feTile",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2459 "feturbulence": "feTurbulence",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2460 "foreignobject": "foreignObject",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2461 "glyphref": "glyphRef",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2462 "lineargradient": "linearGradient",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2463 "radialgradient": "radialGradient",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2464 "textpath": "textPath"}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2465
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2466 if token["name"] in replacements:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2467 token["name"] = replacements[token["name"]]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2468
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2469 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2470 if token["data"] == "\u0000":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2471 token["data"] = "\uFFFD"
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2472 elif (self.parser.framesetOK and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2473 any(char not in spaceCharacters for char in token["data"])):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2474 self.parser.framesetOK = False
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2475 Phase.processCharacters(self, token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2476
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2477 def processStartTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2478 currentNode = self.tree.openElements[-1]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2479 if (token["name"] in self.breakoutElements or
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2480 (token["name"] == "font" and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2481 set(token["data"].keys()) & set(["color", "face", "size"]))):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2482 self.parser.parseError("unexpected-html-element-in-foreign-content",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2483 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2484 while (self.tree.openElements[-1].namespace !=
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2485 self.tree.defaultNamespace and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2486 not self.parser.isHTMLIntegrationPoint(self.tree.openElements[-1]) and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2487 not self.parser.isMathMLTextIntegrationPoint(self.tree.openElements[-1])):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2488 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2489 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2490
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2491 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2492 if currentNode.namespace == namespaces["mathml"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2493 self.parser.adjustMathMLAttributes(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2494 elif currentNode.namespace == namespaces["svg"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2495 self.adjustSVGTagNames(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2496 self.parser.adjustSVGAttributes(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2497 self.parser.adjustForeignAttributes(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2498 token["namespace"] = currentNode.namespace
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2499 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2500 if token["selfClosing"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2501 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2502 token["selfClosingAcknowledged"] = True
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2503
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2504 def processEndTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2505 nodeIndex = len(self.tree.openElements) - 1
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2506 node = self.tree.openElements[-1]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2507 if node.name.translate(asciiUpper2Lower) != token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2508 self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2509
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2510 while True:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2511 if node.name.translate(asciiUpper2Lower) == token["name"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2512 # XXX this isn't in the spec but it seems necessary
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2513 if self.parser.phase == self.parser.phases["inTableText"]:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2514 self.parser.phase.flushCharacters()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2515 self.parser.phase = self.parser.phase.originalPhase
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2516 while self.tree.openElements.pop() != node:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2517 assert self.tree.openElements
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2518 new_token = None
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2519 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2520 nodeIndex -= 1
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2521
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2522 node = self.tree.openElements[nodeIndex]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2523 if node.namespace != self.tree.defaultNamespace:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2524 continue
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2525 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2526 new_token = self.parser.phase.processEndTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2527 break
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2528 return new_token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2529
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2530 class AfterBodyPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2531 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2532 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2533
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2534 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2535 ("html", self.startTagHtml)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2536 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2537 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2538
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2539 self.endTagHandler = _utils.MethodDispatcher([("html", self.endTagHtml)])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2540 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2541
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2542 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2543 # Stop parsing
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2544 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2545
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2546 def processComment(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2547 # This is needed because data is to be appended to the <html> element
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2548 # here and not to whatever is currently open.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2549 self.tree.insertComment(token, self.tree.openElements[0])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2550
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2551 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2552 self.parser.parseError("unexpected-char-after-body")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2553 self.parser.phase = self.parser.phases["inBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2554 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2555
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2556 def startTagHtml(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2557 return self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2558
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2559 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2560 self.parser.parseError("unexpected-start-tag-after-body",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2561 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2562 self.parser.phase = self.parser.phases["inBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2563 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2564
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2565 def endTagHtml(self, name):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2566 if self.parser.innerHTML:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2567 self.parser.parseError("unexpected-end-tag-after-body-innerhtml")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2568 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2569 self.parser.phase = self.parser.phases["afterAfterBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2570
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2571 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2572 self.parser.parseError("unexpected-end-tag-after-body",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2573 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2574 self.parser.phase = self.parser.phases["inBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2575 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2576
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2577 class InFramesetPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2578 # http://www.whatwg.org/specs/web-apps/current-work/#in-frameset
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2579 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2580 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2581
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2582 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2583 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2584 ("frameset", self.startTagFrameset),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2585 ("frame", self.startTagFrame),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2586 ("noframes", self.startTagNoframes)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2587 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2588 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2589
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2590 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2591 ("frameset", self.endTagFrameset)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2592 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2593 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2594
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2595 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2596 if self.tree.openElements[-1].name != "html":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2597 self.parser.parseError("eof-in-frameset")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2598 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2599 assert self.parser.innerHTML
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2600
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2601 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2602 self.parser.parseError("unexpected-char-in-frameset")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2603
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2604 def startTagFrameset(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2605 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2606
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2607 def startTagFrame(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2608 self.tree.insertElement(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2609 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2610
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2611 def startTagNoframes(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2612 return self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2613
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2614 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2615 self.parser.parseError("unexpected-start-tag-in-frameset",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2616 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2617
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2618 def endTagFrameset(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2619 if self.tree.openElements[-1].name == "html":
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2620 # innerHTML case
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2621 self.parser.parseError("unexpected-frameset-in-frameset-innerhtml")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2622 else:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2623 self.tree.openElements.pop()
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2624 if (not self.parser.innerHTML and
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2625 self.tree.openElements[-1].name != "frameset"):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2626 # If we're not in innerHTML mode and the current node is not a
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2627 # "frameset" element (anymore) then switch.
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2628 self.parser.phase = self.parser.phases["afterFrameset"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2629
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2630 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2631 self.parser.parseError("unexpected-end-tag-in-frameset",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2632 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2633
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2634 class AfterFramesetPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2635 # http://www.whatwg.org/specs/web-apps/current-work/#after3
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2636 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2637 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2638
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2639 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2640 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2641 ("noframes", self.startTagNoframes)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2642 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2643 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2644
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2645 self.endTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2646 ("html", self.endTagHtml)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2647 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2648 self.endTagHandler.default = self.endTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2649
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2650 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2651 # Stop parsing
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2652 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2653
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2654 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2655 self.parser.parseError("unexpected-char-after-frameset")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2656
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2657 def startTagNoframes(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2658 return self.parser.phases["inHead"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2659
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2660 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2661 self.parser.parseError("unexpected-start-tag-after-frameset",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2662 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2663
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2664 def endTagHtml(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2665 self.parser.phase = self.parser.phases["afterAfterFrameset"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2666
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2667 def endTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2668 self.parser.parseError("unexpected-end-tag-after-frameset",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2669 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2670
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2671 class AfterAfterBodyPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2672 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2673 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2674
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2675 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2676 ("html", self.startTagHtml)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2677 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2678 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2679
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2680 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2681 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2682
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2683 def processComment(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2684 self.tree.insertComment(token, self.tree.document)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2685
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2686 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2687 return self.parser.phases["inBody"].processSpaceCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2688
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2689 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2690 self.parser.parseError("expected-eof-but-got-char")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2691 self.parser.phase = self.parser.phases["inBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2692 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2693
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2694 def startTagHtml(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2695 return self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2696
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2697 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2698 self.parser.parseError("expected-eof-but-got-start-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2699 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2700 self.parser.phase = self.parser.phases["inBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2701 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2702
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2703 def processEndTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2704 self.parser.parseError("expected-eof-but-got-end-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2705 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2706 self.parser.phase = self.parser.phases["inBody"]
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2707 return token
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2708
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2709 class AfterAfterFramesetPhase(Phase):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2710 def __init__(self, parser, tree):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2711 Phase.__init__(self, parser, tree)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2712
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2713 self.startTagHandler = _utils.MethodDispatcher([
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2714 ("html", self.startTagHtml),
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2715 ("noframes", self.startTagNoFrames)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2716 ])
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2717 self.startTagHandler.default = self.startTagOther
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2718
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2719 def processEOF(self):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2720 pass
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2721
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2722 def processComment(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2723 self.tree.insertComment(token, self.tree.document)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2724
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2725 def processSpaceCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2726 return self.parser.phases["inBody"].processSpaceCharacters(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2727
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2728 def processCharacters(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2729 self.parser.parseError("expected-eof-but-got-char")
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2730
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2731 def startTagHtml(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2732 return self.parser.phases["inBody"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2733
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2734 def startTagNoFrames(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2735 return self.parser.phases["inHead"].processStartTag(token)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2736
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2737 def startTagOther(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2738 self.parser.parseError("expected-eof-but-got-start-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2739 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2740
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2741 def processEndTag(self, token):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2742 self.parser.parseError("expected-eof-but-got-end-tag",
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2743 {"name": token["name"]})
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2744 # pylint:enable=unused-argument
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2745
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2746 return {
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2747 "initial": InitialPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2748 "beforeHtml": BeforeHtmlPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2749 "beforeHead": BeforeHeadPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2750 "inHead": InHeadPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2751 "inHeadNoscript": InHeadNoscriptPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2752 "afterHead": AfterHeadPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2753 "inBody": InBodyPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2754 "text": TextPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2755 "inTable": InTablePhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2756 "inTableText": InTableTextPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2757 "inCaption": InCaptionPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2758 "inColumnGroup": InColumnGroupPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2759 "inTableBody": InTableBodyPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2760 "inRow": InRowPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2761 "inCell": InCellPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2762 "inSelect": InSelectPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2763 "inSelectInTable": InSelectInTablePhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2764 "inForeignContent": InForeignContentPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2765 "afterBody": AfterBodyPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2766 "inFrameset": InFramesetPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2767 "afterFrameset": AfterFramesetPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2768 "afterAfterBody": AfterAfterBodyPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2769 "afterAfterFrameset": AfterAfterFramesetPhase,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2770 # XXX after after frameset
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2771 }
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2772
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2773
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2774 def adjust_attributes(token, replacements):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2775 needs_adjustment = viewkeys(token['data']) & viewkeys(replacements)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2776 if needs_adjustment:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2777 token['data'] = OrderedDict((replacements.get(k, k), v)
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2778 for k, v in token['data'].items())
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2779
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2780
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2781 def impliedTagToken(name, type="EndTag", attributes=None,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2782 selfClosing=False):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2783 if attributes is None:
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2784 attributes = {}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2785 return {"type": tokenTypes[type], "name": name, "data": attributes,
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2786 "selfClosing": selfClosing}
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2787
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2788
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2789 class ParseError(Exception):
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2790 """Error in parsed document"""
9e54283cc701 "planemo upload commit d12c32a45bcd441307e632fca6d9af7d60289d44"
guerler
parents:
diff changeset
2791 pass