Mercurial > repos > shellac > guppy_basecaller

annotate env/lib/python3.7/site-packages/bs4/element.py @ 0:26e78fe6e8c4 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
author shellac
date Sat, 02 May 2020 07:14:21 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1 # Use of this source code is governed by the MIT license.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2 __license__ = "MIT"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
4 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
5 from collections.abc import Callable # Python 3.6
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
6 except ImportError as e:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
7 from collections import Callable
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
8 import re
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
9 import sys
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
10 import warnings
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
11 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
12 import soupsieve
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
13 except ImportError as e:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
14 soupsieve = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
15 warnings.warn(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
16 'The soupsieve package is not installed. CSS selectors cannot be used.'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
17 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
18
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
19 from bs4.formatter import (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
20 Formatter,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
21 HTMLFormatter,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
22 XMLFormatter,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
23 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
24
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
25 DEFAULT_OUTPUT_ENCODING = "utf-8"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
26 PY3K = (sys.version_info[0] > 2)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
27
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
28 nonwhitespace_re = re.compile(r"\S+")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
29
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
30 # NOTE: This isn't used as of 4.7.0. I'm leaving it for a little bit on
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
31 # the off chance someone imported it for their own use.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
32 whitespace_re = re.compile(r"\s+")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
33
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
34 def _alias(attr):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
35 """Alias one attribute name to another for backward compatibility"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
36 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
37 def alias(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
38 return getattr(self, attr)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
39
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
40 @alias.setter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
41 def alias(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
42 return setattr(self, attr)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
43 return alias
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
44
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
45
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
46 class NamespacedAttribute(str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
47 """A namespaced string (e.g. 'xml:lang') that remembers the namespace
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
48 ('xml') and the name ('lang') that were used to create it.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
49 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
50
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
51 def __new__(cls, prefix, name=None, namespace=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
52 if not name:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
53 # This is the default namespace. Its name "has no value"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
54 # per https://www.w3.org/TR/xml-names/#defaulting
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
55 name = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
56
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
57 if name is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
58 obj = str.__new__(cls, prefix)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
59 elif prefix is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
60 # Not really namespaced.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
61 obj = str.__new__(cls, name)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
62 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
63 obj = str.__new__(cls, prefix + ":" + name)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
64 obj.prefix = prefix
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
65 obj.name = name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
66 obj.namespace = namespace
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
67 return obj
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
68
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
69 class AttributeValueWithCharsetSubstitution(str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
70 """A stand-in object for a character encoding specified in HTML."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
71
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
72 class CharsetMetaAttributeValue(AttributeValueWithCharsetSubstitution):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
73 """A generic stand-in for the value of a meta tag's 'charset' attribute.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
74
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
75 When Beautiful Soup parses the markup '<meta charset="utf8">', the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
76 value of the 'charset' attribute will be one of these objects.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
77 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
78
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
79 def __new__(cls, original_value):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
80 obj = str.__new__(cls, original_value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
81 obj.original_value = original_value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
82 return obj
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
83
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
84 def encode(self, encoding):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
85 """When an HTML document is being encoded to a given encoding, the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
86 value of a meta tag's 'charset' is the name of the encoding.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
87 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
88 return encoding
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
89
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
90
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
91 class ContentMetaAttributeValue(AttributeValueWithCharsetSubstitution):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
92 """A generic stand-in for the value of a meta tag's 'content' attribute.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
93
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
94 When Beautiful Soup parses the markup:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
95 <meta http-equiv="content-type" content="text/html; charset=utf8">
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
96
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
97 The value of the 'content' attribute will be one of these objects.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
98 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
99
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
100 CHARSET_RE = re.compile(r"((^|;)\s*charset=)([^;]*)", re.M)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
101
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
102 def __new__(cls, original_value):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
103 match = cls.CHARSET_RE.search(original_value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
104 if match is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
105 # No substitution necessary.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
106 return str.__new__(str, original_value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
107
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
108 obj = str.__new__(cls, original_value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
109 obj.original_value = original_value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
110 return obj
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
111
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
112 def encode(self, encoding):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
113 def rewrite(match):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
114 return match.group(1) + encoding
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
115 return self.CHARSET_RE.sub(rewrite, self.original_value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
116
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
117
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
118 class PageElement(object):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
119 """Contains the navigational information for some part of the page:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
120 that is, its current location in the parse tree.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
121
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
122 NavigableString, Tag, etc. are all subclasses of PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
123 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
124
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
125 def setup(self, parent=None, previous_element=None, next_element=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
126 previous_sibling=None, next_sibling=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
127 """Sets up the initial relations between this element and
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
128 other elements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
129
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
130 :param parent: The parent of this element.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
131
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
132 :param previous_element: The element parsed immediately before
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
133 this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
134
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
135 :param next_element: The element parsed immediately before
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
136 this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
137
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
138 :param previous_sibling: The most recently encountered element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
139 on the same level of the parse tree as this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
140
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
141 :param previous_sibling: The next element to be encountered
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
142 on the same level of the parse tree as this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
143 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
144 self.parent = parent
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
145
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
146 self.previous_element = previous_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
147 if previous_element is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
148 self.previous_element.next_element = self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
149
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
150 self.next_element = next_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
151 if self.next_element is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
152 self.next_element.previous_element = self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
153
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
154 self.next_sibling = next_sibling
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
155 if self.next_sibling is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
156 self.next_sibling.previous_sibling = self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
157
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
158 if (previous_sibling is None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
159 and self.parent is not None and self.parent.contents):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
160 previous_sibling = self.parent.contents[-1]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
161
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
162 self.previous_sibling = previous_sibling
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
163 if previous_sibling is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
164 self.previous_sibling.next_sibling = self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
165
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
166 def format_string(self, s, formatter):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
167 """Format the given string using the given formatter.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
168
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
169 :param s: A string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
170 :param formatter: A Formatter object, or a string naming one of the standard formatters.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
171 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
172 if formatter is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
173 return s
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
174 if not isinstance(formatter, Formatter):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
175 formatter = self.formatter_for_name(formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
176 output = formatter.substitute(s)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
177 return output
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
178
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
179 def formatter_for_name(self, formatter):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
180 """Look up or create a Formatter for the given identifier,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
181 if necessary.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
182
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
183 :param formatter: Can be a Formatter object (used as-is), a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
184 function (used as the entity substitution hook for an
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
185 XMLFormatter or HTMLFormatter), or a string (used to look
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
186 up an XMLFormatter or HTMLFormatter in the appropriate
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
187 registry.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
188 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
189 if isinstance(formatter, Formatter):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
190 return formatter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
191 if self._is_xml:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
192 c = XMLFormatter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
193 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
194 c = HTMLFormatter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
195 if isinstance(formatter, Callable):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
196 return c(entity_substitution=formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
197 return c.REGISTRY[formatter]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
198
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
199 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
200 def _is_xml(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
201 """Is this element part of an XML tree or an HTML tree?
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
202
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
203 This is used in formatter_for_name, when deciding whether an
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
204 XMLFormatter or HTMLFormatter is more appropriate. It can be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
205 inefficient, but it should be called very rarely.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
206 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
207 if self.known_xml is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
208 # Most of the time we will have determined this when the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
209 # document is parsed.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
210 return self.known_xml
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
211
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
212 # Otherwise, it's likely that this element was created by
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
213 # direct invocation of the constructor from within the user's
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
214 # Python code.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
215 if self.parent is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
216 # This is the top-level object. It should have .known_xml set
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
217 # from tree creation. If not, take a guess--BS is usually
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
218 # used on HTML markup.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
219 return getattr(self, 'is_xml', False)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
220 return self.parent._is_xml
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
221
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
222 nextSibling = _alias("next_sibling") # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
223 previousSibling = _alias("previous_sibling") # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
224
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
225 def replace_with(self, replace_with):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
226 """Replace this PageElement with another one, keeping the rest of the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
227 tree the same.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
228
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
229 :param replace_with: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
230 :return: `self`, no longer part of the tree.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
231 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
232 if self.parent is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
233 raise ValueError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
234 "Cannot replace one element with another when the "
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
235 "element to be replaced is not part of a tree.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
236 if replace_with is self:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
237 return
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
238 if replace_with is self.parent:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
239 raise ValueError("Cannot replace a Tag with its parent.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
240 old_parent = self.parent
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
241 my_index = self.parent.index(self)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
242 self.extract(_self_index=my_index)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
243 old_parent.insert(my_index, replace_with)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
244 return self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
245 replaceWith = replace_with # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
246
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
247 def unwrap(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
248 """Replace this PageElement with its contents.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
249
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
250 :return: `self`, no longer part of the tree.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
251 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
252 my_parent = self.parent
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
253 if self.parent is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
254 raise ValueError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
255 "Cannot replace an element with its contents when that"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
256 "element is not part of a tree.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
257 my_index = self.parent.index(self)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
258 self.extract(_self_index=my_index)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
259 for child in reversed(self.contents[:]):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
260 my_parent.insert(my_index, child)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
261 return self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
262 replace_with_children = unwrap
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
263 replaceWithChildren = unwrap # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
264
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
265 def wrap(self, wrap_inside):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
266 """Wrap this PageElement inside another one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
267
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
268 :param wrap_inside: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
269 :return: `wrap_inside`, occupying the position in the tree that used
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
270 to be occupied by `self`, and with `self` inside it.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
271 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
272 me = self.replace_with(wrap_inside)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
273 wrap_inside.append(me)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
274 return wrap_inside
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
275
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
276 def extract(self, _self_index=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
277 """Destructively rips this element out of the tree.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
278
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
279 :param _self_index: The location of this element in its parent's
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
280 .contents, if known. Passing this in allows for a performance
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
281 optimization.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
282
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
283 :return: `self`, no longer part of the tree.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
284 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
285 if self.parent is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
286 if _self_index is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
287 _self_index = self.parent.index(self)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
288 del self.parent.contents[_self_index]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
289
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
290 #Find the two elements that would be next to each other if
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
291 #this element (and any children) hadn't been parsed. Connect
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
292 #the two.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
293 last_child = self._last_descendant()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
294 next_element = last_child.next_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
295
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
296 if (self.previous_element is not None and
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
297 self.previous_element is not next_element):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
298 self.previous_element.next_element = next_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
299 if next_element is not None and next_element is not self.previous_element:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
300 next_element.previous_element = self.previous_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
301 self.previous_element = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
302 last_child.next_element = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
303
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
304 self.parent = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
305 if (self.previous_sibling is not None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
306 and self.previous_sibling is not self.next_sibling):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
307 self.previous_sibling.next_sibling = self.next_sibling
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
308 if (self.next_sibling is not None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
309 and self.next_sibling is not self.previous_sibling):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
310 self.next_sibling.previous_sibling = self.previous_sibling
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
311 self.previous_sibling = self.next_sibling = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
312 return self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
313
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
314 def _last_descendant(self, is_initialized=True, accept_self=True):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
315 """Finds the last element beneath this object to be parsed.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
316
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
317 :param is_initialized: Has `setup` been called on this PageElement
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
318 yet?
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
319 :param accept_self: Is `self` an acceptable answer to the question?
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
320 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
321 if is_initialized and self.next_sibling is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
322 last_child = self.next_sibling.previous_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
323 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
324 last_child = self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
325 while isinstance(last_child, Tag) and last_child.contents:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
326 last_child = last_child.contents[-1]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
327 if not accept_self and last_child is self:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
328 last_child = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
329 return last_child
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
330 # BS3: Not part of the API!
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
331 _lastRecursiveChild = _last_descendant
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
332
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
333 def insert(self, position, new_child):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
334 """Insert a new PageElement in the list of this PageElement's children.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
335
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
336 This works the same way as `list.insert`.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
337
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
338 :param position: The numeric position that should be occupied
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
339 in `self.children` by the new PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
340 :param new_child: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
341 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
342 if new_child is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
343 raise ValueError("Cannot insert None into a tag.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
344 if new_child is self:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
345 raise ValueError("Cannot insert a tag into itself.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
346 if (isinstance(new_child, str)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
347 and not isinstance(new_child, NavigableString)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
348 new_child = NavigableString(new_child)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
349
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
350 from bs4 import BeautifulSoup
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
351 if isinstance(new_child, BeautifulSoup):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
352 # We don't want to end up with a situation where one BeautifulSoup
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
353 # object contains another. Insert the children one at a time.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
354 for subchild in list(new_child.contents):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
355 self.insert(position, subchild)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
356 position += 1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
357 return
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
358 position = min(position, len(self.contents))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
359 if hasattr(new_child, 'parent') and new_child.parent is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
360 # We're 'inserting' an element that's already one
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
361 # of this object's children.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
362 if new_child.parent is self:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
363 current_index = self.index(new_child)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
364 if current_index < position:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
365 # We're moving this element further down the list
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
366 # of this object's children. That means that when
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
367 # we extract this element, our target index will
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
368 # jump down one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
369 position -= 1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
370 new_child.extract()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
371
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
372 new_child.parent = self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
373 previous_child = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
374 if position == 0:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
375 new_child.previous_sibling = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
376 new_child.previous_element = self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
377 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
378 previous_child = self.contents[position - 1]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
379 new_child.previous_sibling = previous_child
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
380 new_child.previous_sibling.next_sibling = new_child
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
381 new_child.previous_element = previous_child._last_descendant(False)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
382 if new_child.previous_element is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
383 new_child.previous_element.next_element = new_child
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
384
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
385 new_childs_last_element = new_child._last_descendant(False)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
386
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
387 if position >= len(self.contents):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
388 new_child.next_sibling = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
389
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
390 parent = self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
391 parents_next_sibling = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
392 while parents_next_sibling is None and parent is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
393 parents_next_sibling = parent.next_sibling
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
394 parent = parent.parent
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
395 if parents_next_sibling is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
396 # We found the element that comes next in the document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
397 break
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
398 if parents_next_sibling is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
399 new_childs_last_element.next_element = parents_next_sibling
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
400 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
401 # The last element of this tag is the last element in
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
402 # the document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
403 new_childs_last_element.next_element = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
404 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
405 next_child = self.contents[position]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
406 new_child.next_sibling = next_child
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
407 if new_child.next_sibling is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
408 new_child.next_sibling.previous_sibling = new_child
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
409 new_childs_last_element.next_element = next_child
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
410
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
411 if new_childs_last_element.next_element is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
412 new_childs_last_element.next_element.previous_element = new_childs_last_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
413 self.contents.insert(position, new_child)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
414
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
415 def append(self, tag):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
416 """Appends the given PageElement to the contents of this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
417
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
418 :param tag: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
419 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
420 self.insert(len(self.contents), tag)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
421
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
422 def extend(self, tags):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
423 """Appends the given PageElements to this one's contents.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
424
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
425 :param tags: A list of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
426 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
427 for tag in tags:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
428 self.append(tag)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
429
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
430 def insert_before(self, *args):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
431 """Makes the given element(s) the immediate predecessor of this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
432
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
433 All the elements will have the same parent, and the given elements
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
434 will be immediately before this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
435
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
436 :param args: One or more PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
437 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
438 parent = self.parent
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
439 if parent is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
440 raise ValueError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
441 "Element has no parent, so 'before' has no meaning.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
442 if any(x is self for x in args):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
443 raise ValueError("Can't insert an element before itself.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
444 for predecessor in args:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
445 # Extract first so that the index won't be screwed up if they
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
446 # are siblings.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
447 if isinstance(predecessor, PageElement):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
448 predecessor.extract()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
449 index = parent.index(self)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
450 parent.insert(index, predecessor)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
451
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
452 def insert_after(self, *args):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
453 """Makes the given element(s) the immediate successor of this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
454
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
455 The elements will have the same parent, and the given elements
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
456 will be immediately after this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
457
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
458 :param args: One or more PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
459 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
460 # Do all error checking before modifying the tree.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
461 parent = self.parent
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
462 if parent is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
463 raise ValueError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
464 "Element has no parent, so 'after' has no meaning.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
465 if any(x is self for x in args):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
466 raise ValueError("Can't insert an element after itself.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
467
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
468 offset = 0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
469 for successor in args:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
470 # Extract first so that the index won't be screwed up if they
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
471 # are siblings.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
472 if isinstance(successor, PageElement):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
473 successor.extract()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
474 index = parent.index(self)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
475 parent.insert(index+1+offset, successor)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
476 offset += 1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
477
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
478 def find_next(self, name=None, attrs={}, text=None, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
479 """Find the first PageElement that matches the given criteria and
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
480 appears later in the document than this PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
481
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
482 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
483 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
484
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
485 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
486 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
487 :param text: A filter for a NavigableString with specific text.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
488 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
489 :return: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
490 :rtype: bs4.element.Tag | bs4.element.NavigableString
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
491 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
492 return self._find_one(self.find_all_next, name, attrs, text, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
493 findNext = find_next # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
494
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
495 def find_all_next(self, name=None, attrs={}, text=None, limit=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
496 **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
497 """Find all PageElements that match the given criteria and appear
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
498 later in the document than this PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
499
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
500 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
501 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
502
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
503 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
504 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
505 :param text: A filter for a NavigableString with specific text.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
506 :param limit: Stop looking after finding this many results.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
507 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
508 :return: A ResultSet containing PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
509 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
510 return self._find_all(name, attrs, text, limit, self.next_elements,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
511 **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
512 findAllNext = find_all_next # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
513
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
514 def find_next_sibling(self, name=None, attrs={}, text=None, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
515 """Find the closest sibling to this PageElement that matches the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
516 given criteria and appears later in the document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
517
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
518 All find_* methods take a common set of arguments. See the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
519 online documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
520
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
521 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
522 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
523 :param text: A filter for a NavigableString with specific text.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
524 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
525 :return: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
526 :rtype: bs4.element.Tag | bs4.element.NavigableString
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
527 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
528 return self._find_one(self.find_next_siblings, name, attrs, text,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
529 **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
530 findNextSibling = find_next_sibling # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
531
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
532 def find_next_siblings(self, name=None, attrs={}, text=None, limit=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
533 **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
534 """Find all siblings of this PageElement that match the given criteria
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
535 and appear later in the document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
536
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
537 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
538 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
539
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
540 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
541 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
542 :param text: A filter for a NavigableString with specific text.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
543 :param limit: Stop looking after finding this many results.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
544 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
545 :return: A ResultSet of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
546 :rtype: bs4.element.ResultSet
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
547 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
548 return self._find_all(name, attrs, text, limit,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
549 self.next_siblings, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
550 findNextSiblings = find_next_siblings # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
551 fetchNextSiblings = find_next_siblings # BS2
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
552
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
553 def find_previous(self, name=None, attrs={}, text=None, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
554 """Look backwards in the document from this PageElement and find the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
555 first PageElement that matches the given criteria.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
556
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
557 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
558 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
559
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
560 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
561 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
562 :param text: A filter for a NavigableString with specific text.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
563 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
564 :return: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
565 :rtype: bs4.element.Tag | bs4.element.NavigableString
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
566 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
567 return self._find_one(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
568 self.find_all_previous, name, attrs, text, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
569 findPrevious = find_previous # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
570
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
571 def find_all_previous(self, name=None, attrs={}, text=None, limit=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
572 **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
573 """Look backwards in the document from this PageElement and find all
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
574 PageElements that match the given criteria.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
575
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
576 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
577 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
578
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
579 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
580 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
581 :param text: A filter for a NavigableString with specific text.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
582 :param limit: Stop looking after finding this many results.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
583 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
584 :return: A ResultSet of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
585 :rtype: bs4.element.ResultSet
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
586 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
587 return self._find_all(name, attrs, text, limit, self.previous_elements,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
588 **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
589 findAllPrevious = find_all_previous # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
590 fetchPrevious = find_all_previous # BS2
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
591
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
592 def find_previous_sibling(self, name=None, attrs={}, text=None, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
593 """Returns the closest sibling to this PageElement that matches the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
594 given criteria and appears earlier in the document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
595
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
596 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
597 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
598
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
599 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
600 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
601 :param text: A filter for a NavigableString with specific text.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
602 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
603 :return: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
604 :rtype: bs4.element.Tag | bs4.element.NavigableString
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
605 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
606 return self._find_one(self.find_previous_siblings, name, attrs, text,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
607 **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
608 findPreviousSibling = find_previous_sibling # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
609
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
610 def find_previous_siblings(self, name=None, attrs={}, text=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
611 limit=None, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
612 """Returns all siblings to this PageElement that match the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
613 given criteria and appear earlier in the document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
614
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
615 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
616 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
617
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
618 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
619 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
620 :param text: A filter for a NavigableString with specific text.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
621 :param limit: Stop looking after finding this many results.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
622 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
623 :return: A ResultSet of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
624 :rtype: bs4.element.ResultSet
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
625 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
626 return self._find_all(name, attrs, text, limit,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
627 self.previous_siblings, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
628 findPreviousSiblings = find_previous_siblings # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
629 fetchPreviousSiblings = find_previous_siblings # BS2
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
630
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
631 def find_parent(self, name=None, attrs={}, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
632 """Find the closest parent of this PageElement that matches the given
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
633 criteria.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
634
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
635 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
636 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
637
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
638 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
639 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
640 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
641
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
642 :return: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
643 :rtype: bs4.element.Tag | bs4.element.NavigableString
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
644 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
645 # NOTE: We can't use _find_one because findParents takes a different
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
646 # set of arguments.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
647 r = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
648 l = self.find_parents(name, attrs, 1, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
649 if l:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
650 r = l[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
651 return r
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
652 findParent = find_parent # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
653
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
654 def find_parents(self, name=None, attrs={}, limit=None, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
655 """Find all parents of this PageElement that match the given criteria.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
656
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
657 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
658 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
659
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
660 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
661 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
662 :param limit: Stop looking after finding this many results.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
663 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
664
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
665 :return: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
666 :rtype: bs4.element.Tag | bs4.element.NavigableString
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
667 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
668 return self._find_all(name, attrs, None, limit, self.parents,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
669 **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
670 findParents = find_parents # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
671 fetchParents = find_parents # BS2
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
672
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
673 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
674 def next(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
675 """The PageElement, if any, that was parsed just after this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
676
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
677 :return: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
678 :rtype: bs4.element.Tag | bs4.element.NavigableString
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
679 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
680 return self.next_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
681
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
682 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
683 def previous(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
684 """The PageElement, if any, that was parsed just before this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
685
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
686 :return: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
687 :rtype: bs4.element.Tag | bs4.element.NavigableString
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
688 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
689 return self.previous_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
690
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
691 #These methods do the real heavy lifting.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
692
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
693 def _find_one(self, method, name, attrs, text, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
694 r = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
695 l = method(name, attrs, text, 1, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
696 if l:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
697 r = l[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
698 return r
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
699
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
700 def _find_all(self, name, attrs, text, limit, generator, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
701 "Iterates over a generator looking for things that match."
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
702
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
703 if text is None and 'string' in kwargs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
704 text = kwargs['string']
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
705 del kwargs['string']
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
706
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
707 if isinstance(name, SoupStrainer):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
708 strainer = name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
709 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
710 strainer = SoupStrainer(name, attrs, text, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
711
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
712 if text is None and not limit and not attrs and not kwargs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
713 if name is True or name is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
714 # Optimization to find all tags.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
715 result = (element for element in generator
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
716 if isinstance(element, Tag))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
717 return ResultSet(strainer, result)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
718 elif isinstance(name, str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
719 # Optimization to find all tags with a given name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
720 if name.count(':') == 1:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
721 # This is a name with a prefix. If this is a namespace-aware document,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
722 # we need to match the local name against tag.name. If not,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
723 # we need to match the fully-qualified name against tag.name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
724 prefix, local_name = name.split(':', 1)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
725 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
726 prefix = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
727 local_name = name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
728 result = (element for element in generator
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
729 if isinstance(element, Tag)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
730 and (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
731 element.name == name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
732 ) or (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
733 element.name == local_name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
734 and (prefix is None or element.prefix == prefix)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
735 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
736 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
737 return ResultSet(strainer, result)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
738 results = ResultSet(strainer)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
739 while True:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
740 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
741 i = next(generator)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
742 except StopIteration:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
743 break
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
744 if i:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
745 found = strainer.search(i)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
746 if found:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
747 results.append(found)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
748 if limit and len(results) >= limit:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
749 break
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
750 return results
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
751
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
752 #These generators can be used to navigate starting from both
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
753 #NavigableStrings and Tags.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
754 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
755 def next_elements(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
756 """All PageElements that were parsed after this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
757
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
758 :yield: A sequence of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
759 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
760 i = self.next_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
761 while i is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
762 yield i
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
763 i = i.next_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
764
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
765 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
766 def next_siblings(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
767 """All PageElements that are siblings of this one but were parsed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
768 later.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
769
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
770 :yield: A sequence of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
771 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
772 i = self.next_sibling
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
773 while i is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
774 yield i
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
775 i = i.next_sibling
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
776
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
777 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
778 def previous_elements(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
779 """All PageElements that were parsed before this one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
780
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
781 :yield: A sequence of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
782 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
783 i = self.previous_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
784 while i is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
785 yield i
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
786 i = i.previous_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
787
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
788 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
789 def previous_siblings(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
790 """All PageElements that are siblings of this one but were parsed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
791 earlier.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
792
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
793 :yield: A sequence of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
794 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
795 i = self.previous_sibling
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
796 while i is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
797 yield i
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
798 i = i.previous_sibling
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
799
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
800 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
801 def parents(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
802 """All PageElements that are parents of this PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
803
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
804 :yield: A sequence of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
805 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
806 i = self.parent
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
807 while i is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
808 yield i
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
809 i = i.parent
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
810
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
811 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
812 def decomposed(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
813 """Check whether a PageElement has been decomposed.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
814
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
815 :rtype: bool
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
816 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
817 return getattr(self, '_decomposed', False) or False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
818
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
819 # Old non-property versions of the generators, for backwards
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
820 # compatibility with BS3.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
821 def nextGenerator(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
822 return self.next_elements
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
823
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
824 def nextSiblingGenerator(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
825 return self.next_siblings
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
826
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
827 def previousGenerator(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
828 return self.previous_elements
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
829
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
830 def previousSiblingGenerator(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
831 return self.previous_siblings
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
832
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
833 def parentGenerator(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
834 return self.parents
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
835
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
836
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
837 class NavigableString(str, PageElement):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
838 """A Python Unicode string that is part of a parse tree.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
839
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
840 When Beautiful Soup parses the markup <b>penguin</b>, it will
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
841 create a NavigableString for the string "penguin".
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
842 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
843
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
844 PREFIX = ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
845 SUFFIX = ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
846
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
847 # We can't tell just by looking at a string whether it's contained
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
848 # in an XML document or an HTML document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
849
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
850 known_xml = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
851
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
852 def __new__(cls, value):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
853 """Create a new NavigableString.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
854
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
855 When unpickling a NavigableString, this method is called with
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
856 the string in DEFAULT_OUTPUT_ENCODING. That encoding needs to be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
857 passed in to the superclass's __new__ or the superclass won't know
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
858 how to handle non-ASCII characters.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
859 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
860 if isinstance(value, str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
861 u = str.__new__(cls, value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
862 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
863 u = str.__new__(cls, value, DEFAULT_OUTPUT_ENCODING)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
864 u.setup()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
865 return u
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
866
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
867 def __copy__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
868 """A copy of a NavigableString has the same contents and class
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
869 as the original, but it is not connected to the parse tree.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
870 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
871 return type(self)(self)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
872
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
873 def __getnewargs__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
874 return (str(self),)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
875
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
876 def __getattr__(self, attr):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
877 """text.string gives you text. This is for backwards
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
878 compatibility for Navigable*String, but for CData* it lets you
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
879 get the string without the CData wrapper."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
880 if attr == 'string':
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
881 return self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
882 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
883 raise AttributeError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
884 "'%s' object has no attribute '%s'" % (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
885 self.__class__.__name__, attr))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
886
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
887 def output_ready(self, formatter="minimal"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
888 """Run the string through the provided formatter.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
889
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
890 :param formatter: A Formatter object, or a string naming one of the standard formatters.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
891 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
892 output = self.format_string(self, formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
893 return self.PREFIX + output + self.SUFFIX
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
894
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
895 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
896 def name(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
897 """Since a NavigableString is not a Tag, it has no .name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
898
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
899 This property is implemented so that code like this doesn't crash
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
900 when run on a mixture of Tag and NavigableString objects:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
901 [x.name for x in tag.children]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
902 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
903 return None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
904
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
905 @name.setter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
906 def name(self, name):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
907 """Prevent NavigableString.name from ever being set."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
908 raise AttributeError("A NavigableString cannot be given a name.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
909
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
910
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
911 class PreformattedString(NavigableString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
912 """A NavigableString not subject to the normal formatting rules.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
913
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
914 This is an abstract class used for special kinds of strings such
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
915 as comments (the Comment class) and CDATA blocks (the CData
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
916 class).
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
917 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
918
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
919 PREFIX = ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
920 SUFFIX = ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
921
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
922 def output_ready(self, formatter=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
923 """Make this string ready for output by adding any subclass-specific
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
924 prefix or suffix.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
925
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
926 :param formatter: A Formatter object, or a string naming one
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
927 of the standard formatters. The string will be passed into the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
928 Formatter, but only to trigger any side effects: the return
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
929 value is ignored.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
930
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
931 :return: The string, with any subclass-specific prefix and
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
932 suffix added on.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
933 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
934 if formatter is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
935 ignore = self.format_string(self, formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
936 return self.PREFIX + self + self.SUFFIX
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
937
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
938 class CData(PreformattedString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
939 """A CDATA block."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
940 PREFIX = '<![CDATA['
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
941 SUFFIX = ']]>'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
942
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
943 class ProcessingInstruction(PreformattedString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
944 """A SGML processing instruction."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
945
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
946 PREFIX = '<?'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
947 SUFFIX = '>'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
948
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
949 class XMLProcessingInstruction(ProcessingInstruction):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
950 """An XML processing instruction."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
951 PREFIX = '<?'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
952 SUFFIX = '?>'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
953
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
954 class Comment(PreformattedString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
955 """An HTML or XML comment."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
956 PREFIX = '<!--'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
957 SUFFIX = '-->'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
958
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
959
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
960 class Declaration(PreformattedString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
961 """An XML declaration."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
962 PREFIX = '<?'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
963 SUFFIX = '?>'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
964
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
965
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
966 class Doctype(PreformattedString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
967 """A document type declaration."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
968 @classmethod
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
969 def for_name_and_ids(cls, name, pub_id, system_id):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
970 """Generate an appropriate document type declaration for a given
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
971 public ID and system ID.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
972
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
973 :param name: The name of the document's root element, e.g. 'html'.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
974 :param pub_id: The Formal Public Identifier for this document type,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
975 e.g. '-//W3C//DTD XHTML 1.1//EN'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
976 :param system_id: The system identifier for this document type,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
977 e.g. 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
978
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
979 :return: A Doctype.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
980 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
981 value = name or ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
982 if pub_id is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
983 value += ' PUBLIC "%s"' % pub_id
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
984 if system_id is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
985 value += ' "%s"' % system_id
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
986 elif system_id is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
987 value += ' SYSTEM "%s"' % system_id
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
988
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
989 return Doctype(value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
990
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
991 PREFIX = '<!DOCTYPE '
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
992 SUFFIX = '>\n'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
993
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
994
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
995 class Stylesheet(NavigableString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
996 """A NavigableString representing an stylesheet (probably
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
997 CSS).
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
998
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
999 Used to distinguish embedded stylesheets from textual content.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1000 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1001 pass
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1002
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1003
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1004 class Script(NavigableString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1005 """A NavigableString representing an executable script (probably
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1006 Javascript).
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1007
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1008 Used to distinguish executable code from textual content.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1009 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1010 pass
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1011
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1012
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1013 class TemplateString(NavigableString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1014 """A NavigableString representing a string found inside an HTML
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1015 template embedded in a larger document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1016
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1017 Used to distinguish such strings from the main body of the document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1018 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1019 pass
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1020
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1021
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1022 class Tag(PageElement):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1023 """Represents an HTML or XML tag that is part of a parse tree, along
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1024 with its attributes and contents.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1025
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1026 When Beautiful Soup parses the markup <b>penguin</b>, it will
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1027 create a Tag object representing the <b> tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1028 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1029
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1030 def __init__(self, parser=None, builder=None, name=None, namespace=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1031 prefix=None, attrs=None, parent=None, previous=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1032 is_xml=None, sourceline=None, sourcepos=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1033 can_be_empty_element=None, cdata_list_attributes=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1034 preserve_whitespace_tags=None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1035 ):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1036 """Basic constructor.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1037
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1038 :param parser: A BeautifulSoup object.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1039 :param builder: A TreeBuilder.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1040 :param name: The name of the tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1041 :param namespace: The URI of this Tag's XML namespace, if any.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1042 :param prefix: The prefix for this Tag's XML namespace, if any.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1043 :param attrs: A dictionary of this Tag's attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1044 :param parent: The PageElement to use as this Tag's parent.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1045 :param previous: The PageElement that was parsed immediately before
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1046 this tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1047 :param is_xml: If True, this is an XML tag. Otherwise, this is an
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1048 HTML tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1049 :param sourceline: The line number where this tag was found in its
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1050 source document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1051 :param sourcepos: The character position within `sourceline` where this
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1052 tag was found.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1053 :param can_be_empty_element: If True, this tag should be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1054 represented as <tag/>. If False, this tag should be represented
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1055 as <tag></tag>.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1056 :param cdata_list_attributes: A list of attributes whose values should
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1057 be treated as CDATA if they ever show up on this tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1058 :param preserve_whitespace_tags: A list of tag names whose contents
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1059 should have their whitespace preserved.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1060 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1061 if parser is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1062 self.parser_class = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1063 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1064 # We don't actually store the parser object: that lets extracted
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1065 # chunks be garbage-collected.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1066 self.parser_class = parser.__class__
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1067 if name is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1068 raise ValueError("No value provided for new tag's name.")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1069 self.name = name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1070 self.namespace = namespace
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1071 self.prefix = prefix
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1072 if ((not builder or builder.store_line_numbers)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1073 and (sourceline is not None or sourcepos is not None)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1074 self.sourceline = sourceline
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1075 self.sourcepos = sourcepos
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1076 if attrs is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1077 attrs = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1078 elif attrs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1079 if builder is not None and builder.cdata_list_attributes:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1080 attrs = builder._replace_cdata_list_attribute_values(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1081 self.name, attrs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1082 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1083 attrs = dict(attrs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1084 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1085 attrs = dict(attrs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1086
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1087 # If possible, determine ahead of time whether this tag is an
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1088 # XML tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1089 if builder:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1090 self.known_xml = builder.is_xml
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1091 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1092 self.known_xml = is_xml
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1093 self.attrs = attrs
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1094 self.contents = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1095 self.setup(parent, previous)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1096 self.hidden = False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1097
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1098 if builder is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1099 # In the absence of a TreeBuilder, use whatever values were
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1100 # passed in here. They're probably None, unless this is a copy of some
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1101 # other tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1102 self.can_be_empty_element = can_be_empty_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1103 self.cdata_list_attributes = cdata_list_attributes
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1104 self.preserve_whitespace_tags = preserve_whitespace_tags
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1105 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1106 # Set up any substitutions for this tag, such as the charset in a META tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1107 builder.set_up_substitutions(self)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1108
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1109 # Ask the TreeBuilder whether this tag might be an empty-element tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1110 self.can_be_empty_element = builder.can_be_empty_element(name)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1111
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1112 # Keep track of the list of attributes of this tag that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1113 # might need to be treated as a list.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1114 #
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1115 # For performance reasons, we store the whole data structure
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1116 # rather than asking the question of every tag. Asking would
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1117 # require building a new data structure every time, and
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1118 # (unlike can_be_empty_element), we almost never need
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1119 # to check this.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1120 self.cdata_list_attributes = builder.cdata_list_attributes
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1121
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1122 # Keep track of the names that might cause this tag to be treated as a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1123 # whitespace-preserved tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1124 self.preserve_whitespace_tags = builder.preserve_whitespace_tags
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1125
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1126 parserClass = _alias("parser_class") # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1127
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1128 def __copy__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1129 """A copy of a Tag is a new Tag, unconnected to the parse tree.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1130 Its contents are a copy of the old Tag's contents.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1131 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1132 clone = type(self)(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1133 None, self.builder, self.name, self.namespace,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1134 self.prefix, self.attrs, is_xml=self._is_xml,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1135 sourceline=self.sourceline, sourcepos=self.sourcepos,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1136 can_be_empty_element=self.can_be_empty_element,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1137 cdata_list_attributes=self.cdata_list_attributes,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1138 preserve_whitespace_tags=self.preserve_whitespace_tags
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1139 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1140 for attr in ('can_be_empty_element', 'hidden'):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1141 setattr(clone, attr, getattr(self, attr))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1142 for child in self.contents:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1143 clone.append(child.__copy__())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1144 return clone
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1145
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1146 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1147 def is_empty_element(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1148 """Is this tag an empty-element tag? (aka a self-closing tag)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1149
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1150 A tag that has contents is never an empty-element tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1151
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1152 A tag that has no contents may or may not be an empty-element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1153 tag. It depends on the builder used to create the tag. If the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1154 builder has a designated list of empty-element tags, then only
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1155 a tag whose name shows up in that list is considered an
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1156 empty-element tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1157
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1158 If the builder has no designated list of empty-element tags,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1159 then any tag with no contents is an empty-element tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1160 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1161 return len(self.contents) == 0 and self.can_be_empty_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1162 isSelfClosing = is_empty_element # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1163
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1164 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1165 def string(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1166 """Convenience property to get the single string within this
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1167 PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1168
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1169 TODO It might make sense to have NavigableString.string return
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1170 itself.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1171
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1172 :return: If this element has a single string child, return
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1173 value is that string. If this element has one child tag,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1174 return value is the 'string' attribute of the child tag,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1175 recursively. If this element is itself a string, has no
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1176 children, or has more than one child, return value is None.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1177 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1178 if len(self.contents) != 1:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1179 return None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1180 child = self.contents[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1181 if isinstance(child, NavigableString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1182 return child
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1183 return child.string
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1184
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1185 @string.setter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1186 def string(self, string):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1187 """Replace this PageElement's contents with `string`."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1188 self.clear()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1189 self.append(string.__class__(string))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1190
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1191 def _all_strings(self, strip=False, types=(NavigableString, CData)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1192 """Yield all strings of certain classes, possibly stripping them.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1193
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1194 :param strip: If True, all strings will be stripped before being
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1195 yielded.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1196
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1197 :types: A tuple of NavigableString subclasses. Any strings of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1198 a subclass not found in this list will be ignored. By
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1199 default, this means only NavigableString and CData objects
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1200 will be considered. So no comments, processing instructions,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1201 etc.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1202
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1203 :yield: A sequence of strings.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1204 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1205 for descendant in self.descendants:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1206 if (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1207 (types is None and not isinstance(descendant, NavigableString))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1208 or
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1209 (types is not None and type(descendant) not in types)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1210 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1211 if strip:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1212 descendant = descendant.strip()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1213 if len(descendant) == 0:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1214 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1215 yield descendant
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1216
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1217 strings = property(_all_strings)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1218
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1219 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1220 def stripped_strings(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1221 """Yield all strings in the document, stripping them first.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1222
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1223 :yield: A sequence of stripped strings.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1224 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1225 for string in self._all_strings(True):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1226 yield string
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1227
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1228 def get_text(self, separator="", strip=False,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1229 types=(NavigableString, CData)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1230 """Get all child strings, concatenated using the given separator.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1231
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1232 :param separator: Strings will be concatenated using this separator.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1233
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1234 :param strip: If True, strings will be stripped before being
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1235 concatenated.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1236
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1237 :types: A tuple of NavigableString subclasses. Any strings of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1238 a subclass not found in this list will be ignored. By
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1239 default, this means only NavigableString and CData objects
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1240 will be considered. So no comments, processing instructions,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1241 stylesheets, etc.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1242
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1243 :return: A string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1244 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1245 return separator.join([s for s in self._all_strings(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1246 strip, types=types)])
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1247 getText = get_text
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1248 text = property(get_text)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1249
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1250 def decompose(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1251 """Recursively destroys this PageElement and its children.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1252
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1253 This element will be removed from the tree and wiped out; so
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1254 will everything beneath it.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1255
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1256 The behavior of a decomposed PageElement is undefined and you
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1257 should never use one for anything, but if you need to _check_
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1258 whether an element has been decomposed, you can use the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1259 `decomposed` property.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1260 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1261 self.extract()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1262 i = self
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1263 while i is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1264 n = i.next_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1265 i.__dict__.clear()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1266 i.contents = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1267 i._decomposed = True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1268 i = n
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1269
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1270 def clear(self, decompose=False):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1271 """Wipe out all children of this PageElement by calling extract()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1272 on them.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1273
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1274 :param decompose: If this is True, decompose() (a more
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1275 destructive method) will be called instead of extract().
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1276 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1277 if decompose:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1278 for element in self.contents[:]:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1279 if isinstance(element, Tag):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1280 element.decompose()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1281 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1282 element.extract()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1283 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1284 for element in self.contents[:]:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1285 element.extract()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1286
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1287 def smooth(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1288 """Smooth out this element's children by consolidating consecutive
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1289 strings.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1290
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1291 This makes pretty-printed output look more natural following a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1292 lot of operations that modified the tree.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1293 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1294 # Mark the first position of every pair of children that need
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1295 # to be consolidated. Do this rather than making a copy of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1296 # self.contents, since in most cases very few strings will be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1297 # affected.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1298 marked = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1299 for i, a in enumerate(self.contents):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1300 if isinstance(a, Tag):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1301 # Recursively smooth children.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1302 a.smooth()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1303 if i == len(self.contents)-1:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1304 # This is the last item in .contents, and it's not a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1305 # tag. There's no chance it needs any work.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1306 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1307 b = self.contents[i+1]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1308 if (isinstance(a, NavigableString)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1309 and isinstance(b, NavigableString)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1310 and not isinstance(a, PreformattedString)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1311 and not isinstance(b, PreformattedString)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1312 ):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1313 marked.append(i)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1314
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1315 # Go over the marked positions in reverse order, so that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1316 # removing items from .contents won't affect the remaining
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1317 # positions.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1318 for i in reversed(marked):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1319 a = self.contents[i]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1320 b = self.contents[i+1]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1321 b.extract()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1322 n = NavigableString(a+b)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1323 a.replace_with(n)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1324
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1325 def index(self, element):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1326 """Find the index of a child by identity, not value.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1327
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1328 Avoids issues with tag.contents.index(element) getting the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1329 index of equal elements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1330
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1331 :param element: Look for this PageElement in `self.contents`.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1332 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1333 for i, child in enumerate(self.contents):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1334 if child is element:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1335 return i
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1336 raise ValueError("Tag.index: element not in tag")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1337
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1338 def get(self, key, default=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1339 """Returns the value of the 'key' attribute for the tag, or
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1340 the value given for 'default' if it doesn't have that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1341 attribute."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1342 return self.attrs.get(key, default)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1343
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1344 def get_attribute_list(self, key, default=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1345 """The same as get(), but always returns a list.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1346
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1347 :param key: The attribute to look for.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1348 :param default: Use this value if the attribute is not present
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1349 on this PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1350 :return: A list of values, probably containing only a single
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1351 value.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1352 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1353 value = self.get(key, default)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1354 if not isinstance(value, list):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1355 value = [value]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1356 return value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1357
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1358 def has_attr(self, key):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1359 """Does this PageElement have an attribute with the given name?"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1360 return key in self.attrs
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1361
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1362 def __hash__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1363 return str(self).__hash__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1364
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1365 def __getitem__(self, key):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1366 """tag[key] returns the value of the 'key' attribute for the Tag,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1367 and throws an exception if it's not there."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1368 return self.attrs[key]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1369
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1370 def __iter__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1371 "Iterating over a Tag iterates over its contents."
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1372 return iter(self.contents)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1373
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1374 def __len__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1375 "The length of a Tag is the length of its list of contents."
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1376 return len(self.contents)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1377
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1378 def __contains__(self, x):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1379 return x in self.contents
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1380
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1381 def __bool__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1382 "A tag is non-None even if it has no contents."
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1383 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1384
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1385 def __setitem__(self, key, value):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1386 """Setting tag[key] sets the value of the 'key' attribute for the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1387 tag."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1388 self.attrs[key] = value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1389
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1390 def __delitem__(self, key):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1391 "Deleting tag[key] deletes all 'key' attributes for the tag."
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1392 self.attrs.pop(key, None)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1393
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1394 def __call__(self, *args, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1395 """Calling a Tag like a function is the same as calling its
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1396 find_all() method. Eg. tag('a') returns a list of all the A tags
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1397 found within this tag."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1398 return self.find_all(*args, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1399
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1400 def __getattr__(self, tag):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1401 """Calling tag.subtag is the same as calling tag.find(name="subtag")"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1402 #print "Getattr %s.%s" % (self.__class__, tag)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1403 if len(tag) > 3 and tag.endswith('Tag'):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1404 # BS3: soup.aTag -> "soup.find("a")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1405 tag_name = tag[:-3]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1406 warnings.warn(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1407 '.%(name)sTag is deprecated, use .find("%(name)s") instead. If you really were looking for a tag called %(name)sTag, use .find("%(name)sTag")' % dict(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1408 name=tag_name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1409 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1410 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1411 return self.find(tag_name)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1412 # We special case contents to avoid recursion.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1413 elif not tag.startswith("__") and not tag == "contents":
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1414 return self.find(tag)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1415 raise AttributeError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1416 "'%s' object has no attribute '%s'" % (self.__class__, tag))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1417
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1418 def __eq__(self, other):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1419 """Returns true iff this Tag has the same name, the same attributes,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1420 and the same contents (recursively) as `other`."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1421 if self is other:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1422 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1423 if (not hasattr(other, 'name') or
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1424 not hasattr(other, 'attrs') or
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1425 not hasattr(other, 'contents') or
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1426 self.name != other.name or
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1427 self.attrs != other.attrs or
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1428 len(self) != len(other)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1429 return False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1430 for i, my_child in enumerate(self.contents):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1431 if my_child != other.contents[i]:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1432 return False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1433 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1434
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1435 def __ne__(self, other):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1436 """Returns true iff this Tag is not identical to `other`,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1437 as defined in __eq__."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1438 return not self == other
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1439
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1440 def __repr__(self, encoding="unicode-escape"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1441 """Renders this PageElement as a string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1442
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1443 :param encoding: The encoding to use (Python 2 only).
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1444 :return: Under Python 2, a bytestring; under Python 3,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1445 a Unicode string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1446 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1447 if PY3K:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1448 # "The return value must be a string object", i.e. Unicode
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1449 return self.decode()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1450 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1451 # "The return value must be a string object", i.e. a bytestring.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1452 # By convention, the return value of __repr__ should also be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1453 # an ASCII string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1454 return self.encode(encoding)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1455
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1456 def __unicode__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1457 """Renders this PageElement as a Unicode string."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1458 return self.decode()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1459
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1460 def __str__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1461 """Renders this PageElement as a generic string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1462
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1463 :return: Under Python 2, a UTF-8 bytestring; under Python 3,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1464 a Unicode string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1465 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1466 if PY3K:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1467 return self.decode()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1468 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1469 return self.encode()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1470
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1471 if PY3K:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1472 __str__ = __repr__ = __unicode__
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1473
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1474 def encode(self, encoding=DEFAULT_OUTPUT_ENCODING,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1475 indent_level=None, formatter="minimal",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1476 errors="xmlcharrefreplace"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1477 """Render a bytestring representation of this PageElement and its
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1478 contents.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1479
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1480 :param encoding: The destination encoding.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1481 :param indent_level: Each line of the rendering will be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1482 indented this many spaces. Used internally in
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1483 recursive calls while pretty-printing.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1484 :param formatter: A Formatter object, or a string naming one of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1485 the standard formatters.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1486 :param errors: An error handling strategy such as
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1487 'xmlcharrefreplace'. This value is passed along into
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1488 encode() and its value should be one of the constants
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1489 defined by Python.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1490 :return: A bytestring.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1491
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1492 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1493 # Turn the data structure into Unicode, then encode the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1494 # Unicode.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1495 u = self.decode(indent_level, encoding, formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1496 return u.encode(encoding, errors)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1497
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1498 def decode(self, indent_level=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1499 eventual_encoding=DEFAULT_OUTPUT_ENCODING,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1500 formatter="minimal"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1501 """Render a Unicode representation of this PageElement and its
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1502 contents.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1503
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1504 :param indent_level: Each line of the rendering will be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1505 indented this many spaces. Used internally in
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1506 recursive calls while pretty-printing.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1507 :param eventual_encoding: The tag is destined to be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1508 encoded into this encoding. This method is _not_
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1509 responsible for performing that encoding. This information
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1510 is passed in so that it can be substituted in if the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1511 document contains a <META> tag that mentions the document's
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1512 encoding.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1513 :param formatter: A Formatter object, or a string naming one of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1514 the standard formatters.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1515 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1516
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1517 # First off, turn a non-Formatter `formatter` into a Formatter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1518 # object. This will stop the lookup from happening over and
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1519 # over again.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1520 if not isinstance(formatter, Formatter):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1521 formatter = self.formatter_for_name(formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1522 attributes = formatter.attributes(self)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1523 attrs = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1524 for key, val in attributes:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1525 if val is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1526 decoded = key
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1527 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1528 if isinstance(val, list) or isinstance(val, tuple):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1529 val = ' '.join(val)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1530 elif not isinstance(val, str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1531 val = str(val)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1532 elif (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1533 isinstance(val, AttributeValueWithCharsetSubstitution)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1534 and eventual_encoding is not None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1535 ):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1536 val = val.encode(eventual_encoding)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1537
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1538 text = formatter.attribute_value(val)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1539 decoded = (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1540 str(key) + '='
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1541 + formatter.quoted_attribute_value(text))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1542 attrs.append(decoded)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1543 close = ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1544 closeTag = ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1545
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1546 prefix = ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1547 if self.prefix:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1548 prefix = self.prefix + ":"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1549
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1550 if self.is_empty_element:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1551 close = formatter.void_element_close_prefix or ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1552 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1553 closeTag = '</%s%s>' % (prefix, self.name)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1554
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1555 pretty_print = self._should_pretty_print(indent_level)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1556 space = ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1557 indent_space = ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1558 if indent_level is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1559 indent_space = (' ' * (indent_level - 1))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1560 if pretty_print:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1561 space = indent_space
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1562 indent_contents = indent_level + 1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1563 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1564 indent_contents = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1565 contents = self.decode_contents(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1566 indent_contents, eventual_encoding, formatter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1567 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1568
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1569 if self.hidden:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1570 # This is the 'document root' object.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1571 s = contents
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1572 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1573 s = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1574 attribute_string = ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1575 if attrs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1576 attribute_string = ' ' + ' '.join(attrs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1577 if indent_level is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1578 # Even if this particular tag is not pretty-printed,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1579 # we should indent up to the start of the tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1580 s.append(indent_space)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1581 s.append('<%s%s%s%s>' % (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1582 prefix, self.name, attribute_string, close))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1583 if pretty_print:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1584 s.append("\n")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1585 s.append(contents)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1586 if pretty_print and contents and contents[-1] != "\n":
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1587 s.append("\n")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1588 if pretty_print and closeTag:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1589 s.append(space)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1590 s.append(closeTag)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1591 if indent_level is not None and closeTag and self.next_sibling:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1592 # Even if this particular tag is not pretty-printed,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1593 # we're now done with the tag, and we should add a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1594 # newline if appropriate.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1595 s.append("\n")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1596 s = ''.join(s)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1597 return s
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1598
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1599 def _should_pretty_print(self, indent_level):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1600 """Should this tag be pretty-printed?
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1601
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1602 Most of them should, but some (such as <pre> in HTML
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1603 documents) should not.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1604 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1605 return (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1606 indent_level is not None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1607 and (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1608 not self.preserve_whitespace_tags
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1609 or self.name not in self.preserve_whitespace_tags
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1610 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1611 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1612
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1613 def prettify(self, encoding=None, formatter="minimal"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1614 """Pretty-print this PageElement as a string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1615
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1616 :param encoding: The eventual encoding of the string. If this is None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1617 a Unicode string will be returned.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1618 :param formatter: A Formatter object, or a string naming one of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1619 the standard formatters.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1620 :return: A Unicode string (if encoding==None) or a bytestring
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1621 (otherwise).
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1622 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1623 if encoding is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1624 return self.decode(True, formatter=formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1625 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1626 return self.encode(encoding, True, formatter=formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1627
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1628 def decode_contents(self, indent_level=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1629 eventual_encoding=DEFAULT_OUTPUT_ENCODING,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1630 formatter="minimal"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1631 """Renders the contents of this tag as a Unicode string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1632
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1633 :param indent_level: Each line of the rendering will be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1634 indented this many spaces. Used internally in
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1635 recursive calls while pretty-printing.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1636
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1637 :param eventual_encoding: The tag is destined to be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1638 encoded into this encoding. decode_contents() is _not_
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1639 responsible for performing that encoding. This information
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1640 is passed in so that it can be substituted in if the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1641 document contains a <META> tag that mentions the document's
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1642 encoding.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1643
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1644 :param formatter: A Formatter object, or a string naming one of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1645 the standard Formatters.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1646 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1647 # First off, turn a string formatter into a Formatter object. This
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1648 # will stop the lookup from happening over and over again.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1649 if not isinstance(formatter, Formatter):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1650 formatter = self.formatter_for_name(formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1651
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1652 pretty_print = (indent_level is not None)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1653 s = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1654 for c in self:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1655 text = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1656 if isinstance(c, NavigableString):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1657 text = c.output_ready(formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1658 elif isinstance(c, Tag):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1659 s.append(c.decode(indent_level, eventual_encoding,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1660 formatter))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1661 preserve_whitespace = (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1662 self.preserve_whitespace_tags and self.name in self.preserve_whitespace_tags
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1663 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1664 if text and indent_level and not preserve_whitespace:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1665 text = text.strip()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1666 if text:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1667 if pretty_print and not preserve_whitespace:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1668 s.append(" " * (indent_level - 1))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1669 s.append(text)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1670 if pretty_print and not preserve_whitespace:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1671 s.append("\n")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1672 return ''.join(s)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1673
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1674 def encode_contents(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1675 self, indent_level=None, encoding=DEFAULT_OUTPUT_ENCODING,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1676 formatter="minimal"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1677 """Renders the contents of this PageElement as a bytestring.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1678
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1679 :param indent_level: Each line of the rendering will be
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1680 indented this many spaces. Used internally in
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1681 recursive calls while pretty-printing.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1682
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1683 :param eventual_encoding: The bytestring will be in this encoding.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1684
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1685 :param formatter: A Formatter object, or a string naming one of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1686 the standard Formatters.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1687
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1688 :return: A bytestring.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1689 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1690 contents = self.decode_contents(indent_level, encoding, formatter)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1691 return contents.encode(encoding)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1692
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1693 # Old method for BS3 compatibility
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1694 def renderContents(self, encoding=DEFAULT_OUTPUT_ENCODING,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1695 prettyPrint=False, indentLevel=0):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1696 """Deprecated method for BS3 compatibility."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1697 if not prettyPrint:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1698 indentLevel = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1699 return self.encode_contents(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1700 indent_level=indentLevel, encoding=encoding)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1701
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1702 #Soup methods
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1703
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1704 def find(self, name=None, attrs={}, recursive=True, text=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1705 **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1706 """Look in the children of this PageElement and find the first
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1707 PageElement that matches the given criteria.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1708
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1709 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1710 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1711
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1712 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1713 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1714 :param recursive: If this is True, find() will perform a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1715 recursive search of this PageElement's children. Otherwise,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1716 only the direct children will be considered.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1717 :param limit: Stop looking after finding this many results.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1718 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1719 :return: A PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1720 :rtype: bs4.element.Tag | bs4.element.NavigableString
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1721 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1722 r = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1723 l = self.find_all(name, attrs, recursive, text, 1, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1724 if l:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1725 r = l[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1726 return r
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1727 findChild = find
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1728
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1729 def find_all(self, name=None, attrs={}, recursive=True, text=None,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1730 limit=None, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1731 """Look in the children of this PageElement and find all
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1732 PageElements that match the given criteria.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1733
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1734 All find_* methods take a common set of arguments. See the online
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1735 documentation for detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1736
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1737 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1738 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1739 :param recursive: If this is True, find_all() will perform a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1740 recursive search of this PageElement's children. Otherwise,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1741 only the direct children will be considered.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1742 :param limit: Stop looking after finding this many results.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1743 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1744 :return: A ResultSet of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1745 :rtype: bs4.element.ResultSet
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1746 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1747 generator = self.descendants
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1748 if not recursive:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1749 generator = self.children
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1750 return self._find_all(name, attrs, text, limit, generator, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1751 findAll = find_all # BS3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1752 findChildren = find_all # BS2
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1753
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1754 #Generator methods
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1755 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1756 def children(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1757 """Iterate over all direct children of this PageElement.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1758
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1759 :yield: A sequence of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1760 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1761 # return iter() to make the purpose of the method clear
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1762 return iter(self.contents) # XXX This seems to be untested.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1763
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1764 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1765 def descendants(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1766 """Iterate over all children of this PageElement in a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1767 breadth-first sequence.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1768
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1769 :yield: A sequence of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1770 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1771 if not len(self.contents):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1772 return
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1773 stopNode = self._last_descendant().next_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1774 current = self.contents[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1775 while current is not stopNode:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1776 yield current
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1777 current = current.next_element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1778
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1779 # CSS selector code
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1780 def select_one(self, selector, namespaces=None, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1781 """Perform a CSS selection operation on the current element.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1782
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1783 :param selector: A CSS selector.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1784
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1785 :param namespaces: A dictionary mapping namespace prefixes
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1786 used in the CSS selector to namespace URIs. By default,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1787 Beautiful Soup will use the prefixes it encountered while
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1788 parsing the document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1789
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1790 :param kwargs: Keyword arguments to be passed into SoupSieve's
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1791 soupsieve.select() method.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1792
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1793 :return: A Tag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1794 :rtype: bs4.element.Tag
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1795 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1796 value = self.select(selector, namespaces, 1, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1797 if value:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1798 return value[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1799 return None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1800
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1801 def select(self, selector, namespaces=None, limit=None, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1802 """Perform a CSS selection operation on the current element.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1803
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1804 This uses the SoupSieve library.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1805
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1806 :param selector: A string containing a CSS selector.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1807
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1808 :param namespaces: A dictionary mapping namespace prefixes
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1809 used in the CSS selector to namespace URIs. By default,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1810 Beautiful Soup will use the prefixes it encountered while
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1811 parsing the document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1812
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1813 :param limit: After finding this number of results, stop looking.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1814
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1815 :param kwargs: Keyword arguments to be passed into SoupSieve's
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1816 soupsieve.select() method.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1817
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1818 :return: A ResultSet of Tags.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1819 :rtype: bs4.element.ResultSet
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1820 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1821 if namespaces is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1822 namespaces = self._namespaces
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1823
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1824 if limit is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1825 limit = 0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1826 if soupsieve is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1827 raise NotImplementedError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1828 "Cannot execute CSS selectors because the soupsieve package is not installed."
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1829 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1830
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1831 results = soupsieve.select(selector, self, namespaces, limit, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1832
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1833 # We do this because it's more consistent and because
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1834 # ResultSet.__getattr__ has a helpful error message.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1835 return ResultSet(None, results)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1836
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1837 # Old names for backwards compatibility
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1838 def childGenerator(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1839 """Deprecated generator."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1840 return self.children
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1841
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1842 def recursiveChildGenerator(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1843 """Deprecated generator."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1844 return self.descendants
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1845
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1846 def has_key(self, key):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1847 """Deprecated method. This was kind of misleading because has_key()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1848 (attributes) was different from __in__ (contents).
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1849
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1850 has_key() is gone in Python 3, anyway.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1851 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1852 warnings.warn('has_key is deprecated. Use has_attr("%s") instead.' % (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1853 key))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1854 return self.has_attr(key)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1855
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1856 # Next, a couple classes to represent queries and their results.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1857 class SoupStrainer(object):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1858 """Encapsulates a number of ways of matching a markup element (tag or
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1859 string).
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1860
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1861 This is primarily used to underpin the find_* methods, but you can
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1862 create one yourself and pass it in as `parse_only` to the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1863 `BeautifulSoup` constructor, to parse a subset of a large
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1864 document.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1865 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1866
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1867 def __init__(self, name=None, attrs={}, text=None, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1868 """Constructor.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1869
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1870 The SoupStrainer constructor takes the same arguments passed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1871 into the find_* methods. See the online documentation for
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1872 detailed explanations.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1873
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1874 :param name: A filter on tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1875 :param attrs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1876 :param text: A filter for a NavigableString with specific text.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1877 :kwargs: A dictionary of filters on attribute values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1878 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1879 self.name = self._normalize_search_value(name)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1880 if not isinstance(attrs, dict):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1881 # Treat a non-dict value for attrs as a search for the 'class'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1882 # attribute.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1883 kwargs['class'] = attrs
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1884 attrs = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1885
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1886 if 'class_' in kwargs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1887 # Treat class_="foo" as a search for the 'class'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1888 # attribute, overriding any non-dict value for attrs.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1889 kwargs['class'] = kwargs['class_']
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1890 del kwargs['class_']
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1891
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1892 if kwargs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1893 if attrs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1894 attrs = attrs.copy()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1895 attrs.update(kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1896 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1897 attrs = kwargs
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1898 normalized_attrs = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1899 for key, value in list(attrs.items()):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1900 normalized_attrs[key] = self._normalize_search_value(value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1901
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1902 self.attrs = normalized_attrs
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1903 self.text = self._normalize_search_value(text)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1904
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1905 def _normalize_search_value(self, value):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1906 # Leave it alone if it's a Unicode string, a callable, a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1907 # regular expression, a boolean, or None.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1908 if (isinstance(value, str) or isinstance(value, Callable) or hasattr(value, 'match')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1909 or isinstance(value, bool) or value is None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1910 return value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1911
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1912 # If it's a bytestring, convert it to Unicode, treating it as UTF-8.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1913 if isinstance(value, bytes):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1914 return value.decode("utf8")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1915
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1916 # If it's listlike, convert it into a list of strings.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1917 if hasattr(value, '__iter__'):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1918 new_value = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1919 for v in value:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1920 if (hasattr(v, '__iter__') and not isinstance(v, bytes)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1921 and not isinstance(v, str)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1922 # This is almost certainly the user's mistake. In the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1923 # interests of avoiding infinite loops, we'll let
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1924 # it through as-is rather than doing a recursive call.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1925 new_value.append(v)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1926 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1927 new_value.append(self._normalize_search_value(v))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1928 return new_value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1929
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1930 # Otherwise, convert it into a Unicode string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1931 # The unicode(str()) thing is so this will do the same thing on Python 2
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1932 # and Python 3.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1933 return str(str(value))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1934
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1935 def __str__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1936 """A human-readable representation of this SoupStrainer."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1937 if self.text:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1938 return self.text
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1939 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1940 return "%s|%s" % (self.name, self.attrs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1941
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1942 def search_tag(self, markup_name=None, markup_attrs={}):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1943 """Check whether a Tag with the given name and attributes would
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1944 match this SoupStrainer.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1945
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1946 Used prospectively to decide whether to even bother creating a Tag
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1947 object.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1948
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1949 :param markup_name: A tag name as found in some markup.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1950 :param markup_attrs: A dictionary of attributes as found in some markup.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1951
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1952 :return: True if the prospective tag would match this SoupStrainer;
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1953 False otherwise.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1954 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1955 found = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1956 markup = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1957 if isinstance(markup_name, Tag):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1958 markup = markup_name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1959 markup_attrs = markup
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1960 call_function_with_tag_data = (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1961 isinstance(self.name, Callable)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1962 and not isinstance(markup_name, Tag))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1963
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1964 if ((not self.name)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1965 or call_function_with_tag_data
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1966 or (markup and self._matches(markup, self.name))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1967 or (not markup and self._matches(markup_name, self.name))):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1968 if call_function_with_tag_data:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1969 match = self.name(markup_name, markup_attrs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1970 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1971 match = True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1972 markup_attr_map = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1973 for attr, match_against in list(self.attrs.items()):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1974 if not markup_attr_map:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1975 if hasattr(markup_attrs, 'get'):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1976 markup_attr_map = markup_attrs
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1977 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1978 markup_attr_map = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1979 for k, v in markup_attrs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1980 markup_attr_map[k] = v
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1981 attr_value = markup_attr_map.get(attr)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1982 if not self._matches(attr_value, match_against):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1983 match = False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1984 break
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1985 if match:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1986 if markup:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1987 found = markup
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1988 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1989 found = markup_name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1990 if found and self.text and not self._matches(found.string, self.text):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1991 found = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1992 return found
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1993
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1994 # For BS3 compatibility.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1995 searchTag = search_tag
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1996
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1997 def search(self, markup):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1998 """Find all items in `markup` that match this SoupStrainer.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1999
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2000 Used by the core _find_all() method, which is ultimately
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2001 called by all find_* methods.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2002
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2003 :param markup: A PageElement or a list of them.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2004 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2005 # print 'looking for %s in %s' % (self, markup)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2006 found = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2007 # If given a list of items, scan it for a text element that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2008 # matches.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2009 if hasattr(markup, '__iter__') and not isinstance(markup, (Tag, str)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2010 for element in markup:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2011 if isinstance(element, NavigableString) \
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2012 and self.search(element):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2013 found = element
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2014 break
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2015 # If it's a Tag, make sure its name or attributes match.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2016 # Don't bother with Tags if we're searching for text.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2017 elif isinstance(markup, Tag):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2018 if not self.text or self.name or self.attrs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2019 found = self.search_tag(markup)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2020 # If it's text, make sure the text matches.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2021 elif isinstance(markup, NavigableString) or \
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2022 isinstance(markup, str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2023 if not self.name and not self.attrs and self._matches(markup, self.text):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2024 found = markup
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2025 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2026 raise Exception(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2027 "I don't know how to match against a %s" % markup.__class__)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2028 return found
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2029
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2030 def _matches(self, markup, match_against, already_tried=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2031 # print u"Matching %s against %s" % (markup, match_against)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2032 result = False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2033 if isinstance(markup, list) or isinstance(markup, tuple):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2034 # This should only happen when searching a multi-valued attribute
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2035 # like 'class'.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2036 for item in markup:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2037 if self._matches(item, match_against):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2038 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2039 # We didn't match any particular value of the multivalue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2040 # attribute, but maybe we match the attribute value when
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2041 # considered as a string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2042 if self._matches(' '.join(markup), match_against):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2043 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2044 return False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2045
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2046 if match_against is True:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2047 # True matches any non-None value.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2048 return markup is not None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2049
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2050 if isinstance(match_against, Callable):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2051 return match_against(markup)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2052
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2053 # Custom callables take the tag as an argument, but all
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2054 # other ways of matching match the tag name as a string.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2055 original_markup = markup
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2056 if isinstance(markup, Tag):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2057 markup = markup.name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2058
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2059 # Ensure that `markup` is either a Unicode string, or None.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2060 markup = self._normalize_search_value(markup)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2061
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2062 if markup is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2063 # None matches None, False, an empty string, an empty list, and so on.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2064 return not match_against
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2065
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2066 if (hasattr(match_against, '__iter__')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2067 and not isinstance(match_against, str)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2068 # We're asked to match against an iterable of items.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2069 # The markup must be match at least one item in the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2070 # iterable. We'll try each one in turn.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2071 #
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2072 # To avoid infinite recursion we need to keep track of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2073 # items we've already seen.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2074 if not already_tried:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2075 already_tried = set()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2076 for item in match_against:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2077 if item.__hash__:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2078 key = item
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2079 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2080 key = id(item)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2081 if key in already_tried:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2082 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2083 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2084 already_tried.add(key)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2085 if self._matches(original_markup, item, already_tried):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2086 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2087 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2088 return False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2089
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2090 # Beyond this point we might need to run the test twice: once against
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2091 # the tag's name and once against its prefixed name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2092 match = False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2093
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2094 if not match and isinstance(match_against, str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2095 # Exact string match
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2096 match = markup == match_against
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2097
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2098 if not match and hasattr(match_against, 'search'):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2099 # Regexp match
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2100 return match_against.search(markup)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2101
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2102 if (not match
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2103 and isinstance(original_markup, Tag)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2104 and original_markup.prefix):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2105 # Try the whole thing again with the prefixed tag name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2106 return self._matches(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2107 original_markup.prefix + ':' + original_markup.name, match_against
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2108 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2109
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2110 return match
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2111
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2112
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2113 class ResultSet(list):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2114 """A ResultSet is just a list that keeps track of the SoupStrainer
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2115 that created it."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2116 def __init__(self, source, result=()):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2117 """Constructor.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2118
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2119 :param source: A SoupStrainer.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2120 :param result: A list of PageElements.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2121 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2122 super(ResultSet, self).__init__(result)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2123 self.source = source
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2124
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2125 def __getattr__(self, key):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2126 """Raise a helpful exception to explain a common code fix."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2127 raise AttributeError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2128 "ResultSet object has no attribute '%s'. You're probably treating a list of elements like a single element. Did you call find_all() when you meant to call find()?" % key
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2129 )