annotate planemo/lib/python3.7/site-packages/bs4/tests/test_htmlparser.py @ 0:d30785e31577 draft

"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
author guerler
date Fri, 31 Jul 2020 00:18:57 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
1 """Tests to ensure that the html.parser tree builder generates good
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
2 trees."""
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
3
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
4 from pdb import set_trace
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
5 import pickle
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
6 from bs4.testing import SoupTest, HTMLTreeBuilderSmokeTest
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
7 from bs4.builder import HTMLParserTreeBuilder
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
8 from bs4.builder._htmlparser import BeautifulSoupHTMLParser
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
9
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
10 class HTMLParserTreeBuilderSmokeTest(SoupTest, HTMLTreeBuilderSmokeTest):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
11
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
12 default_builder = HTMLParserTreeBuilder
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
13
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
14 def test_namespaced_system_doctype(self):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
15 # html.parser can't handle namespaced doctypes, so skip this one.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
16 pass
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
17
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
18 def test_namespaced_public_doctype(self):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
19 # html.parser can't handle namespaced doctypes, so skip this one.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
20 pass
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
21
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
22 def test_builder_is_pickled(self):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
23 """Unlike most tree builders, HTMLParserTreeBuilder and will
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
24 be restored after pickling.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
25 """
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
26 tree = self.soup("<a><b>foo</a>")
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
27 dumped = pickle.dumps(tree, 2)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
28 loaded = pickle.loads(dumped)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
29 self.assertTrue(isinstance(loaded.builder, type(tree.builder)))
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
30
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
31 def test_redundant_empty_element_closing_tags(self):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
32 self.assertSoupEquals('<br></br><br></br><br></br>', "<br/><br/><br/>")
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
33 self.assertSoupEquals('</br></br></br>', "")
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
34
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
35 def test_empty_element(self):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
36 # This verifies that any buffered data present when the parser
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
37 # finishes working is handled.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
38 self.assertSoupEquals("foo &# bar", "foo &amp;# bar")
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
39
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
40 def test_tracking_line_numbers(self):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
41 # The html.parser TreeBuilder keeps track of line number and
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
42 # position of each element.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
43 markup = "\n <p>\n\n<sourceline>\n<b>text</b></sourceline><sourcepos></p>"
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
44 soup = self.soup(markup)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
45 self.assertEqual(2, soup.p.sourceline)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
46 self.assertEqual(3, soup.p.sourcepos)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
47 self.assertEqual("sourceline", soup.p.find('sourceline').name)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
48
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
49 # You can deactivate this behavior.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
50 soup = self.soup(markup, store_line_numbers=False)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
51 self.assertEqual("sourceline", soup.p.sourceline.name)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
52 self.assertEqual("sourcepos", soup.p.sourcepos.name)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
53
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
54 def test_on_duplicate_attribute(self):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
55 # The html.parser tree builder has a variety of ways of
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
56 # handling a tag that contains the same attribute multiple times.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
57
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
58 markup = '<a class="cls" href="url1" href="url2" href="url3" id="id">'
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
59
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
60 # If you don't provide any particular value for
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
61 # on_duplicate_attribute, later values replace earlier values.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
62 soup = self.soup(markup)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
63 self.assertEqual("url3", soup.a['href'])
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
64 self.assertEqual(["cls"], soup.a['class'])
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
65 self.assertEqual("id", soup.a['id'])
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
66
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
67 # You can also get this behavior explicitly.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
68 def assert_attribute(on_duplicate_attribute, expected):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
69 soup = self.soup(
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
70 markup, on_duplicate_attribute=on_duplicate_attribute
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
71 )
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
72 self.assertEqual(expected, soup.a['href'])
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
73
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
74 # Verify that non-duplicate attributes are treated normally.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
75 self.assertEqual(["cls"], soup.a['class'])
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
76 self.assertEqual("id", soup.a['id'])
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
77 assert_attribute(None, "url3")
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
78 assert_attribute(BeautifulSoupHTMLParser.REPLACE, "url3")
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
79
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
80 # You can ignore subsequent values in favor of the first.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
81 assert_attribute(BeautifulSoupHTMLParser.IGNORE, "url1")
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
82
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
83 # And you can pass in a callable that does whatever you want.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
84 def accumulate(attrs, key, value):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
85 if not isinstance(attrs[key], list):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
86 attrs[key] = [attrs[key]]
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
87 attrs[key].append(value)
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
88 assert_attribute(accumulate, ["url1", "url2", "url3"])
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
89
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
90
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
91 class TestHTMLParserSubclass(SoupTest):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
92 def test_error(self):
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
93 """Verify that our HTMLParser subclass implements error() in a way
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
94 that doesn't cause a crash.
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
95 """
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
96 parser = BeautifulSoupHTMLParser()
d30785e31577 "planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff changeset
97 parser.error("don't crash")