Mercurial > cpython-lru_cache-weakref
annotate Lib/test/test_htmlparser.py @ 25838:8540887bcec6 2.2
Backport of 2.183:
Fix memory leak in add_subclass() found by NealN with valgrind.
| author | Guido van Rossum <guido@python.org> |
|---|---|
| date | Fri, 18 Oct 2002 13:42:21 +0000 |
| parents | e592472d55a3 |
| children | aa0d2d51fbde |
| rev | line source |
|---|---|
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
1 """Tests for HTMLParser.py.""" |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
2 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
3 import HTMLParser |
|
19410
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
4 import pprint |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
5 import sys |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
6 import test_support |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
7 import unittest |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
8 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
9 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
10 class EventCollector(HTMLParser.HTMLParser): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
11 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
12 def __init__(self): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
13 self.events = [] |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
14 self.append = self.events.append |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
15 HTMLParser.HTMLParser.__init__(self) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
16 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
17 def get_events(self): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
18 # Normalize the list of events so that buffer artefacts don't |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
19 # separate runs of contiguous characters. |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
20 L = [] |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
21 prevtype = None |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
22 for event in self.events: |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
23 type = event[0] |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
24 if type == prevtype == "data": |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
25 L[-1] = ("data", L[-1][1] + event[1]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
26 else: |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
27 L.append(event) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
28 prevtype = type |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
29 self.events = L |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
30 return L |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
31 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
32 # structure markup |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
33 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
34 def handle_starttag(self, tag, attrs): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
35 self.append(("starttag", tag, attrs)) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
36 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
37 def handle_startendtag(self, tag, attrs): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
38 self.append(("startendtag", tag, attrs)) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
39 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
40 def handle_endtag(self, tag): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
41 self.append(("endtag", tag)) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
42 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
43 # all other markup |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
44 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
45 def handle_comment(self, data): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
46 self.append(("comment", data)) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
47 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
48 def handle_charref(self, data): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
49 self.append(("charref", data)) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
50 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
51 def handle_data(self, data): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
52 self.append(("data", data)) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
53 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
54 def handle_decl(self, data): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
55 self.append(("decl", data)) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
56 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
57 def handle_entityref(self, data): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
58 self.append(("entityref", data)) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
59 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
60 def handle_pi(self, data): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
61 self.append(("pi", data)) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
62 |
|
19683
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
63 def unknown_decl(self, decl): |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
64 self.append(("unknown decl", decl)) |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
65 |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
66 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
67 class EventCollectorExtra(EventCollector): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
68 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
69 def handle_starttag(self, tag, attrs): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
70 EventCollector.handle_starttag(self, tag, attrs) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
71 self.append(("starttag_text", self.get_starttag_text())) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
72 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
73 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
74 class TestCaseBase(unittest.TestCase): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
75 |
|
19683
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
76 def _run_check(self, source, expected_events, collector=EventCollector): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
77 parser = collector() |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
78 for s in source: |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
79 parser.feed(s) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
80 parser.close() |
|
19410
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
81 events = parser.get_events() |
|
19683
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
82 if events != expected_events: |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
83 self.fail("received events did not match expected events\n" |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
84 "Expected:\n" + pprint.pformat(expected_events) + |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
85 "\nReceived:\n" + pprint.pformat(events)) |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
86 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
87 def _run_check_extra(self, source, events): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
88 self._run_check(source, events, EventCollectorExtra) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
89 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
90 def _parse_error(self, source): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
91 def parse(source=source): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
92 parser = HTMLParser.HTMLParser() |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
93 parser.feed(source) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
94 parser.close() |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
95 self.assertRaises(HTMLParser.HTMLParseError, parse) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
96 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
97 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
98 class HTMLParserTestCase(TestCaseBase): |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
99 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
100 def test_processing_instruction_only(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
101 self._run_check("<?processing instruction>", [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
102 ("pi", "processing instruction"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
103 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
104 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
105 def test_simple_html(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
106 self._run_check(""" |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
107 <!DOCTYPE html PUBLIC 'foo'> |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
108 <HTML>&entity;  |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
109 <!--comment1a |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
110 -></foo><bar><<?pi?></foo<bar |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
111 comment1b--> |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
112 <Img sRc='Bar' isMAP>sample |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
113 text |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
114 “ |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
115 <!--comment2a-- --comment2b--> |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
116 </Html> |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
117 """, [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
118 ("data", "\n"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
119 ("decl", "DOCTYPE html PUBLIC 'foo'"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
120 ("data", "\n"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
121 ("starttag", "html", []), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
122 ("entityref", "entity"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
123 ("charref", "32"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
124 ("data", "\n"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
125 ("comment", "comment1a\n-></foo><bar><<?pi?></foo<bar\ncomment1b"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
126 ("data", "\n"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
127 ("starttag", "img", [("src", "Bar"), ("ismap", None)]), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
128 ("data", "sample\ntext\n"), |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
129 ("charref", "x201C"), |
|
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
130 ("data", "\n"), |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
131 ("comment", "comment2a-- --comment2b"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
132 ("data", "\n"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
133 ("endtag", "html"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
134 ("data", "\n"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
135 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
136 |
|
21157
e592472d55a3
Add a test that makes sure unclosed entity references are handled consitently.
Fred Drake <fdrake@acm.org>
parents:
20069
diff
changeset
|
137 def test_unclosed_entityref(self): |
|
e592472d55a3
Add a test that makes sure unclosed entity references are handled consitently.
Fred Drake <fdrake@acm.org>
parents:
20069
diff
changeset
|
138 self._run_check("&entityref foo", [ |
|
e592472d55a3
Add a test that makes sure unclosed entity references are handled consitently.
Fred Drake <fdrake@acm.org>
parents:
20069
diff
changeset
|
139 ("entityref", "entityref"), |
|
e592472d55a3
Add a test that makes sure unclosed entity references are handled consitently.
Fred Drake <fdrake@acm.org>
parents:
20069
diff
changeset
|
140 ("data", " foo"), |
|
e592472d55a3
Add a test that makes sure unclosed entity references are handled consitently.
Fred Drake <fdrake@acm.org>
parents:
20069
diff
changeset
|
141 ]) |
|
e592472d55a3
Add a test that makes sure unclosed entity references are handled consitently.
Fred Drake <fdrake@acm.org>
parents:
20069
diff
changeset
|
142 |
|
19410
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
143 def test_doctype_decl(self): |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
144 inside = """\ |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
145 DOCTYPE html [ |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
146 <!ELEMENT html - O EMPTY> |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
147 <!ATTLIST html |
|
19683
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
148 version CDATA #IMPLIED |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
149 profile CDATA 'DublinCore'> |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
150 <!NOTATION datatype SYSTEM 'http://xml.python.org/notations/python-module'> |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
151 <!ENTITY myEntity 'internal parsed entity'> |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
152 <!ENTITY anEntity SYSTEM 'http://xml.python.org/entities/something.xml'> |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
153 <!ENTITY % paramEntity 'name|name|name'> |
|
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
154 %paramEntity; |
|
19410
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
155 <!-- comment --> |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
156 ]""" |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
157 self._run_check("<!%s>" % inside, [ |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
158 ("decl", inside), |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
159 ]) |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
160 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
161 def test_bad_nesting(self): |
|
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
162 # Strangely, this *is* supposed to test that overlapping |
|
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
163 # elements are allowed. HTMLParser is more geared toward |
|
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
164 # lexing the input that parsing the structure. |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
165 self._run_check("<a><b></a></b>", [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
166 ("starttag", "a", []), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
167 ("starttag", "b", []), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
168 ("endtag", "a"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
169 ("endtag", "b"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
170 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
171 |
|
19410
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
172 def test_bare_ampersands(self): |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
173 self._run_check("this text & contains & ampersands &", [ |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
174 ("data", "this text & contains & ampersands &"), |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
175 ]) |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
176 |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
177 def test_bare_pointy_brackets(self): |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
178 self._run_check("this < text > contains < bare>pointy< brackets", [ |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
179 ("data", "this < text > contains < bare>pointy< brackets"), |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
180 ]) |
|
c403af003884
Deal more appropriately with bare ampersands and pointy brackets; this
Fred Drake <fdrake@acm.org>
parents:
18985
diff
changeset
|
181 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
182 def test_attr_syntax(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
183 output = [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
184 ("starttag", "a", [("b", "v"), ("c", "v"), ("d", "v"), ("e", None)]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
185 ] |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
186 self._run_check("""<a b='v' c="v" d=v e>""", output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
187 self._run_check("""<a b = 'v' c = "v" d = v e>""", output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
188 self._run_check("""<a\nb\n=\n'v'\nc\n=\n"v"\nd\n=\nv\ne>""", output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
189 self._run_check("""<a\tb\t=\t'v'\tc\t=\t"v"\td\t=\tv\te>""", output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
190 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
191 def test_attr_values(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
192 self._run_check("""<a b='xxx\n\txxx' c="yyy\t\nyyy" d='\txyz\n'>""", |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
193 [("starttag", "a", [("b", "xxx\n\txxx"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
194 ("c", "yyy\t\nyyy"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
195 ("d", "\txyz\n")]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
196 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
197 self._run_check("""<a b='' c="">""", [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
198 ("starttag", "a", [("b", ""), ("c", "")]), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
199 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
200 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
201 def test_attr_entity_replacement(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
202 self._run_check("""<a b='&><"''>""", [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
203 ("starttag", "a", [("b", "&><\"'")]), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
204 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
205 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
206 def test_attr_funky_names(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
207 self._run_check("""<a a.b='v' c:d=v e-f=v>""", [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
208 ("starttag", "a", [("a.b", "v"), ("c:d", "v"), ("e-f", "v")]), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
209 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
210 |
|
19683
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
211 def test_illegal_declarations(self): |
|
19687
3731ad1d9033
HTMLParser is allowed to be more strict than sgmllib, so let's not
Fred Drake <fdrake@acm.org>
parents:
19683
diff
changeset
|
212 self._parse_error('<!spacer type="block" height="25">') |
|
19683
8cbc74f5586b
Enhanced the test for DOCTYPE declarations, added a test for dealing with
Fred Drake <fdrake@acm.org>
parents:
19410
diff
changeset
|
213 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
214 def test_starttag_end_boundary(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
215 self._run_check("""<a b='<'>""", [("starttag", "a", [("b", "<")])]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
216 self._run_check("""<a b='>'>""", [("starttag", "a", [("b", ">")])]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
217 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
218 def test_buffer_artefacts(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
219 output = [("starttag", "a", [("b", "<")])] |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
220 self._run_check(["<a b='<'>"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
221 self._run_check(["<a ", "b='<'>"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
222 self._run_check(["<a b", "='<'>"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
223 self._run_check(["<a b=", "'<'>"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
224 self._run_check(["<a b='<", "'>"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
225 self._run_check(["<a b='<'", ">"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
226 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
227 output = [("starttag", "a", [("b", ">")])] |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
228 self._run_check(["<a b='>'>"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
229 self._run_check(["<a ", "b='>'>"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
230 self._run_check(["<a b", "='>'>"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
231 self._run_check(["<a b=", "'>'>"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
232 self._run_check(["<a b='>", "'>"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
233 self._run_check(["<a b='>'", ">"], output) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
234 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
235 def test_starttag_junk_chars(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
236 self._parse_error("</>") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
237 self._parse_error("</$>") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
238 self._parse_error("</") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
239 self._parse_error("</a") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
240 self._parse_error("<a<a>") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
241 self._parse_error("</a<a>") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
242 self._parse_error("<!") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
243 self._parse_error("<a $>") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
244 self._parse_error("<a") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
245 self._parse_error("<a foo='bar'") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
246 self._parse_error("<a foo='bar") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
247 self._parse_error("<a foo='>'") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
248 self._parse_error("<a foo='>") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
249 self._parse_error("<a foo=>") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
250 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
251 def test_declaration_junk_chars(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
252 self._parse_error("<!DOCTYPE foo $ >") |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
253 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
254 def test_startendtag(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
255 self._run_check("<p/>", [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
256 ("startendtag", "p", []), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
257 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
258 self._run_check("<p></p>", [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
259 ("starttag", "p", []), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
260 ("endtag", "p"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
261 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
262 self._run_check("<p><img src='foo' /></p>", [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
263 ("starttag", "p", []), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
264 ("startendtag", "img", [("src", "foo")]), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
265 ("endtag", "p"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
266 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
267 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
268 def test_get_starttag_text(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
269 s = """<foo:bar \n one="1"\ttwo=2 >""" |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
270 self._run_check_extra(s, [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
271 ("starttag", "foo:bar", [("one", "1"), ("two", "2")]), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
272 ("starttag_text", s)]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
273 |
|
18985
ba9e4f55133f
Fix stupid bug: when migrating these tests from the Zope repository, the
Fred Drake <fdrake@acm.org>
parents:
18634
diff
changeset
|
274 def test_cdata_content(self): |
|
17854
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
275 s = """<script> <!-- not a comment --> ¬-an-entity-ref; </script>""" |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
276 self._run_check(s, [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
277 ("starttag", "script", []), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
278 ("data", " <!-- not a comment --> ¬-an-entity-ref; "), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
279 ("endtag", "script"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
280 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
281 s = """<script> <not a='start tag'> </script>""" |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
282 self._run_check(s, [ |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
283 ("starttag", "script", []), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
284 ("data", " <not a='start tag'> "), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
285 ("endtag", "script"), |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
286 ]) |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
287 |
|
028a1d39efb6
Added test suite for the new HTMLParser module, originally from the
Fred Drake <fdrake@acm.org>
parents:
diff
changeset
|
288 |
|
20069
90a4f23abda7
Adapt to use the test_main() approach.
Fred Drake <fdrake@acm.org>
parents:
19687
diff
changeset
|
289 def test_main(): |
|
90a4f23abda7
Adapt to use the test_main() approach.
Fred Drake <fdrake@acm.org>
parents:
19687
diff
changeset
|
290 test_support.run_unittest(HTMLParserTestCase) |
|
90a4f23abda7
Adapt to use the test_main() approach.
Fred Drake <fdrake@acm.org>
parents:
19687
diff
changeset
|
291 |
|
90a4f23abda7
Adapt to use the test_main() approach.
Fred Drake <fdrake@acm.org>
parents:
19687
diff
changeset
|
292 |
|
90a4f23abda7
Adapt to use the test_main() approach.
Fred Drake <fdrake@acm.org>
parents:
19687
diff
changeset
|
293 if __name__ == "__main__": |
|
90a4f23abda7
Adapt to use the test_main() approach.
Fred Drake <fdrake@acm.org>
parents:
19687
diff
changeset
|
294 test_main() |
