python-2.5.2/win32/Lib/test/test_htmlparser.py
author Deepak Modgil <Deepak.Modgil@Nokia.com>
Fri, 03 Apr 2009 17:19:34 +0100
changeset 0 ae805ac0140d
permissions -rw-r--r--
DP tools release version Revision: 200912
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
     1
"""Tests for HTMLParser.py."""
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
     2
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
     3
import HTMLParser
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
     4
import pprint
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
     5
import sys
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
     6
import unittest
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
     7
from test import test_support
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
     8
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
     9
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    10
class EventCollector(HTMLParser.HTMLParser):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    11
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    12
    def __init__(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    13
        self.events = []
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    14
        self.append = self.events.append
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    15
        HTMLParser.HTMLParser.__init__(self)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    16
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    17
    def get_events(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    18
        # Normalize the list of events so that buffer artefacts don't
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    19
        # separate runs of contiguous characters.
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    20
        L = []
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    21
        prevtype = None
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    22
        for event in self.events:
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    23
            type = event[0]
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    24
            if type == prevtype == "data":
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    25
                L[-1] = ("data", L[-1][1] + event[1])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    26
            else:
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    27
                L.append(event)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    28
            prevtype = type
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    29
        self.events = L
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    30
        return L
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    31
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    32
    # structure markup
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    33
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    34
    def handle_starttag(self, tag, attrs):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    35
        self.append(("starttag", tag, attrs))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    36
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    37
    def handle_startendtag(self, tag, attrs):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    38
        self.append(("startendtag", tag, attrs))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    39
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    40
    def handle_endtag(self, tag):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    41
        self.append(("endtag", tag))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    42
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    43
    # all other markup
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    44
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    45
    def handle_comment(self, data):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    46
        self.append(("comment", data))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    47
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    48
    def handle_charref(self, data):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    49
        self.append(("charref", data))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    50
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    51
    def handle_data(self, data):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    52
        self.append(("data", data))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    53
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    54
    def handle_decl(self, data):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    55
        self.append(("decl", data))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    56
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    57
    def handle_entityref(self, data):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    58
        self.append(("entityref", data))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    59
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    60
    def handle_pi(self, data):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    61
        self.append(("pi", data))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    62
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    63
    def unknown_decl(self, decl):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    64
        self.append(("unknown decl", decl))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    65
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    66
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    67
class EventCollectorExtra(EventCollector):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    68
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    69
    def handle_starttag(self, tag, attrs):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    70
        EventCollector.handle_starttag(self, tag, attrs)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    71
        self.append(("starttag_text", self.get_starttag_text()))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    72
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    73
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    74
class TestCaseBase(unittest.TestCase):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    75
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    76
    def _run_check(self, source, expected_events, collector=EventCollector):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    77
        parser = collector()
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    78
        for s in source:
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    79
            parser.feed(s)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    80
        parser.close()
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    81
        events = parser.get_events()
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    82
        if events != expected_events:
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    83
            self.fail("received events did not match expected events\n"
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    84
                      "Expected:\n" + pprint.pformat(expected_events) +
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    85
                      "\nReceived:\n" + pprint.pformat(events))
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    86
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    87
    def _run_check_extra(self, source, events):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    88
        self._run_check(source, events, EventCollectorExtra)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    89
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    90
    def _parse_error(self, source):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    91
        def parse(source=source):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    92
            parser = HTMLParser.HTMLParser()
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    93
            parser.feed(source)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    94
            parser.close()
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    95
        self.assertRaises(HTMLParser.HTMLParseError, parse)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    96
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    97
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    98
class HTMLParserTestCase(TestCaseBase):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
    99
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   100
    def test_processing_instruction_only(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   101
        self._run_check("<?processing instruction>", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   102
            ("pi", "processing instruction"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   103
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   104
        self._run_check("<?processing instruction ?>", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   105
            ("pi", "processing instruction ?"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   106
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   107
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   108
    def test_simple_html(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   109
        self._run_check("""
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   110
<!DOCTYPE html PUBLIC 'foo'>
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   111
<HTML>&entity;&#32;
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   112
<!--comment1a
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   113
-></foo><bar>&lt;<?pi?></foo<bar
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   114
comment1b-->
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   115
<Img sRc='Bar' isMAP>sample
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   116
text
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   117
&#x201C;
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   118
<!--comment2a-- --comment2b--><!>
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   119
</Html>
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   120
""", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   121
    ("data", "\n"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   122
    ("decl", "DOCTYPE html PUBLIC 'foo'"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   123
    ("data", "\n"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   124
    ("starttag", "html", []),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   125
    ("entityref", "entity"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   126
    ("charref", "32"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   127
    ("data", "\n"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   128
    ("comment", "comment1a\n-></foo><bar>&lt;<?pi?></foo<bar\ncomment1b"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   129
    ("data", "\n"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   130
    ("starttag", "img", [("src", "Bar"), ("ismap", None)]),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   131
    ("data", "sample\ntext\n"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   132
    ("charref", "x201C"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   133
    ("data", "\n"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   134
    ("comment", "comment2a-- --comment2b"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   135
    ("data", "\n"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   136
    ("endtag", "html"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   137
    ("data", "\n"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   138
    ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   139
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   140
    def test_unclosed_entityref(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   141
        self._run_check("&entityref foo", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   142
            ("entityref", "entityref"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   143
            ("data", " foo"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   144
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   145
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   146
    def test_doctype_decl(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   147
        inside = """\
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   148
DOCTYPE html [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   149
  <!ELEMENT html - O EMPTY>
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   150
  <!ATTLIST html
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   151
      version CDATA #IMPLIED
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   152
      profile CDATA 'DublinCore'>
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   153
  <!NOTATION datatype SYSTEM 'http://xml.python.org/notations/python-module'>
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   154
  <!ENTITY myEntity 'internal parsed entity'>
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   155
  <!ENTITY anEntity SYSTEM 'http://xml.python.org/entities/something.xml'>
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   156
  <!ENTITY % paramEntity 'name|name|name'>
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   157
  %paramEntity;
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   158
  <!-- comment -->
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   159
]"""
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   160
        self._run_check("<!%s>" % inside, [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   161
            ("decl", inside),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   162
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   163
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   164
    def test_bad_nesting(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   165
        # Strangely, this *is* supposed to test that overlapping
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   166
        # elements are allowed.  HTMLParser is more geared toward
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   167
        # lexing the input that parsing the structure.
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   168
        self._run_check("<a><b></a></b>", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   169
            ("starttag", "a", []),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   170
            ("starttag", "b", []),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   171
            ("endtag", "a"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   172
            ("endtag", "b"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   173
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   174
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   175
    def test_bare_ampersands(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   176
        self._run_check("this text & contains & ampersands &", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   177
            ("data", "this text & contains & ampersands &"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   178
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   179
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   180
    def test_bare_pointy_brackets(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   181
        self._run_check("this < text > contains < bare>pointy< brackets", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   182
            ("data", "this < text > contains < bare>pointy< brackets"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   183
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   184
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   185
    def test_attr_syntax(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   186
        output = [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   187
          ("starttag", "a", [("b", "v"), ("c", "v"), ("d", "v"), ("e", None)])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   188
          ]
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   189
        self._run_check("""<a b='v' c="v" d=v e>""", output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   190
        self._run_check("""<a  b = 'v' c = "v" d = v e>""", output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   191
        self._run_check("""<a\nb\n=\n'v'\nc\n=\n"v"\nd\n=\nv\ne>""", output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   192
        self._run_check("""<a\tb\t=\t'v'\tc\t=\t"v"\td\t=\tv\te>""", output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   193
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   194
    def test_attr_values(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   195
        self._run_check("""<a b='xxx\n\txxx' c="yyy\t\nyyy" d='\txyz\n'>""",
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   196
                        [("starttag", "a", [("b", "xxx\n\txxx"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   197
                                            ("c", "yyy\t\nyyy"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   198
                                            ("d", "\txyz\n")])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   199
                         ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   200
        self._run_check("""<a b='' c="">""", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   201
            ("starttag", "a", [("b", ""), ("c", "")]),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   202
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   203
        # Regression test for SF patch #669683.
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   204
        self._run_check("<e a=rgb(1,2,3)>", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   205
            ("starttag", "e", [("a", "rgb(1,2,3)")]),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   206
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   207
        # Regression test for SF bug #921657.
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   208
        self._run_check("<a href=mailto:xyz@example.com>", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   209
            ("starttag", "a", [("href", "mailto:xyz@example.com")]),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   210
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   211
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   212
    def test_attr_entity_replacement(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   213
        self._run_check("""<a b='&amp;&gt;&lt;&quot;&apos;'>""", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   214
            ("starttag", "a", [("b", "&><\"'")]),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   215
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   216
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   217
    def test_attr_funky_names(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   218
        self._run_check("""<a a.b='v' c:d=v e-f=v>""", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   219
            ("starttag", "a", [("a.b", "v"), ("c:d", "v"), ("e-f", "v")]),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   220
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   221
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   222
    def test_illegal_declarations(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   223
        self._parse_error('<!spacer type="block" height="25">')
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   224
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   225
    def test_starttag_end_boundary(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   226
        self._run_check("""<a b='<'>""", [("starttag", "a", [("b", "<")])])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   227
        self._run_check("""<a b='>'>""", [("starttag", "a", [("b", ">")])])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   228
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   229
    def test_buffer_artefacts(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   230
        output = [("starttag", "a", [("b", "<")])]
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   231
        self._run_check(["<a b='<'>"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   232
        self._run_check(["<a ", "b='<'>"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   233
        self._run_check(["<a b", "='<'>"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   234
        self._run_check(["<a b=", "'<'>"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   235
        self._run_check(["<a b='<", "'>"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   236
        self._run_check(["<a b='<'", ">"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   237
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   238
        output = [("starttag", "a", [("b", ">")])]
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   239
        self._run_check(["<a b='>'>"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   240
        self._run_check(["<a ", "b='>'>"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   241
        self._run_check(["<a b", "='>'>"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   242
        self._run_check(["<a b=", "'>'>"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   243
        self._run_check(["<a b='>", "'>"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   244
        self._run_check(["<a b='>'", ">"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   245
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   246
        output = [("comment", "abc")]
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   247
        self._run_check(["", "<!--abc-->"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   248
        self._run_check(["<", "!--abc-->"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   249
        self._run_check(["<!", "--abc-->"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   250
        self._run_check(["<!-", "-abc-->"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   251
        self._run_check(["<!--", "abc-->"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   252
        self._run_check(["<!--a", "bc-->"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   253
        self._run_check(["<!--ab", "c-->"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   254
        self._run_check(["<!--abc", "-->"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   255
        self._run_check(["<!--abc-", "->"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   256
        self._run_check(["<!--abc--", ">"], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   257
        self._run_check(["<!--abc-->", ""], output)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   258
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   259
    def test_starttag_junk_chars(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   260
        self._parse_error("</>")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   261
        self._parse_error("</$>")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   262
        self._parse_error("</")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   263
        self._parse_error("</a")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   264
        self._parse_error("<a<a>")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   265
        self._parse_error("</a<a>")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   266
        self._parse_error("<!")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   267
        self._parse_error("<a $>")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   268
        self._parse_error("<a")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   269
        self._parse_error("<a foo='bar'")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   270
        self._parse_error("<a foo='bar")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   271
        self._parse_error("<a foo='>'")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   272
        self._parse_error("<a foo='>")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   273
        self._parse_error("<a foo=>")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   274
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   275
    def test_declaration_junk_chars(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   276
        self._parse_error("<!DOCTYPE foo $ >")
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   277
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   278
    def test_startendtag(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   279
        self._run_check("<p/>", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   280
            ("startendtag", "p", []),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   281
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   282
        self._run_check("<p></p>", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   283
            ("starttag", "p", []),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   284
            ("endtag", "p"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   285
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   286
        self._run_check("<p><img src='foo' /></p>", [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   287
            ("starttag", "p", []),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   288
            ("startendtag", "img", [("src", "foo")]),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   289
            ("endtag", "p"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   290
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   291
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   292
    def test_get_starttag_text(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   293
        s = """<foo:bar   \n   one="1"\ttwo=2   >"""
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   294
        self._run_check_extra(s, [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   295
            ("starttag", "foo:bar", [("one", "1"), ("two", "2")]),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   296
            ("starttag_text", s)])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   297
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   298
    def test_cdata_content(self):
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   299
        s = """<script> <!-- not a comment --> &not-an-entity-ref; </script>"""
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   300
        self._run_check(s, [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   301
            ("starttag", "script", []),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   302
            ("data", " <!-- not a comment --> &not-an-entity-ref; "),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   303
            ("endtag", "script"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   304
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   305
        s = """<script> <not a='start tag'> </script>"""
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   306
        self._run_check(s, [
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   307
            ("starttag", "script", []),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   308
            ("data", " <not a='start tag'> "),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   309
            ("endtag", "script"),
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   310
            ])
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   311
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   312
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   313
def test_main():
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   314
    test_support.run_unittest(HTMLParserTestCase)
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   315
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   316
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   317
if __name__ == "__main__":
ae805ac0140d DP tools release version Revision: 200912
Deepak Modgil <Deepak.Modgil@Nokia.com>
parents:
diff changeset
   318
    test_main()