symbian-qemu-0.9.1-12/python-2.6.1/Lib/test/test_htmlparser.py
author William Roberts <williamr@symbian.org>
Tue, 04 Aug 2009 11:50:27 +0100
changeset 3 c2946f91d81f
parent 1 2fb8b9db1c86
permissions -rw-r--r--
Automated conversion of relative paths to new source layout
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     1
"""Tests for HTMLParser.py."""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     2
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     3
import HTMLParser
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     4
import pprint
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     5
import unittest
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     6
from test import test_support
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     7
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     8
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     9
class EventCollector(HTMLParser.HTMLParser):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    10
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    11
    def __init__(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    12
        self.events = []
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    13
        self.append = self.events.append
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    14
        HTMLParser.HTMLParser.__init__(self)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    15
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    16
    def get_events(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    17
        # Normalize the list of events so that buffer artefacts don't
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    18
        # separate runs of contiguous characters.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    19
        L = []
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    20
        prevtype = None
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    21
        for event in self.events:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    22
            type = event[0]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    23
            if type == prevtype == "data":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    24
                L[-1] = ("data", L[-1][1] + event[1])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    25
            else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    26
                L.append(event)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    27
            prevtype = type
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    28
        self.events = L
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    29
        return L
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    30
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    31
    # structure markup
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    32
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    33
    def handle_starttag(self, tag, attrs):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    34
        self.append(("starttag", tag, attrs))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    35
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    36
    def handle_startendtag(self, tag, attrs):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    37
        self.append(("startendtag", tag, attrs))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    38
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    39
    def handle_endtag(self, tag):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    40
        self.append(("endtag", tag))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    41
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    42
    # all other markup
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    43
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    44
    def handle_comment(self, data):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    45
        self.append(("comment", data))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    46
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    47
    def handle_charref(self, data):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    48
        self.append(("charref", data))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    49
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    50
    def handle_data(self, data):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    51
        self.append(("data", data))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    52
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    53
    def handle_decl(self, data):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    54
        self.append(("decl", data))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    55
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    56
    def handle_entityref(self, data):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    57
        self.append(("entityref", data))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    58
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    59
    def handle_pi(self, data):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    60
        self.append(("pi", data))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    61
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    62
    def unknown_decl(self, decl):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    63
        self.append(("unknown decl", decl))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    64
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    65
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    66
class EventCollectorExtra(EventCollector):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    67
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    68
    def handle_starttag(self, tag, attrs):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    69
        EventCollector.handle_starttag(self, tag, attrs)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    70
        self.append(("starttag_text", self.get_starttag_text()))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    71
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    72
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    73
class TestCaseBase(unittest.TestCase):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    74
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    75
    def _run_check(self, source, expected_events, collector=EventCollector):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    76
        parser = collector()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    77
        for s in source:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    78
            parser.feed(s)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    79
        parser.close()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    80
        events = parser.get_events()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    81
        if events != expected_events:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    82
            self.fail("received events did not match expected events\n"
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    83
                      "Expected:\n" + pprint.pformat(expected_events) +
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    84
                      "\nReceived:\n" + pprint.pformat(events))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    85
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    86
    def _run_check_extra(self, source, events):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    87
        self._run_check(source, events, EventCollectorExtra)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    88
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    89
    def _parse_error(self, source):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    90
        def parse(source=source):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    91
            parser = HTMLParser.HTMLParser()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    92
            parser.feed(source)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    93
            parser.close()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    94
        self.assertRaises(HTMLParser.HTMLParseError, parse)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    95
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    96
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    97
class HTMLParserTestCase(TestCaseBase):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    98
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    99
    def test_processing_instruction_only(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   100
        self._run_check("<?processing instruction>", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   101
            ("pi", "processing instruction"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   102
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   103
        self._run_check("<?processing instruction ?>", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   104
            ("pi", "processing instruction ?"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   105
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   106
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   107
    def test_simple_html(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   108
        self._run_check("""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   109
<!DOCTYPE html PUBLIC 'foo'>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   110
<HTML>&entity;&#32;
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   111
<!--comment1a
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   112
-></foo><bar>&lt;<?pi?></foo<bar
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   113
comment1b-->
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   114
<Img sRc='Bar' isMAP>sample
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   115
text
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   116
&#x201C;
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   117
<!--comment2a-- --comment2b--><!>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   118
</Html>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   119
""", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   120
    ("data", "\n"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   121
    ("decl", "DOCTYPE html PUBLIC 'foo'"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   122
    ("data", "\n"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   123
    ("starttag", "html", []),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   124
    ("entityref", "entity"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   125
    ("charref", "32"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   126
    ("data", "\n"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   127
    ("comment", "comment1a\n-></foo><bar>&lt;<?pi?></foo<bar\ncomment1b"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   128
    ("data", "\n"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   129
    ("starttag", "img", [("src", "Bar"), ("ismap", None)]),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   130
    ("data", "sample\ntext\n"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   131
    ("charref", "x201C"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   132
    ("data", "\n"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   133
    ("comment", "comment2a-- --comment2b"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   134
    ("data", "\n"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   135
    ("endtag", "html"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   136
    ("data", "\n"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   137
    ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   138
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   139
    def test_unclosed_entityref(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   140
        self._run_check("&entityref foo", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   141
            ("entityref", "entityref"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   142
            ("data", " foo"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   143
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   144
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   145
    def test_doctype_decl(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   146
        inside = """\
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   147
DOCTYPE html [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   148
  <!ELEMENT html - O EMPTY>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   149
  <!ATTLIST html
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   150
      version CDATA #IMPLIED
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   151
      profile CDATA 'DublinCore'>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   152
  <!NOTATION datatype SYSTEM 'http://xml.python.org/notations/python-module'>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   153
  <!ENTITY myEntity 'internal parsed entity'>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   154
  <!ENTITY anEntity SYSTEM 'http://xml.python.org/entities/something.xml'>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   155
  <!ENTITY % paramEntity 'name|name|name'>
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   156
  %paramEntity;
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   157
  <!-- comment -->
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   158
]"""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   159
        self._run_check("<!%s>" % inside, [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   160
            ("decl", inside),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   161
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   162
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   163
    def test_bad_nesting(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   164
        # Strangely, this *is* supposed to test that overlapping
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   165
        # elements are allowed.  HTMLParser is more geared toward
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   166
        # lexing the input that parsing the structure.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   167
        self._run_check("<a><b></a></b>", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   168
            ("starttag", "a", []),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   169
            ("starttag", "b", []),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   170
            ("endtag", "a"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   171
            ("endtag", "b"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   172
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   173
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   174
    def test_bare_ampersands(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   175
        self._run_check("this text & contains & ampersands &", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   176
            ("data", "this text & contains & ampersands &"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   177
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   178
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   179
    def test_bare_pointy_brackets(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   180
        self._run_check("this < text > contains < bare>pointy< brackets", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   181
            ("data", "this < text > contains < bare>pointy< brackets"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   182
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   183
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   184
    def test_attr_syntax(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   185
        output = [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   186
          ("starttag", "a", [("b", "v"), ("c", "v"), ("d", "v"), ("e", None)])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   187
          ]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   188
        self._run_check("""<a b='v' c="v" d=v e>""", output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   189
        self._run_check("""<a  b = 'v' c = "v" d = v e>""", output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   190
        self._run_check("""<a\nb\n=\n'v'\nc\n=\n"v"\nd\n=\nv\ne>""", output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   191
        self._run_check("""<a\tb\t=\t'v'\tc\t=\t"v"\td\t=\tv\te>""", output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   192
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   193
    def test_attr_values(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   194
        self._run_check("""<a b='xxx\n\txxx' c="yyy\t\nyyy" d='\txyz\n'>""",
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   195
                        [("starttag", "a", [("b", "xxx\n\txxx"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   196
                                            ("c", "yyy\t\nyyy"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   197
                                            ("d", "\txyz\n")])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   198
                         ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   199
        self._run_check("""<a b='' c="">""", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   200
            ("starttag", "a", [("b", ""), ("c", "")]),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   201
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   202
        # Regression test for SF patch #669683.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   203
        self._run_check("<e a=rgb(1,2,3)>", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   204
            ("starttag", "e", [("a", "rgb(1,2,3)")]),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   205
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   206
        # Regression test for SF bug #921657.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   207
        self._run_check("<a href=mailto:xyz@example.com>", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   208
            ("starttag", "a", [("href", "mailto:xyz@example.com")]),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   209
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   210
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   211
    def test_attr_entity_replacement(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   212
        self._run_check("""<a b='&amp;&gt;&lt;&quot;&apos;'>""", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   213
            ("starttag", "a", [("b", "&><\"'")]),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   214
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   215
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   216
    def test_attr_funky_names(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   217
        self._run_check("""<a a.b='v' c:d=v e-f=v>""", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   218
            ("starttag", "a", [("a.b", "v"), ("c:d", "v"), ("e-f", "v")]),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   219
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   220
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   221
    def test_illegal_declarations(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   222
        self._parse_error('<!spacer type="block" height="25">')
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   223
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   224
    def test_starttag_end_boundary(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   225
        self._run_check("""<a b='<'>""", [("starttag", "a", [("b", "<")])])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   226
        self._run_check("""<a b='>'>""", [("starttag", "a", [("b", ">")])])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   227
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   228
    def test_buffer_artefacts(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   229
        output = [("starttag", "a", [("b", "<")])]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   230
        self._run_check(["<a b='<'>"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   231
        self._run_check(["<a ", "b='<'>"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   232
        self._run_check(["<a b", "='<'>"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   233
        self._run_check(["<a b=", "'<'>"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   234
        self._run_check(["<a b='<", "'>"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   235
        self._run_check(["<a b='<'", ">"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   236
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   237
        output = [("starttag", "a", [("b", ">")])]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   238
        self._run_check(["<a b='>'>"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   239
        self._run_check(["<a ", "b='>'>"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   240
        self._run_check(["<a b", "='>'>"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   241
        self._run_check(["<a b=", "'>'>"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   242
        self._run_check(["<a b='>", "'>"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   243
        self._run_check(["<a b='>'", ">"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   244
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   245
        output = [("comment", "abc")]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   246
        self._run_check(["", "<!--abc-->"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   247
        self._run_check(["<", "!--abc-->"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   248
        self._run_check(["<!", "--abc-->"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   249
        self._run_check(["<!-", "-abc-->"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   250
        self._run_check(["<!--", "abc-->"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   251
        self._run_check(["<!--a", "bc-->"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   252
        self._run_check(["<!--ab", "c-->"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   253
        self._run_check(["<!--abc", "-->"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   254
        self._run_check(["<!--abc-", "->"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   255
        self._run_check(["<!--abc--", ">"], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   256
        self._run_check(["<!--abc-->", ""], output)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   257
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   258
    def test_starttag_junk_chars(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   259
        self._parse_error("</>")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   260
        self._parse_error("</$>")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   261
        self._parse_error("</")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   262
        self._parse_error("</a")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   263
        self._parse_error("<a<a>")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   264
        self._parse_error("</a<a>")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   265
        self._parse_error("<!")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   266
        self._parse_error("<a $>")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   267
        self._parse_error("<a")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   268
        self._parse_error("<a foo='bar'")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   269
        self._parse_error("<a foo='bar")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   270
        self._parse_error("<a foo='>'")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   271
        self._parse_error("<a foo='>")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   272
        self._parse_error("<a foo=>")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   273
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   274
    def test_declaration_junk_chars(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   275
        self._parse_error("<!DOCTYPE foo $ >")
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   276
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   277
    def test_startendtag(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   278
        self._run_check("<p/>", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   279
            ("startendtag", "p", []),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   280
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   281
        self._run_check("<p></p>", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   282
            ("starttag", "p", []),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   283
            ("endtag", "p"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   284
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   285
        self._run_check("<p><img src='foo' /></p>", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   286
            ("starttag", "p", []),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   287
            ("startendtag", "img", [("src", "foo")]),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   288
            ("endtag", "p"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   289
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   290
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   291
    def test_get_starttag_text(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   292
        s = """<foo:bar   \n   one="1"\ttwo=2   >"""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   293
        self._run_check_extra(s, [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   294
            ("starttag", "foo:bar", [("one", "1"), ("two", "2")]),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   295
            ("starttag_text", s)])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   296
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   297
    def test_cdata_content(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   298
        s = """<script> <!-- not a comment --> &not-an-entity-ref; </script>"""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   299
        self._run_check(s, [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   300
            ("starttag", "script", []),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   301
            ("data", " <!-- not a comment --> &not-an-entity-ref; "),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   302
            ("endtag", "script"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   303
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   304
        s = """<script> <not a='start tag'> </script>"""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   305
        self._run_check(s, [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   306
            ("starttag", "script", []),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   307
            ("data", " <not a='start tag'> "),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   308
            ("endtag", "script"),
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   309
            ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   310
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   311
    def test_entityrefs_in_attributes(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   312
        self._run_check("<html foo='&euro;&amp;&#97;&#x61;&unsupported;'>", [
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   313
                ("starttag", "html", [("foo", u"\u20AC&aa&unsupported;")])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   314
                ])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   315
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   316
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   317
def test_main():
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   318
    test_support.run_unittest(HTMLParserTestCase)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   319
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   320
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   321
if __name__ == "__main__":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   322
    test_main()