cubicweb/devtools/htmlparser.py
author Jérémy Bobbio <jeremy.bobbio@irq7.fr>
Tue, 09 Jul 2019 11:13:08 +0200
changeset 12684 d464495452aa
parent 12505 55014a79b2a5
permissions -rw-r--r--
[pkg] Run all unit tests in autopkgtest We previously forgot to copy `tox.ini` alongside the tests. As this file configures pytest pattern matching rules, we were not running quite a few tests. This is now fixed. The added tests required some adjustments in the Debian dependencies. Sadly, not all tests currently pass. It seems there are some ordering dependencies between the tests in regards to how data are loaded. Fixing this probably would probably be better debugged by ensuring the test suite does not fail when using pytest random-order plugin.
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
8937
8a1809c9a043 [htmlparser] add missing deprecation message
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8695
diff changeset
     1
# copyright 2003-2013 LOGILAB S.A. (Paris, FRANCE), all rights reserved.
5421
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
     2
# contact http://www.logilab.fr/ -- mailto:contact@logilab.fr
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
     3
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
     4
# This file is part of CubicWeb.
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
     5
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
     6
# CubicWeb is free software: you can redistribute it and/or modify it under the
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
     7
# terms of the GNU Lesser General Public License as published by the Free
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
     8
# Software Foundation, either version 2.1 of the License, or (at your option)
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
     9
# any later version.
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
    10
#
5424
8ecbcbff9777 replace logilab-common by CubicWeb in disclaimer
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5421
diff changeset
    11
# CubicWeb is distributed in the hope that it will be useful, but WITHOUT
5421
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
    12
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
    13
# FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public License for more
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
    14
# details.
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
    15
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
    16
# You should have received a copy of the GNU Lesser General Public License along
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5276
diff changeset
    17
# with CubicWeb.  If not, see <http://www.gnu.org/licenses/>.
6771
da71f1ad1721 minor code cleanup
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
    18
"""defines a validating HTML parser used in web application tests"""
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    19
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    20
import re
3325
44caeccd2db9 fix sys import
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 3151
diff changeset
    21
import sys
8977
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    22
from xml import sax
10618
3274a1648c7e [py3k] io.BytesIO
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10588
diff changeset
    23
from io import BytesIO
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    24
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    25
from lxml import etree
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    26
1421
77ee26df178f doc type handling refactoring: do the ext substitution at the module level
sylvain.thenault@logilab.fr
parents: 1132
diff changeset
    27
from cubicweb.view import STRICT_DOCTYPE, TRANSITIONAL_DOCTYPE
6771
da71f1ad1721 minor code cleanup
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
    28
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
    29
STRICT_DOCTYPE = str(STRICT_DOCTYPE)
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
    30
TRANSITIONAL_DOCTYPE = str(TRANSITIONAL_DOCTYPE)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    31
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    32
ERR_COUNT = 0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    33
10696
4ba4be5553cf [py3k] unicode vs str vs bytes vs the world
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10662
diff changeset
    34
_REM_SCRIPT_RGX = re.compile(br"<script[^>]*>.*?</script>", re.M|re.I|re.S)
8977
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    35
def _remove_script_tags(data):
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    36
    """Remove the script (usually javascript) tags to help the lxml
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    37
    XMLParser / HTMLParser do their job. Without that, they choke on
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    38
    tags embedded in JS strings.
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    39
    """
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    40
    # Notice we may want to use lxml cleaner, but it's far too intrusive:
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    41
    #
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    42
    # cleaner = Cleaner(scripts=True,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    43
    #                   javascript=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    44
    #                   comments=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    45
    #                   style=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    46
    #                   links=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    47
    #                   meta=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    48
    #                   page_structure=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    49
    #                   processing_instructions=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    50
    #                   embedded=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    51
    #                   frames=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    52
    #                   forms=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    53
    #                   annoying_tags=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    54
    #                   remove_tags=(),
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    55
    #                   remove_unknown_tags=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    56
    #                   safe_attrs_only=False,
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    57
    #                   add_nofollow=False)
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    58
    # >>> cleaner.clean_html('<body></body>')
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    59
    # '<span></span>'
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    60
    # >>> cleaner.clean_html('<!DOCTYPE html><body></body>')
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    61
    # '<html><body></body></html>'
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    62
    # >>> cleaner.clean_html('<body><div/></body>')
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    63
    # '<div></div>'
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    64
    # >>> cleaner.clean_html('<html><body><div/><br></body><html>')
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    65
    # '<html><body><div></div><br></body></html>'
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    66
    # >>> cleaner.clean_html('<html><body><div/><br><span></body><html>')
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    67
    # '<html><body><div></div><br><span></span></body></html>'
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    68
    #
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    69
    # using that, we'll miss most actual validation error we want to
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    70
    # catch. For now, use dumb regexp
10696
4ba4be5553cf [py3k] unicode vs str vs bytes vs the world
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10662
diff changeset
    71
    return _REM_SCRIPT_RGX.sub(b'', data)
8977
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    72
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    73
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    74
class Validator(object):
8977
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
    75
    """ base validator API """
8973
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
    76
    parser = None
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
    77
8973
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
    78
    def parse_string(self, source):
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
    79
        etree = self._parse(self.preprocess_data(source))
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
    80
        return PageInfo(source, etree)
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
    81
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
    82
    def preprocess_data(self, data):
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
    83
        return data
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
    84
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
    85
    def _parse(self, pdata):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    86
        try:
8973
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
    87
            return etree.fromstring(pdata, self.parser)
8695
358d8bed9626 [toward-py3k] rewrite to "except AnException as exc:" (part of #2711624)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7014
diff changeset
    88
        except etree.XMLSyntaxError as exc:
8941
7b26fe71404f drop xhtml content-type support (closes #2065651)
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 8940
diff changeset
    89
            new_exc = AssertionError(u'invalid document: %s' % exc)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    90
            new_exc.position = exc.position
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    91
            raise new_exc
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    92
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    93
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    94
class DTDValidator(Validator):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    95
    def __init__(self):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    96
        Validator.__init__(self)
3151
5d45c0945bd3 note about this test under windows
Aurelien Campeas
parents: 1977
diff changeset
    97
        # XXX understand what's happening under windows
6771
da71f1ad1721 minor code cleanup
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
    98
        self.parser = etree.XMLParser(dtd_validation=sys.platform != 'win32')
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    99
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   100
    def preprocess_data(self, data):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   101
        """used to fix potential blockquote mess generated by docutils"""
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   102
        if STRICT_DOCTYPE not in data:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   103
            return data
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   104
        # parse using transitional DTD
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   105
        data = data.replace(STRICT_DOCTYPE, TRANSITIONAL_DOCTYPE)
8979
8f5416b1562a [devtools] use self._parse so AssertionError is properly raised instead of lxml error (test failure introduced in 6711f78c18be)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8977
diff changeset
   106
        tree = self._parse(data)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   107
        namespace = tree.nsmap.get(None)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   108
        # this is the list of authorized child tags for <blockquote> nodes
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   109
        expected = 'p h1 h2 h3 h4 h5 h6 div ul ol dl pre hr blockquote address ' \
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   110
                   'fieldset table form noscript ins del script'.split()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   111
        if namespace:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   112
            blockquotes = tree.findall('.//{%s}blockquote' % namespace)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   113
            expected = ['{%s}%s' % (namespace, tag) for tag in expected]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   114
        else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   115
            blockquotes = tree.findall('.//blockquote')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   116
        # quick and dirty approach: remove all blockquotes
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   117
        for blockquote in blockquotes:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   118
            parent = blockquote.getparent()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   119
            parent.remove(blockquote)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   120
        data = etree.tostring(tree)
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   121
        return '<?xml version="1.0" encoding="UTF-8"?>%s\n%s' % (
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   122
            STRICT_DOCTYPE, data)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   123
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   124
8938
198fdadafed6 [htmlparser] rename SaxOnlyValidator to XMLValidator
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8937
diff changeset
   125
class XMLValidator(Validator):
8977
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   126
    """XML validator, checks that XML is well-formed and used XMLNS are defined"""
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   127
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   128
    def __init__(self):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   129
        Validator.__init__(self)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   130
        self.parser = etree.XMLParser()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   131
8977
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   132
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   133
class XMLSyntaxValidator(Validator):
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   134
    """XML syntax validator, check XML is well-formed"""
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   135
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   136
    class MySaxErrorHandler(sax.ErrorHandler):
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   137
        """override default handler to avoid choking because of unknown entity"""
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   138
        def fatalError(self, exception):
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   139
            # XXX check entity in htmlentitydefs
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   140
            if not str(exception).endswith('undefined entity'):
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   141
                raise exception
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   142
    _parser = sax.make_parser()
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   143
    _parser.setContentHandler(sax.handler.ContentHandler())
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   144
    _parser.setErrorHandler(MySaxErrorHandler())
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   145
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   146
    def __init__(self):
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   147
        super(XMLSyntaxValidator, self).__init__()
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   148
        # XMLParser() wants xml namespaces defined
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   149
        # XMLParser(recover=True) will accept almost anything
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   150
        #
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   151
        # -> use the later but preprocess will check xml well-formness using a
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   152
        #    dumb SAX parser
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   153
        self.parser = etree.XMLParser(recover=True)
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   154
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   155
    def preprocess_data(self, data):
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   156
        return _remove_script_tags(data)
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   157
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   158
    def _parse(self, data):
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   159
        inpsrc = sax.InputSource()
10618
3274a1648c7e [py3k] io.BytesIO
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10588
diff changeset
   160
        inpsrc.setByteStream(BytesIO(data))
8977
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   161
        try:
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   162
            self._parser.parse(inpsrc)
10588
fdaa0e4b7eaf [py3k] except as
Samuel Trégouët <samuel.tregouet@logilab.fr>
parents: 10066
diff changeset
   163
        except sax.SAXParseException as exc:
8977
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   164
            new_exc = AssertionError(u'invalid document: %s' % exc)
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   165
            new_exc.position = (exc._linenum, exc._colnum)
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   166
            raise new_exc
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   167
        return super(XMLSyntaxValidator, self)._parse(data)
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   168
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   169
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   170
class HTMLValidator(Validator):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   171
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   172
    def __init__(self):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   173
        Validator.__init__(self)
8940
ae898a084da2 [htmlparser] exclude <script> tag from html source
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8939
diff changeset
   174
        self.parser = etree.HTMLParser(recover=False)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   175
8940
ae898a084da2 [htmlparser] exclude <script> tag from html source
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8939
diff changeset
   176
    def preprocess_data(self, data):
8977
57e564c0118e [testlib] introduce a validator that check xml-well formness
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8973
diff changeset
   177
        return _remove_script_tags(data)
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   178
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   179
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   180
class PageInfo(object):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   181
    """holds various informations on the view's output"""
8973
6711f78c18be [testlib] unspaghettify Validator / PageInfo api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8972
diff changeset
   182
    def __init__(self, source, root):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   183
        self.source = source
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   184
        self.etree = root
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   185
        self.raw_text = u''.join(root.xpath('//text()'))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   186
        self.namespace = self.etree.nsmap
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   187
        self.default_ns = self.namespace.get(None)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   188
        self.a_tags = self.find_tag('a')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   189
        self.h1_tags = self.find_tag('h1')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   190
        self.h2_tags = self.find_tag('h2')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   191
        self.h3_tags = self.find_tag('h3')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   192
        self.h4_tags = self.find_tag('h4')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   193
        self.input_tags = self.find_tag('input')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   194
        self.title_tags = [self.h1_tags, self.h2_tags, self.h3_tags, self.h4_tags]
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   195
7014
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   196
    def _iterstr(self, tag):
6977
cb78108bf603 [testlib] new method on page info object to ensure some tag with arbitrary attributes is found
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6772
diff changeset
   197
        if self.default_ns is None:
cb78108bf603 [testlib] new method on page info object to ensure some tag with arbitrary attributes is found
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6772
diff changeset
   198
            return ".//%s" % tag
cb78108bf603 [testlib] new method on page info object to ensure some tag with arbitrary attributes is found
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6772
diff changeset
   199
        else:
cb78108bf603 [testlib] new method on page info object to ensure some tag with arbitrary attributes is found
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6772
diff changeset
   200
            return ".//{%s}%s" % (self.default_ns, tag)
cb78108bf603 [testlib] new method on page info object to ensure some tag with arbitrary attributes is found
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6772
diff changeset
   201
7014
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   202
    def matching_nodes(self, tag, **attrs):
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   203
        for elt in self.etree.iterfind(self._iterstr(tag)):
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   204
            eltattrs  = elt.attrib
10662
10942ed172de [py3k] dict.iteritems → dict.items
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10618
diff changeset
   205
            for attr, value in attrs.items():
7014
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   206
                try:
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   207
                    if eltattrs[attr] != value:
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   208
                        break
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   209
                except KeyError:
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   210
                    break
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   211
            else: # all attributes match
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   212
                yield elt
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   213
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   214
    def has_tag(self, tag, nboccurs=1, **attrs):
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   215
        """returns True if tag with given attributes appears in the page
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   216
        `nbtimes` (any if None)
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   217
        """
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   218
        for elt in self.matching_nodes(tag, **attrs):
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   219
            if nboccurs is None: # no need to check number of occurences
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   220
                return True
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   221
            if not nboccurs: # too much occurences
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   222
                return False
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   223
            nboccurs -= 1
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   224
        if nboccurs == 0: # correct number of occurences
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   225
            return True
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   226
        return False # no matching tag/attrs
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   227
1945
2b59d9ae17ae new argument telling if we want text or (text / attrs), keeping bw compat
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 1485
diff changeset
   228
    def find_tag(self, tag, gettext=True):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   229
        """return a list which contains text of all "tag" elements """
7014
7e3e80f4179a [testlib pageinfo] extract matching_node method from has_tag to ease looking for a node with a given set of attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6977
diff changeset
   230
        iterstr = self._iterstr(tag)
1945
2b59d9ae17ae new argument telling if we want text or (text / attrs), keeping bw compat
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 1485
diff changeset
   231
        if not gettext or tag in ('a', 'input'):
6977
cb78108bf603 [testlib] new method on page info object to ensure some tag with arbitrary attributes is found
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6772
diff changeset
   232
            return [(elt.text, elt.attrib)
cb78108bf603 [testlib] new method on page info object to ensure some tag with arbitrary attributes is found
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6772
diff changeset
   233
                    for elt in self.etree.iterfind(iterstr)]
cb78108bf603 [testlib] new method on page info object to ensure some tag with arbitrary attributes is found
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6772
diff changeset
   234
        return [u''.join(elt.xpath('.//text()'))
cb78108bf603 [testlib] new method on page info object to ensure some tag with arbitrary attributes is found
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6772
diff changeset
   235
                for elt in self.etree.iterfind(iterstr)]
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   236
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   237
    def appears(self, text):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   238
        """returns True if <text> appears in the page"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   239
        return text in self.raw_text
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   240
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   241
    def __contains__(self, text):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   242
        return text in self.source
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   243
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   244
    def has_title(self, text, level=None):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   245
        """returns True if <h?>text</h?>
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   246
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   247
        :param level: the title's level (1 for h1, 2 for h2, etc.)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   248
        """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   249
        if level is None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   250
            for hlist in self.title_tags:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   251
                if text in hlist:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   252
                    return True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   253
            return False
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   254
        else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   255
            hlist = self.title_tags[level - 1]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   256
            return text in hlist
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   257
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   258
    def has_title_regexp(self, pattern, level=None):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   259
        """returns True if <h?>pattern</h?>"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   260
        sre = re.compile(pattern)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   261
        if level is None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   262
            for hlist in self.title_tags:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   263
                for title in hlist:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   264
                    if sre.match(title):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   265
                        return True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   266
            return False
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   267
        else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   268
            hlist = self.title_tags[level - 1]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   269
            for title in hlist:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   270
                if sre.match(title):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   271
                    return True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   272
            return False
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   273
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   274
    def has_link(self, text, url=None):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   275
        """returns True if <a href=url>text</a> was found in the page"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   276
        for link_text, attrs in self.a_tags:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   277
            if text == link_text:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   278
                if url is None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   279
                    return True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   280
                try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   281
                    href = attrs['href']
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   282
                    if href == url:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   283
                        return True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   284
                except KeyError:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   285
                    continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   286
        return False
1485
4d532f3c012e nicer fix
sylvain.thenault@logilab.fr
parents: 1480
diff changeset
   287
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   288
    def has_link_regexp(self, pattern, url=None):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   289
        """returns True if <a href=url>pattern</a> was found in the page"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   290
        sre = re.compile(pattern)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   291
        for link_text, attrs in self.a_tags:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   292
            if sre.match(link_text):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   293
                if url is None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   294
                    return True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   295
                try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   296
                    href = attrs['href']
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   297
                    if href == url:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   298
                        return True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   299
                except KeyError:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   300
                    continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   301
        return False
2773
b2530e3e0afb [testlib] #345052 and #344207: major test lib refactoring/cleanup + update usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 1977
diff changeset
   302
8972
771337c3a754 [testlib] update htmlparsers.VALMAP: stop using SaxOnlyValidator and add an entry for html
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8951
diff changeset
   303
VALMAP = {None: None,
771337c3a754 [testlib] update htmlparsers.VALMAP: stop using SaxOnlyValidator and add an entry for html
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8951
diff changeset
   304
          'dtd': DTDValidator,
771337c3a754 [testlib] update htmlparsers.VALMAP: stop using SaxOnlyValidator and add an entry for html
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8951
diff changeset
   305
          'xml': XMLValidator,
771337c3a754 [testlib] update htmlparsers.VALMAP: stop using SaxOnlyValidator and add an entry for html
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8951
diff changeset
   306
          'html': HTMLValidator,
771337c3a754 [testlib] update htmlparsers.VALMAP: stop using SaxOnlyValidator and add an entry for html
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8951
diff changeset
   307
          }