uilib.py
author Nicolas Chauvat <nicolas.chauvat@logilab.fr>
Mon, 06 Jun 2011 19:54:37 +0200
changeset 7464 9f116f9b3be8
parent 7411 238da9684f99
child 7575 335f14e8e5a7
permissions -rw-r--r--
[datetime] by default display 'UTC' string when displaying TZDatetime
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
     1
# -*- coding: utf-8 -*-
7411
238da9684f99 cleanup and update some copyright
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7166
diff changeset
     2
# copyright 2003-2011 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: 4466
diff changeset
     3
# 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: 4466
diff changeset
     4
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4466
diff changeset
     5
# 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: 4466
diff changeset
     6
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4466
diff changeset
     7
# 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: 4466
diff changeset
     8
# 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: 4466
diff changeset
     9
# 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: 4466
diff changeset
    10
# any later version.
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4466
diff changeset
    11
#
5424
8ecbcbff9777 replace logilab-common by CubicWeb in disclaimer
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5421
diff changeset
    12
# 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: 4466
diff changeset
    13
# 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: 4466
diff changeset
    14
# 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: 4466
diff changeset
    15
# details.
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4466
diff changeset
    16
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4466
diff changeset
    17
# 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: 4466
diff changeset
    18
# with CubicWeb.  If not, see <http://www.gnu.org/licenses/>.
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    19
"""user interface libraries
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    20
5730
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
    21
contains some functions designed to help implementation of cubicweb user
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
    22
interface.
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
    23
"""
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
__docformat__ = "restructuredtext en"
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    26
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    27
import csv
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    28
import re
1635
866563e2d0fc don't depends on simplejson outside web/
sylvain.thenault@logilab.fr
parents: 1623
diff changeset
    29
from StringIO import StringIO
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    30
2312
af4d8f75c5db use xml_escape
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2208
diff changeset
    31
from logilab.mtconverter import xml_escape, html_unescape
4466
8b0ca7904820 moved generic datetime manipulation function to lgc
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
    32
from logilab.common.date import ustrftime
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    33
5949
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
    34
from cubicweb.utils import json_dumps
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
    35
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    36
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    37
def rql_for_eid(eid):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    38
    """return the rql query necessary to fetch entity with the given eid.  This
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    39
    function should only be used to generate link with rql inside, not to give
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    40
    to cursor.execute (in which case you won't benefit from rql cache).
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    41
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    42
    :Parameters:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    43
      - `eid`: the eid of the entity we should search
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    44
    :rtype: str
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    45
    :return: the rql query
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    46
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    47
    return 'Any X WHERE X eid %s' % eid
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    48
6106
1e6d93f70d14 [selectors] fix match_transition & introduce new edited_attribute selector
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 5951
diff changeset
    49
def eid_param(name, eid):
7058
ea22892e82d4 [uilib] fix eid_param implementation
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6688
diff changeset
    50
    assert name is not None
6106
1e6d93f70d14 [selectors] fix match_transition & introduce new edited_attribute selector
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 5951
diff changeset
    51
    assert eid is not None
1e6d93f70d14 [selectors] fix match_transition & introduce new edited_attribute selector
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 5951
diff changeset
    52
    return '%s:%s' % (name, eid)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    53
3212
07d11bacfefe displaytime attribute should not have been removed from there
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3165
diff changeset
    54
def printable_value(req, attrtype, value, props=None, displaytime=True):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    55
    """return a displayable value (i.e. unicode string)"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    56
    if value is None or attrtype == 'Bytes':
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    57
        return u''
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    58
    if attrtype == 'String':
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    59
        # don't translate empty value if you don't want strange results
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    60
        if props is not None and value and props.get('internationalizable'):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    61
            return req._(value)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    62
        return value
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    63
    if attrtype == 'Date':
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    64
        return ustrftime(value, req.property_value('ui.date-format'))
7464
9f116f9b3be8 [datetime] by default display 'UTC' string when displaying TZDatetime
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7411
diff changeset
    65
    if attrtype == 'Time':
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    66
        return ustrftime(value, req.property_value('ui.time-format'))
7464
9f116f9b3be8 [datetime] by default display 'UTC' string when displaying TZDatetime
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7411
diff changeset
    67
    if attrtype == 'TZTime':
9f116f9b3be8 [datetime] by default display 'UTC' string when displaying TZDatetime
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7411
diff changeset
    68
        return ustrftime(value, req.property_value('ui.time-format')) + u' UTC'
9f116f9b3be8 [datetime] by default display 'UTC' string when displaying TZDatetime
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7411
diff changeset
    69
    if attrtype == 'Datetime':
3212
07d11bacfefe displaytime attribute should not have been removed from there
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3165
diff changeset
    70
        if displaytime:
07d11bacfefe displaytime attribute should not have been removed from there
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3165
diff changeset
    71
            return ustrftime(value, req.property_value('ui.datetime-format'))
07d11bacfefe displaytime attribute should not have been removed from there
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3165
diff changeset
    72
        return ustrftime(value, req.property_value('ui.date-format'))
7464
9f116f9b3be8 [datetime] by default display 'UTC' string when displaying TZDatetime
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7411
diff changeset
    73
    if attrtype == 'TZDatetime':
9f116f9b3be8 [datetime] by default display 'UTC' string when displaying TZDatetime
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7411
diff changeset
    74
        if displaytime:
9f116f9b3be8 [datetime] by default display 'UTC' string when displaying TZDatetime
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7411
diff changeset
    75
            return ustrftime(value, req.property_value('ui.datetime-format')) + u' UTC'
9f116f9b3be8 [datetime] by default display 'UTC' string when displaying TZDatetime
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7411
diff changeset
    76
        return ustrftime(value, req.property_value('ui.date-format'))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    77
    if attrtype == 'Boolean':
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    78
        if value:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    79
            return req._('yes')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    80
        return req._('no')
7166
dde161937d3e [time zone] support for TZDatetime and TZTime data type
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7058
diff changeset
    81
    if attrtype in ('Float', 'Decimal'):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    82
        value = req.property_value('ui.float-format') % value
7166
dde161937d3e [time zone] support for TZDatetime and TZTime data type
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7058
diff changeset
    83
    # XXX Interval
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    84
    return unicode(value)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    85
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    86
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    87
# text publishing #############################################################
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    88
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    89
try:
6491
ee9a10b6620e pylint option update
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6106
diff changeset
    90
    from cubicweb.ext.rest import rest_publish # pylint: disable=W0611
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    91
except ImportError:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    92
    def rest_publish(entity, data):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    93
        """default behaviour if docutils was not found"""
2312
af4d8f75c5db use xml_escape
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2208
diff changeset
    94
        return xml_escape(data)
1581
80ee6397c087 fix rest import, html_escape in null rest_publish
sylvain.thenault@logilab.fr
parents: 1263
diff changeset
    95
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    96
TAG_PROG = re.compile(r'</?.*?>', re.U)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    97
def remove_html_tags(text):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    98
    """Removes HTML tags from text
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
    >>> remove_html_tags('<td>hi <a href="http://www.google.fr">world</a></td>')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   101
    'hi world'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   102
    >>>
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   103
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   104
    return TAG_PROG.sub('', text)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   105
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   106
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   107
REF_PROG = re.compile(r"<ref\s+rql=([\'\"])([^\1]*?)\1\s*>([^<]*)</ref>", re.U)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   108
def _subst_rql(view, obj):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   109
    delim, rql, descr = obj.groups()
3418
7b49fa7e942d [api] use _cw, cw_row, cw_col, cw_rset etc.
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 3369
diff changeset
   110
    return u'<a href="%s">%s</a>' % (view._cw.build_url(rql=rql), descr)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   111
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   112
def html_publish(view, text):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   113
    """replace <ref rql=''> links by <a href="...">"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   114
    if not text:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   115
        return u''
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   116
    return REF_PROG.sub(lambda obj, view=view:_subst_rql(view, obj), text)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   117
277
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   118
# fallback implementation, nicer one defined below if lxml> 2.0 is available
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   119
def safe_cut(text, length):
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   120
    """returns a string of length <length> based on <text>, removing any html
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   121
    tags from given text if cut is necessary."""
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   122
    if text is None:
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   123
        return u''
362
a6a319f000c3 use mtconverter's html_unescape rather than saxutils' escape to deal with any html entity
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 350
diff changeset
   124
    noenttext = html_unescape(text)
350
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   125
    text_nohtml = remove_html_tags(noenttext)
277
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   126
    # try to keep html tags if text is short enough
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   127
    if len(text_nohtml) <= length:
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   128
        return text
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   129
    # else if un-tagged text is too long, cut it
2312
af4d8f75c5db use xml_escape
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2208
diff changeset
   130
    return xml_escape(text_nohtml[:length] + u'...')
350
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   131
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   132
fallback_safe_cut = safe_cut
277
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   133
5730
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   134
REM_ROOT_HTML_TAGS = re.compile('</(body|html)>', re.U)
6685
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   135
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   136
try:
6685
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   137
    from lxml import etree, html
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   138
    from lxml.html import clean, defs
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   139
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   140
    ALLOWED_TAGS = (defs.general_block_tags | defs.list_tags | defs.table_tags |
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   141
                    defs.phrase_tags | defs.font_style_tags |
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   142
                    set(('span', 'a', 'br', 'img', 'map', 'area', 'sub', 'sup'))
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   143
                    )
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   144
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   145
    CLEANER = clean.Cleaner(allow_tags=ALLOWED_TAGS, remove_unknown_tags=False,
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   146
                            style=True, safe_attrs_only=True,
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   147
                            add_nofollow=False,
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   148
                            )
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   149
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   150
    def soup2xhtml(data, encoding):
6685
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   151
        """tidy html soup by allowing some element tags and return the result
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   152
        """
5730
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   153
        # remove spurious </body> and </html> tags, then normalize line break
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   154
        # (see http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.7.1)
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   155
        data = REM_ROOT_HTML_TAGS.sub('', u'\n'.join(data.splitlines()))
6685
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   156
        xmltree = etree.HTML(CLEANER.clean_html('<div>%s</div>' % data))
6688
51ddb4842c56 update note
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6685
diff changeset
   157
        # NOTE: lxml 2.0 does support encoding='unicode', but last time I (syt)
51ddb4842c56 update note
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6685
diff changeset
   158
        # tried I got weird results (lxml 2.2.8)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   159
        body = etree.tostring(xmltree[0], encoding=encoding)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   160
        # remove <body> and </body> and decode to unicode
5730
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   161
        snippet = body[6:-7].decode(encoding)
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   162
        # take care to bad xhtml (for instance starting with </div>) which
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   163
        # may mess with the <div> we added below. Only remove it if it's
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   164
        # still there...
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   165
        if snippet.startswith('<div>') and snippet.endswith('</div>'):
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   166
            snippet = snippet[5:-6]
784025c15a3c [xhtml] fix soup2xhtml to deal with malformed div,body and html tags which may leads to malformed return value
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   167
        return snippet
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   168
6685
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   169
        # lxml.Cleaner envelops text elements by internal logic (not accessible)
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   170
        # see http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.7.1
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   171
        # TODO drop attributes in elements
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   172
        # TODO add policy configuration (content only, embedded content, ...)
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   173
        # XXX this is buggy for "<p>text1</p><p>text2</p>"...
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   174
        # XXX drop these two snippets action and follow the lxml behaviour
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   175
        # XXX (tests need to be updated)
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   176
        # if snippet.startswith('<div>') and snippet.endswith('</div>'):
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   177
        #     snippet = snippet[5:-6]
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   178
        # if snippet.startswith('<p>') and snippet.endswith('</p>'):
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   179
        #     snippet = snippet[3:-4]
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   180
        return snippet.decode(encoding)
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   181
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   182
except (ImportError, AttributeError):
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   183
    # gae environment: lxml not available
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   184
    # fallback implementation
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   185
    def soup2xhtml(data, encoding):
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   186
        # normalize line break
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   187
        # see http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.7.1
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   188
        return u'\n'.join(data.splitlines())
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   189
else:
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   190
eeedb3575d25 [uilib] soup2xhtml uses now lxml.html.Cleaner
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 6683
diff changeset
   191
    if hasattr(etree.HTML('<div>test</div>'), 'iter'): # XXX still necessary?
165
c5ff97312b8a cleaning code
Laure Bourgois <Laure.Bourgois@logilab.fr>
parents: 164
diff changeset
   192
277
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   193
        def safe_cut(text, length):
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   194
            """returns an html document of length <length> based on <text>,
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   195
            and cut is necessary.
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   196
            """
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   197
            if text is None:
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   198
                return u''
350
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   199
            dom = etree.HTML(text)
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   200
            curlength = 0
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   201
            add_ellipsis = False
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   202
            for element in dom.iter():
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   203
                if curlength >= length:
277
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   204
                    parent = element.getparent()
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   205
                    parent.remove(element)
350
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   206
                    if curlength == length and (element.text or element.tail):
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   207
                        add_ellipsis = True
277
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   208
                else:
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   209
                    if element.text is not None:
350
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   210
                        element.text = cut(element.text, length - curlength)
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   211
                        curlength += len(element.text)
277
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   212
                    if element.tail is not None:
350
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   213
                        if curlength < length:
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   214
                            element.tail = cut(element.tail, length - curlength)
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   215
                            curlength += len(element.tail)
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   216
                        elif curlength == length:
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   217
                            element.tail = '...'
277
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   218
                        else:
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   219
                            element.tail = ''
350
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   220
            text = etree.tounicode(dom[0])[6:-7] # remove wrapping <body></body>
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   221
            if add_ellipsis:
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   222
                return text + u'...'
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   223
            return text
1157
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   224
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   225
def text_cut(text, nbwords=30, gotoperiod=True):
350
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   226
    """from the given plain text, return a text with at least <nbwords> words,
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   227
    trying to go to the end of the current sentence.
277
a11a3c231050 fix lxml is available, we can have a nicer version of soup2xhtml even if its lxml < 2.0
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 228
diff changeset
   228
1157
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   229
    :param nbwords: the minimum number of words required
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   230
    :param gotoperiod: specifies if the function should try to go to
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   231
                       the first period after the cut (i.e. finish
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   232
                       the sentence if possible)
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   233
350
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   234
    Note that spaces are normalized.
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   235
    """
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   236
    if text is None:
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   237
        return u''
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   238
    words = text.split()
927
bfcc610c3d5e text_cut must return unicode not string
Stephanie Marcu <stephanie.marcu@logilab.fr>
parents: 525
diff changeset
   239
    text = u' '.join(words) # normalize spaces
1157
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   240
    textlength = minlength = len(' '.join(words[:nbwords]))
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   241
    if gotoperiod:
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   242
        textlength = text.find('.', minlength) + 1
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   243
        if textlength == 0: # no period found
81a383cdda5c text_cut() accepts a gotoperiod parameter
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 927
diff changeset
   244
            textlength = minlength
350
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   245
    return text[:textlength]
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   246
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   247
def cut(text, length):
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   248
    """returns a string of a maximum length <length> based on <text>
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   249
    (approximatively, since if text has been  cut, '...' is added to the end of the string,
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   250
    resulting in a string of len <length> + 3)
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   251
    """
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   252
    if text is None:
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   253
        return u''
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   254
    if len(text) <= length:
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   255
        return text
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   256
    # else if un-tagged text is too long, cut it
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   257
    return text[:length] + u'...'
f34ef2c64605 cleanup/fix cut variants
Sylvain Thenault <sylvain.thenault@logilab.fr>
parents: 277
diff changeset
   258
165
c5ff97312b8a cleaning code
Laure Bourgois <Laure.Bourgois@logilab.fr>
parents: 164
diff changeset
   259
1581
80ee6397c087 fix rest import, html_escape in null rest_publish
sylvain.thenault@logilab.fr
parents: 1263
diff changeset
   260
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   261
# HTML generation helper functions ############################################
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   262
5949
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   263
class _JSId(object):
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   264
    def __init__(self, id, parent=None):
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   265
        self.id = id
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   266
        self.parent = parent
5951
6026582ae4f1 [uilib] js objects implements __unicode__
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5949
diff changeset
   267
    def __unicode__(self):
5949
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   268
        if self.parent:
5951
6026582ae4f1 [uilib] js objects implements __unicode__
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5949
diff changeset
   269
            return u'%s.%s' % (self.parent, self.id)
6026582ae4f1 [uilib] js objects implements __unicode__
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5949
diff changeset
   270
        return unicode(self.id)
6026582ae4f1 [uilib] js objects implements __unicode__
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5949
diff changeset
   271
    def __str__(self):
6026582ae4f1 [uilib] js objects implements __unicode__
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5949
diff changeset
   272
        return unicode(self).encode('utf8')
5949
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   273
    def __getattr__(self, attr):
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   274
        return _JSId(attr, self)
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   275
    def __call__(self, *args):
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   276
        return _JSCallArgs(args, self)
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   277
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   278
class _JSCallArgs(_JSId):
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   279
    def __init__(self, args, parent=None):
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   280
        assert isinstance(args, tuple)
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   281
        self.args = args
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   282
        self.parent = parent
5951
6026582ae4f1 [uilib] js objects implements __unicode__
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5949
diff changeset
   283
    def __unicode__(self):
6026582ae4f1 [uilib] js objects implements __unicode__
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5949
diff changeset
   284
        args = u','.join(json_dumps(arg) for arg in self.args)
5949
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   285
        if self.parent:
5951
6026582ae4f1 [uilib] js objects implements __unicode__
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5949
diff changeset
   286
            return u'%s(%s)' % (self.parent, args)
5949
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   287
        return args
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   288
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   289
class _JS(object):
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   290
    def __getattr__(self, attr):
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   291
        return _JSId(attr)
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   292
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   293
"""magic object to return strings suitable to call some javascript function with
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   294
the given arguments (which should be correctly typed).
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   295
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   296
>>> str(js.pouet(1, "2"))
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   297
'pouet(1,"2")'
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   298
>>> str(js.cw.pouet(1, "2"))
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   299
'cw.pouet(1,"2")'
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   300
>>> str(js.cw.pouet(1, "2").pouet(None))
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   301
'cw.pouet(1,"2").pouet(null)')
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   302
"""
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   303
js = _JS()
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   304
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   305
def domid(string):
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   306
    """return a valid DOM id from a string (should also be usable in jQuery
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   307
    search expression...)
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   308
    """
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   309
    return string.replace('.', '_').replace('-', '_')
2a273c896a38 [box] provide a new generic base box class to edit relation to simple entities, backported from the 'tag' cube
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5730
diff changeset
   310
2398
a8d18e320ef3 a standard-conformant fix for sgml_tags
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 2375
diff changeset
   311
HTML4_EMPTY_TAGS = frozenset(('base', 'meta', 'link', 'hr', 'br', 'param',
a8d18e320ef3 a standard-conformant fix for sgml_tags
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 2375
diff changeset
   312
                              'img', 'area', 'input', 'col'))
a8d18e320ef3 a standard-conformant fix for sgml_tags
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 2375
diff changeset
   313
2516
b58826130680 extract function: sgml_attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2458
diff changeset
   314
def sgml_attributes(attrs):
b58826130680 extract function: sgml_attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2458
diff changeset
   315
    return u' '.join(u'%s="%s"' % (attr, xml_escape(unicode(value)))
b58826130680 extract function: sgml_attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2458
diff changeset
   316
                     for attr, value in sorted(attrs.items())
b58826130680 extract function: sgml_attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2458
diff changeset
   317
                     if value is not None)
b58826130680 extract function: sgml_attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2458
diff changeset
   318
1623
sylvain.thenault@logilab.fr
parents: 1581
diff changeset
   319
def simple_sgml_tag(tag, content=None, escapecontent=True, **attrs):
525
bd4e03297cf0 function to make generation of a simple sgml tag
sylvain.thenault@logilab.fr
parents: 362
diff changeset
   320
    """generation of a simple sgml tag (eg without children tags) easier
bd4e03297cf0 function to make generation of a simple sgml tag
sylvain.thenault@logilab.fr
parents: 362
diff changeset
   321
2399
68799e25f893 a standard-conformant fix for sgml_tags
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 2375
diff changeset
   322
    content and attri butes will be escaped
525
bd4e03297cf0 function to make generation of a simple sgml tag
sylvain.thenault@logilab.fr
parents: 362
diff changeset
   323
    """
bd4e03297cf0 function to make generation of a simple sgml tag
sylvain.thenault@logilab.fr
parents: 362
diff changeset
   324
    value = u'<%s' % tag
bd4e03297cf0 function to make generation of a simple sgml tag
sylvain.thenault@logilab.fr
parents: 362
diff changeset
   325
    if attrs:
980
59552ba2015f more tags, map klass -> class avoiding needs to use **dict
sylvain.thenault@logilab.fr
parents: 862
diff changeset
   326
        try:
59552ba2015f more tags, map klass -> class avoiding needs to use **dict
sylvain.thenault@logilab.fr
parents: 862
diff changeset
   327
            attrs['class'] = attrs.pop('klass')
59552ba2015f more tags, map klass -> class avoiding needs to use **dict
sylvain.thenault@logilab.fr
parents: 862
diff changeset
   328
        except KeyError:
59552ba2015f more tags, map klass -> class avoiding needs to use **dict
sylvain.thenault@logilab.fr
parents: 862
diff changeset
   329
            pass
2516
b58826130680 extract function: sgml_attributes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2458
diff changeset
   330
        value += u' ' + sgml_attributes(attrs)
525
bd4e03297cf0 function to make generation of a simple sgml tag
sylvain.thenault@logilab.fr
parents: 362
diff changeset
   331
    if content:
1623
sylvain.thenault@logilab.fr
parents: 1581
diff changeset
   332
        if escapecontent:
2312
af4d8f75c5db use xml_escape
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2208
diff changeset
   333
            content = xml_escape(unicode(content))
1623
sylvain.thenault@logilab.fr
parents: 1581
diff changeset
   334
        value += u'>%s</%s>' % (content, tag)
525
bd4e03297cf0 function to make generation of a simple sgml tag
sylvain.thenault@logilab.fr
parents: 362
diff changeset
   335
    else:
2398
a8d18e320ef3 a standard-conformant fix for sgml_tags
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 2375
diff changeset
   336
        if tag in HTML4_EMPTY_TAGS:
a8d18e320ef3 a standard-conformant fix for sgml_tags
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 2375
diff changeset
   337
            value += u' />'
a8d18e320ef3 a standard-conformant fix for sgml_tags
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 2375
diff changeset
   338
        else:
a8d18e320ef3 a standard-conformant fix for sgml_tags
Aurelien Campeas <aurelien.campeas@logilab.fr>
parents: 2375
diff changeset
   339
            value += u'></%s>' % tag
525
bd4e03297cf0 function to make generation of a simple sgml tag
sylvain.thenault@logilab.fr
parents: 362
diff changeset
   340
    return value
bd4e03297cf0 function to make generation of a simple sgml tag
sylvain.thenault@logilab.fr
parents: 362
diff changeset
   341
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   342
def tooltipize(text, tooltip, url=None):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   343
    """make an HTML tooltip"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   344
    url = url or '#'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   345
    return u'<a href="%s" title="%s">%s</a>' % (url, tooltip, text)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   346
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   347
def toggle_action(nodeid):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   348
    """builds a HTML link that uses the js toggleVisibility function"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   349
    return u"javascript: toggleVisibility('%s')" % nodeid
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   350
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   351
def toggle_link(nodeid, label):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   352
    """builds a HTML link that uses the js toggleVisibility function"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   353
    return u'<a href="%s">%s</a>' % (toggle_action(nodeid), label)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   354
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   355
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   356
def ureport_as_html(layout):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   357
    from logilab.common.ureports import HTMLWriter
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   358
    formater = HTMLWriter(True)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   359
    stream = StringIO() #UStringIO() don't want unicode assertion
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   360
    formater.format(layout, stream)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   361
    res = stream.getvalue()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   362
    if isinstance(res, str):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   363
        res = unicode(res, 'UTF8')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   364
    return res
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   365
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   366
# traceback formatting ########################################################
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   367
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   368
import traceback
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   369
6683
68cfebd3b9f3 fix #724689: exception's display during ajax call
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6491
diff changeset
   370
def exc_message(ex, encoding):
68cfebd3b9f3 fix #724689: exception's display during ajax call
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6491
diff changeset
   371
    try:
68cfebd3b9f3 fix #724689: exception's display during ajax call
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6491
diff changeset
   372
        return unicode(ex)
68cfebd3b9f3 fix #724689: exception's display during ajax call
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6491
diff changeset
   373
    except:
68cfebd3b9f3 fix #724689: exception's display during ajax call
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6491
diff changeset
   374
        try:
68cfebd3b9f3 fix #724689: exception's display during ajax call
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6491
diff changeset
   375
            return unicode(str(ex), encoding, 'replace')
68cfebd3b9f3 fix #724689: exception's display during ajax call
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6491
diff changeset
   376
        except:
68cfebd3b9f3 fix #724689: exception's display during ajax call
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6491
diff changeset
   377
            return unicode(repr(ex), encoding, 'replace')
68cfebd3b9f3 fix #724689: exception's display during ajax call
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6491
diff changeset
   378
68cfebd3b9f3 fix #724689: exception's display during ajax call
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6491
diff changeset
   379
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   380
def rest_traceback(info, exception):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   381
    """return a ReST formated traceback"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   382
    res = [u'Traceback\n---------\n::\n']
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   383
    for stackentry in traceback.extract_tb(info[2]):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   384
        res.append(u'\tFile %s, line %s, function %s' % tuple(stackentry[:3]))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   385
        if stackentry[3]:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   386
            res.append(u'\t  %s' % stackentry[3].decode('utf-8', 'replace'))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   387
    res.append(u'\n')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   388
    try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   389
        res.append(u'\t Error: %s\n' % exception)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   390
    except:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   391
        pass
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   392
    return u'\n'.join(res)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   393
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   394
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   395
def html_traceback(info, exception, title='',
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   396
                   encoding='ISO-8859-1', body=''):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   397
    """ return an html formatted traceback from python exception infos.
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   398
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   399
    tcbk = info[2]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   400
    stacktb = traceback.extract_tb(tcbk)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   401
    strings = []
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   402
    if body:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   403
        strings.append(u'<div class="error_body">')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   404
        # FIXME
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   405
        strings.append(body)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   406
        strings.append(u'</div>')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   407
    if title:
2312
af4d8f75c5db use xml_escape
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2208
diff changeset
   408
        strings.append(u'<h1 class="error">%s</h1>'% xml_escape(title))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   409
    try:
2312
af4d8f75c5db use xml_escape
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2208
diff changeset
   410
        strings.append(u'<p class="error">%s</p>' % xml_escape(str(exception)).replace("\n","<br />"))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   411
    except UnicodeError:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   412
        pass
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   413
    strings.append(u'<div class="error_traceback">')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   414
    for index, stackentry in enumerate(stacktb):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   415
        strings.append(u'<b>File</b> <b class="file">%s</b>, <b>line</b> '
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   416
                       u'<b class="line">%s</b>, <b>function</b> '
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   417
                       u'<b class="function">%s</b>:<br/>'%(
2312
af4d8f75c5db use xml_escape
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2208
diff changeset
   418
            xml_escape(stackentry[0]), stackentry[1], xml_escape(stackentry[2])))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   419
        if stackentry[3]:
2312
af4d8f75c5db use xml_escape
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2208
diff changeset
   420
            string = xml_escape(stackentry[3]).decode('utf-8', 'replace')
2996
866a2c135c33 B #345282 xhtml requires to use &#160; instead of &nbsp;
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 2516
diff changeset
   421
            strings.append(u'&#160;&#160;%s<br/>\n' % (string))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   422
        # add locals info for each entry
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   423
        try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   424
            local_context = tcbk.tb_frame.f_locals
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   425
            html_info = []
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   426
            chars = 0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   427
            for name, value in local_context.iteritems():
2312
af4d8f75c5db use xml_escape
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2208
diff changeset
   428
                value = xml_escape(repr(value))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   429
                info = u'<span class="name">%s</span>=%s, ' % (name, value)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   430
                line_length = len(name) + len(value)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   431
                chars += line_length
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   432
                # 150 is the result of *years* of research ;-) (CSS might be helpful here)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   433
                if chars > 150:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   434
                    info = u'<br/>' + info
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   435
                    chars = line_length
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   436
                html_info.append(info)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   437
            boxid = 'ctxlevel%d' % index
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   438
            strings.append(u'[%s]' % toggle_link(boxid, '+'))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   439
            strings.append(u'<div id="%s" class="pycontext hidden">%s</div>' %
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   440
                           (boxid, ''.join(html_info)))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   441
            tcbk = tcbk.tb_next
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   442
        except Exception:
1581
80ee6397c087 fix rest import, html_escape in null rest_publish
sylvain.thenault@logilab.fr
parents: 1263
diff changeset
   443
            pass # doesn't really matter if we have no context info
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   444
    strings.append(u'</div>')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   445
    return '\n'.join(strings)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   446
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   447
# csv files / unicode support #################################################
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   448
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   449
class UnicodeCSVWriter:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   450
    """proxies calls to csv.writer.writerow to be able to deal with unicode"""
1581
80ee6397c087 fix rest import, html_escape in null rest_publish
sylvain.thenault@logilab.fr
parents: 1263
diff changeset
   451
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   452
    def __init__(self, wfunc, encoding, **kwargs):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   453
        self.writer = csv.writer(self, **kwargs)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   454
        self.wfunc = wfunc
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   455
        self.encoding = encoding
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   456
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   457
    def write(self, data):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   458
        self.wfunc(data)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   459
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   460
    def writerow(self, row):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   461
        csvrow = []
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   462
        for elt in row:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   463
            if isinstance(elt, unicode):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   464
                csvrow.append(elt.encode(self.encoding))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   465
            else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   466
                csvrow.append(str(elt))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   467
        self.writer.writerow(csvrow)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   468
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   469
    def writerows(self, rows):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   470
        for row in rows:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   471
            self.writerow(row)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   472
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   473
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   474
# some decorators #############################################################
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   475
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   476
class limitsize(object):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   477
    def __init__(self, maxsize):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   478
        self.maxsize = maxsize
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   479
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   480
    def __call__(self, function):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   481
        def newfunc(*args, **kwargs):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   482
            ret = function(*args, **kwargs)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   483
            if isinstance(ret, basestring):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   484
                return ret[:self.maxsize]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   485
            return ret
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   486
        return newfunc
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   487
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   488
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   489
def htmlescape(function):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   490
    def newfunc(*args, **kwargs):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   491
        ret = function(*args, **kwargs)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   492
        assert isinstance(ret, basestring)
2312
af4d8f75c5db use xml_escape
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2208
diff changeset
   493
        return xml_escape(ret)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   494
    return newfunc