Philipp Hörist pushed to branch master at gajim / gajim

Commits:

7 changed files:

Changes:

  • gajim/common/helpers.py
    ... ... @@ -80,6 +80,8 @@
    80 80 
     from gajim.common.const import URIAction
    81 81 
     from gajim.common.const import GIO_TLS_ERRORS
    82 82 
     from gajim.common.const import SHOW_LIST
    83 
    +from gajim.common.regex import INVALID_XML_CHARS_REGEX
    84 
    +from gajim.common.regex import STH_AT_STH_DOT_STH_REGEX
    83 85 
     from gajim.common.structs import URI
    84 86
    85 87
    ... ... @@ -632,7 +634,7 @@ def get_auth_sha(sid, initiator, target):
    632 634
    633 635 
     def remove_invalid_xml_chars(string_):
    634 636 
         if string_:
    635 
    -        string_ = re.sub(app.interface.invalid_XML_chars_re, '', string_)
    637 
    +        string_ = re.sub(INVALID_XML_CHARS_REGEX, '', string_)
    636 638 
         return string_
    637 639
    638 640 
     def get_random_string(count=16):
    ... ... @@ -1067,7 +1069,7 @@ def parse_uri(uri):
    1067 1069 
             uri = uri[4:]
    1068 1070 
             return URI(type=URIType.TEL, data=uri)
    1069 1071
    1070 
    -    if app.interface.sth_at_sth_dot_sth_re.match(uri):
    1072 
    +    if STH_AT_STH_DOT_STH_REGEX.match(uri):
    1071 1073 
             return URI(type=URIType.AT, data=uri)
    1072 1074
    1073 1075 
         if uri.startswith('geo:'):

  • gajim/common/regex.py
    1 
    +import re
    2 
    +
    3 
    +def _get_link_pattern():
    4 
    +    # regexp meta characters are:  . ^ $ * + ? { } [ ] \ | ( )
    5 
    +    # one escapes the metachars with \
    6 
    +    # \S matches anything but ' ' '\t' '\n' '\r' '\f' and '\v'
    7 
    +    # \s matches any whitespace character
    8 
    +    # \w any alphanumeric character
    9 
    +    # \W any non-alphanumeric character
    10 
    +    # \b means word boundary. This is a zero-width assertion that
    11 
    +    #    matches only at the beginning or end of a word.
    12 
    +    # ^ matches at the beginning of lines
    13 
    +    #
    14 
    +    # * means 0 or more times
    15 
    +    # + means 1 or more times
    16 
    +    # ? means 0 or 1 time
    17 
    +    # | means or
    18 
    +    # [^*] anything but '*' (inside [] you don't have to escape metachars)
    19 
    +    # [^\s*] anything but whitespaces and '*'
    20 
    +    # (?<!\S) is a one char lookbehind assertion and asks for any leading
    21 
    +    #         whitespace
    22 
    +    # and matches beginning of lines so we have correct formatting detection
    23 
    +    # even if the text is just '*foo*'
    24 
    +    # (?!\S) is the same thing but it's a lookahead assertion
    25 
    +    # \S*[^\s\W] --> in the matching string don't match ? or ) etc.. if at
    26 
    +    #                the end
    27 
    +    # so http://be) will match http://be and http://be)be) will match
    28 
    +    # http://be)be
    29 
    +
    30 
    +    legacy_prefixes = r"((?<=\()(www|ftp)\.([A-Za-z0-9\.\-_~:/\?#\[\]@!\$"\
    31 
    +        r"&'\(\)\*\+,;=]|%[A-Fa-f0-9]{2})+(?=\)))"\
    32 
    +        r"|((www|ftp)\.([A-Za-z0-9\.\-_~:/\?#\[\]@!\$&'\(\)\*\+,;=]"\
    33 
    +        r"|%[A-Fa-f0-9]{2})+"\
    34 
    +        r"\.([A-Za-z0-9\.\-_~:/\?#\[\]@!\$&'\(\)\*\+,;=]|%[A-Fa-f0-9]{2})+)"
    35 
    +    # NOTE: it's ok to catch www.gr such stuff exist!
    36 
    +
    37 
    +    # FIXME: recognize xmpp: and treat it specially
    38 
    +    links = r"((?<=\()[A-Za-z][A-Za-z0-9\+\.\-]*:"\
    39 
    +        r"([\w\.\-_~:/\?#\[\]@!\$&'\(\)\*\+,;=]|%[A-Fa-f0-9]{2})+"\
    40 
    +        r"(?=\)))|(\w[\w\+\.\-]*:([^<>\s]|%[A-Fa-f0-9]{2})+)"
    41 
    +
    42 
    +    # 2nd one: at_least_one_char@at_least_one_char.at_least_one_char
    43 
    +    mail = r'\bmailto:\S*[^\s\W]|' r'\b\S+@\S+\.\S*[^\s\W]'
    44 
    +
    45 
    +    link_pattern = links + '|' + mail + '|' + legacy_prefixes
    46 
    +    return link_pattern
    47 
    +
    48 
    +def _get_basic_pattern():
    49 
    +    basic_pattern = _get_link_pattern()
    50 
    +    # detects eg. *b* *bold* *bold bold* test *bold* *bold*! (*bold*)
    51 
    +    # doesn't detect (it's a feature :P) * bold* *bold * * bold * test*bold*
    52 
    +    formatting = r'|(?<!\w)' r'\*[^\s*]' r'([^*]*[^\s*])?' r'\*(?!\w)|'\
    53 
    +        r'(?<!\S)' r'~[^\s~]' r'([^~]*[^\s~])?' r'~(?!\S)|'\
    54 
    +        r'(?<!\w)' r'_[^\s_]' r'([^_]*[^\s_])?' r'_(?!\w)'
    55 
    +    return basic_pattern + formatting
    56 
    +
    57 
    +def _get_emot_and_basic_pattern(use_ascii_formatting=True):
    58 
    +    from gajim.gui.emoji_data import emoji_data
    59 
    +    # because emoticons match later (in the string) they need to be after
    60 
    +    # basic matches that may occur earlier
    61 
    +    emoticons = emoji_data.get_regex()
    62 
    +
    63 
    +    if use_ascii_formatting:
    64 
    +        pattern = _get_basic_pattern()
    65 
    +    else:
    66 
    +        pattern = _get_link_pattern()
    67 
    +
    68 
    +    return '%s|%s' % (pattern, emoticons)
    69 
    +
    70 
    +LINK_REGEX = re.compile(_get_link_pattern(), re.I | re.U)
    71 
    +
    72 
    +# link pattern + ASCII formatting
    73 
    +BASIC_REGEX = re.compile(_get_basic_pattern(), re.IGNORECASE)
    74 
    +
    75 
    +# emoticons + link pattern
    76 
    +EMOT_AND_LINK_REGEX = re.compile(_get_emot_and_basic_pattern(False),
    77 
    +                                          re.IGNORECASE)
    78 
    +
    79 
    +# emoticons + link pattern + ASCII formatting
    80 
    +EMOT_AND_BASIC_REGEX = re.compile(_get_emot_and_basic_pattern(True),
    81 
    +                                          re.IGNORECASE)
    82 
    +
    83 
    +INVALID_XML_CHARS_REGEX = re.compile(
    84 
    +    '[\x00-\x08]|[\x0b-\x0c]|[\x0e-\x1f]|[\ud800-\udfff]|[\ufffe-\uffff]')
    85 
    +
    86 
    +# at least one character in 3 parts (before @, after @, after .)
    87 
    +STH_AT_STH_DOT_STH_REGEX = re.compile(
    88 
    +    r'^[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+$')

  • gajim/conversation_textview.py
    ... ... @@ -44,6 +44,11 @@
    44 44 
     from gajim.common.const import Trust
    45 45 
     from gajim.common.const import URI_SCHEMES
    46 46 
     from gajim.common.helpers import to_user_string
    47 
    +from gajim.common.regex import STH_AT_STH_DOT_STH_REGEX
    48 
    +from gajim.common.regex import BASIC_REGEX
    49 
    +from gajim.common.regex import LINK_REGEX
    50 
    +from gajim.common.regex import EMOT_AND_BASIC_REGEX
    51 
    +from gajim.common.regex import EMOT_AND_LINK_REGEX
    47 52
    48 53 
     from gajim.gui import util
    49 54 
     from gajim.gui.util import get_cursor
    ... ... @@ -584,9 +589,17 @@ def detect_and_print_special_text(self, otext, other_tags, graphics=True,
    584 589 
             # basic: links + mail + formatting is always checked (we like that)
    585 590 
             if app.settings.get('emoticons_theme') and graphics:
    586 591 
                 # search for emoticons & urls
    587 
    -            iterator = app.interface.emot_and_basic_re.finditer(otext)
    588 
    -        else: # search for just urls + mail + formatting
    589 
    -            iterator = app.interface.basic_pattern_re.finditer(otext)
    592 
    +            if app.settings.get('ascii_formatting'):
    593 
    +                regex = EMOT_AND_BASIC_REGEX
    594 
    +            else:
    595 
    +                regex = EMOT_AND_LINK_REGEX
    596 
    +        else:
    597 
    +            if app.settings.get('ascii_formatting'):
    598 
    +                # search for just urls + mail + formatting
    599 
    +                regex = BASIC_REGEX
    600 
    +            else: # search for just urls + mail
    601 
    +                regex = LINK_REGEX
    602 
    +        iterator = regex.finditer(otext)
    590 603 
             if iter_:
    591 604 
                 end_iter = iter_
    592 605 
             else:
    ... ... @@ -693,7 +706,7 @@ def print_special_text(self, special_text, other_tags, graphics=True,
    693 706 
                 tags.append('mail')
    694 707 
             elif special_text.startswith('xmpp:') and not is_xhtml_link:
    695 708 
                 tags.append('xmpp')
    696 
    -        elif app.interface.sth_at_sth_dot_sth_re.match(special_text) and\
    709 
    +        elif STH_AT_STH_DOT_STH_REGEX.match(special_text) and \
    697 710 
             not is_xhtml_link:
    698 711 
                 # it's a JID or mail
    699 712 
                 tags.append('sth_at_sth')

  • gajim/gtk/message_input.py
    ... ... @@ -28,6 +28,7 @@
    28 28 
     from gajim.common import app
    29 29 
     from gajim.common.i18n import _
    30 30 
     from gajim.common.const import StyleAttr
    31 
    +from gajim.common.regex import LINK_REGEX
    31 32
    32 33 
     from .util import scroll_to_end
    33 34
    ... ... @@ -207,7 +208,7 @@ def make_clickable_urls(self, text):
    207 208 
             index = 0
    208 209
    209 210 
             new_text = ''
    210 
    -        iterator = app.interface.link_pattern_re.finditer(text)
    211 
    +        iterator = LINK_REGEX.finditer(text)
    211 212 
             for match in iterator:
    212 213 
                 start, end = match.span()
    213 214 
                 url = text[start:end]

  • gajim/gui_interface.py
    ... ... @@ -33,7 +33,6 @@
    33 33
    34 34 
     import os
    35 35 
     import sys
    36 
    -import re
    37 36 
     import time
    38 37 
     import json
    39 38 
     import logging
    ... ... @@ -101,7 +100,6 @@
    101 100 
     from gajim.gui.dialogs import InputDialog
    102 101 
     from gajim.gui.dialogs import PassphraseDialog
    103 102 
     from gajim.gui.filechoosers import FileChooserDialog
    104 
    -from gajim.gui.emoji_data import emoji_data
    105 103 
     from gajim.gui.filetransfer import FileTransfersWindow
    106 104 
     from gajim.gui.filetransfer_progress import FileTransferProgress
    107 105 
     from gajim.gui.roster_item_exchange import RosterItemExchangeWindow
    ... ... @@ -1296,112 +1294,6 @@ def handle_event(self, account, fjid, type_):
    1296 1294 
                 if isinstance(ctrl, ChatControlBase):
    1297 1295 
                     ctrl.scroll_to_end()
    1298 1296
    1299 
    -################################################################################
    1300 
    -### Methods dealing with emoticons
    1301 
    -################################################################################
    1302 
    -
    1303 
    -    @property
    1304 
    -    def basic_pattern_re(self):
    1305 
    -        if not self._basic_pattern_re:
    1306 
    -            self._basic_pattern_re = re.compile(self.basic_pattern,
    1307 
    -                re.IGNORECASE)
    1308 
    -        return self._basic_pattern_re
    1309 
    -
    1310 
    -    @property
    1311 
    -    def emot_and_basic_re(self):
    1312 
    -        if not self._emot_and_basic_re:
    1313 
    -            self._emot_and_basic_re = re.compile(
    1314 
    -                self.emot_and_basic, re.IGNORECASE)
    1315 
    -        return self._emot_and_basic_re
    1316 
    -
    1317 
    -    @property
    1318 
    -    def sth_at_sth_dot_sth_re(self):
    1319 
    -        if not self._sth_at_sth_dot_sth_re:
    1320 
    -            self._sth_at_sth_dot_sth_re = re.compile(self.sth_at_sth_dot_sth)
    1321 
    -        return self._sth_at_sth_dot_sth_re
    1322 
    -
    1323 
    -    @property
    1324 
    -    def invalid_XML_chars_re(self):
    1325 
    -        if not self._invalid_XML_chars_re:
    1326 
    -            self._invalid_XML_chars_re = re.compile(self.invalid_XML_chars)
    1327 
    -        return self._invalid_XML_chars_re
    1328 
    -
    1329 
    -    def make_regexps(self):
    1330 
    -        # regexp meta characters are:  . ^ $ * + ? { } [ ] \ | ( )
    1331 
    -        # one escapes the metachars with \
    1332 
    -        # \S matches anything but ' ' '\t' '\n' '\r' '\f' and '\v'
    1333 
    -        # \s matches any whitespace character
    1334 
    -        # \w any alphanumeric character
    1335 
    -        # \W any non-alphanumeric character
    1336 
    -        # \b means word boundary. This is a zero-width assertion that
    1337 
    -        #    matches only at the beginning or end of a word.
    1338 
    -        # ^ matches at the beginning of lines
    1339 
    -        #
    1340 
    -        # * means 0 or more times
    1341 
    -        # + means 1 or more times
    1342 
    -        # ? means 0 or 1 time
    1343 
    -        # | means or
    1344 
    -        # [^*] anything but '*' (inside [] you don't have to escape metachars)
    1345 
    -        # [^\s*] anything but whitespaces and '*'
    1346 
    -        # (?<!\S) is a one char lookbehind assertion and asks for any leading
    1347 
    -        #         whitespace
    1348 
    -        # and matches beginning of lines so we have correct formatting detection
    1349 
    -        # even if the text is just '*foo*'
    1350 
    -        # (?!\S) is the same thing but it's a lookahead assertion
    1351 
    -        # \S*[^\s\W] --> in the matching string don't match ? or ) etc.. if at
    1352 
    -        #                the end
    1353 
    -        # so http://be) will match http://be and http://be)be) will match
    1354 
    -        # http://be)be
    1355 
    -
    1356 
    -        self._basic_pattern_re = None
    1357 
    -        self._emot_and_basic_re = None
    1358 
    -        self._sth_at_sth_dot_sth_re = None
    1359 
    -        self._invalid_XML_chars_re = None
    1360 
    -
    1361 
    -        legacy_prefixes = r"((?<=\()(www|ftp)\.([A-Za-z0-9\.\-_~:/\?#\[\]@!\$"\
    1362 
    -            r"&'\(\)\*\+,;=]|%[A-Fa-f0-9]{2})+(?=\)))"\
    1363 
    -            r"|((www|ftp)\.([A-Za-z0-9\.\-_~:/\?#\[\]@!\$&'\(\)\*\+,;=]"\
    1364 
    -            r"|%[A-Fa-f0-9]{2})+"\
    1365 
    -            r"\.([A-Za-z0-9\.\-_~:/\?#\[\]@!\$&'\(\)\*\+,;=]|%[A-Fa-f0-9]{2})+)"
    1366 
    -        # NOTE: it's ok to catch www.gr such stuff exist!
    1367 
    -
    1368 
    -        # FIXME: recognize xmpp: and treat it specially
    1369 
    -        links = r"((?<=\()[A-Za-z][A-Za-z0-9\+\.\-]*:"\
    1370 
    -            r"([\w\.\-_~:/\?#\[\]@!\$&'\(\)\*\+,;=]|%[A-Fa-f0-9]{2})+"\
    1371 
    -            r"(?=\)))|(\w[\w\+\.\-]*:([^<>\s]|%[A-Fa-f0-9]{2})+)"
    1372 
    -
    1373 
    -        # 2nd one: at_least_one_char@at_least_one_char.at_least_one_char
    1374 
    -        mail = r'\bmailto:\S*[^\s\W]|' r'\b\S+@\S+\.\S*[^\s\W]'
    1375 
    -
    1376 
    -        # detects eg. *b* *bold* *bold bold* test *bold* *bold*! (*bold*)
    1377 
    -        # doesn't detect (it's a feature :P) * bold* *bold * * bold * test*bold*
    1378 
    -        formatting = r'|(?<!\w)' r'\*[^\s*]' r'([^*]*[^\s*])?' r'\*(?!\w)|'\
    1379 
    -            r'(?<!\S)' r'~[^\s~]' r'([^~]*[^\s~])?' r'~(?!\S)|'\
    1380 
    -            r'(?<!\w)' r'_[^\s_]' r'([^_]*[^\s_])?' r'_(?!\w)'
    1381 
    -
    1382 
    -        basic_pattern = links + '|' + mail + '|' + legacy_prefixes
    1383 
    -
    1384 
    -        link_pattern = basic_pattern
    1385 
    -        self.link_pattern_re = re.compile(link_pattern, re.I | re.U)
    1386 
    -
    1387 
    -        if app.settings.get('ascii_formatting'):
    1388 
    -            basic_pattern += formatting
    1389 
    -        self.basic_pattern = basic_pattern
    1390 
    -
    1391 
    -        # because emoticons match later (in the string) they need to be after
    1392 
    -        # basic matches that may occur earlier
    1393 
    -        emoticons = emoji_data.get_regex()
    1394 
    -
    1395 
    -        self.emot_and_basic = '%s|%s' % (basic_pattern, emoticons)
    1396 
    -
    1397 
    -        # at least one character in 3 parts (before @, after @, after .)
    1398 
    -        self.sth_at_sth_dot_sth = r'^[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+$'
    1399 
    -
    1400 
    -        # Invalid XML chars
    1401 
    -        self.invalid_XML_chars = '[\x00-\x08]|[\x0b-\x0c]|[\x0e-\x1f]|'\
    1402 
    -            '[\ud800-\udfff]|[\ufffe-\uffff]'
    1403 
    -
    1404 
    -
    1405 1297 
     ################################################################################
    1406 1298 
     ### Methods for opening new messages controls
    1407 1299 
     ################################################################################
    ... ... @@ -2114,15 +2006,6 @@ def __init__(self):
    2114 2006
    2115 2007 
             self.handlers = {}
    2116 2008 
             self.roster = None
    2117 
    -        self._invalid_XML_chars_re = None
    2118 
    -        self._basic_pattern_re = None
    2119 
    -        self._emot_and_basic_re = None
    2120 
    -        self._sth_at_sth_dot_sth_re = None
    2121 
    -        self.link_pattern_re = None
    2122 
    -        self.invalid_XML_chars = None
    2123 
    -        self.basic_pattern = None
    2124 
    -        self.emot_and_basic = None
    2125 
    -        self.sth_at_sth_dot_sth = None
    2126 2009
    2127 2010 
             self.avatar_storage = AvatarStorage()
    2128 2011
    ... ... @@ -2199,8 +2082,6 @@ def __init__(self):
    2199 2082 
                 from gajim.gui.emoji_chooser import emoji_chooser
    2200 2083 
                 emoji_chooser.load()
    2201 2084
    2202 
    -        self.make_regexps()
    2203 
    -
    2204 2085 
             self.last_ftwindow_update = 0
    2205 2086
    2206 2087 
             self._network_monitor = Gio.NetworkMonitor.get_default()

  • test/no_gui/test_regex.py
    1 
    +import unittest
    2 
    +
    3 
    +from gajim import gui
    4 
    +gui.init('gtk')
    5 
    +import gajim.common.regex as regex
    6 
    +
    7 
    +class Test(unittest.TestCase):
    8 
    +    def test_links_regexp_entire(self):
    9 
    +        def assert_matches_all(str_):
    10 
    +            m = regex.BASIC_REGEX.match(str_)
    11 
    +
    12 
    +            # the match should equal the string
    13 
    +            str_span = (0, len(str_))
    14 
    +            self.assertEqual(m.span(), str_span)
    15 
    +
    16 
    +        # these entire strings should be parsed as links
    17 
    +        assert_matches_all('http://google.com/')
    18 
    +        assert_matches_all('http://google.com')
    19 
    +        assert_matches_all('http://www.google.ca/search?q=xmpp')
    20 
    +
    21 
    +        assert_matches_all('http://tools.ietf.org/html/draft-saintandre-rfc3920bis-05#section-12.3')
    22 
    +
    23 
    +        assert_matches_all('http://en.wikipedia.org/wiki/Protocol_(computing)')
    24 
    +        assert_matches_all(
    25 
    +                'http://en.wikipedia.org/wiki/Protocol_%28computing%29')
    26 
    +
    27 
    +        assert_matches_all('mailto:test@example.org')
    28 
    +
    29 
    +        assert_matches_all('xmpp:example-node@example.com')
    30 
    +        assert_matches_all('xmpp:example-node@example.com/some-resource')
    31 
    +        assert_matches_all('xmpp:example-node@example.com?message')
    32 
    +        assert_matches_all('xmpp://guest@example.com/support@example.com?message')
    33 
    +
    34 
    +
    35 
    +if __name__ == "__main__":
    36 
    +    unittest.main()

  • test/unit/test_gui_interface.py
    ... ... @@ -24,33 +24,6 @@ def close_app():
    24 24 
             GLib.idle_add(close_app)
    25 25 
             app.app.run()
    26 26
    27 
    -    def test_links_regexp_entire(self):
    28 
    -        sut = Interface()
    29 
    -        def assert_matches_all(str_):
    30 
    -            m = sut.basic_pattern_re.match(str_)
    31 
    -
    32 
    -            # the match should equal the string
    33 
    -            str_span = (0, len(str_))
    34 
    -            self.assertEqual(m.span(), str_span)
    35 
    -
    36 
    -        # these entire strings should be parsed as links
    37 
    -        assert_matches_all('http://google.com/')
    38 
    -        assert_matches_all('http://google.com')
    39 
    -        assert_matches_all('http://www.google.ca/search?q=xmpp')
    40 
    -
    41 
    -        assert_matches_all('http://tools.ietf.org/html/draft-saintandre-rfc3920bis-05#section-12.3')
    42 
    -
    43 
    -        assert_matches_all('http://en.wikipedia.org/wiki/Protocol_(computing)')
    44 
    -        assert_matches_all(
    45 
    -                'http://en.wikipedia.org/wiki/Protocol_%28computing%29')
    46 
    -
    47 
    -        assert_matches_all('mailto:test@example.org')
    48 
    -
    49 
    -        assert_matches_all('xmpp:example-node@example.com')
    50 
    -        assert_matches_all('xmpp:example-node@example.com/some-resource')
    51 
    -        assert_matches_all('xmpp:example-node@example.com?message')
    52 
    -        assert_matches_all('xmpp://guest@example.com/support@example.com?message')
    53 
    -
    54 27
    55 28 
     if __name__ == "__main__":
    56 29 
         unittest.main()

    • —
    View it on GitLab.
    You're receiving this email because of your account on dev.gajim.org. If you'd like to receive fewer emails, you can adjust your notification settings.