Changeset 9898

Timestamp:

07/08/08 20:49:51 (2 months ago)

Author:

steve-e

Message:

[Florob] Improved URL matching.

We don't match all valid URIs like blub:blaa but include common used (non)-urls like www.google.de

Location:

trunk/src

Files:

• common/config.py (modified) (1 diff)
• conversation_textview.py (modified) (3 diffs)
• gajim.py (modified) (2 diffs)

trunk/src/common/config.py

@@ -252 +252 @@
                 'check_idle_every_foo_seconds': [opt_int, 2, _('Choose interval between 2 checks of idleness.')],
                 'latex_png_dpi': [opt_str, '108',_('Change the value to change the size of latex formulas displayed. The higher is larger.') ],
+                'uri_schemes': [opt_str, 'http https ftp ftps gopher news ed2k irc magnet sip', _('Valid uri schemes. Only schemes in this list will be made accepted as "real" uri.')],
         }
 

trunk/src/conversation_textview.py

@@ -908 +908 @@
                         # add with possible animation
                         self.tv.add_child_at_anchor(img, anchor)
-                #FIXME: one day, somehow sync with regexp in gajim.py
-                elif special_text.startswith('http://') or \
-                        special_text.startswith('www.') or \
-                        special_text.startswith('ftp://') or \
-                        special_text.startswith('ftp.') or \
-                        special_text.startswith('https://') or \
-                        special_text.startswith('gopher://') or \
-                        special_text.startswith('news://') or \
-                        special_text.startswith('ed2k://') or \
-                        special_text.startswith('irc://') or \
-                        special_text.startswith('sip:') or \
-                        special_text.startswith('magnet:'):
-                        # it's a url
+                elif special_text.startswith('www.') or \
+                special_text.startswith('ftp.'):
                         tags.append('url')
                         use_other_tags = False
@@ -992 +981 @@
                         use_other_tags = False
                 else:
-                        #it's a url
-                        tags.append('url')
-                        use_other_tags = False
+                        # Check if we accept this as an uri
+                        schemes = gajim.config.get('uri_schemes').split()
+                        for scheme in schemes:
+                                if special_text.startswith(scheme + ':'):
+                                        tags.append('url')
+                                        use_other_tags = False
+                        # It's not a accepted uri
+                        if use_other_tags:
+                                end_iter = buffer.get_end_iter()
+                                buffer.insert_with_tags_by_name(end_iter, special_text, *other_tags)
 
                 if len(tags) > 0:
@@ -1172 +1168 @@
                 if name and (text.startswith('/me ') or text.startswith('/me\n')):
                         text = '* ' + name + text[3:]
+                        text_tags.append('italic')
                 # detect urls formatting and if the user has it on emoticons
                 index = self.detect_and_print_special_text(text, text_tags)

trunk/src/gajim.py

@@ -2275 +2275 @@
                 # so http://be) will match http://be and http://be)be) will match http://be)be
 
-                prefixes = '|'.join((r'http://', r'https://', r'gopher://', r'news://',
-                        r'ftp://', r'ed2k://', r'irc://', r'magnet:', r'sip:', r'www\.',
-                        r'ftp\.'))
+                legacy_prefixes = r"((?<=\()(www|ftp)\.([A-Za-z0-9\.\-_~:/\?#\[\]@!\$&'\(\)\*\+,;=]|%[A-Fa-f0-9]{2})+(?=\)))"\
+                                r"|((www|ftp)\.([A-Za-z0-9\.\-_~:/\?#\[\]@!\$&'\(\)\*\+,;=]|%[A-Fa-f0-9]{2})+"\
+                                r"\.([A-Za-z0-9\.\-_~:/\?#\[\]@!\$&'\(\)\*\+,;=]|%[A-Fa-f0-9]{2})+)"
                 # NOTE: it's ok to catch www.gr such stuff exist!
 
@@ -2296 +2296 @@
                 latex = r'|\$\$[^$\\]*?([\]\[0-9A-Za-z()|+*/-]|[\\][\]\[0-9A-Za-z()|{}$])(.*?[^\\])?\$\$'
 
-                basic_pattern = links + '|' + mail
+                basic_pattern = links + '|' + mail + '|' + legacy_prefixes
 
                 if gajim.config.get('use_latex'):