GitLab

at 2022-06-19T23:51:31+02:00

fix: Styling: Parse text correctly which contains uris and addresses

Use one combined regex for uris and addresses so we get no overlaps

Fixes #10974

fix: Styling: Improve uri regex

Add more url delimiters

fix: Styling: Improve address regex

Allow "-" in domain part

chore: Styling: Add more tests

 BLOCK_NESTED_RX = re.compile(PRE_NESTED_RX + '|' + QUOTE_RX, re.S | re.M)

 UNQUOTE_RX = re.compile(r'^> |^>', re.M)

 

-URI_RX = r'((?P<protocol>[\w-]+://?|www[.])[\S()<>]+?(?=[,]?(\s|\Z)+))'

-URI_RX = re.compile(URI_RX)

+URI_RX = r'(?P<uri>([\w-]+://|www[.])[\S()<>]+?(?=[,>]?(\s|\Z)+))'

+ADDRESS_RX = r'(?P<address>\b((xmpp|mailto):)?[\w-]*@([\w-]*?\.)+[\w]+([\?].*?(?=([\s\),]|$)))?)'  # noqa: E501

+URI_ADDRESS_RX = URI_RX + '|' + ADDRESS_RX

 

-ADDRESS_RX = r'(\b(?P<protocol>(xmpp|mailto):)?[\w-]*@(\w*?\.)+[\w]+([\?].*?(?=([\s\),]|$)))?)'  # noqa: E501

+URI_RX = re.compile(URI_RX)

 ADDRESS_RX = re.compile(ADDRESS_RX)

+URI_ADDRESS_RX = re.compile(URI_ADDRESS_RX)

 

 EMOJI_RX = emoji_data.get_regex()

 EMOJI_RX = re.compile(EMOJI_RX)

                           offset_bytes,

                           is_address)

 

-    for match in URI_RX.finditer(line):

-        uri = make(match, False)

-        uris.append(uri)

-

-    for match in ADDRESS_RX.finditer(line):

-        uri = make(match, True)

+    for match in URI_ADDRESS_RX.finditer(line):

+        uri = make(match, bool(match.group('address')))

         uris.append(uri)

 

     return uris

 EMAILS = [

     'asd@asd.at',

     'asd@asd.asd.at',

+    'asd@asd.asd-asd.at',

+    'asd-asd@asd.asdasd.at',

     'mailto:foo@bar.com.uk',

 ]

 

 URL_WITH_TEXT = [

     ('see this http://userid@example.com/ link', 'http://userid@example.com/'),

     ('see this http://userid@example.com/, and ..', 'http://userid@example.com/'),

+    ('<http://userid@example.com/>', 'http://userid@example.com/'),

 ]

 

 XMPP_URIS = [

             self.assertTrue(text[start:end] == result)

 

 

-

 if __name__ == "__main__":

     unittest.main()

Philipp Hörist pushed to branch master at gajim / gajim

Commits:

2 changed files:

Changes: