If acronym contains a hyphen, it is not linked to correct page

[Ryan J Ollos]

As described in ​comment:3#857, the following example:

||XXX    || XXX Page     || XXXPage    || ||
||YYY-XXX|| YYY-XXX Page || YYY-XXXPage|| ||

Results in:

[Ryan J Ollos]

I've traced this to the regular expression not matching XXX-YYY. We'll need to modify the regular expression:

valid_acronym = re.compile('^\w+$')

[Ryan J Ollos]

I've added the UNICODE flag so that acronyms with unicode characters classified as alphanumeric will be matched. An alternative would be to set the LOCALE flag, in which case characters classified as alphanumeric in the environment's locale would be matched. I'm not sure which is better.

hasienda, is this something you'd like to test out, since you have done a lot of work with locales?

[Ryan J Ollos]

(In [9585]) Refs #8266:

• Some minor refactoring.
• Set the UNICODE flag when compiling the regular expression used to match acronyms.

[Steffen Hoffmann]

Replying to rjollos:

hasienda, is this something you'd like to test out, since you have done a lot of work with locales?

Will do and report back here; thank you for the hint.

[Ryan J Ollos]

Received a hint about this in #5938, and will submit a fix shortly.

[Ryan J Ollos]

(In [9662]) Use \S in the regular expression that extracts acronym definitions from the /wiki/acronym page. \S will match any non-whitespace character, whereas \w only matches alphanumeric characters and the underscore. Fixes #8266.

[Steffen Hoffmann]

I'm re-iterating through issues for this plugin now while preparing for an upcoming Trac application.

The regexp change sucessfully solved another issue for me: acronyms with Unicode characters like German umlauts. Before coming to this ticket I've done own experiments on this matter. Results have been rather confusing to me: re.U flag for that r'^\w+$' expression didn't result in expected matches:

Python 2.6.6 (r266:84292, Dec 27 2010, 00:02:40) 
[GCC 4.4.5] on linux2
>>> import re
>>> RE = re.compile(r'^\w+$', re.U)
>>> RE.match('ä')
>>> RE.match('ö')
>>> RE.match('ü')
<_sre.SRE_Match object at 0xb7359d08>
>>> RE.match('ß')
>>> RE.match('Ä')
>>> RE.match('Ö')
>>> RE.match('Ü')

The re.L flag didn't change matches at all. So the very general \S match is the best I can see right now. Still it troubles me, I may not understand that flags correctly...