Last modified: 2013-04-08 12:45:42 UTC
On a French wiki, some French translation of months have trouble to be parsed. For instance, "janvier" (january) works well, as do the other months without any special character, but not "février" (february) or "décembre" (december). A small warning sign is displayed on the right of the date. [[Date de sortie::21 mars 2012]] -> ok [[Date de sortie::21 février 2011]] -> not ok [[Date de sortie::21 fév 2011]] (short version) -> not ok [[Date de sortie::21 août 2012]] -> not ok This problem was already there in SMW 1.6 but it wasn't in a more distant past.
I have found the culprit: $matches = preg_split( "/([T]?[0-2]?[0-9]:[\:0-9]+[+\-]?[0-2]?[0-9\:]+|[a-z,A-Z]+|[0-9]+|[ ])/u", $parsevalue , -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY ); in includes/datavalues/SMW_DV_Time.php:202 (SMW 1.8.0.4). For instance, with the date "8 décembre 1999" (december), $matches is: Array ( [0] => 8 [1] => [2] => d [3] => \xc3\xa9 [4] => cembre [5] => [6] => 1999 ) It should be something like: Array ( [0] => 8 [1] => [2] => d\xc3\xa9cembre [3] => [4] => 1999 )
I replaced [a-z,A-Z] by [\p{L}] to match any Unicode letter character (see <http://www.php.net/manual/en/regexp.reference.unicode.php>) and it works fine. I think it's a harmless change. Any opinion?
That looks good, thanks for reporting, investigating and providing a patch :) I will try it out now.
https://gerrit.wikimedia.org/r/#/c/57422/
