Extra semicolons causing overly broad text matching

[Jakob Petsovits]

Hi,

I ended up researching the root cause of https://phabricator.kde.org/D28458 - why would KRunner translate "1/2" to all kinds of units in German, but not in English? And after a few rounds of digging, I found the answer:

https://websvn.kde.org/*checkout*/trunk/l10n-kf5/de/messages/kunitconversion/kunitconversion5.po

Specifically these two entries:

#: volume.cpp:96
#, kde-format
msgctxt "unit synonyms for matching user input"
msgid "cubic hectometer;cubic hectometers;hm³;hm/-3;hm^3;hm3"
msgstr "Kubikhektometer;;hm³;hm^3;hm3"

#: force.cpp:192
#, kde-format
msgctxt "unit synonyms for matching user input"
msgid "dyne;dynes;dyn"
msgstr "dyn;"

The code treats anything before, between or after a semicolon as a synonym for the main unit name. After "Kubikhektometer" there are two semicolons, so the code thinks the empty space in between is a synonym for "cubic hectometer". Similarly, "dyn;" has a semicolon at the end and the code treats the empty space after the semicolon as a synonym for "dyn".

In both cases, removing the extra semicolon would resolve this issue. I'm also trying to put in a safeguard in the unit conversion code itself as https://invent.kde.org/frameworks/kunitconversion/-/merge_requests/4 but it's better to resolve it in both places if possible.

Please have a look if you can. Thanks!
- Jakob