#992 Extend check_rules.groovy with ability to sort by error type

[Roman Mironov]

The check_rules.groovy is an extremely valuable script for running quality assurance on translations. Although it lacks some of the common functions right now, it could be used as a basis for creating a full-fledged QA feature like the ones found in commercial CAT tools. One step toward this would be adding the ability to sort by error type (or more generally column), which is crucial for efficient QA: errors are usually sorted by type, not by segment, which makes it possible to focus on one error type at a stretch.

It appears that the sorting mechanism created in the feature request [#972] might be a solution to this new feature request as well.

[cienislaw]

can do this during weekend.

[Kos Ivantsov]

I've added "Equal source and target" and "Untranslated/Empty target" rules (the updated version of the script is attached here). I can try to incorporate what cienislaw did for the spell-check script for this one too, and it would be a wonderful monkey-programming practice for me =). May I and should I?

[cienislaw]

No objections from my side.

[Kos Ivantsov]

Ok, it took me a while but I managed to implement cienislaw's code into check_rules.groovy
What has been added/improved:

• Results can be sorted, with the default sort column being the Rule/Error column;
• Results can be refreshed which eliminates the need to run the script multiple times while correcting QA errors;
• The script can run globally or in the current file only (default is current file);
• Four new rules added: check the number of tags, spaces around tags, tags sequence, and number inconsistencies between source and target;
• Rules can be enabled/disabled through the GUI on an individual basis. All rules are enabled by default;
• The script is completely localized through .properties. Currently there are English, Ukrainian and Russian localizations;
• Rules/Errors are given user-friendly names that can be localized.

I included cienislaw as the author, as much of his work has been almost verbatim copied here. Thank you very much, Piotr; playing with your code was very educational.

Attached are the new script and 3 .properties files (the script fails to execute if no .property is found).

[Kos Ivantsov]

Thanks to the prompt response from Roman, new rule was added, previous post edited, updated files attached anew.

[Didier Briel]

Rules can be enabled/disabled through the GUI on an individual basis. Checking for untranslated segments, target = source segments, and new tag rules are disabled by default;

Have you changed your mind about what is disabled by default? In the script I downloaded, everything was checked.

Didier

[Kos Ivantsov]

Have you changed your mind about what is disabled by default? In the script I downloaded, everything was checked.

Yes, sorry for not updating that part of the post. I personally prefer not to have those enabled by default, but Roman and several other people who tested the script, wanted to have all of the options enabled. Defaults are easily changeable, of course.

[Didier Briel]

No problem. I just wanted to be sure the documentation was consistent with the script. I edit your post accordingly.

Didier

[Didier Briel]

Implemented in SVN (/trunk).

I added a French localisation.

Didier

[Didier Briel]

Implemented in the released version 3.1.2 of OmegaT.

Didier