Usage of "target at import time" as source for tags: Only for bilingual tasks

Please talk with Marc before starting this.

Current behavior is,

• "target at import time" column in the editor is used as source for handling of tags
  • if it has been populated by pre-translation or by a pre-populated bilingual file at time of import
  • and not if it has been left empty at import time for what ever reason

Future behavior should be

• target at import time" column in the editor is used as source for handling of tags (and only then)

• • if it has been populated by a pre-populated bilingual file at time of import
  • but NOT if it has been populated by pre-translation within translate5

Handling of tags of tags refers to

• tag check in front-end
• tag check in autoQA in back-end
• taking over tags via short-cut in front-end

Reasons:

• Some clients pre-populate segments also with Fuzzy matches in pre-translation (by setting pre-translation match rate e. g. to 80%)
• Even 100% matches or pre-translations from MT may have tag errors
• So in both scenarios it does not make sense to use the pre-translated segment as a source for the tag check

Also please change the description of the following config, so that it really reflects, what it does.

Current text:

Use Source text as reference for review task
If enabled 'Source text' filed is used as reference for checking tags, otherwise 'Target text at time of import' is used. This is only for review tasks.

Please change it to

Use Source text as reference source for tags for review task
If enabled 'Source text' field is used as reference for checking and copying tags, otherwise 'Target text at time of import' is used. This is only for review tasks, so tasks where directly the bilingual files have been imported to translate5 and even for those tasks only for segments, that had already content in the target of the imported bilingual segments.

Also until recently we had a bug, that populated "target at import time", when a match was taken over by the user. This is fixed now, but the data is still there. So we should empty the "target at import time" field for all segments, that

• do not stem from a pre-translation
• AND where the target had been empty in the imported xliff file