daniel-sc/ng-extract-i18n-merge

Give a hint if the same ID occurs more than once

ftaffelt opened this issue · 3 comments

ftaffelt commented

During a faulty merge in our source versioning system, duplicate ID's occurred in the messages files for one language. This condition was not detected for a long time and led to a long search for causes on different levels before the problem was found in duplicate message IDs.

I know the angular property "i18nDuplicateTranslation" but this doesn't seem to give a warning or error message here either.

Would it be possible to integrate a check for duplicate ID's?

daniel-sc commented

@ftaffelt what would be the desired result?
Some logging vs. error + exit vs. remove duplicates?

(not really sure if / how this fits into this lib - some angular option affecting/aborting the build like i18nDuplicateTranslation would seem a better fit?)

ftaffelt commented

Good question. I think that an automatic removal of duplicate values is not possible, because you don't know what is the right one to keep. Such a warning would be helpful.

In my case, I wrote an XSL stylesheet that embeds the duplicate values in the transunit:

<trans-unit duplicates="2" id="1220142055650021084" datatype="html">
   <source>Change number of items</source>
   <target state="new">Change number of items</target>
   <note priority="1" from="description">change page size title</note>
   <note priority="1" from="meaning">Resultset</note>
   <duplicates>
      <target state="new">Change number of items</target>
      <target state="new">Change number of hits</target>
   </duplicates>
</trans-unit>

Then it is a manual/human job to correct the values and remove the entries under duplicates/values again.

But maybe you are right and the topic is a bit too special and not in the scope of this lib.

daniel-sc commented

After some thought, this can probably be done with a "simple" bash command (not tested) - which could be a verification step in your CI:

grep -Eo 'id="[^"]+"' messages*.xlf | sort | uniq -d

Based on this, I'd rather keep this additional complexity out of this lib.
In case you - or others - think this is relevant to more users, please reopen/comment!