well I wanted to use deepl to translate a Japanese game in batches but sometimes it doesn't translate everything , in another case I also used deepl to translate in parts but it doesn't translate everything and also sometimes the same sentence was repeated , so I'm just using google to do my translation so at least the game is understood.
Yeah, the "solution" I've found is to run multiple passes with a script stripping out the already translated lines (because annoyingly enough the "ignore translated lines" also doesn't work with DeepL) between them, though inevitably a few lines it won't be able to translate no matter what, which I just finish off with Google.
I think it's generally caused by problems matching up the lines in the DeepL output to the original lines, which causes it to discard the entire batch assumedly. I try to counter this by also chopping every text block into individual lines beforehand, though I'm not sure how much it helps, but I assume it should be at least a bit less complex matching X number of individual lines compared to matching Y number of text blocks with Z lines each... You can find the script I use here:
You must be registered to see the links
if you want, though there's probably still some bugs in it since I sometimes run into odd behaviors like the number of untranslated lines increasing after a pass which would indicate to me that some lines didn't get merged back in properly... but I haven't put in the time to actually trace it yet since it's at least at the point that it "mostly" works on all the games I've tried recently. Also note that any lines with any Japanese text left in them at all get discarded, so if you have things like speaker names left untranslated due to escape rules your whole line will get tossed as a result.