Remove a bad import from a translation memory
The wrong client's TMX went into the master TM, or an MT-pretranslated file was imported as if it were reviewed. CAT tools import in one click and offer nothing to take it back — unless you still have the file you imported, which you usually do.
━━ the problem
After the import, the bad units carry the same languages and fields as everything else. Deleting them by hand means searching one sentence at a time.
Where the bad file had the same source with a different translation, you want to keep yours. A crude "delete everything with this source" throws away good work.
Without the file, the only clue is when it happened: every unit created or changed that afternoon.
━━ how TM Cleaner handles it
Export the TM to TMX and drop it in, then choose “+ Remove what's also in another TM” and add the file you imported. Every pair found in it is removed — pair by pair, so your own translation of the same source stays. No file? Remove everything changed after a date under Translators & dates.
- ✓matches whole pairs, ignoring case and spacing differences
- ✓your translation of the same source stays when the bad file had a different one
- ✓every removed unit is listed as “in the other TM” and kept in removed.tmx
- ✓works with zips and spreadsheets on either side
━━ frequently asked
what if I no longer have the file I imported?+
use the date filter in Translators & dates: remove every unit changed after the import. the preview shows exactly which units that is before anything goes.
does the other TM need the same language codes?+
no. it's read with your TM's source language, so en_US against en-US, or a header that says *all*, still lines up.
how do I get the result back into my CAT tool?+
import the cleaned TMX into a new, empty TM (or replace the old one's content), the usual way.
━━ related pages
free · no account up to 50 MB · files up to 2 GB · try the demo