How do I merge POT and PO files so that I exclude entries that are not in the POT file?

Question

In short, I am trying to find a way to create a new PO file from a new POT and an existing PO file - but I want to exclude any strings (and their translations) that are not in the POT file.

Every time we change the wording on our cakePHP site, we generate a new POT file that contains all the translatable strings in the site. But when we merge it with the existing PO file (using POEdit), the merge process only adds the POT entries to the PO file. It doesn't remove the translations we no longer need. We have over 12k unneeded translations in our PO files. This makes our translator very unhappy. She has taken to just looking at the site and sending me translations to add manually, which makes me very unhappy.

I've looked around for tools that do this destructive merge, but I haven't been successful finding one. Before I head off to write one...is there something I missed?

(Sorry if this belongs on a different exchange, I will move this post to a better exchange if anyone tells me which one).

Answer 1

What you describe as "destructive" merge is the standard, normal merge operation in gettext and what everybody wants — you'd have to go out of your way to accomplish non-destructive versions, and I'm not even sure how.

From this it's safe to conclude that (1) you must be doing some weird steps not described above, or (2) your POT file contains more than you think it does (e.g. because you append to it instead of replacing it), or (3) you or the tools you use misinterpret the resulting PO file.

To merge using GNU gettext command line tools:

msgmerge -U your_old_translation.po latest_strings.pot

To merge using Poedit (notice the spelling):

1. Open PO file with the (now outdated) translations.
2. Use Catalog → Update from POT file…
3. Choose the newly regenerated POT file.

Notice that by default, outdated translations are kept in the PO file as backup. In Poedit, you can purge them (see Catalog → Purge deleted translations). However, these obsolete entries are stored in a different way in the PO file, as specially formatted comments, and are not visible or editable in Poedit or any conforming PO editing tool.

If I were to bet, I'd say (3) is the most likely cause (in which case, use a better editor like, ahem, Poedit), or perhaps (2) (should be easy to review by searching the POT for now-unused strings).

But merging really does the right thing that you expect it to do.