Catalogs: Get completeness of translation

[buhtz]

I was looking in to 'Catalogs' class but couldn't find the right method or attribute.

Can I calculate how complete a translation is when loading a specific po file?

[agammann]

You can calculate it by iterating over the Message objects from read_po(). It helps to define what counts as complete: the example below requires a non-fuzzy translation, with every required plural form filled in, and counts each message once.

from babel.messages.pofile import read_po

with open("messages.po", "rb") as f:
    catalog = read_po(f)

messages = [message for message in catalog if message.id != ""]

def is_complete(message):
    if message.fuzzy:
        return False
    if message.pluralizable:
        forms = message.string
        return len(forms) == catalog.num_plurals and all(forms)
    return bool(message.string)

translated = sum(is_complete(message) for message in messages)
print(f"{translated}/{len(messages)} translated")
if messages:
    print(f"{100 * translated / len(messages):.1f}%")

The empty msgid is the PO header, so it is excluded. Obsolete entries are stored separately in catalog.obsolete and aren't included by ordinary iteration. Fuzzy entries and partially translated plural entries remain in the denominator but don't count as complete here.

This measures coverage of the active entries in that PO file; it doesn't validate the translations or check for messages missing from an outdated catalog. Merge against the current POT first if you want project-wide coverage. I checked the example with Babel 2.18.0, including fuzzy, untranslated, plural, obsolete, and empty-catalog cases.

References: PO file loading, Catalog and Message API.

[buhtz]

Why not using PoFile.percent_translated() ?

[agammann]

Yes—if you're using polib, POFile.percent_translated() is the simpler option:

import polib
print(polib.pofile("messages.po").percent_translated())

My example kept the calculation within Babel, since the question linked to its Catalog API. Babel's read_po() returns a Catalog, which doesn't provide that method in Babel 2.18.0. I checked the snippet above with polib 1.2.0.