compare_translations.py

from pathlib import Path
from babel.messages import pofile
from babel.messages.catalog import Catalog

missing = Catalog("ar")
changed = Catalog("ar")
added = Catalog("ar")

asu_django = pofile.read_po(Path("asu/django.po").open())
edx_django = pofile.read_po(Path("edx/django.po").open())

edx_dict = {message.id: message.string for message in edx_django}


def add_to_catalog(catalog, message):
    catalog.add(
        message.id,
        message.string,
        message.locations,
        message.flags,
        message.auto_comments,
        message.user_comments,
        message.previous_id,
        message.lineno,
        message.context,
    )


for message in asu_django:
    # Just skip blank translations
    if not message.string:
        continue

    if message.id in edx_dict:
        edx_msg = edx_dict[message.id]
        if not any(edx_msg):
            # edX doesn't have this translation at all
            add_to_catalog(missing, message)
        elif edx_msg != message.string:
            # edX has this translation, but it is different
            print("Conflicting translation", message.id)
            print("edx version", edx_msg)
            print("asu version", message.string)
            add_to_catalog(changed, message)
        else:
            # edX this translation and it is the same as ASU, skip
            continue
    else:
        # edx doesn't even have this message id, so it's probably for a
        # string only used by ASU's branch
        add_to_catalog(added, message)


pofile.write_po(Path("missing_django.po").open("wb"), missing)
pofile.write_po(Path("changed_django.po").open("wb"), changed)
pofile.write_po(Path("added_django.po").open("wb"), added)