tm.py

tm.py

import time
from pontoon.base.models import TranslationMemoryEntry, Translation
from bulk_update.helper import bulk_update

tm_entries_to_update = []

"""
Generate a dict of TM entries for which translation ForeignKey is null
  - key: serialized TM entry data
  - value: reference to a TM entry
"""
tm_entries_dict = {}
tm_entries_without_translation = TranslationMemoryEntry.objects.filter(translation__isnull=True)
tm_entries_without_translation_pks = list(tm_entries_without_translation.values_list("pk"))
tm_entries_without_translation_values = list(tm_entries_without_translation.values_list("entity", "locale", "source", "target"))

for i, value in enumerate(tm_entries_without_translation_values):
  tm_entries_dict[unicode(value[0]) + "\x04" + unicode(value[1]) + "\x04" + unicode(value[2]) + "\x04" + unicode(value[3])] = tm_entries_without_translation_pks[i]

"""
Loop over all translations without TM entries, find matching TM entries and assign translations to them
"""
translations_not_in_tm = Translation.objects.filter(approved=True, memory_entries__isnull=True)
translations_not_in_tm_pks = list(translations_not_in_tm.values_list("pk"))
translations_not_in_tm_values = list(translations_not_in_tm.values_list("entity", "locale", "entity__string", "string"))

for i, value in enumerate(translations_not_in_tm_values):
  try:
    tm_entry = tm_entries_dict[unicode(value[0]) + "\x04" + unicode(value[1]) + "\x04" + unicode(value[2]) + "\x04" + unicode(value[3])]
    tm_entries_to_update.append(TranslationMemoryEntry(
      pk=tm_entry,
      translation_id=translations_not_in_tm_pks[i][0]
    ))
  except KeyError:
    pass

for i in range(0, len(tm_entries_to_update)/1000):
  print i
  bulk_update(tm_entries_to_update[i*1000:(i*1000+1000)], update_fields=['translation'])
  time.sleep(1)

"""
len(tm_entries_without_translation_pks)
len(translations_not_in_tm_pks)
len(tm_entries_to_update)
"""

I would order_by() those queries to be sure that indexes will match.