sepehr/office_translator
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(glossaries): update script to translate missing terms in any glossary, even if target_language is already 'multi'
All checks were successful
Deploy to Production / Build and Deploy (push) Successful in 2m43s
Details

Browse Source
This commit is contained in:
sepehr
2026-06-28 10:46:48 +02:00
parent 7398cae359
commit dde80f6bc3
1 changed files with 26 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

44
scripts/migrate_glossaries_to_multilingual_by_translation.py
View File

@@ -36,39 +36,41 @@ def translate_term(adapter, text_to_translate: str, src_lang: str, target_langs:
return translations
def main():
logger.info("Connecting to database and fetching non-multilingual glossaries...")
logger.info("Connecting to database and fetching all glossaries...")
adapter = get_legacy_google_adapter()
with get_sync_session() as session:
# Find glossaries where target_language != 'multi'
non_multi_glossaries = session.query(Glossary).filter(Glossary.target_language != 'multi').all()
# Find all glossaries
all_glossaries = session.query(Glossary).all()
if not non_multi_glossaries:
logger.info("All glossaries in database are already multilingual ('multi').")
if not all_glossaries:
logger.info("No glossaries found in database.")
return 0
logger.info(f"Found {len(non_multi_glossaries)} non-multilingual glossaries to convert.")
logger.info(f"Found {len(all_glossaries)} glossaries to process.")
for glossary in non_multi_glossaries:
for glossary in all_glossaries:
logger.info(f"Processing glossary '{glossary.name}' (ID: {glossary.id})")
# Change target_language to 'multi'
glossary.target_language = 'multi'
# Change target_language to 'multi' if not already
if glossary.target_language != 'multi':
glossary.target_language = 'multi'
# Rename the glossary to indicate it's now multilingual (e.g. replace "Anglais" with "Multilingue")
if "Anglais" in glossary.name:
glossary.name = glossary.name.replace("Anglais", "Multilingue")
elif "English" in glossary.name:
glossary.name = glossary.name.replace("English", "Multilingue")
else:
glossary.name = f"{glossary.name} → Multilingue"
# Rename the glossary to indicate it's now multilingual (e.g. replace "Anglais" with "Multilingue")
if "Anglais" in glossary.name:
glossary.name = glossary.name.replace("Anglais", "Multilingue")
elif "English" in glossary.name:
glossary.name = glossary.name.replace("English", "Multilingue")
else:
glossary.name = f"{glossary.name} → Multilingue"
src_lang = glossary.source_language or "fr"
terms = session.query(GlossaryTerm).filter(GlossaryTerm.glossary_id == glossary.id).all()
logger.info(f"Translating {len(terms)} terms for '{glossary.name}'...")
logger.info(f"Checking translations for {len(terms)} terms in '{glossary.name}'...")
updated_terms_count = 0
for idx, term in enumerate(terms):
translations = term.translations or {}
@@ -80,6 +82,7 @@ def main():
logger.info(f" [{idx+1}/{len(terms)}] Translating '{term.source}' to {missing_langs}...")
new_translations = translate_term(adapter, term.source, src_lang, missing_langs)
translations.update(new_translations)
updated_terms_count += 1
# Ensure the original default target (e.g. English translation) is in the translations dict under 'en'
if 'en' not in translations or not translations['en']:
@@ -87,8 +90,11 @@ def main():
term.translations = translations
session.commit()
logger.info(f"Glossary '{glossary.name}' successfully converted to multilingual.")
if updated_terms_count > 0:
session.commit()
logger.info(f"Glossary '{glossary.name}' updated with {updated_terms_count} translated terms.")
else:
logger.info(f"No translations were missing for glossary '{glossary.name}'.")
logger.info("Migration complete!")
return 0