sepehr
adc3583358
All checks were successful
Deploy to Production / Build and Deploy (push) Successful in 2m57s
106 lines
3.3 KiB
Python
106 lines
3.3 KiB
Python
"""Clean up glossaries: delete old FR→EN imports, rename multilingual ones
|
|
|
|
Revision ID: c8d9e0f1a2b3
|
|
Revises: b7c8d9e0f1a2
|
|
Create Date: 2026-05-31
|
|
|
|
Old glossaries imported before multilingual enrichment have target_language='en'
|
|
and no translations. They are stale duplicates of the enriched templates.
|
|
This migration deletes them so users re-import the enriched versions.
|
|
"""
|
|
|
|
from typing import Sequence, Union
|
|
|
|
from alembic import op
|
|
import sqlalchemy as sa
|
|
|
|
# revision identifiers
|
|
revision = "c8d9e0f1a2b3"
|
|
down_revision = "b7c8d9e0f1a2"
|
|
branch_labels: Union[str, Sequence[str], None] = None
|
|
depends_on: Union[str, Sequence[str], None] = None
|
|
|
|
|
|
def upgrade() -> None:
|
|
conn = op.get_bind()
|
|
dialect = conn.dialect.name
|
|
|
|
if dialect == "sqlite":
|
|
# SQLite test databases are created fresh and have no legacy glossaries to clean up.
|
|
# The PostgreSQL-specific jsonb_object_keys logic is not needed here.
|
|
return
|
|
|
|
# 1. Delete old FR→EN glossaries that don't have multilingual translations
|
|
# (imported before enrichment, they are stale duplicates)
|
|
op.execute("""
|
|
DELETE FROM glossary_terms
|
|
WHERE glossary_id IN (
|
|
SELECT g.id
|
|
FROM glossaries g
|
|
WHERE g.target_language = 'en'
|
|
AND g.source_language = 'fr'
|
|
AND g.id IN (
|
|
SELECT g2.id
|
|
FROM glossaries g2
|
|
JOIN glossary_terms gt ON gt.glossary_id = g2.id
|
|
WHERE gt.translations IS NULL
|
|
OR jsonb_typeof(gt.translations) = 'null'
|
|
OR (SELECT count(*) FROM jsonb_object_keys(COALESCE(gt.translations, '{}'))) = 0
|
|
LIMIT 1
|
|
)
|
|
)
|
|
""")
|
|
|
|
op.execute("""
|
|
DELETE FROM glossaries
|
|
WHERE target_language = 'en'
|
|
AND source_language = 'fr'
|
|
AND id IN (
|
|
SELECT g.id
|
|
FROM glossaries g
|
|
LEFT JOIN glossary_terms gt ON gt.glossary_id = g.id
|
|
WHERE gt.translations IS NULL
|
|
OR jsonb_typeof(COALESCE(gt.translations, 'null')) IN ('null', 'null')
|
|
LIMIT 1
|
|
)
|
|
""")
|
|
|
|
# 2. Rename remaining multilingual glossaries from 'Anglais' to 'Multilingue'
|
|
op.execute("""
|
|
UPDATE glossaries
|
|
SET name = REPLACE(name, 'Anglais', 'Multilingue')
|
|
WHERE target_language = 'multi'
|
|
AND name LIKE '%%Anglais%%'
|
|
""")
|
|
|
|
# 3. Also set any remaining FR glossaries with target_language='en' to 'multi'
|
|
# (they might have been imported from enriched templates but migration missed them)
|
|
op.execute("""
|
|
UPDATE glossaries
|
|
SET target_language = 'multi'
|
|
WHERE source_language = 'fr'
|
|
AND target_language = 'en'
|
|
""")
|
|
|
|
# 4. Rename those too
|
|
op.execute("""
|
|
UPDATE glossaries
|
|
SET name = REPLACE(name, 'Anglais', 'Multilingue')
|
|
WHERE target_language = 'multi'
|
|
AND name LIKE '%%Anglais%%'
|
|
""")
|
|
|
|
|
|
def downgrade() -> None:
|
|
# Cannot restore deleted glossaries, just revert names and target_language
|
|
op.execute("""
|
|
UPDATE glossaries
|
|
SET name = REPLACE(name, 'Multilingue', 'Anglais')
|
|
WHERE name LIKE '%%Multilingue%%'
|
|
""")
|
|
op.execute("""
|
|
UPDATE glossaries
|
|
SET target_language = 'en'
|
|
WHERE target_language = 'multi'
|
|
""")
|