feat(translation): per-field language detection for mixed-language posts
CI / lint (push) Successful in 3s
CI / frontend-build (push) Successful in 20s
CI / backend-lint-and-test (push) Successful in 33s
CI / integration (push) Successful in 3m43s

_translate_one translated [title, description] in ONE Interpreter call and keyed
the whole-post passthrough on the aggregate detected_lang (the FIRST item). So an
English title + non-English description detected "en" and marked the post handled,
leaving the description untranslated. Now each field is translated independently
(its own detected_lang / passthrough) and the non-target field is stored on its
own; translated_source_lang reflects the translated field's language.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-07-07 23:31:04 -04:00
parent 06f98acf3e
commit 3c7ab44e74
2 changed files with 58 additions and 19 deletions
+29
View File
@@ -115,6 +115,35 @@ def test_translate_posts_passthrough_english_marks_handled(db_sync, monkeypatch)
assert p.post_title_translated is None # ...but nothing to show
def test_translate_posts_mixed_language_translates_nonenglish_field(db_sync, monkeypatch):
# English title + Japanese description: the description is still translated
# even though the title is already English (per-field detection, not the old
# aggregate first-item bail). Source lang comes from the translated field.
_patch(monkeypatch, db_sync)
monkeypatch.setattr("backend.app.tasks.translation.ic.health", lambda *a, **k: True)
def fake_translate(texts, **k):
t = texts[0]
if t.isascii(): # already English → passthrough
return {"translations": [t], "detected_lang": "en",
"engine": "none", "engine_version": None}
return {"translations": [f"EN:{t}"], "detected_lang": "ja",
"engine": "llm", "engine_version": "v9"}
monkeypatch.setattr("backend.app.tasks.translation.ic.translate", fake_translate)
a = _artist(db_sync)
p = _post(db_sync, a.id, title="hello", desc="ねこ")
_enable(db_sync)
db_sync.commit()
assert "translated=1" in translate_posts()
db_sync.refresh(p)
assert p.post_title_translated is None # English title untouched
assert p.description_translated == "EN:ねこ" # Japanese desc translated
assert p.translated_source_lang == "ja" # from the translated field
assert p.translation_engine_version == "v9"
def test_translate_posts_disabled_is_noop(db_sync, monkeypatch):
_patch(monkeypatch, db_sync)
a = _artist(db_sync)