feat(i18n): manifest translation extract/merge tool + en catalog

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
BenStullsBets
2026-06-29 18:49:52 -07:00
parent 1977e679ee
commit dcdc3d62f7
4 changed files with 1339 additions and 0 deletions
+51
View File
@@ -0,0 +1,51 @@
import copy
import pytest
from tools.i18n.translate_manifest import extract_en_catalog, merge_catalog
MANIFEST = {
"clips": [
{"id": "cosmos", "strings": {"en": {
"detected.star": ["star", "young star"],
"measure.distance": "≈7,500 ly",
}}},
]
}
def test_extract_pulls_every_en_string():
cat = extract_en_catalog(MANIFEST)
assert cat == {"cosmos": {"detected.star": ["star", "young star"], "measure.distance": "≈7,500 ly"}}
def test_merge_writes_lang_and_leaves_en():
m = copy.deepcopy(MANIFEST)
cat = {"cosmos": {"detected.star": ["estrella", "estrella joven"], "measure.distance": "≈7.500 al"}}
merge_catalog(m, "es", cat)
s = m["clips"][0]["strings"]
assert s["en"]["measure.distance"] == "≈7,500 ly" # en untouched
assert s["es"]["detected.star"] == ["estrella", "estrella joven"]
def test_merge_is_idempotent():
m = copy.deepcopy(MANIFEST)
cat = {"cosmos": {"detected.star": ["estrella", "estrella joven"], "measure.distance": "≈7.500 al"}}
merge_catalog(m, "es", cat)
once = copy.deepcopy(m)
merge_catalog(m, "es", cat)
assert m == once
def test_merge_rejects_tier_length_mismatch():
m = copy.deepcopy(MANIFEST)
bad = {"cosmos": {"detected.star": ["estrella"], "measure.distance": "≈7.500 al"}} # 1 tier vs en's 2
with pytest.raises(ValueError):
merge_catalog(m, "es", bad)
def test_merge_rejects_type_mismatch():
m = copy.deepcopy(MANIFEST)
bad = {"cosmos": {"detected.star": "estrella", "measure.distance": "≈7.500 al"}} # str vs en's list
with pytest.raises(ValueError):
merge_catalog(m, "es", bad)
View File
File diff suppressed because it is too large Load Diff
+79
View File
@@ -0,0 +1,79 @@
"""Extract a manifest's English annotation strings and merge authored
translations back into each clip's `strings` dict (no schema change).
Workflow:
1. `extract` writes tools/i18n/catalogs/en.json — the English strings.
2. A translator (human or LLM) authors es.json / fr.json / ja.json with the
SAME structure (lists stay lists of the same length).
3. `merge` validates structure and writes strings[lang] into the manifest.
"""
from __future__ import annotations
import argparse
import json
from pathlib import Path
def extract_en_catalog(manifest: dict) -> dict:
out = {}
for clip in manifest.get("clips", []):
en = (clip.get("strings") or {}).get("en")
if en:
out[clip["id"]] = en
return out
def _check_shape(key: str, en_val, tr_val):
if isinstance(en_val, list):
if not isinstance(tr_val, list) or len(tr_val) != len(en_val):
raise ValueError(f"{key}: expected list of {len(en_val)} tiers, got {tr_val!r}")
elif not isinstance(tr_val, str):
raise ValueError(f"{key}: expected string, got {tr_val!r}")
def merge_catalog(manifest: dict, lang: str, catalog: dict) -> dict:
if lang == "en":
raise ValueError("refusing to overwrite the en source")
for clip in manifest.get("clips", []):
en = (clip.get("strings") or {}).get("en")
tr = catalog.get(clip["id"])
if not en or not tr:
continue
for key, en_val in en.items():
if key not in tr:
raise ValueError(f"{clip['id']}/{key}: missing in {lang} catalog")
_check_shape(f"{clip['id']}/{key}", en_val, tr[key])
clip["strings"][lang] = {k: tr[k] for k in en} # en key order, lang values
return manifest
def _main(argv=None):
ap = argparse.ArgumentParser(description=__doc__)
ap.add_argument("command", choices=["extract", "merge"])
ap.add_argument("--manifest", default="simulator/sample_media/manifest.json")
ap.add_argument("--catalog-dir", default="tools/i18n/catalogs")
ap.add_argument("--lang", help="merge: target language code")
args = ap.parse_args(argv)
mpath = Path(args.manifest)
manifest = json.loads(mpath.read_text(encoding="utf-8"))
cdir = Path(args.catalog_dir)
if args.command == "extract":
cdir.mkdir(parents=True, exist_ok=True)
(cdir / "en.json").write_text(
json.dumps(extract_en_catalog(manifest), ensure_ascii=False, indent=1) + "\n",
encoding="utf-8",
)
print(f"wrote {cdir / 'en.json'}")
else:
if not args.lang:
ap.error("merge requires --lang")
catalog = json.loads((cdir / f"{args.lang}.json").read_text(encoding="utf-8"))
merge_catalog(manifest, args.lang, catalog)
mpath.write_text(json.dumps(manifest, ensure_ascii=False, indent=1) + "\n", encoding="utf-8")
print(f"merged {args.lang} into {mpath}")
if __name__ == "__main__":
_main()