Momento/memento-note/scripts/sync_locales_from_en.py

#!/usr/bin/env python3
"""
Synchronise locales/*.json avec locales/en.json (référence des clés).
- Garde les chaînes déjà présentes et non vides dans la locale cible.
- Traduit les clés manquantes (Google via deep-translator).
Usage : depuis memento-note/ avec le venv :
  .venv-i18n/bin/python scripts/sync_locales_from_en.py
"""
from __future__ import annotations

import json
import sys
import time
from pathlib import Path

from deep_translator import GoogleTranslator

ROOT = Path(__file__).resolve().parents[1]
LOCALES = ROOT / "locales"

# Codes Google Translate pour deep-translator (source=en)
LANG_TARGETS = {
    "fr": "fr",
    "es": "es",
    "de": "de",
    "fa": "fa",
    "it": "it",
    "pt": "pt",
    "ru": "ru",
    "zh": "zh-CN",
    "ja": "ja",
    "ko": "ko",
    "ar": "ar",
    "hi": "hi",
    "nl": "nl",
    "pl": "pl",
}


def flatten_leaves(obj: dict, prefix: str = "") -> dict[str, str]:
    out: dict[str, str] = {}
    for k, v in obj.items():
        path = f"{prefix}.{k}" if prefix else k
        if isinstance(v, dict):
            out.update(flatten_leaves(v, path))
        elif isinstance(v, str):
            out[path] = v
        else:
            raise TypeError(f"Valeur non supportée à {path}: {type(v)}")
    return out


def unflatten_leaves(flat: dict[str, str]) -> dict:
    root: dict = {}
    for path, val in flat.items():
        parts = path.split(".")
        cur = root
        for p in parts[:-1]:
            cur = cur.setdefault(p, {})
        cur[parts[-1]] = val
    return root


def translate_unique(texts: list[str], target_code: str, batch_size: int = 35) -> dict[str, str]:
    translator = GoogleTranslator(source="en", target=target_code)
    mapping: dict[str, str] = {}
    for i in range(0, len(texts), batch_size):
        batch = texts[i : i + batch_size]
        try:
            outs = translator.translate_batch(batch)
        except Exception as e:
            print(f"  batch erreur ({e}), traduction unitaire…", flush=True)
            outs = []
            for t in batch:
                try:
                    outs.append(translator.translate(t))
                except Exception:
                    outs.append(t)
                time.sleep(0.15)
        if len(outs) != len(batch):
            outs = batch  # fallback
        for src, dst in zip(batch, outs):
            mapping[src] = dst if isinstance(dst, str) else src
        time.sleep(0.6)
        print(f"  … {min(i + batch_size, len(texts))}/{len(texts)}", flush=True)
    return mapping


def merge_locale(en_flat: dict[str, str], loc_flat: dict[str, str], target_code: str) -> dict[str, str]:
    text_to_keys: dict[str, list[str]] = {}
    result: dict[str, str] = {}

    for key, en_val in en_flat.items():
        loc_val = loc_flat.get(key)
        if isinstance(loc_val, str) and loc_val.strip():
            result[key] = loc_val
        else:
            text_to_keys.setdefault(en_val, []).append(key)

    if not text_to_keys:
        return result

    unique = list(text_to_keys.keys())
    print(f"  {len(unique)} textes uniques à traduire ({sum(len(v) for v in text_to_keys.values())} clés)", flush=True)
    trans_map = translate_unique(unique, target_code)

    for src, keys in text_to_keys.items():
        tr = trans_map.get(src, src)
        for k in keys:
            result[k] = tr
    return result


def main() -> int:
    en_path = LOCALES / "en.json"
    if not en_path.exists():
        print("locales/en.json introuvable", file=sys.stderr)
        return 1

    en_obj = json.loads(en_path.read_text(encoding="utf-8"))
    en_flat = flatten_leaves(en_obj)
    print(f"Référence en.json : {len(en_flat)} clés feuilles", flush=True)

    skip = {"en"}
    for code, google_target in LANG_TARGETS.items():
        if code in skip:
            continue
        path = LOCALES / f"{code}.json"
        if not path.exists():
            print(f"Absence de {path.name}, ignoré", flush=True)
            continue

        loc_obj = json.loads(path.read_text(encoding="utf-8"))
        loc_flat = flatten_leaves(loc_obj)

        before_missing = sum(1 for k in en_flat if k not in loc_flat or not str(loc_flat.get(k, "")).strip())
        if before_missing == 0:
            print(f"\n=== {code}.json — déjà complet ({len(en_flat)} clés), rien à faire", flush=True)
            continue

        print(f"\n=== {code}.json ({google_target}) — manquantes avant: {before_missing}", flush=True)

        merged_flat = merge_locale(en_flat, loc_flat, google_target)

        # Couverture complète des clés en
        out_flat = dict(en_flat)
        out_flat.update(merged_flat)

        missing_after = sum(1 for k in en_flat if k not in out_flat or not str(out_flat[k]).strip())
        if missing_after:
            print(f"ERREUR: encore {missing_after} clés vides pour {code}", file=sys.stderr)
            return 1

        tree = unflatten_leaves(out_flat)
        path.write_text(json.dumps(tree, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
        print(f"  Écrit {path.name}", flush=True)

    print("\nTerminé.", flush=True)
    return 0


if __name__ == "__main__":
    raise SystemExit(main())