add a retranslate chinese function

sentence_api/repository.py

@@ -2,7 +2,7 @@ import json
 import sqlite3
 from datetime import datetime, timedelta, timezone
 from pathlib import Path
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List, Optional, Sequence, Tuple
 import uuid
 
 from .auth import hash_password, hash_token, new_token, verify_password
@@ -202,6 +202,14 @@ class VideoRepository:
             if "zh_text" not in sentence_columns:
                 connection.execute("ALTER TABLE sentences ADD COLUMN zh_text TEXT")
 
+        with self._connect() as connection:
+            video_columns = {
+                row["name"]
+                for row in connection.execute("PRAGMA table_info(videos)")
+            }
+            if "translation_glossary" not in video_columns:
+                connection.execute("ALTER TABLE videos ADD COLUMN translation_glossary TEXT")
+
     def upsert_upload(
         self,
         *,
@@ -833,6 +841,36 @@ class VideoRepository:
 
         return [_sentence_from_row(row) for row in result_rows]
 
+    def get_translation_glossary(self, video_hash: str) -> List[Tuple[str, str]]:
+        video = self.get_video(video_hash)
+        if video is None:
+            return []
+        return _parse_translation_glossary(video.get("translation_glossary"))
+
+    def save_translation_glossary(
+        self,
+        video_hash: str,
+        terms: Sequence[Any],
+    ) -> bool:
+        normalized_hash = normalize_video_hash(video_hash)
+        payload = json.dumps(
+            [
+                {"source": source, "zh": zh}
+                for source, zh in _normalize_translation_glossary(terms)
+            ],
+            ensure_ascii=False,
+        )
+        with self._connect() as connection:
+            cursor = connection.execute(
+                """
+                UPDATE videos
+                SET translation_glossary = ?, updated_at = ?
+                WHERE video_hash = ?
+                """,
+                (payload, utc_now(), normalized_hash),
+            )
+            return cursor.rowcount > 0
+
     def clear_sentence_translations(self, video_hash: str) -> None:
         with self._connect() as connection:
             connection.execute(
@@ -1345,3 +1383,43 @@ def _moss_segment_row(video_hash: str, index: int, segment: Dict[str, Any]) -> t
         str(speaker) if speaker is not None else None,
         words_json,
     )
+
+
+def _normalize_translation_glossary(terms: Sequence[Any]) -> List[Tuple[str, str]]:
+    cleaned: List[Tuple[str, str]] = []
+    seen: Dict[str, int] = {}
+    for item in terms or []:
+        source = ""
+        zh = ""
+        if isinstance(item, dict):
+            source = str(item.get("source") or "")
+            zh = str(item.get("zh") or "")
+        elif isinstance(item, (tuple, list)) and len(item) >= 2:
+            source = str(item[0] or "")
+            zh = str(item[1] or "")
+        source = source.strip()[:120]
+        zh = zh.strip()[:120]
+        if not source or not zh:
+            continue
+        key = source.casefold()
+        if key in seen:
+            cleaned[seen[key]] = (source, zh)
+            continue
+        seen[key] = len(cleaned)
+        cleaned.append((source, zh))
+        if len(cleaned) >= 50:
+            break
+    return cleaned
+
+
+def _parse_translation_glossary(raw: Any) -> List[Tuple[str, str]]:
+    if not raw:
+        return []
+    if isinstance(raw, str):
+        try:
+            raw = json.loads(raw)
+        except json.JSONDecodeError:
+            return []
+    if not isinstance(raw, list):
+        return []
+    return _normalize_translation_glossary(raw)