pc translate sentences

sentence_api/repository.py

@@ -355,6 +355,98 @@ class VideoRepository:
                     ],
                 )
 
+    def import_local_subtitle_document(
+        self,
+        document: SentenceBoundaryDocument,
+    ) -> SentenceBoundaryDocument:
+        """Save locally generated English subtitles for translation.
+
+        Official course videos are left unchanged. A previous local-only
+        import is replaced when the English cues changed.
+        """
+        normalized_hash = normalize_video_hash(document.video_hash)
+        document = document.model_copy(update={"video_hash": normalized_hash})
+        existing_video = self.get_video(normalized_hash)
+        existing_document = self.get_document(normalized_hash)
+        if existing_video is not None and existing_video.get("status") != "local_subtitles":
+            if existing_document is None:
+                raise KeyError(normalized_hash)
+            return existing_document
+        if existing_document is not None and _english_fingerprint(existing_document) == _english_fingerprint(document):
+            return existing_document
+
+        now = utc_now()
+        language = next((sentence.language for sentence in document.sentences if sentence.language), None)
+        with self._connect() as connection:
+            if existing_video is None:
+                connection.execute(
+                    """
+                    INSERT INTO videos (
+                        video_hash, title, filename, stored_filename, content_type,
+                        size_bytes, duration_ms, language, status, error_message,
+                        algorithm_version, transcription, remote_url, created_at, updated_at
+                    ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, 'local_subtitles', NULL, ?, NULL, NULL, ?, ?)
+                    """,
+                    (
+                        normalized_hash,
+                        "Local subtitles",
+                        "local-subtitles.json",
+                        "",
+                        "application/json",
+                        0,
+                        document.duration_ms,
+                        language,
+                        document.algorithm_version,
+                        now,
+                        now,
+                    ),
+                )
+            else:
+                connection.execute(
+                    """
+                    UPDATE videos
+                    SET duration_ms = ?, language = COALESCE(?, language),
+                        algorithm_version = ?, status = 'local_subtitles',
+                        error_message = NULL, updated_at = ?
+                    WHERE video_hash = ?
+                    """,
+                    (
+                        document.duration_ms,
+                        language,
+                        document.algorithm_version,
+                        now,
+                        normalized_hash,
+                    ),
+                )
+            connection.execute(
+                "DELETE FROM sentences WHERE video_hash = ?",
+                (normalized_hash,),
+            )
+            connection.executemany(
+                """
+                INSERT INTO sentences (
+                    video_hash, sentence_index, start_ms, end_ms, text,
+                    language, zh_text, reference_speech_duration_ms
+                ) VALUES (?, ?, ?, ?, ?, ?, NULL, ?)
+                """,
+                [
+                    (
+                        normalized_hash,
+                        sentence.index,
+                        sentence.start_ms,
+                        sentence.end_ms,
+                        sentence.text,
+                        sentence.language,
+                        sentence.reference_speech_duration_ms,
+                    )
+                    for sentence in document.sentences
+                ],
+            )
+        imported = self.get_document(normalized_hash)
+        if imported is None:
+            raise KeyError(normalized_hash)
+        return imported
+
     def mark_oss_uploaded(self, video_hash: str, remote_url: str) -> None:
         with self._connect() as connection:
             connection.execute(
@@ -389,6 +481,7 @@ class VideoRepository:
                 SELECT v.*, COUNT(s.sentence_index) AS sentence_count
                 FROM videos v
                 LEFT JOIN sentences s ON s.video_hash = v.video_hash
+                WHERE v.status != 'local_subtitles'
                 GROUP BY v.video_hash
                 ORDER BY v.created_at DESC
                 """
@@ -1316,6 +1409,13 @@ class VideoRepository:
 
 
 
+def _english_fingerprint(document: SentenceBoundaryDocument) -> tuple:
+    return tuple(
+        (sentence.index, sentence.start_ms, sentence.end_ms, (sentence.text or "").strip())
+        for sentence in document.sentences
+    )
+
+
 def _sentence_from_row(row: sqlite3.Row) -> SentenceBoundary:
     keys = set(row.keys())
     zh_text = row["zh_text"] if "zh_text" in keys else None