add Moss result to db

sentence_api/processing.py

@@ -4,7 +4,7 @@ import subprocess
 import tempfile
 import uuid
 from pathlib import Path
-from typing import List, Optional
+from typing import Any, Dict, List, Optional
 
 from .audio_metrics import (
     AudioAnalysisError,
@@ -124,7 +124,11 @@ class VideoProcessor:
                 )
                 if not document.sentences:
                     raise RuntimeError("MOSS returned no timestamped speech segments.")
-                self._finish_processing(document, transcript.text)
+                self._finish_processing(
+                    document,
+                    transcript.text,
+                    moss_segments=_serialize_moss_segments(transcript),
+                )
             else:
                 entry, _ = make_entry(media_path, video_hash=video_hash)
                 document = SentenceBoundaryDocument(
@@ -145,13 +149,18 @@ class VideoProcessor:
         self,
         document: SentenceBoundaryDocument,
         transcription: Optional[str],
+        moss_segments: Optional[List[Dict[str, Any]]] = None,
     ) -> None:
         if self.oss_uploader is None or not self.oss_uploader.enabled:
-            self.repository.save_processing_result(document, transcription, ready=True)
+            self.repository.save_processing_result(
+                document, transcription, ready=True, moss_segments=moss_segments
+            )
             return
 
         # Keep the course invisible while the mandatory OSS upload is running.
-        self.repository.save_processing_result(document, transcription, ready=False)
+        self.repository.save_processing_result(
+            document, transcription, ready=False, moss_segments=moss_segments
+        )
         video = self.repository.get_video(document.video_hash)
         if video is None:
             raise ValueError(f"Unknown video: {document.video_hash}")
@@ -539,3 +548,32 @@ def _media_duration_ms(path: Path) -> int:
     from .generate_boundaries import media_duration_ms
 
     return media_duration_ms(path)
+
+
+def _serialize_moss_segments(transcript: Transcript) -> List[Dict[str, Any]]:
+    rows: List[Dict[str, Any]] = []
+    for segment in transcript.segments:
+        text = segment.text.strip()
+        if not text:
+            continue
+        words = None
+        if segment.words:
+            words = [
+                {
+                    "start_seconds": word.start_seconds,
+                    "end_seconds": word.end_seconds,
+                    "text": word.text,
+                }
+                for word in segment.words
+            ]
+        rows.append(
+            {
+                "start_seconds": segment.start_seconds,
+                "end_seconds": segment.end_seconds,
+                "text": text,
+                "speaker": segment.speaker,
+                "words": words,
+            }
+        )
+    return rows
+