later process

sentence_api/tests/test_api.py

@@ -790,6 +790,7 @@ def test_admin_can_merge_adjacent_sentences_and_reindexes_following_sentences(tm
         VIDEO_HASH,
         {0: "之前。", 1: "第一段", 2: "第二段", 3: "下一句。"},
     )
+    (settings.videos_dir / f"{VIDEO_HASH}.mp4").write_bytes(b"video")
 
     class UnavailableTranscriber:
         available = False
@@ -833,6 +834,108 @@ def test_admin_can_merge_adjacent_sentences_and_reindexes_following_sentences(tm
 
 
 
+def test_admin_batch_applies_merge_and_split_once(tmp_path, monkeypatch):
+    settings = replace(Settings.from_env(), data_dir=tmp_path / "data")
+    settings.ensure_directories()
+    repository = VideoRepository(settings.database_path)
+    repository.upsert_upload(
+        video_hash=VIDEO_HASH,
+        title="Batch edit lesson",
+        filename="lesson.mp4",
+        stored_filename=f"{VIDEO_HASH}.mp4",
+        content_type="video/mp4",
+        size_bytes=5,
+        language="en",
+    )
+    repository.save_processing_result(
+        SentenceBoundaryDocument(
+            video_hash=VIDEO_HASH,
+            duration_ms=12000,
+            algorithm_version="test",
+            sentences=[
+                SentenceBoundary(index=0, start_ms=0, end_ms=2000, text="Before.", language="en", reference_speech_duration_ms=1800),
+                SentenceBoundary(index=1, start_ms=2000, end_ms=5000, text="First part", language="en", reference_speech_duration_ms=2800),
+                SentenceBoundary(index=2, start_ms=5000, end_ms=9000, text="second part", language="en", reference_speech_duration_ms=3600),
+                SentenceBoundary(index=3, start_ms=9000, end_ms=12000, text="Next sentence.", language="en", reference_speech_duration_ms=2800),
+            ],
+        ),
+        None,
+    )
+    repository.save_sentence_translations(
+        VIDEO_HASH,
+        {0: "之前。", 1: "第一段", 2: "第二段", 3: "下一句。"},
+    )
+    (settings.videos_dir / f"{VIDEO_HASH}.mp4").write_bytes(b"video")
+
+    calls = []
+    from sentence_api.transcription import Transcript, TranscriptionSegment, WordTimestamp
+
+    class OnceTranscriber:
+        available = True
+
+        def transcribe(self, audio_path, language=None):
+            calls.append(audio_path)
+            return Transcript(
+                text="Next sentence.",
+                segments=[
+                    TranscriptionSegment(
+                        start_seconds=0.0,
+                        end_seconds=1.0,
+                        text="Next sentence.",
+                        words=[
+                            WordTimestamp(0.0, 0.2, "Next"),
+                            WordTimestamp(0.2, 1.0, "sentence."),
+                        ],
+                    )
+                ],
+            )
+
+    monkeypatch.setattr(
+        "sentence_api.processing.extract_audio_segment",
+        lambda media_path, output_path, *, start_ms, duration_ms: output_path.write_bytes(b"audio"),
+    )
+    client = TestClient(
+        create_app(
+            BoundaryStore(tmp_path / "boundaries.json"),
+            settings=settings,
+            repository=repository,
+            transcriber=OnceTranscriber(),
+        )
+    )
+    response = client.put(
+        f"/api/v1/admin/videos/{VIDEO_HASH}/sentences/batch",
+        json={
+            "sentences": [
+                {"start_ms": 0, "end_ms": 2000, "text": "Before.", "language": "en", "reference_speech_duration_ms": 1800},
+                {"start_ms": 2000, "end_ms": 9000, "text": "First part second part", "language": "en", "reference_speech_duration_ms": 6400},
+                {"start_ms": 9000, "end_ms": 9200, "text": "Next", "language": "en", "reference_speech_duration_ms": 1000, "source_start_ms": 9000, "source_end_ms": 12000},
+                {"start_ms": 9200, "end_ms": 12000, "text": "sentence.", "language": "en", "reference_speech_duration_ms": 1000, "source_start_ms": 9000, "source_end_ms": 12000},
+            ]
+        },
+    )
+
+    assert response.status_code == 200, response.text
+    payload = response.json()
+    assert [item["text"] for item in payload] == [
+        "Before.",
+        "First part second part",
+        "Next",
+        "sentence.",
+    ]
+    assert [item["start_ms"] for item in payload] == [0, 2000, 9000, 9600]
+    assert [item["end_ms"] for item in payload] == [2000, 9000, 9600, 12000]
+    assert [item["reference_speech_duration_ms"] for item in payload] == [1800, 6400, 862, 1938]
+    assert len(calls) == 1
+    document = repository.get_document(VIDEO_HASH)
+    assert [item.text for item in document.sentences] == [
+        "Before.",
+        "First part second part",
+        "Next",
+        "sentence.",
+    ]
+    assert all(not (item.zh_text or "").strip() for item in document.sentences)
+
+
 def test_admin_sentence_merge_restarts_subtitle_translation(tmp_path, monkeypatch):
     client = make_client(tmp_path)
     _ready_video(client.app.state.video_repository)