add merging sentences action

sentence_api/tests/test_api.py

@@ -277,6 +277,102 @@ def test_admin_can_split_sentence_and_reindexes_following_sentences(tmp_path):
     assert [item.index for item in document.sentences] == [0, 1, 2, 3]
 
 
+def test_admin_can_merge_adjacent_sentences_and_reindexes_following_sentences(tmp_path):
+    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="Manual merge 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,
+    )
+
+    class UnavailableTranscriber:
+        available = False
+
+    client = TestClient(
+        create_app(
+            BoundaryStore(tmp_path / "boundaries.json"),
+            settings=settings,
+            repository=repository,
+            transcriber=UnavailableTranscriber(),
+        )
+    )
+    response = client.put(
+        f"/api/v1/admin/videos/{VIDEO_HASH}/sentences/1/merge",
+    )
+
+    assert response.status_code == 200
+    merged = response.json()
+    assert merged["index"] == 1
+    assert merged["text"] == "First part second part"
+    assert merged["start_ms"] == 2000
+    assert merged["end_ms"] == 9000
+    assert merged["reference_speech_duration_ms"] == 6400
+
+    document = repository.get_document(VIDEO_HASH)
+    assert [item.text for item in document.sentences] == [
+        "Before.",
+        "First part second part",
+        "Next sentence.",
+    ]
+    assert [item.index for item in document.sentences] == [0, 1, 2]
+    assert [item.start_ms for item in document.sentences] == [0, 2000, 9000]
+    assert [item.end_ms for item in document.sentences] == [2000, 9000, 12000]
+
+    last_response = client.put(
+        f"/api/v1/admin/videos/{VIDEO_HASH}/sentences/2/merge",
+    )
+    assert last_response.status_code == 400
+    assert last_response.json()["detail"] == "No next sentence to merge with."
+
+
 def test_admin_boundary_adjustment_updates_linked_next_start(tmp_path):
     settings = replace(Settings.from_env(), data_dir=tmp_path / "data")
     settings.ensure_directories()