更新

app/tests/test_mock_repository.py

@@ -232,6 +232,69 @@ def test_demo_call_lifecycle_mutates_record(repository: DemoDoctorRepository) ->
     assert ended["room_id"] == "room-501"
 
 
+def test_demo_transcript_upsert_and_finish_round_trip_in_call_records(
+    repository: DemoDoctorRepository,
+) -> None:
+    """Demo replay reads expose one finalized segment for a repeated segment ID."""
+
+    started = repository.start_call(501, 301)
+    call_record_id = started["id"]
+    repository.start_call_transcription(501, call_record_id, "session-1")
+    repository.upsert_call_transcript_segments(
+        501,
+        call_record_id,
+        "session-1",
+        [
+            {
+                "segment_id": "seg-1",
+                "speaker_user_id": "patient_301",
+                "speaker_role": "patient",
+                "timestamp": 1200,
+                "text": "draft words",
+            }
+        ],
+    )
+    repository.upsert_call_transcript_segments(
+        501,
+        call_record_id,
+        "session-1",
+        [
+            {
+                "segment_id": "seg-1",
+                "speaker_user_id": "patient_301",
+                "speaker_role": "patient",
+                "timestamp": 1200,
+                "text": "final words",
+            }
+        ],
+    )
+    repository.finish_call_transcription(
+        501,
+        call_record_id,
+        "session-1",
+        expected_segment_count=1,
+        status="completed",
+    )
+    repository.end_call(501)
+
+    record = next(
+        row for row in repository.list_call_records(501) if row["id"] == call_record_id
+    )
+    assert record["status"] == 2
+    assert record["transcription_status"] == "completed"
+    assert record["transcription_segment_count"] == 1
+    assert record["transcript_segments"] == [
+        {
+            "segment_id": "seg-1",
+            "speaker_role": "patient",
+            "speaker_user_id": "patient_301",
+            "timestamp": 1200,
+            "text": "final words",
+        }
+    ]
+    assert "final words" in record["transcript_text"]
+
+
 def test_tolerant_page_parsing_accepts_aliases_and_bad_rows() -> None:
     """List parsing handles nullable fields, aliases and non-object rows safely."""
 
@@ -315,6 +378,8 @@ class _StubApiClient:
                 "userSig": "short-lived",
                 "patientUserId": "patient_2",
             }
+        if endpoint == "tcm.diagnosis/startCall":
+            return {"call_record_id": 901}
         return {"ok": True}
 
 

app/tests/test_repository_parity.py

@@ -7,6 +7,7 @@ from typing import Any
 
 import pytest
 
+from doctor_workstation.core.errors import ApiProtocolError
 from doctor_workstation.core.models import Appointment, Consultation, PageResult, Prescription
 from doctor_workstation.services.mock_repository import DemoDoctorRepository
 from doctor_workstation.services.repository import (
@@ -76,6 +77,8 @@ class RecordingClient:
         self.post_calls.append((endpoint, body))
         if endpoint in {"tcm.prescription/add", "tcm.prescriptionOrder/create"}:
             return {"id": 88}
+        if endpoint == "tcm.diagnosis/startCall":
+            return {"call_record_id": 901}
         return {"ok": True}
 
 
@@ -253,6 +256,106 @@ def test_remote_new_contracts_use_exact_admin_endpoints_and_dtos() -> None:
     } <= get_endpoints
 
 
+def test_remote_transcription_endpoints_use_exact_normalized_dtos() -> None:
+    """Realtime transcript persistence stays within the three audited POST DTOs."""
+
+    client = RecordingClient()
+    repository = RemoteDoctorRepository(client)  # type: ignore[arg-type]
+
+    repository.start_call_transcription(501, 901, " session-1 ", language=" zh-CN ")
+    repository.upsert_call_transcript_segments(
+        501,
+        901,
+        "session-1",
+        [
+            {
+                "segment_id": "seg-1",
+                "speaker_user_id": "patient_301",
+                "speaker_role": "patient",
+                "timestamp": "1200",
+                "text": " patient words ",
+            }
+        ],
+    )
+    repository.finish_call_transcription(
+        501,
+        901,
+        "session-1",
+        expected_segment_count=1,
+        status="completed",
+    )
+
+    assert client.post_calls == [
+        (
+            "tcm.diagnosis/startCallTranscription",
+            {
+                "diagnosis_id": 501,
+                "call_record_id": 901,
+                "transcription_session_id": "session-1",
+                "language": "zh-CN",
+            },
+        ),
+        (
+            "tcm.diagnosis/upsertCallTranscriptSegments",
+            {
+                "diagnosis_id": 501,
+                "call_record_id": 901,
+                "transcription_session_id": "session-1",
+                "segments": [
+                    {
+                        "segment_id": "seg-1",
+                        "speaker_user_id": "patient_301",
+                        "speaker_role": "patient",
+                        "timestamp": 1200,
+                        "text": "patient words",
+                    }
+                ],
+            },
+        ),
+        (
+            "tcm.diagnosis/finishCallTranscription",
+            {
+                "diagnosis_id": 501,
+                "call_record_id": 901,
+                "transcription_session_id": "session-1",
+                "expected_segment_count": 1,
+                "status": "completed",
+            },
+        ),
+    ]
+
+
+def test_remote_start_call_requires_and_normalizes_current_record_id() -> None:
+    client = RecordingClient()
+    repository = RemoteDoctorRepository(client)  # type: ignore[arg-type]
+
+    assert repository.start_call(501, 301) == {"call_record_id": 901}
+    assert client.post_calls == [
+        (
+            "tcm.diagnosis/startCall",
+            {"diagnosis_id": 501, "patient_id": 301, "call_type": 2},
+        )
+    ]
+
+
+@pytest.mark.parametrize(
+    "response",
+    [None, {}, {"ok": True}, {"call_record_id": 0}, {"callRecordId": -1}, {"id": True}],
+)
+def test_remote_start_call_rejects_missing_or_invalid_record_id(response: Any) -> None:
+    class StartCallClient(RecordingClient):
+        def post(self, endpoint: str, payload: dict[str, Any] | None = None) -> Any:
+            if endpoint == "tcm.diagnosis/startCall":
+                self.post_calls.append((endpoint, dict(payload or {})))
+                return response
+            return super().post(endpoint, payload)
+
+    repository = RemoteDoctorRepository(StartCallClient())  # type: ignore[arg-type]
+
+    with pytest.raises(ApiProtocolError, match="call_record"):
+        repository.start_call(501, 301)
+
+
 @pytest.mark.parametrize(
     "unsafe_reference",
     [r"C:\records\tongue.jpg", r"\\server\share\report.pdf", "file:///tmp/a.jpg"],

app/tests/test_video_contract.py

@@ -188,10 +188,13 @@ def test_call_lifecycle_is_fifo_daemon_and_never_blocks_caller() -> None:
     release_start = threading.Event()
 
     class Repository:
-        def start_call(self, diagnosis_id: int, patient_id: int, *, call_type: int) -> None:
+        def start_call(
+            self, diagnosis_id: int, patient_id: int, *, call_type: int
+        ) -> dict[str, int]:
             start_entered.set()
             assert release_start.wait(2)
             events.append(("start", diagnosis_id, patient_id, call_type))
+            return {"call_record_id": 900}
 
         def bind_call_room(self, diagnosis_id: int, room_id: str) -> None:
             events.append(("bind", diagnosis_id, room_id))
@@ -237,6 +240,208 @@ def test_call_lifecycle_is_fifo_daemon_and_never_blocks_caller() -> None:
     ]
 
 
+def test_transcription_lifecycle_uses_start_record_id_and_remains_fifo() -> None:
+    events: list[tuple[object, ...]] = []
+
+    class Repository:
+        def start_call(
+            self, diagnosis_id: int, patient_id: int, *, call_type: int
+        ) -> dict[str, object]:
+            events.append(("start", diagnosis_id, patient_id, call_type))
+            return {"data": {"callRecordId": 901}}
+
+        def start_call_transcription(
+            self,
+            diagnosis_id: int,
+            call_record_id: int,
+            transcription_session_id: str,
+            *,
+            language: str,
+        ) -> None:
+            events.append(
+                (
+                    "transcription-start",
+                    diagnosis_id,
+                    call_record_id,
+                    transcription_session_id,
+                    language,
+                )
+            )
+
+        def upsert_call_transcript_segments(
+            self,
+            diagnosis_id: int,
+            call_record_id: int,
+            transcription_session_id: str,
+            segments: list[dict[str, object]],
+        ) -> None:
+            events.append(
+                (
+                    "segment",
+                    diagnosis_id,
+                    call_record_id,
+                    transcription_session_id,
+                    segments,
+                )
+            )
+
+        def finish_call_transcription(
+            self,
+            diagnosis_id: int,
+            call_record_id: int,
+            transcription_session_id: str,
+            expected_segment_count: int,
+            *,
+            status: str,
+        ) -> None:
+            events.append(
+                (
+                    "finish",
+                    diagnosis_id,
+                    call_record_id,
+                    transcription_session_id,
+                    expected_segment_count,
+                    status,
+                )
+            )
+
+        def end_call(self, diagnosis_id: int) -> None:
+            events.append(("end", diagnosis_id))
+
+    request = VideoCallRequest(
+        sdk_app_id=1400123456,
+        user_id="doctor_42",
+        user_sig="short-lived-ticket",
+        target_user_id="patient_8",
+        diagnosis_id=123,
+        patient_id=8,
+    )
+    lifecycle = OrderedCallLifecycle(request, Repository(), logging.getLogger(__name__))
+
+    start = lifecycle.start()
+    transcription_start = lifecycle.start_transcription("session-1")
+    segment = lifecycle.save_transcript_segment(
+        {
+            "segment_id": "seg-1",
+            "speaker_user_id": "patient_8",
+            "speaker_role": "patient",
+            "timestamp": 1200,
+            "text": "patient words",
+        }
+    )
+    duplicate = lifecycle.save_transcript_segment(
+        {"segment_id": "seg-1", "text": "must not produce another write"}
+    )
+    finish = lifecycle.finish_transcription(status="completed")
+    end = lifecycle.end("test")
+
+    assert duplicate is segment
+    assert start.result(timeout=2) is True
+    assert transcription_start.result(timeout=2) is True
+    assert segment.result(timeout=2) is True
+    assert finish.result(timeout=2) is True
+    assert end.result(timeout=2) is True
+    assert lifecycle.wait(1) is True
+    assert lifecycle.call_record_id == 901
+    assert events == [
+        ("start", 123, 8, 2),
+        ("transcription-start", 123, 901, "session-1", "zh-CN"),
+        (
+            "segment",
+            123,
+            901,
+            "session-1",
+            [
+                {
+                    "segment_id": "seg-1",
+                    "transcription_session_id": "session-1",
+                    "speaker_user_id": "patient_8",
+                    "speaker_role": "patient",
+                    "timestamp": 1200,
+                    "text": "patient words",
+                }
+            ],
+        ),
+        ("finish", 123, 901, "session-1", 1, "completed"),
+        ("end", 123),
+    ]
+
+
+def test_end_auto_finishes_active_transcription_as_partial() -> None:
+    events: list[tuple[object, ...]] = []
+
+    class Repository:
+        def start_call(
+            self, diagnosis_id: int, patient_id: int, *, call_type: int
+        ) -> dict[str, int]:
+            events.append(("start", diagnosis_id, patient_id, call_type))
+            return {"call_record_id": 902}
+
+        def start_call_transcription(
+            self,
+            diagnosis_id: int,
+            call_record_id: int,
+            transcription_session_id: str,
+            *,
+            language: str,
+        ) -> None:
+            events.append(
+                (
+                    "transcription-start",
+                    diagnosis_id,
+                    call_record_id,
+                    transcription_session_id,
+                    language,
+                )
+            )
+
+        def finish_call_transcription(
+            self,
+            diagnosis_id: int,
+            call_record_id: int,
+            transcription_session_id: str,
+            expected_segment_count: int,
+            *,
+            status: str,
+        ) -> None:
+            events.append(
+                (
+                    "finish",
+                    diagnosis_id,
+                    call_record_id,
+                    transcription_session_id,
+                    expected_segment_count,
+                    status,
+                )
+            )
+
+        def end_call(self, diagnosis_id: int) -> None:
+            events.append(("end", diagnosis_id))
+
+    request = VideoCallRequest(
+        sdk_app_id=1400123456,
+        user_id="doctor_42",
+        user_sig="short-lived-ticket",
+        target_user_id="patient_8",
+        diagnosis_id=123,
+        patient_id=8,
+    )
+    lifecycle = OrderedCallLifecycle(request, Repository(), logging.getLogger(__name__))
+
+    lifecycle.start()
+    lifecycle.start_transcription("session-auto-partial")
+    ended = lifecycle.end("window-closed")
+
+    assert ended.result(timeout=2) is True
+    assert lifecycle.wait(1) is True
+    assert events == [
+        ("start", 123, 8, 2),
+        ("transcription-start", 123, 902, "session-auto-partial", "zh-CN"),
+        ("finish", 123, 902, "session-auto-partial", 0, "partial"),
+        ("end", 123),
+    ]
+
+
 def test_failed_start_prevents_bind_and_end_writes() -> None:
     events: list[str] = []
 
@@ -278,8 +483,9 @@ def test_video_screenshot_is_uploaded_and_appended_to_patient_tongue_images() ->
     events: list[tuple[object, ...]] = []
 
     class Repository:
-        def start_call(self, diagnosis_id: int, *, call_type: int) -> None:
+        def start_call(self, diagnosis_id: int, *, call_type: int) -> dict[str, int]:
             events.append(("start", diagnosis_id, call_type))
+            return {"call_record_id": 903}
 
         def upload_material_bytes(
             self,
@@ -325,6 +531,28 @@ def test_video_screenshot_is_uploaded_and_appended_to_patient_tongue_images() ->
     ]
 
 
+def test_start_rejects_missing_record_id_without_using_ticket_fallback() -> None:
+    class Repository:
+        def start_call(self, diagnosis_id: int, *, call_type: int) -> dict[str, object]:
+            del diagnosis_id, call_type
+            return {}
+
+    request = VideoCallRequest(
+        sdk_app_id=1400123456,
+        user_id="doctor_42",
+        user_sig="short-lived-ticket",
+        target_user_id="patient_8",
+        diagnosis_id=123,
+        call_record_id=77,
+    )
+    lifecycle = OrderedCallLifecycle(request, Repository(), logging.getLogger(__name__))
+
+    with pytest.raises(ValueError, match="startCall response did not include"):
+        lifecycle.start().result(timeout=2)
+
+    assert lifecycle.started is False
+
+
 def test_https_document_policy_is_exact_and_origin_scoped() -> None:
     policy = TrustedDocumentPolicy.from_url(
         "https://RTC.Example.com/doctor-call/index.html?tenant=a#boot",