Skip to content

:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.

feat(data-warehouse): add demodesk recording details and transcripts tables (#119237)

Tom Owers2a1228206aec
diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/COVERAGE_GAPS_APPENDIX.md b/products/warehouse_sources/backend/temporal/data_imports/sources/COVERAGE_GAPS_APPENDIX.md
index b2f317980bf8..e954181e87d0 100644
--- a/products/warehouse_sources/backend/temporal/data_imports/sources/COVERAGE_GAPS_APPENDIX.md
+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/COVERAGE_GAPS_APPENDIX.md
@@ -3663,12 +3663,12 @@ Note: DeepSource is a Relay-style GraphQL API at https://api.deepsource.com/grap
## Demodesk — gaps
-Today (6): `demo_templates`, `demos`, `recording_scorecards`, `recording_summaries`, `recordings`, `users`
+Today (8): `demo_templates`, `demos`, `recording_details`, `recording_scorecards`, `recording_summaries`, `recording_transcripts`, `recordings`, `users`
Diffed against: <https://demodesk.com/api/docs/v2/generated.yaml>
-- [ ] `Recording transcripts (GET /api/v2/recordings/{token}/transcript)` — analyze conversation content, speaker contributions, and topics across imported meetings (high)
-- [ ] `Recording details, participants, and statistics (GET /api/v2/recordings/{token})` — analyze participant attendance, engagement scores, and feedback scores absent from recording listings (medium)
+- [x] `Recording transcripts (GET /api/v2/recordings/{token}/transcript)` — analyze conversation content, speaker contributions, and topics across imported meetings (high)
+- [x] `Recording details, participants, and statistics (GET /api/v2/recordings/{token})` — analyze participant attendance, engagement scores, and feedback scores absent from recording listings (medium)
Note: Parsed the v2 OpenAPI paths and response schemas, then read the official legacy v1 reference. Existing tables cover meetings, templates, users, recordings, summaries, and scorecards. Missing content needs recording-token fan-out; transcript responses contain paragraph and sentence collections. Auth, webhooks, and writes were excluded; detailed participant and statistics objects were checked separately from list metadata.
diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/canonical_descriptions.py b/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/canonical_descriptions.py
index 12a6795ee3a5..6930a7062312 100644
--- a/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/canonical_descriptions.py
+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/canonical_descriptions.py
@@ -98,4 +98,39 @@ CANONICAL_DESCRIPTIONS: CanonicalDescriptions = {
"updatedAt": "When the scorecard was last updated.",
},
},
+ "recording_details": {
+ "description": "Detailed metadata for a recording, including the host, participants, attendance, and engagement and feedback statistics.",
+ "docs_url": "https://demodesk.com/api/docs/index.html",
+ "columns": {
+ "recordingId": "Internal recording id.",
+ "recordingToken": "Public recording token.",
+ "demoId": "Internal id of the recorded meeting.",
+ "userId": "Id of the meeting host user.",
+ "name": "Recording name.",
+ "status": "Recording status: pending, ready, failure, cancelled, or expired.",
+ "postprocessingStatus": "Post-processing status (transcription, statistics, summaries): pending, done, failed, or too_short.",
+ "access": "Access permission level: company_wide, selected_groups, host_and_participants, or host_only.",
+ "createdAt": "When the recording was created.",
+ "updatedAt": "When the recording was last updated.",
+ "demoStartDate": "Start time of the recorded meeting.",
+ "durationMs": "Recording duration in milliseconds.",
+ "host": "The meeting host, or null when no host is assigned.",
+ "audioOnly": "Whether the recording has audio only.",
+ "attendeeClassification": "Whether the meeting was internal or external.",
+ "meetingLocation": "Where the meeting took place.",
+ "participants": "Meeting participants, each with display name, email, permission, whether they attended, and CRM links.",
+ "groups": "Groups assigned to the meeting. Only set when access is selected_groups.",
+ "statistics": "Derived meeting statistics: engagement score (0 to 10) and feedback score.",
+ "crmLinks": "CRM records the meeting is linked to.",
+ },
+ },
+ "recording_transcripts": {
+ "description": "The transcript of a recording, grouped into speaker paragraphs and sentences.",
+ "docs_url": "https://demodesk.com/api/docs/index.html",
+ "columns": {
+ "recordingToken": "Token of the recording the transcript belongs to.",
+ "language": "Language of the transcript.",
+ "paragraphs": "Transcript paragraphs, each with start and end time in seconds, speaker, and sentences.",
+ },
+ },
}
diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/demodesk.py b/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/demodesk.py
index 4e4c8d4971e3..8c937aec60c9 100644
--- a/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/demodesk.py
+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/demodesk.py
@@ -42,7 +42,7 @@ class DemodeskResumeConfig:
# v1 endpoints checkpoint a page number, v2 endpoints an opaque cursor; one field is set.
page: int | None = None
cursor: str | None = None
- # `recording_summaries`/`recording_scorecards` fan out over `recordings`; this is the
+ # The `recording_*` endpoints fan out over `recordings`; this is the
# `build_dependent_resource` checkpoint (which parents finished, where the current one
# stopped), mutually exclusive with `page`/`cursor` above.
fanout_state: dict[str, Any] | None = None
@@ -116,9 +116,9 @@ def demodesk_client_config(api_key: str, api_version: DemodeskApiVersion) -> Cli
# `Authorization` is, so `allow_redirects=False` is the only thing that keeps either auth
# scheme's credential from following a spoofed 3xx off `BASE_URL`.
"allow_redirects": False,
- # `recording_summaries` bodies carry free-text AI meeting summaries and `recordings`
- # carries signed `temporaryDirectUrl` links; keep both out of the shared HTTP sample
- # store (still metered and logged) rather than trust the name-based scrubbers on
+ # `recording_summaries` and `recording_transcripts` bodies carry free-text meeting content
+ # and `recordings` carries signed `temporaryDirectUrl` links; keep them out of the shared
+ # HTTP sample store (still metered and logged) rather than trust the name-based scrubbers on
# unstructured content.
"capture": False,
}
diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/settings.py b/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/settings.py
index 1d8834eb4bff..5c1afe3796d2 100644
--- a/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/settings.py
+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/settings.py
@@ -115,6 +115,42 @@ DEMODESK_ENDPOINTS: dict[str, DemodeskEndpointConfig] = {
child_response_actions=[{"status_code": 404, "action": "ignore"}],
),
),
+ "recording_details": DemodeskEndpointConfig(
+ name="recording_details",
+ path="/v2/recordings/{recording_token}",
+ api_version="v2",
+ primary_keys=["recordingId"],
+ partition_key="createdAt",
+ # Full refresh only: CRM link changes on the recording or its participants do not advance
+ # `updatedAt`, so bounding the parent listing by that watermark would leave them stale.
+ fanout=DependentEndpointConfig(
+ parent_name="recordings",
+ resolve_param="recording_token",
+ resolve_field="recordingToken",
+ # The detail row already carries its own `recordingToken`.
+ include_from_parent=[],
+ child_response_actions=[{"status_code": 404, "action": "ignore"}],
+ ),
+ ),
+ "recording_transcripts": DemodeskEndpointConfig(
+ name="recording_transcripts",
+ path="/v2/recordings/{recording_token}/transcript",
+ api_version="v2",
+ # One transcript per recording. It carries no timestamp, so the table is full refresh only.
+ primary_keys=["recordingToken"],
+ fanout=DependentEndpointConfig(
+ parent_name="recordings",
+ resolve_param="recording_token",
+ resolve_field="recordingToken",
+ include_from_parent=["recordingToken"],
+ parent_field_renames={"recordingToken": "recordingToken"},
+ child_response_actions=[
+ {"status_code": 404, "action": "ignore"},
+ # 202 means the transcript is still being processed; the next sync picks it up.
+ {"status_code": 202, "action": "ignore"},
+ ],
+ ),
+ ),
}
ENDPOINTS = tuple(DEMODESK_ENDPOINTS.keys())
diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/tests/test_demodesk.py b/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/tests/test_demodesk.py
index cb2e036694c8..0497320c8ea4 100644
--- a/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/tests/test_demodesk.py
+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/demodesk/tests/test_demodesk.py
@@ -159,6 +159,37 @@ class TestDemodeskFanout:
assert result.paths == ["/api/v2/recordings", "/api/v2/recordings/tokB/summaries"]
assert [row["summaryId"] for row in result.rows] == ["sB"]
+ def test_recording_transcripts_skip_unready_transcripts(self) -> None:
+ result = _driver().run(
+ "recording_transcripts",
+ route(
+ {
+ "/v2/recordings": [
+ ScriptedResponse(
+ json={
+ "data": [
+ {"recordingToken": "tokA"},
+ {"recordingToken": "tokB"},
+ {"recordingToken": "tokC"},
+ ],
+ "meta": {"hasNext": False, "limit": 100},
+ }
+ )
+ ],
+ "/v2/recordings/tokA/transcript": [
+ ScriptedResponse(json={"data": {"language": "en", "paragraphs": []}})
+ ],
+ "/v2/recordings/tokB/transcript": [
+ ScriptedResponse(status=202, json={"error": {"message": "processing"}})
+ ],
+ "/v2/recordings/tokC/transcript": [ScriptedResponse(status=204)],
+ }
+ ),
+ )
+
+ assert result.raised is None
+ assert [(row["recordingToken"], row["language"]) for row in result.rows] == [("tokA", "en")]
+
class TestValidateCredentials:
@pytest.mark.parametrize(