wip

Fix broken speaker assignement tests following #1522.

CHANGELOG.md

@@ -18,6 +18,7 @@ and this project adheres to
 - ⬆️(frontend) upgrade i18next from 26.3.1 to 26.3.2
 - ⬆️(frontend) upgrade posthog-js from 1.391.2 to 1.395.0
 - ⬆️(frontend) upgrade @tanstack/react-query from 5.101.0 to 5.101.1
+- ⬆️(frontend) upgrade livekit-client from 2.19.2 to 2.20.0
 
 ## [1.24.0] - 2026-07-21
 

src/backend/core/utils.py

@@ -423,6 +423,31 @@ def detect_mimetype(file_buffer: bytes, filename: str | None = None) -> str:
     return mimetype_from_content or "application/octet-stream"
 
 
+def _get_s3_client(*, override_domain: bool = True):
+    """Return an S3 client, honoring the AWS_S3_DOMAIN_REPLACE endpoint override.
+
+    AWS_S3_DOMAIN_REPLACE is used when the backend and frontend reach object
+    storage under different domains (this is the case in the docker compose stack
+    used in development: the frontend connects to the object storage on localhost
+    while the backend uses the object storage service name declared in the stack).
+    The domain name is used to compute the signature, so it can't be changed
+    dynamically by the frontend; we build a dedicated boto3 client pointed at that
+    endpoint. Otherwise we reuse the default storage client.
+    """
+    if settings.AWS_S3_DOMAIN_REPLACE and override_domain:
+        return boto3.client(
+            "s3",
+            aws_access_key_id=settings.AWS_S3_ACCESS_KEY_ID,
+            aws_secret_access_key=settings.AWS_S3_SECRET_ACCESS_KEY,
+            endpoint_url=settings.AWS_S3_DOMAIN_REPLACE,
+            config=botocore.client.Config(
+                region_name=settings.AWS_S3_REGION_NAME,
+                signature_version=settings.AWS_S3_SIGNATURE_VERSION,
+            ),
+        )
+    return default_storage.connection.meta.client
+
+
 def generate_upload_policy(file):
     """
     Generate a S3 upload policy for a given file.
@@ -433,26 +458,7 @@ def generate_upload_policy(file):
 
     key = file.temporary_file_key
 
-    # This settings should be used if the backend application and the frontend application
-    # can't connect to the object storage with the same domain. This is the case in the
-    # docker compose stack used in development. The frontend application will use localhost
-    # to connect to the object storage while the backend application will use the object storage
-    # service name declared in the docker compose stack.
-    # This is needed because the domain name is used to compute the signature. So it can't be
-    # changed dynamically by the frontend application.
-    if settings.AWS_S3_DOMAIN_REPLACE:
-        s3_client = boto3.client(
-            "s3",
-            aws_access_key_id=settings.AWS_S3_ACCESS_KEY_ID,
-            aws_secret_access_key=settings.AWS_S3_SECRET_ACCESS_KEY,
-            endpoint_url=settings.AWS_S3_DOMAIN_REPLACE,
-            config=botocore.client.Config(
-                region_name=settings.AWS_S3_REGION_NAME,
-                signature_version=settings.AWS_S3_SIGNATURE_VERSION,
-            ),
-        )
-    else:
-        s3_client = default_storage.connection.meta.client
+    s3_client = _get_s3_client()
 
     # Generate the policy
     policy = s3_client.generate_presigned_url(
@@ -473,26 +479,7 @@ def generate_download_s3_url(
     if not key:
         raise ValueError("key cannot be empty")
 
-    # This setting should be used if the backend application and the frontend application
-    # can't connect to the object storage with the same domain. This is the case in the
-    # docker compose stack used in development. The frontend application will use localhost
-    # to connect to the object storage while the backend application will use the object storage
-    # service name declared in the docker compose stack.
-    # This is needed because the domain name is used to compute the signature. So it can't be
-    # changed dynamically by the frontend application.
-    if settings.AWS_S3_DOMAIN_REPLACE and override_domain:
-        s3_client = boto3.client(
-            "s3",
-            aws_access_key_id=settings.AWS_S3_ACCESS_KEY_ID,
-            aws_secret_access_key=settings.AWS_S3_SECRET_ACCESS_KEY,
-            endpoint_url=settings.AWS_S3_DOMAIN_REPLACE,
-            config=botocore.client.Config(
-                region_name=settings.AWS_S3_REGION_NAME,
-                signature_version=settings.AWS_S3_SIGNATURE_VERSION,
-            ),
-        )
-    else:
-        s3_client = default_storage.connection.meta.client
+    s3_client = _get_s3_client(override_domain=override_domain)
 
     return s3_client.generate_presigned_url(
         ClientMethod="get_object",

src/frontend/package-lock.json

@@ -29,7 +29,7 @@
         "i18next-browser-languagedetector": "8.2.1",
         "i18next-parser": "9.4.0",
         "i18next-resources-to-backend": "1.2.1",
-        "livekit-client": "2.19.2",
+        "livekit-client": "2.20.0",
         "posthog-js": "1.395.0",
         "react": "18.3.1",
         "react-aria": "3.50.0",
@@ -1028,9 +1028,9 @@
       "license": "Apache-2.0"
     },
     "node_modules/@livekit/protocol": {
-      "version": "1.45.8",
-      "resolved": "https://registry.npmjs.org/@livekit/protocol/-/protocol-1.45.8.tgz",
-      "integrity": "sha512-Q+l57E7w/xxOBFVWzdX5rkAZO7ffyF+rlDzNUYq2SU114+5aTyCq+PK4unaEVDNd4952Af7wteKr3sOgasGuaA==",
+      "version": "1.46.6",
+      "resolved": "https://registry.npmjs.org/@livekit/protocol/-/protocol-1.46.6.tgz",
+      "integrity": "sha512-upzlHP1vi/kZ/QqALZTFskQ0ifqc2f15RKucHYOsIHJsaXvEYanG75mAb7o+Yomfs4XhQ4BaRsdY+TFHXpaqrg==",
       "license": "Apache-2.0",
       "dependencies": {
         "@bufbuild/protobuf": "^1.10.0"
@@ -8125,13 +8125,13 @@
       "license": "MIT"
     },
     "node_modules/livekit-client": {
-      "version": "2.19.2",
-      "resolved": "https://registry.npmjs.org/livekit-client/-/livekit-client-2.19.2.tgz",
-      "integrity": "sha512-Kvk07QYDWRAbmYNLRll04ZIuxMQobW/oLPYnmR1kCy8GGHpU0gqyHf704Rz+29zfy8IJZRjKqeVbzGSKn9sumw==",
+      "version": "2.20.0",
+      "resolved": "https://registry.npmjs.org/livekit-client/-/livekit-client-2.20.0.tgz",
+      "integrity": "sha512-RIJcpvBmOmwz3jTj3rmdY6Dzr55HrhcaJjMgY+HSmoEM+yIRyA40m7r8UKv0hnZWM3z/AYhP1q8C8ciz5UWFKQ==",
       "license": "Apache-2.0",
       "dependencies": {
         "@livekit/mutex": "1.1.1",
-        "@livekit/protocol": "1.45.8",
+        "@livekit/protocol": "1.46.6",
         "events": "^3.3.0",
         "jose": "^6.1.0",
         "loglevel": "^1.9.2",

src/frontend/package.json

@@ -36,7 +36,7 @@
     "i18next-browser-languagedetector": "8.2.1",
     "i18next-parser": "9.4.0",
     "i18next-resources-to-backend": "1.2.1",
-    "livekit-client": "2.19.2",
+    "livekit-client": "2.20.0",
     "posthog-js": "1.395.0",
     "react": "18.3.1",
     "react-aria": "3.50.0",

src/summary/tests/unit/test_user_assign.py

@@ -1,7 +1,6 @@
 """Tests for the speaker-to-user assignment service."""
 
 import math
-from dataclasses import dataclass, field
 from datetime import datetime
 
 from summary.core import user_assign
@@ -17,11 +16,9 @@ from summary.core.user_assign import (
 )
 
 
-@dataclass
-class FakeTranscription:
-    """Mimics the OpenAI Transcription pydantic model for testing."""
-
-    segments: list = field(default_factory=list)
+def make_transcription(segments: list | None = None) -> dict:
+    """Build a WhisperX transcription dict, matching WhisperXResponse.model_dump()."""
+    return {"segments": segments or []}
 
 
 RECORDING_START = datetime.fromisoformat("2026-03-17T15:30:33.000001")
@@ -68,7 +65,7 @@ METADATA_SINGLE_USER = {
     ],
 }
 
-DIARIZATION_SINGLE_SPEAKER = FakeTranscription(
+DIARIZATION_SINGLE_SPEAKER = make_transcription(
     segments=[
         {
             "start": 1.363,
@@ -182,7 +179,7 @@ class TestBuildSpeakerTimelines:
 
     def test_segment_without_words_falls_back_to_segment_bounds(self):
         """Segments missing a `words` key use the segment start/end as one interval."""
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[{"start": 1.5, "end": 3.5, "speaker": "SPEAKER_00"}],
         )
         result = _build_speaker_timelines(transcription)
@@ -190,7 +187,7 @@ class TestBuildSpeakerTimelines:
 
     def test_segment_with_only_none_word_timestamps_falls_back(self):
         """If every word has None start/end, fall back to segment bounds."""
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {
                     "start": 1.0,
@@ -208,7 +205,7 @@ class TestBuildSpeakerTimelines:
 
     def test_short_words_only_uses_segment_start_and_last_word_end(self):
         """With no overly long words, the interval runs segment start to end."""
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {
                     "start": 1.0,
@@ -231,7 +228,7 @@ class TestBuildSpeakerTimelines:
         max_word_duration = (
             user_assign.settings.resolve_speaker_identities_max_word_duration
         )
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {
                     "start": 0.0,
@@ -252,7 +249,7 @@ class TestBuildSpeakerTimelines:
 
     def test_long_word_in_middle_splits_segment(self):
         """Short words around a long word produce two intervals (before-cap + after)."""
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {
                     "start": 0.0,
@@ -275,7 +272,7 @@ class TestBuildSpeakerTimelines:
 
     def test_tail_word_is_capped_at_max_duration(self):
         """The trailing word's end is capped at word.start + max_word_duration."""
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {
                     "start": 0.0,
@@ -302,7 +299,7 @@ class TestBuildSpeakerTimelines:
                 update={"resolve_speaker_identities_enable_split_on_words": False},
             ),
         )
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {
                     "start": 0.0,
@@ -322,7 +319,7 @@ class TestBuildSpeakerTimelines:
 
     def test_multiple_speakers_keep_separate_timelines(self):
         """Segments from different speakers populate independent timeline entries."""
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {
                     "start": 0.0,
@@ -402,7 +399,7 @@ class TestResolveSpeakerIdentities:
             ],
             "participants": [{"participantId": "user-a", "name": "Shared Mic"}],
         }
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {"start": 1.0, "end": 3.0, "speaker": "SPEAKER_00"},
                 {"start": 5.0, "end": 7.0, "speaker": "SPEAKER_01"},
@@ -445,7 +442,7 @@ class TestResolveSpeakerIdentities:
                 {"participantId": "user-b", "name": "Bob"},
             ],
         }
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {"start": 1.5, "end": 3.5, "speaker": "SPEAKER_00"},
                 {"start": 5.5, "end": 7.5, "speaker": "SPEAKER_01"},
@@ -493,7 +490,7 @@ class TestResolveSpeakerIdentities:
                 {"participantId": "user-b", "name": "Bob"},
             ],
         }
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {"start": 1.5, "end": 5.5, "speaker": "SPEAKER_00"},
                 {"start": 4.0, "end": 7.5, "speaker": "SPEAKER_01"},
@@ -529,7 +526,7 @@ class TestResolveSpeakerIdentities:
             ],
             "participants": [{"participantId": "user-a", "name": "Brief User"}],
         }
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {"start": 1.0, "end": 10.0, "speaker": "SPEAKER_00"},
             ],
@@ -561,7 +558,7 @@ class TestResolveSpeakerIdentities:
             ],
             "participants": [{"participantId": "user-a", "name": "Early User"}],
         }
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {"start": 0.0, "end": 3.0, "speaker": "SPEAKER_00"},
             ],
@@ -576,7 +573,7 @@ class TestResolveSpeakerIdentities:
         """No segments produces empty result."""
         result = resolve_speaker_identities(
             METADATA_SINGLE_USER,
-            FakeTranscription(segments=[]),
+            make_transcription(segments=[]),
             RECORDING_START,
             RECORDING_END,
         )
@@ -584,7 +581,7 @@ class TestResolveSpeakerIdentities:
 
     def test_segment_without_speaker_ignored(self):
         """Segments missing speaker key are skipped."""
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {"start": 1.0, "end": 3.0, "text": "no speaker"},
             ],
@@ -608,7 +605,7 @@ class TestResolveSpeakerIdentities:
             ],
             "participants": [{"participantId": "user-a", "name": "Still Talking"}],
         }
-        transcription = FakeTranscription(
+        transcription = make_transcription(
             segments=[
                 {"start": 2.0, "end": 9.0, "speaker": "SPEAKER_00"},
             ],