✅(summary) add unit and API tests for summary service

Summary service currently has no tests. Add unit and API tests to
summary service.

CONTRIBUTING.md

@@ -57,7 +57,9 @@ It is nice to add information about the purpose of the pull request to help revi
 ### Don't forget to:
 - check your commits
 - check the linting: `make lint && make frontend-lint`
-- check the tests: `make test`
+- check the tests: 
+    - backend: `make test`
+    - summary: `make test-summary`
 - add a changelog entry
 
 Once all the required tests have passed, you can request a review from the project maintainers.

Makefile

@@ -203,6 +203,11 @@ test-back-parallel: ## run all back-end tests in parallel
 	bin/pytest -n auto $${args:-${1}}
 .PHONY: test-back-parallel
 
+test-summary: ## run summary service tests
+	@args="$(filter-out $@,$(MAKECMDGOALS))" && \
+	bin/pytest-summary $${args:-${1}}
+.PHONY: test-summary
+
 makemigrations:  ## run django makemigrations for the Meet project.
 	@echo "$(BOLD)Running makemigrations$(RESET)"
 	@$(COMPOSE) up -d postgresql

bin/pytest

@@ -1,8 +1,9 @@
 #!/usr/bin/env bash
+# NB: this file is used locally only. In CI, it is overwritten by pytest install
 
 source "$(dirname "${BASH_SOURCE[0]}")/_config.sh"
 
 _dc_run \
     -e DJANGO_CONFIGURATION=Test \
     app-dev \
-    pytest "$@"
+    pytest "$@" 

bin/pytest-summary

@@ -0,0 +1,7 @@
+#!/usr/bin/env bash
+
+source "$(dirname "${BASH_SOURCE[0]}")/_config.sh"
+
+_dc_run \
+    app-summary-dev \
+    python -m pytest "$@"

src/summary/pyproject.toml

@@ -21,8 +21,17 @@ dependencies = [
 [project.optional-dependencies]
 dev = [
     "ruff==0.14.4",
+    "pytest==9.0.2",
+    "responses>=0.25.8",
 ]
 
+[tool.pytest.ini_options]
+markers = [
+    "unit: Test individual components",
+    "integration: Test the API",
+]
+testpaths = ["tests"]
+
 [build-system]
 requires = ["setuptools>=61.0"]
 build-backend = "setuptools.build_meta"
@@ -49,11 +58,10 @@ select = [
     "T20", # flake8-print
     "W",  # pycodestyle warning
 ]
+ignore= ["DJ001", "PLR2004"]
 
 [tool.ruff.lint.per-file-ignores]
-"tests/*" = [
-    "S101", # use of assert
-]
+"tests/*" = ["S", "SLF"]
 
 [tool.ruff.lint.pydocstyle]
 # Use Google-style docstrings.

src/summary/summary/core/celery_worker.py

@@ -140,7 +140,7 @@ def format_transcript(
     )
 
 
-def format_actions(llm_output: dict) -> str:
+def _format_actions(llm_output: dict) -> str:
     """Format the actions from the LLM output into a markdown list.
 
     fomat:
@@ -328,7 +328,7 @@ def summarize_transcription(
         response_format=FORMAT_NEXT_STEPS,
     )
 
-    next_steps = format_actions(json.loads(next_steps))
+    next_steps = _format_actions(json.loads(next_steps))
 
     logger.info("Next steps generated")
 

src/summary/tests/__init__.py

@@ -0,0 +1 @@
+"""Tests for the summary service."""

src/summary/tests/conftest.py

@@ -0,0 +1,26 @@
+"""Shared test fixtures and environment setup for the summary service tests."""
+
+import os
+
+# Set required environment variables before any summary module imports.
+# This is necessary because several modules call get_settings() at module level,
+# which validates env vars via Pydantic Settings.
+os.environ.setdefault("APP_API_TOKEN", "test-api-token")
+os.environ.setdefault("AWS_STORAGE_BUCKET_NAME", "test-bucket")
+os.environ.setdefault("AWS_S3_ENDPOINT_URL", "http://localhost:9000")
+os.environ.setdefault("AWS_S3_ACCESS_KEY_ID", "test-access-key")
+os.environ.setdefault("AWS_S3_SECRET_ACCESS_KEY", "test-secret-key")
+os.environ.setdefault("AWS_S3_SECURE_ACCESS", "false")
+os.environ.setdefault("WHISPERX_API_KEY", "test-whisperx-key")
+os.environ.setdefault("WHISPERX_BASE_URL", "http://localhost:8000/v1")
+os.environ.setdefault("LLM_BASE_URL", "http://localhost:8001/v1")
+os.environ.setdefault("LLM_API_KEY", "test-llm-key")
+os.environ.setdefault("LLM_MODEL", "test-model")
+os.environ.setdefault("WEBHOOK_API_TOKEN", "test-webhook-token")
+os.environ.setdefault("WEBHOOK_URL", "http://localhost:8002/webhook")
+os.environ.setdefault("CELERY_BROKER_URL", "memory://")
+os.environ.setdefault("CELERY_RESULT_BACKEND", "cache+memory://")
+os.environ.setdefault("POSTHOG_ENABLED", "false")
+os.environ.setdefault("SENTRY_IS_ENABLED", "false")
+os.environ.setdefault("LANGFUSE_ENABLED", "false")
+os.environ.setdefault("TASK_TRACKER_REDIS_URL", "redis://localhost:6379/0")

src/summary/tests/integration/__init__.py

@@ -0,0 +1 @@
+"""Integration tests for the summary service."""

src/summary/tests/integration/conftest.py

@@ -0,0 +1,23 @@
+"""Integration test configuration. Provides shared fixtures."""
+
+import pytest
+from fastapi.testclient import TestClient
+
+from summary.core.celery_worker import celery
+from summary.main import app
+
+
+@pytest.fixture()
+def client():
+    """Provide a FastAPI TestClient for integration tests."""
+    return TestClient(app)
+
+
+@pytest.fixture()
+def eager_celery():
+    """Run Celery tasks synchronously in the same process."""
+    celery.conf.task_always_eager = True
+    celery.conf.task_eager_propagates = True
+    yield
+    celery.conf.task_always_eager = False
+    celery.conf.task_eager_propagates = False

src/summary/tests/integration/test_api_health.py

@@ -0,0 +1,21 @@
+"""Integration tests for the health check endpoints."""
+
+
+class TestHeartbeat:
+    """Tests for the /__heartbeat__ endpoint."""
+
+    def test_returns_200(self, client):
+        """The heartbeat endpoint responds with 200 OK without a token."""
+        response = client.get("/__heartbeat__")
+
+        assert response.status_code == 200
+
+
+class TestLBHeartbeat:
+    """Tests for the /__lbheartbeat__ endpoint."""
+
+    def test_returns_200(self, client):
+        """The load-balancer heartbeat endpoint responds with 200 OK without a token."""
+        response = client.get("/__lbheartbeat__")
+
+        assert response.status_code == 200

src/summary/tests/integration/test_api_tasks.py

@@ -0,0 +1,130 @@
+"""Integration test for the transcribe-and-summarize task flow via the API."""
+
+import json
+from contextlib import contextmanager
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+import responses
+
+from summary.core.config import get_settings
+
+settings = get_settings()
+
+API_PREFIX = "/api/v1"
+AUTH_HEADER = {"Authorization": f"Bearer {settings.app_api_token.get_secret_value()}"}
+WEBHOOK_URL = settings.webhook_url
+
+
+class TestTranscribeSummarizeFlow:
+    """End-to-end test: POST /tasks/ triggers transcription and summary via webhook."""
+
+    @responses.activate
+    @patch("summary.core.celery_worker.analytics")
+    @patch("summary.core.celery_worker.LLMObservability")
+    @patch("summary.core.celery_worker.LLMService")
+    @patch("summary.core.celery_worker.metadata_manager")
+    @patch("summary.core.celery_worker.openai")
+    @patch("summary.core.celery_worker.file_service")
+    def test_transcription_and_summary_are_submitted(  # noqa: PLR0913
+        self,
+        mock_file_service,
+        mock_openai,
+        mock_metadata,
+        mock_llm_cls,
+        mock_observability_cls,
+        mock_analytics,
+        client,
+        eager_celery,
+    ):
+        """Creating a task produces a transcription and summary sent to the webhook."""
+        # Stub file service
+        fake_audio = MagicMock()
+
+        @contextmanager
+        def fake_prepare(filename):
+            yield fake_audio, {"duration": 60.0}
+
+        mock_file_service.prepare_audio_file = fake_prepare
+
+        # Stub WhisperX transcription
+        fake_transcription = SimpleNamespace(
+            segments=[
+                {"speaker": "SPEAKER_00", "text": "Hello everyone."},
+                {"speaker": "SPEAKER_01", "text": "Let's discuss the roadmap."},
+            ],
+        )
+        mock_client = MagicMock()
+        mock_client.audio.transcriptions.create.return_value = fake_transcription
+        mock_openai.OpenAI.return_value = mock_client
+
+        # Stub analytics to enable summary
+        mock_analytics.is_feature_enabled.return_value = True
+
+        # Stub LLM for summarization
+        mock_llm = MagicMock()
+        mock_llm_cls.return_value = mock_llm
+
+        plan_json = json.dumps({"titles": ["Roadmap"]})
+        next_steps_json = json.dumps(
+            {
+                "actions": [
+                    {
+                        "title": "Draft roadmap",
+                        "assignees": ["Aleb"],
+                        "due_date": "2026-03-04",
+                    }
+                ]
+            }
+        )
+        mock_llm.call.side_effect = [
+            "### TL;DR\nQuick overview.",  # tldr
+            plan_json,  # parts plan
+            "### Roadmap\nDetails.",  # part content
+            next_steps_json,  # next steps
+            "Cleaned summary.",  # cleaning
+        ]
+
+        mock_observability_cls.return_value = MagicMock()
+
+        # Stub webhook (called twice: transcription + summary)
+        responses.post(WEBHOOK_URL, json={"id": "doc-1"}, status=200)
+        responses.post(WEBHOOK_URL, json={"id": "doc-2"}, status=200)
+
+        payload = {
+            "owner_id": "owner-1",
+            "filename": "recording.webm",
+            "email": "user@example.com",
+            "sub": "user-sub-id",
+            "room": "Visio room",
+            "recording_date": "2026-03-04",
+            "recording_time": "09:00",
+            "language": "en",
+            "download_link": "https://example.com/rec.webm",
+            "context_language": "en",
+        }
+
+        response = client.post(
+            f"{API_PREFIX}/tasks/", json=payload, headers=AUTH_HEADER
+        )
+
+        assert response.status_code == 200
+        body = response.json()
+        assert "id" in body
+        assert body["message"] == "Task created"
+
+        # Verify the webhook received the transcription
+        assert len(responses.calls) >= 1
+        transcript_payload = json.loads(responses.calls[0].request.body)
+        assert "SPEAKER_00" in transcript_payload["content"]
+        assert "Hello everyone." in transcript_payload["content"]
+        assert "Visio room" in transcript_payload["title"]
+        assert transcript_payload["email"] == "user@example.com"
+        assert transcript_payload["sub"] == "user-sub-id"
+
+        # Verify the webhook received the summary
+        assert len(responses.calls) == 2
+        summary_payload = json.loads(responses.calls[1].request.body)
+        assert "TL;DR" in summary_payload["content"]
+        assert "Cleaned summary." in summary_payload["content"]
+        assert "Draft roadmap" in summary_payload["content"]

src/summary/tests/unit/__init__.py

@@ -0,0 +1 @@
+"""Unit tests for the summary service."""

src/summary/tests/unit/test_celery_worker.py

@@ -0,0 +1,251 @@
+"""Tests for the celery_worker module."""
+
+import json
+from contextlib import contextmanager
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+import pytest
+import responses
+
+from summary.core.celery_worker import (
+    format_transcript,
+    summarize_transcription,
+    transcribe_audio,
+)
+from summary.core.config import get_settings
+from summary.core.file_service import FileServiceException
+
+settings = get_settings()
+
+WEBHOOK_URL = settings.webhook_url
+
+# ---------------------------------------------------------------------------
+# transcribe_audio
+# ---------------------------------------------------------------------------
+
+
+class TestTranscribeAudio:
+    """Tests for the transcribe_audio function."""
+
+    @patch("summary.core.celery_worker.metadata_manager")
+    @patch("summary.core.celery_worker.openai")
+    @patch("summary.core.celery_worker.file_service")
+    def test_success(self, mock_file_service, mock_openai, mock_metadata):
+        """Transcription succeeds and returns the transcription object."""
+        fake_audio = MagicMock()
+        fake_metadata = {"duration": 120.5}
+
+        @contextmanager
+        def fake_prepare(filename):
+            yield fake_audio, fake_metadata
+
+        mock_file_service.prepare_audio_file = fake_prepare
+
+        fake_transcription = SimpleNamespace(
+            segments=[{"speaker": "SPEAKER_00", "text": "Hello"}],
+        )
+        mock_client = MagicMock()
+        mock_client.audio.transcriptions.create.return_value = fake_transcription
+        mock_openai.OpenAI.return_value = mock_client
+
+        result = transcribe_audio("task-1", "recording.ogg", "en")
+
+        assert result is fake_transcription
+        mock_client.audio.transcriptions.create.assert_called_once()
+        call_kwargs = mock_client.audio.transcriptions.create.call_args
+        assert call_kwargs.kwargs["language"] == "en"
+        assert call_kwargs.kwargs["file"] is fake_audio
+
+    @patch("summary.core.celery_worker.metadata_manager")
+    @patch("summary.core.celery_worker.openai")
+    @patch("summary.core.celery_worker.file_service")
+    def test_file_service_error_returns_none(
+        self, mock_file_service, mock_openai, mock_metadata
+    ):
+        """Returns None when the file cannot be retrieved."""
+
+        @contextmanager
+        def failing_prepare(filename):
+            raise FileServiceException("download failed")
+            yield  # NOSONAR - yield required for contextmanager
+
+        mock_file_service.prepare_audio_file = failing_prepare
+
+        result = transcribe_audio("task-1", "recording.ogg", "en")
+
+        assert result is None
+        mock_openai.OpenAI.return_value.audio.transcriptions.create.assert_not_called()
+
+
+# ---------------------------------------------------------------------------
+# format_transcript
+# ---------------------------------------------------------------------------
+
+
+class TestFormatTranscript:
+    """Tests for the format_transcript function."""
+
+    def test_with_segments(self):
+        """Formats a transcription with segments into content and title."""
+        transcription = {
+            "segments": [
+                {"speaker": "SPEAKER_00", "text": "Hello everyone."},
+                {"speaker": "SPEAKER_01", "text": "Good morning."},
+            ],
+        }
+
+        content, title = format_transcript(
+            transcription,
+            context_language="en",
+            language="en",
+            room="Daily standup",
+            recording_date="2026-03-04",
+            recording_time="09:00",
+            download_link="https://example.com/rec.ogg",
+        )
+
+        assert "SPEAKER_00" in content
+        assert "Hello everyone." in content
+        assert "SPEAKER_01" in content
+        assert "Good morning." in content
+        assert "Daily standup" in title
+        assert "2026-03-04" in title
+        assert "09:00" in title
+
+    @pytest.mark.parametrize(
+        "context_language, expected_string",
+        [
+            ("en", "Download your recording"),
+            ("fr", "Télécharger votre enregistrement"),
+            ("de", "diesem Link folgen"),
+            ("nl", "Download uw opname door"),
+        ],
+    )
+    def test_context_language(self, context_language, expected_string):
+        """Context language parameter modifies output."""
+        transcription = {
+            "segments": [
+                {"speaker": "SPEAKER_00", "text": "Hello everyone."},
+            ],
+        }
+
+        content, _ = format_transcript(
+            transcription,
+            context_language=context_language,
+            language="en",
+            room="Daily standup",
+            recording_date="2026-03-04",
+            recording_time="09:00",
+            download_link="https://example.com/rec.ogg",
+        )
+
+        assert expected_string in content
+
+    def test_empty_segments(self):
+        """Returns empty-transcription message when there are no segments."""
+        transcription = {"segments": []}
+
+        content, title = format_transcript(
+            transcription,
+            context_language="en",
+            language="en",
+            room=None,
+            recording_date=None,
+            recording_time=None,
+            download_link=None,
+        )
+
+        assert "No audio content" in content or "Transcription" in title
+
+
+# ---------------------------------------------------------------------------
+# summarize_transcription
+# ---------------------------------------------------------------------------
+
+
+class TestSummarizeTranscription:
+    """Tests for the summarize_transcription Celery task."""
+
+    @responses.activate
+    @patch("summary.core.celery_worker.LLMService")
+    @patch("summary.core.celery_worker.LLMObservability")
+    @patch("summary.core.celery_worker.analytics")
+    def test_generates_and_submits_summary(
+        self, mock_analytics, mock_observability_cls, mock_llm_cls
+    ):
+        """Assembles TLDR + parts + next steps + cleaning, then submits."""
+        mock_analytics.is_feature_enabled.return_value = False
+
+        # Mock the webhook HTTP endpoint
+        responses.post(
+            WEBHOOK_URL,
+            json={"id": "doc-42"},
+            status=200,
+        )
+
+        mock_llm = MagicMock()
+        mock_llm_cls.return_value = mock_llm
+
+        plan_json = json.dumps({"titles": ["Topic A", "Topic B"]})
+        next_steps_json = json.dumps(
+            {
+                "actions": [
+                    {
+                        "title": "What's nice about Visio",
+                        "assignees": ["Aleb"],
+                        "due_date": "2026-03-04",
+                    }
+                ]
+            }
+        )
+
+        # LLM calls in order: tldr, parts (plan), part A, part B, next-steps, cleaning
+        mock_llm.call.side_effect = [
+            "### TL;DR\nShort summary.",  # tldr
+            plan_json,  # parts plan
+            "### Topic A\nDetails about A.",  # part A
+            "### Topic B\nDetails about B.",  # part B
+            next_steps_json,  # next steps
+            "Cleaned summary content.",  # cleaning
+        ]
+
+        mock_observability = MagicMock()
+        mock_observability_cls.return_value = mock_observability
+
+        # Push a fake request context so self.request.id is available
+        summarize_transcription.push_request(id="summary-task-1")
+        try:
+            summarize_transcription.run(
+                "owner-1",
+                "Full transcript text",
+                "user@example.com",
+                "oidc-sub-123",
+                "99.999% uptime. Is it reasonable ?",
+            )
+        finally:
+            summarize_transcription.pop_request()
+
+        # Verify the webhook was called with the assembled summary
+        assert len(responses.calls) == 1
+        webhook_request = responses.calls[0]
+        submitted_payload = json.loads(webhook_request.request.body)
+
+        assert "TL;DR" in submitted_payload["content"]
+        assert "Cleaned summary content." in submitted_payload["content"]
+        assert "What's nice about Visio" in submitted_payload["content"]
+        assert "99.999% uptime. Is it reasonable ?" in submitted_payload["title"]
+        assert submitted_payload["email"] == "user@example.com"
+        assert submitted_payload["sub"] == "oidc-sub-123"
+
+        # Verify auth header was sent
+        assert (
+            webhook_request.request.headers["Authorization"]
+            == f"Bearer {settings.webhook_api_token.get_secret_value()}"
+        )
+
+        # LLM was called for: tldr, plan, part A, part B, next-steps, cleaning
+        expected_llm_calls = 6
+        assert mock_llm.call.call_count == expected_llm_calls
+
+        mock_observability.flush.assert_called_once()