Skip to content

Speakr is a personal, self-hosted web application designed for transcribing audio recordings

Python

CI: run full pytest + vitest suite on push/PR with isolated test DB

Murtaza Nasir385b4ca05c56
diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
new file mode 100644
index 00000000..fa1a0781
--- /dev/null
+++ b/.github/workflows/test.yml
@@ -0,0 +1,77 @@
+name: Tests
+
+# Runs the full automated suite on every push and pull request. The existing
+# docker-publish.yml only ran two migration tests on release, so nothing
+# guarded day-to-day changes; this workflow gives the ~378 Python tests + the
+# Vitest JS tests real per-commit regression protection.
+#
+# Syntax is standard GitHub Actions and also runs on Forgejo Actions
+# (Forgejo reads .github/workflows by default). The Python suite isolates
+# itself to a throwaway SQLite DB via conftest.py, so it needs no app config.
+
+on:
+ push:
+ pull_request:
+ workflow_dispatch:
+
+jobs:
+ python-tests:
+ runs-on: ubuntu-latest
+
+ services:
+ postgres:
+ image: postgres:16
+ env:
+ POSTGRES_USER: speakr
+ POSTGRES_PASSWORD: speakr
+ POSTGRES_DB: speakr_test
+ ports:
+ - 5432:5432
+ options: >-
+ --health-cmd "pg_isready -U speakr"
+ --health-interval 10s
+ --health-timeout 5s
+ --health-retries 5
+
+ steps:
+ - name: Checkout repository
+ uses: actions/checkout@v5
+
+ - name: Set up Python
+ uses: actions/setup-python@v6
+ with:
+ python-version: '3.11'
+ cache: pip
+
+ - name: Install ffmpeg (needed by stitch / ffprobe tests)
+ run: sudo apt-get update && sudo apt-get install -y ffmpeg
+
+ - name: Install dependencies
+ run: pip install -r requirements.txt pytest
+
+ - name: Run test suite
+ env:
+ # The app tests run against an isolated SQLite DB (see conftest.py).
+ # test_postgres_migrations.py reads TEST_DATABASE_URI separately so it
+ # exercises real PostgreSQL compatibility against the service above.
+ TEST_DATABASE_URI: postgresql://speakr:speakr@localhost:5432/speakr_test
+ run: pytest tests/ -q
+
+ js-tests:
+ runs-on: ubuntu-latest
+
+ steps:
+ - name: Checkout repository
+ uses: actions/checkout@v5
+
+ - name: Set up Node
+ uses: actions/setup-node@v4
+ with:
+ node-version: '20'
+ cache: npm
+
+ - name: Install dependencies
+ run: npm ci
+
+ - name: Run Vitest
+ run: npm test
diff --git a/conftest.py b/conftest.py
new file mode 100644
index 00000000..7bcd9159
--- /dev/null
+++ b/conftest.py
@@ -0,0 +1,58 @@
+"""Pytest configuration for the Speakr test suite.
+
+This file is imported by pytest BEFORE it collects any test module, so the
+environment it sets here is in place before any `from src.app import app`
+binds the Flask app to a database.
+
+Goal: isolation. Point the app at a throwaway temporary SQLite database so
+the test suite never reads or mutates the developer's real
+`instance/transcriptions.db`. Running the suite with `pytest` therefore
+needs no services and is safe on any machine / in CI.
+
+Note: the legacy standalone invocation (`python tests/test_foo.py`) does NOT
+load conftest.py, so this changes nothing for that workflow — it only affects
+`pytest` runs.
+"""
+
+import os
+import tempfile
+
+# --- Isolated database (set BEFORE src.app is imported) ---------------------
+# Use a real file (not :memory:) because the app and tests open multiple
+# connections / app contexts, which an in-memory SQLite DB would not share.
+_TEST_DIR = tempfile.mkdtemp(prefix="speakr_pytest_")
+_TEST_DB_PATH = os.path.join(_TEST_DIR, "test.db")
+os.environ["SQLALCHEMY_DATABASE_URI"] = f"sqlite:///{_TEST_DB_PATH}"
+
+# The app os.makedirs(UPLOAD_FOLDER) at import time; default is /data/uploads
+# which isn't writable in CI / outside Docker. Point it at the temp dir.
+os.environ["UPLOAD_FOLDER"] = os.path.join(_TEST_DIR, "uploads")
+
+# --- Quiet, deterministic defaults for tests --------------------------------
+# Only set if the caller hasn't already chosen a value.
+os.environ.setdefault("SECRET_KEY", "pytest-secret-key")
+os.environ.setdefault("ENABLE_AUTO_PROCESSING", "false") # no black-hole file monitor
+os.environ.setdefault("TEXT_MODEL_API_KEY", "test-key") # avoid config hard-fails
+# NB: do NOT force WEBHOOK_GLOBAL_ENABLED=false here — the webhook suite needs
+# delivery enabled (it mocks the actual HTTP POST), so leave the default (true).
+
+import pytest # noqa: E402
+
+
[email protected](scope="session", autouse=True)
+def _initialize_test_database():
+ """Create the full schema in the throwaway DB once per session."""
+ from src.app import app
+ from src.init_db import initialize_database
+
+ with app.app_context():
+ initialize_database(app)
+
+ yield
+
+ # Best-effort cleanup of the temp DB file after the session.
+ for path in (_TEST_DB_PATH, _TEST_DB_PATH + "-wal", _TEST_DB_PATH + "-shm"):
+ try:
+ os.remove(path)
+ except OSError:
+ pass
diff --git a/tests/test_csrf_token_bypass_fix.py b/tests/test_csrf_token_bypass_fix.py
index b8df47de..45364593 100644
--- a/tests/test_csrf_token_bypass_fix.py
+++ b/tests/test_csrf_token_bypass_fix.py
@@ -424,9 +424,13 @@ def test_api_v1_post_with_query_string_token_still_works():
db.session.add(rec)
db.session.commit()
- # Pure-query-string authentication on a v1 POST endpoint
+ # Pure-query-string authentication on a v1 POST endpoint.
+ # The v1 route delegates to src.api.recordings.regenerate_title,
+ # which calls _generate_ai_title (a real LLM call). Mock that so
+ # the test exercises the auth/CSRF path without needing a live LLM
+ # — otherwise it returns 500 in CI and masks what we're asserting.
from unittest.mock import patch
- with patch('src.api.api_v1._enqueue_title_regeneration', create=True), \
+ with patch('src.tasks.processing._generate_ai_title', return_value='Mock Title'), \
patch('src.services.job_queue.job_queue.enqueue', return_value=999):
resp = client.post(
f'/api/v1/recordings/{rec.id}/regenerate_title?token={plaintext}',