ci(coverage): measure test coverage in CI, report-only (#1278 item 4)

.coveragerc

@@ -0,0 +1,35 @@
+# Coverage.py configuration (#1278 item 4 — testing roadmap).
+#
+# We measure the `website` app only: that's the code we own and want to
+# backfill. Django itself, third-party packages, and the vendored in-repo forks
+# (image_cropping/, sortedm2m_filter_horizontal_widget/) are deliberately out of
+# scope. Coverage is REPORT-ONLY in CI — there is no --fail-under gate yet; the
+# percentage is a signal for targeting backfill (item 5), not a pass/fail bar.
+#
+# Run locally (inside the container, after `pip install -r requirements-dev.txt`):
+#   coverage run manage.py test website --settings=makeabilitylab.settings_test
+#   coverage report          # text table with missing line numbers
+#   coverage html            # browsable htmlcov/ report
+
+[run]
+source = website
+# Branch coverage: also tracks whether both sides of an if/else were taken,
+# not just whether a line ran. More honest signal for view/utility logic.
+branch = True
+omit =
+    # The tests themselves aren't production code.
+    website/tests/*
+    # Per-environment migrations are gitignored and disabled in the test
+    # settings (MIGRATION_MODULES = {'website': None}), so they never execute.
+    website/migrations/*
+
+[report]
+# Show the line numbers that aren't covered, so the report points at the gaps.
+show_missing = True
+# exclude_also ADDS to coverage's built-in defaults (which already include
+# "pragma: no cover"), rather than replacing them.
+exclude_also =
+    def __repr__
+    raise NotImplementedError
+    if __name__ == .__main__.:
+    if TYPE_CHECKING:

.github/workflows/test.yml

@@ -56,8 +56,29 @@ jobs:
       - name: Install Python dependencies
         run: pip install -r requirements.txt
 
-      - name: Run tests
-        run: python manage.py test website --settings=makeabilitylab.settings_test --verbosity=2
+      # coverage.py is a test-only tool (lives in requirements-dev.txt) — install
+      # it directly here rather than pulling all of requirements-dev.txt, which
+      # would drag in Playwright. Keep this pin in sync with requirements-dev.txt.
+      - name: Install coverage
+        run: pip install coverage==7.14.1
+
+      # `coverage run` wraps the same test command and propagates its exit code,
+      # so a test failure still fails this step (red ✗ + email). Config is in
+      # .coveragerc (measures the `website` app, branch coverage on).
+      - name: Run tests with coverage
+        run: coverage run manage.py test website --settings=makeabilitylab.settings_test --verbosity=2
+
+      # Report-only: no --fail-under gate. Print to the log and also publish a
+      # table to the run's Summary so the number is visible without opening logs.
+      # Skipped automatically if the test step above failed.
+      - name: Coverage report
+        run: |
+          coverage report
+          {
+            echo "## Coverage — website app"
+            echo
+            coverage report --format=markdown
+          } >> "$GITHUB_STEP_SUMMARY"
 
   # Browser end-to-end tests (member-page "Load more"/"Load all", bio toggle,
   # section nav). Kept in a SEPARATE job from `test` so the fast unit/integration

.gitignore

@@ -48,3 +48,10 @@ docker_makelab_build.log
 a11y-report.json
 repomix-output.xml
 codebase-analysis.md
+
+# coverage.py artifacts (#1278 item 4) — the data file and HTML/XML reports are
+# written into the repo root (the container's CWD is the bind-mounted /code).
+.coverage
+.coverage.*
+htmlcov/
+coverage.xml

CLAUDE.md

@@ -43,6 +43,7 @@ A superuser is required to use `/admin` and add content; create one with `python
   - When fixing a bug reachable through a real queryset, URL, or view, add a regression test in the matching style before applying the fix (matches the tests-first workflow).
   - **Always use the `--settings=makeabilitylab.settings_test` shim.** It sets `MIGRATION_MODULES = {'website': None}` so the test DB is built directly from the current models, sidestepping the gitignored, per-environment `website/migrations/` history. This is the durable fix for #1267 — without it, a fresh test DB can fail at creation with `column "..." already exists` (old workaround: `docker exec makeabilitylabwebsite-db-1 psql -U admin -d postgres -c "DROP DATABASE IF EXISTS test_makeability;"`).
   - **CI:** `.github/workflows/test.yml` runs this same command on every push to `master` and every PR (free/unlimited for this public repo). It reports a green ✓ / red ✗ — it does not block pushes or the deploy. See the testing roadmap in #1278.
+  - **Coverage (#1278 item 4):** the CI `test` job wraps the suite in `coverage run` and publishes a table to the run's Summary; config is in `.coveragerc` (measures the `website` app, branch coverage on). It's **report-only** — no `--fail-under` gate; the number targets backfill. Run locally inside the container: `pip install -r requirements-dev.txt`, then `coverage run manage.py test website --settings=makeabilitylab.settings_test && coverage report` (or `coverage html` for a browsable `htmlcov/`).
 - Accessibility (Pa11y CI + Axe, WCAG 2.0 AA): start the site, then `docker-compose -f docker-compose-local-dev.yml --profile testing run --rm a11y`. URLs to scan are configured in `.pa11yci.json`. Run this before submitting UI changes.
 
 ## Deployment

CONTRIBUTING.md

@@ -22,6 +22,8 @@ This document outlines how to set up your local development environment and our
   - [Running the Test Suite](#running-the-test-suite)
     - [When to add a test](#when-to-add-a-test)
     - [Troubleshooting](#troubleshooting-tests)
+    - [Continuous integration](#continuous-integration)
+    - [Test coverage](#test-coverage)
   - [Accessibility Testing](#accessibility-testing)
     - [Running Accessibility Checks](#running-accessibility-checks)
     - [Configuring Tests](#configuring-tests)
@@ -290,7 +292,21 @@ If you forget the shim and the legacy `manage.py test website` fails at test-DB
 
 ### Continuous integration
 
-`.github/workflows/test.yml` runs the suite (with the test-settings shim, against a Postgres 16 service container) on every push to `master` and every pull request. GitHub Actions is free and unlimited for this public repo. A failing run shows a red ✗ on the commit/PR and emails the author — it **reports** status, it does not block the push or the test-server deploy. The broader testing roadmap (coverage, Pa11y-in-CI, test backfill) is tracked in [#1278](https://github.com/makeabilitylab/makeabilitylabwebsite/issues/1278).
+`.github/workflows/test.yml` runs the suite (with the test-settings shim, against a Postgres 16 service container) on every push to `master` and every pull request. GitHub Actions is free and unlimited for this public repo. A failing run shows a red ✗ on the commit/PR and emails the author — it **reports** status, it does not block the push or the test-server deploy. Each run also measures coverage (see [Test coverage](#test-coverage)). The remaining testing roadmap (Pa11y-in-CI, test backfill) is tracked in [#1278](https://github.com/makeabilitylab/makeabilitylabwebsite/issues/1278).
+
+### Test coverage
+
+The CI `test` job wraps the suite in [`coverage`](https://coverage.readthedocs.io/) and posts a per-file table to the run's **Summary** tab. Coverage is **report-only** — there is no `--fail-under` gate, so a low number never fails CI; it's a signal for deciding what to backfill next (#1278 item 5). Configuration is in `.coveragerc`: it measures the `website` app with branch coverage on, and leaves the vendored forks (`image_cropping/`, `sortedm2m_filter_horizontal_widget/`) and third-party packages out of scope.
+
+To see coverage locally, inside the container:
+
+```bash
+# coverage lives in requirements-dev.txt (test-only, not in the prod image)
+pip install -r requirements-dev.txt
+coverage run manage.py test website --settings=makeabilitylab.settings_test
+coverage report   # text table with the uncovered line numbers
+coverage html     # browsable htmlcov/index.html
+```
 
 ## Accessibility Testing
 

README.md

@@ -38,6 +38,7 @@ makeabilitylabwebsite/
 | [CONTRIBUTING.md](CONTRIBUTING.md) | Local development setup, workflow, and pull request guidelines |
 | [docs/DEPLOYMENT.md](docs/DEPLOYMENT.md) | Production servers, logging, and data management |
 | [.pa11yci.json](.pa11yci.json) | Accessibility test configuration (Pa11y + Axe) |
+| [.coveragerc](.coveragerc) | Test-coverage configuration (coverage.py; reported in CI) |
 | [Troubleshooting Wiki](https://github.com/makeabilitylab/makeabilitylabwebsite/wiki/Troubleshooting) | Common issues and solutions |
 
 ## 🤝 Contributing

requirements-dev.txt

@@ -15,3 +15,9 @@
 # lockstep (a mismatch makes `playwright install` download the wrong revision).
 -r requirements.txt
 playwright==1.60.0
+
+# coverage.py — test-coverage measurement (#1278 item 4). Used locally
+# (`coverage run manage.py test ...`, see .coveragerc) and pinned identically in
+# the CI `test` job, which installs it directly rather than pulling all of
+# requirements-dev.txt (it has no need for Playwright). Keep the two pins in sync.
+coverage==7.14.1