backend-tests

name: backend-tests description: Run backend tests and code quality checks for OPRE OPS. Covers ops_api pytest, data_tools pytest, and nox linting/formatting sessions. Use this skill when the user wants to run backend tests, check code quality, lint Python code, run pytest, or verify their backend changes pass CI checks — even if they just say "run the tests" or "does this pass". argument-hint: "[api | data-tools | lint | format | all | ]" allowed-tools: Read, Grep, Glob, Bash, Edit, Write disable-model-invocation: true

Backend Test Runner

You run tests and code quality checks for the two backend Python packages in OPRE OPS: ops_api (the Flask API) and data_tools (the ETL/data loading scripts). Each has its own pipenv environment, test suite, and nox sessions.

How to Determine What to Do

Interpret $ARGUMENTS to decide the action:

1. API Tests: $ARGUMENTS is api or starts with api

Run the ops_api test suite.

cd backend/ops_api

# Run all API tests
pipenv run pytest

# Run with verbose output and short summary
pipenv run pytest -v --tb=short

If $ARGUMENTS includes a specific path or test name after api (e.g., api test_agreements):

cd backend/ops_api

# Find matching test files
find tests -name "*<pattern>*" -type f

# Run the matching test file
pipenv run pytest tests/ops/resources/test_agreements.py

Test files mirror the source structure: ops/resources/agreements.py -> tests/ops/resources/test_agreements.py.

Report results:

• Total tests: passed, failed, errors, skipped
• For failures: show the test name and the assertion error
• Suggest fixes if the failure is straightforward

2. Data Tools Tests: $ARGUMENTS is data-tools or data or etl

Run the data_tools test suite. These tests use pytest-docker to spin up a Postgres instance.

Pre-flight check — Docker must be available:

docker info > /dev/null 2>&1 && echo "Docker: OK" || echo "Docker: NOT RUNNING"

cd backend/data_tools

# Run all data tools tests
pipenv run pytest

# Run with verbose output
pipenv run pytest -v --tb=short

If a specific test path is provided:

cd backend/data_tools
pipenv run pytest tests/load_projects/test_load_projects.py

Data tools tests use the loaded_db fixture which:

• Starts Postgres via pytest-docker
• Applies history triggers
• Rolls back and cleans ops_db_history after each test

Report results the same as API tests.

3. Lint: $ARGUMENTS is lint

Run linting across both packages:

echo "=== ops_api linting ==="
cd backend/ops_api
pipenv run nox -s lint

echo ""
echo "=== data_tools linting ==="
cd backend/data_tools
pipenv run nox -s lint

Report any linting violations with file, line number, and rule code.

4. Format Check: $ARGUMENTS is format or format-check

Check formatting (Black + isort) without modifying files:

echo "=== ops_api format check ==="
cd backend/ops_api
pipenv run nox -s format_check

echo ""
echo "=== data_tools format check ==="
cd backend/data_tools
pipenv run nox -s format_check

If formatting issues are found, offer to auto-fix:

cd backend/ops_api
pipenv run nox -s black

cd backend/data_tools
pipenv run nox -s black

5. Specific Test Path: $ARGUMENTS is a file path or test identifier

If the argument looks like a file path or test name (contains /, .py, or ::):

Determine which package it belongs to:

• If it contains data_tools or load_ -> run in backend/data_tools/
• Otherwise -> run in backend/ops_api/

# For ops_api tests
cd backend/ops_api
pipenv run pytest <test-path> -v --tb=short

# For data_tools tests
cd backend/data_tools
pipenv run pytest <test-path> -v --tb=short

6. All Checks: $ARGUMENTS is all or ci

Run the full backend CI check suite — everything that CI would catch.

echo "=== Step 1/4: ops_api lint + format ==="
cd backend/ops_api
pipenv run nox -s lint
pipenv run nox -s format_check

echo ""
echo "=== Step 2/4: ops_api tests ==="
cd backend/ops_api
pipenv run pytest --tb=short

echo ""
echo "=== Step 3/4: data_tools lint + format ==="
cd backend/data_tools
pipenv run nox -s lint
pipenv run nox -s format_check

echo ""
echo "=== Step 4/4: data_tools tests ==="
cd backend/data_tools
pipenv run pytest --tb=short

Note on data_tools lint: The data_tools package has known pre-existing lint violations (unused imports, f-string placeholders, complexity warnings). These are technical debt, not new failures. Only report lint issues that are in files you or the user have changed.

Report a final summary:

Backend CI Check Summary:
  ops_api lint:         PASS/FAIL
  ops_api format:       PASS/FAIL
  ops_api tests:        X passed, Y failed
  data_tools lint:      PASS/FAIL
  data_tools format:    PASS/FAIL
  data_tools tests:     X passed, Y failed

7. Default: $ARGUMENTS is empty or unrecognized

Show help:

Backend Test Skill - Available Commands:

  /backend-tests api                Run ops_api test suite
  /backend-tests api <pattern>      Run matching ops_api tests (e.g., "api agreements")
  /backend-tests data-tools         Run data_tools test suite (needs Docker)
  /backend-tests lint               Run flake8 linting on both packages
  /backend-tests format             Check Black/isort formatting on both packages
  /backend-tests all                Run full CI check suite (lint + format + tests)
  /backend-tests <test-path>        Run a specific test file or function

Package locations:
  ops_api:    backend/ops_api/    (Flask API tests)
  data_tools: backend/data_tools/ (ETL tests, needs Docker for Postgres)

Key File Locations

• ops_api tests: backend/ops_api/tests/
• ops_api conftest: backend/ops_api/tests/conftest.py (fixtures: app, client, auth_client, loaded_db)
• ops_api noxfile: backend/ops_api/noxfile.py
• data_tools tests: backend/data_tools/tests/
• data_tools conftest: backend/data_tools/tests/conftest.py (fixtures: db_service, loaded_db)
• data_tools noxfile: backend/data_tools/noxfile.py
• Shared models: backend/models/ (used by both packages)

Common Issues

• "ModuleNotFoundError: No module named 'models'": You're running from the wrong directory. ops_api tests run from backend/ops_api/, data_tools tests from backend/data_tools/.
• data_tools tests hang: Docker might not be running. Check with docker info.
• Fixture errors: The loaded_db fixture in ops_api loads test data from data_tools/initial_data/. If seed data has changed, the fixture may need updating.
• Formatting differs between packages: Each package has its own Black/isort config. Run formatting per-package, not globally.