GitLab CI Pipeline for Python Projects

Setting Up Python CI in GitLab

Python CI pipelines in GitLab benefit from pip caching and JUnit-formatted test reports that GitLab can render natively in merge requests.

Pipeline Configuration

stages:
  - lint
  - test

variables:
  PIP_CACHE_DIR: "$CI_PROJECT_DIR/.cache/pip"

flake8:
  stage: lint
  image: python:3.12-slim
  script:
    - pip install flake8
    - flake8 . --max-line-length=120 --statistics
  cache:
    key: pip-cache
    paths:
      - .cache/pip/
    policy: pull-push

pytest:
  stage: test
  image: python:3.12-slim
  script:
    - pip install -r requirements.txt
    - pytest --junitxml=report.xml --cov=src --cov-report=html
  cache:
    key: pip-cache
    paths:
      - .cache/pip/
    policy: pull-push
  artifacts:
    paths:
      - report.xml
      - htmlcov/
    expire_in: 30 days
    reports:
      junit: report.xml

Pip Cache Configuration

Setting PIP_CACHE_DIR to a project-relative path is essential. By default, pip caches to ~/.cache/pip, which is outside the project directory and cannot be cached by GitLab CI. By redirecting to .cache/pip/ inside the project, the cache keyword can preserve downloaded wheels across pipeline runs.

JUnit Reports

The reports: junit: report.xml directive tells GitLab to parse the JUnit XML and display test results directly in merge requests. Developers see which tests failed without opening the job log.

Coverage Reports

Using --cov-report=html generates an HTML coverage report saved as an artifact. GitLab can also extract coverage percentages from job logs if you add a coverage regex to the job configuration.