​
GitLab CI/CD Workflow

This gitlab-ci.yml configuration runs on every push, executes tests, generates a coverage report, and uploads it to CodeAnt AI: 
image: python:3.11

stages:
  - test

variables:
  VIRTUAL_ENV: ".venv"
  ACCESS_TOKEN: "<your_access_token>"
  API_BASE: "https://api.codeant.ai"

coverage:
  stage: test
  script:
    - python -m venv $VIRTUAL_ENV
    - source $VIRTUAL_ENV/bin/activate
    - pip install --upgrade pip
    - pip install coverage pytest uv

    - coverage run -m pytest tests/
    - coverage xml -o coverage.xml
    - cat coverage.xml

    - curl -sS -X GET "${API_BASE}/pr/analysis/coverage/script/get" \
         --output upload_coverage.sh.b64
    - base64 -d upload_coverage.sh.b64 > upload_coverage.sh
    - chmod +x upload_coverage.sh

    - bash upload_coverage.sh \
        -t "${ACCESS_TOKEN}" \
        -r "${CI_PROJECT_PATH}" \
        -c "${CI_COMMIT_SHA}" \
        -f coverage.xml \
        -p gitlab \
        -b "${CI_COMMIT_REF_NAME}" \
        -m <module_name>
  rules:
    - if: $CI_PIPELINE_SOURCE == "push"
      when: always
Note: If using GitHub Enterprise or self-hosted GitLab/Azure DevOps, add -u <your_vcs_base_url> to the upload script (e.g., -u https://github.enterprise.com). Don’t forget to add your access token. The access token should have write access to the repo. The <module_name> is optional. It is used to specify modules in the case of monorepo structure.

​
Coverage config file

You have to create a .coveragerc file in the project’s root folder to include all the source files in the test coverage calculation. Example: 
# include every Python file under the repo root
source = .

# exclude tests, virtualenvs, build artifacts, etc.
omit =
    */tests/*
    */.venv/*
    */build/*
    */dist/*
When you assign source to ”.” , It checks for every python file in the root folder and its sub directories. You can omit some directories by placing them in the omit section of the file.

​
How it works

With the above configuration:
  1. coverage run -m pytest tests/ will count every .py under the workspace as “valid” lines except for those in the omitted directories.
  2. Lines actually executed by your tests are marked “covered.”
  3. coverage xml -o coverage.xml produces a Cobertura-style report reflecting true coverage over the entire codebase.
  4. Using this coverage xml, we calculate the coverage percentage and the status check will be done on every new push to the branch.