.gitlab-ci.yml

variables:
  GIT_CLONE_PATH: $CI_BUILDS_DIR/rover-soft
  DEVELOP_IMAGE: $CI_REGISTRY_IMAGE/rover-soft:develop
  MAIN_IMAGE: $CI_REGISTRY_IMAGE/rover-soft:stable

# This job is used to check the code.
# It invokes c++ and python linters checking for abidance to our coding guidelines.
# Then builds project checking for build errors.
#
# Uses build cache created by latest pipeline run on develop branch.
# This way we skip rebuilding majority of the packages (those that didn't change since branching from develop) each run.
#
# Different job is executed for the main branch, which doesn't use any cache.
run-code-checks:
  image: $DEVELOP_IMAGE
  cache:
    key: ${CI_COMMIT_REF_SLUG}
    paths:
      - ccache
      - build
      - devel
  stage: build
  script:
    # Prepare the environment:
    - export CCACHE_BASEDIR="${CI_PROJECT_DIR}"
    - export CCACHE_DIR="${CI_PROJECT_DIR}/ccache"
    - export PATH="/usr/lib/ccache:$PATH"
    - if ! ( test -d $CI_PROJECT_DIR/ccache && [ "$(ls -A $CI_PROJECT_DIR/ccache)" ] ); then
    - cp -r /rover-soft/ccache $CI_PROJECT_DIR
    - fi
    - cd $CI_PROJECT_DIR
    - apt-get update
    - rosdep update
    - rosdep install --from-paths src -iy
    # Invoke linters:
    - catkin_lint -W2 --output=explain --color=always src/
    # Automatically format code
    - curl https://raw.githubusercontent.com/Sarcasm/run-clang-format/master/run-clang-format.py | python3 - -r src --clang-format-executable clang-format-11
    - if [[ -n $(find ./src/ -type f -name "*.py") ]] ; then yapf3 --diff --recursive --style ./.yapf-format ./src/ ; fi
    # Try to build the project and save cache on success
    - catkin build --cmake-args -DCMAKE_CXX_COMPILER_LAUNCHER=ccache -DCMAKE_CC_COMPILER_LAUNCHER=ccache
    - ccache -s
  except:
    - main

# This job creates docker image based on develop branch and publishes it to project's container registry.
# The image is used as a base in run-code-checks job so that packages aren't rebuild if they didn't change.
#
# run-code-checks depends on image created by this job, but it is run after run-code-checks succeedes.
# This means that, when starting to use the pipeline, we need to manually create "first" image in the repository
# so that first stage can be run.
deploy-develop-image:
  image: docker:20.10
  services:
    - docker:20.10-dind
  variables:
    DOCKER_DRIVER: overlay2
    DOCKER_HOST: tcp://docker:2376
    DOCKER_TLS_CERTDIR: "/certs"
  stage: deploy
  needs: [run-code-checks]
  script:
    - docker login -u $CI_REGISTRY_USER -p $CI_REGISTRY_PASSWORD $CI_REGISTRY
    - docker build --pull -f ci/Dockerfile.develop -t $DEVELOP_IMAGE $GIT_CLONE_PATH
    - docker push $DEVELOP_IMAGE
  only:
    - develop

# This job creates docker image based on main branch and publishes it to project's container registry.
# It doesn't use any cache and performs clean build, so it may take longer than other jobs.
deploy-main-image:
  image: jdrouet/docker-with-buildx
  services:
    - docker:20.10-dind
  variables:
    DOCKER_DRIVER: overlay2
    DOCKER_HOST: tcp://docker:2375
    DOCKER_TLS_CERTDIR: ""
    PLATFORM: linux/amd64,linux/arm64
  stage: deploy
  script:
    - echo "$CI_REGISTRY_PASSWORD" | docker login $CI_REGISTRY --username $CI_REGISTRY_USER --password-stdin
    - docker run --rm --privileged multiarch/qemu-user-static --reset -p yes
    - docker buildx create --use
    - docker buildx build --push -f ci/Dockerfile.stable -t $MAIN_IMAGE --platform "$PLATFORM" --cache-from $MAIN_IMAGE --build-arg BUILDKIT_INLINE_CACHE=1 $GIT_CLONE_PATH
  only:
    - main