archived-osu-wiki/.github/workflows/continuous-integration.yml

name: osu-wiki continuous integration

on:
  push:
    branches:
      - master
  pull_request:
    branches:
      - master
    types:
      - opened
      - reopened
      - synchronize
      - edited

jobs:
  ci:
    name: changed files
    runs-on: ubuntu-latest
    steps:
      - name: sparse checkout
        shell: bash
        run: |
          # repository url, utilising provided github credentials
          REPOSITORY="https://${{ github.actor }}:${{ github.token }}@github.com/${{ github.repository }}.git"
          # merge commit ref name (with refs/heads/ stripped out)
          BRANCH="${GITHUB_REF/#refs\/heads\//}"

          git version
          # clone without blobs; don't checkout to avoid fetching them anyway
          git clone --filter=blob:none --no-checkout ${REPOSITORY} .
          git config --local gc.auto 0

          # set up sparse checkout
          git sparse-checkout init
          git sparse-checkout set '**/*.md' '**/.remark*' '**/*.json' '**/*.yaml' '.github/scripts' 'wiki/**/*'

          # fetch the merge commit ref
          git -c protocol.version=2 fetch --no-tags --prune --progress --depth=2 origin +${GITHUB_SHA}:refs/remotes/origin/${BRANCH}
          git checkout --progress --force -B $BRANCH refs/remotes/origin/$BRANCH

      - name: inspect binary file sizes
        shell: bash
        run: |
          WARN_ON_SIZE=500000
          ERROR_ON_SIZE=1000000
          EXIT=0

          while read file
          do
              echo "Checking ${file}..."
              # git ls-tree will output:
              # (file mode) (file type) (blob hash)<TAB>(file name)
              # we're interested in the hash to pull the file's size using cat-file
              hash=`git ls-tree ${{ github.sha }} "${file}" | awk -F ' ' '{ print $3 }'`
              filesize=`git cat-file -s ${hash} 2>/dev/null`

              if [[ ${filesize} -ge ${ERROR_ON_SIZE} ]]; then
                  echo "::error file=${file}::The size of the file exceeds 1MB. Compress it to optimise performance."
                  EXIT=1
              elif [[ ${filesize} -ge ${WARN_ON_SIZE} ]]; then
                  echo "::warning file=${file}::The size of the file exceeds 500kB. Consider compressing it to optimise performance."
              else
                  echo "::debug::File ${file} is ok."
              fi
          done < <(git diff --numstat --no-renames --diff-filter=d ${{ github.sha }}^ ${{ github.sha }} | grep -Poe '-\t-\t\K.+')
          # git diff --numstat will output -<TAB>-<TAB>$filename for blobs

          exit ${EXIT}

      - name: setup node
        uses: actions/setup-node@v1

      - name: install remark
        run: npm install

      - name: run remark on changed files
        # xargs is ran with -d '\n' to properly handle single and double quotes
        # stdout is discarded (remark prints files being checked there)
        run: git diff --diff-filter=d --name-only ${{ github.sha }}^ ${{ github.sha }} '*.md' | xargs -d '\n' npx remark -qf --no-stdout --silently-ignore --report=vfile-reporter-position --color

      - name: setup Python
        uses: actions/setup-python@v2
        with:
          python-version: "3.x"

      - name: setup and run yamllint on .yaml and .md files
        run: |
          pip install yamllint
          python .github/scripts/ci/run_yamllint.py --config .yamllint.yaml

      - name: find broken wikilinks
        shell: bash
        env:
          PULL_REQUEST_TAG: 'SKIP_WIKILINK_CHECK'
        run: |
          if ${{ contains(github.event.pull_request.body, env.PULL_REQUEST_TAG) }}; then
            echo "::notice::Broken wikilink check suppressed ($PULL_REQUEST_TAG tag found in the pull request)"
            exit 0
          fi

          ARTICLES=$( git diff --diff-filter=d --name-only ${{ github.sha }}^ ${{ github.sha }} "wiki/**/*.md" "news/*.md" )
          python .github/scripts/ci/find_broken_wikilinks.py --target $ARTICLES

      - name: check if translations are marked as outdated
        shell: bash
        env:
          MASK: '---\n(.*\n)*?((outdated: |outdated_since: ).+\n)(.*\n)*?---'
          PULL_REQUEST_TAG: 'SKIP_OUTDATED_CHECK'
        run: |
          # https://en.wikipedia.org/wiki/ANSI_escape_code#3-bit_and_4-bit
          function echo_red () { echo -e "\e[0;31m$1\e[m"; }
          function echo_green () { echo -e "\e[0;32m$1\e[m"; }

          function print_error () {
            echo "::error::You have edited some original articles (en.md), but did not outdate their translations:"
            while read FILENAME; do
              echo_red "* $FILENAME"
            done <<< "$1"

            echo -e "\nIf your changes DON'T NEED to be added to the translations, add $( echo_red $PULL_REQUEST_TAG ) anywhere in the description of your pull request."
            echo -e "Otherwise, add the following to each article's front matter (https://osu.ppy.sh/wiki/en/Article_styling_criteria/Formatting#front-matter):\n"
            echo_green "---"
            echo_green "outdated: true"
            if [[ -n "$2" ]]; then
              echo_green "outdated_since: $2"
            fi
            echo_green "---"
          }

          function diff_files () { git diff --diff-filter=d --name-only ${{ github.sha }}^ ${{ github.sha }} "$@"; }

          TRANSLATIONS=$( diff_files 'wiki/**/*.md' ':(exclude)*/en.md' )
          ORIGINAL_ARTICLES=$( diff_files 'wiki/**/en.md' )
          if [[ -z "$ORIGINAL_ARTICLES" ]]; then
            echo "::notice::No original articles are edited, exiting"
            exit 0
          fi
          if ${{ contains(github.event.pull_request.body, env.PULL_REQUEST_TAG) }}; then
            echo "::notice::Outdated articles check suppressed ($PULL_REQUEST_TAG tag found in the pull request)"
            exit 0
          fi

          # obtain directories of the modified en.md files, then list translations in them which:
          # - are not modified in the PR (assuming that the author took care of them)
          # - don't have the outdated/outdated_since markers

          MISSED_TRANSLATIONS=$( echo "$ORIGINAL_ARTICLES" | tr \\n \\0 | xargs -0 dirname | sort -u | {
            while read DIRECTORY; do
              for ARTICLE_NAME in "$DIRECTORY"/*.md; do
                if ! [[ "$ARTICLE_NAME" =~ \/(..|..-..)\.md ]] || [[ "$ARTICLE_NAME" =~ \/en\.md ]]; then
                  continue
                fi
                if ! [[ "$TRANSLATIONS" =~ "$ARTICLE_NAME" ]]; then
                  grep -LzP -e "$MASK" "$ARTICLE_NAME" || true  # print only names of non-matching files, with multiline match (considering \n)
                fi
              done
            done
          })

          if [[ -n "$MISSED_TRANSLATIONS" ]]; then
            # get the first commit of the branch associated with the PR; GitHub's ubuntu-latest has curl/jq: https://github.com/actions/virtual-environments
            FIRST_COMMIT_HASH=$( curl -sS ${{ github.event.pull_request.commits_url }}?per_page=1 | jq -r '.[0].sha' || true )
            print_error "$MISSED_TRANSLATIONS" "$FIRST_COMMIT_HASH"
            exit 1
          else
            echo -e "::notice::Either you have edited no original articles, or all translations are properly outdated"
          fi