docs/.github/workflows/create-translation-batch-pr.yml

name: Create translation Batch Pull Request

# **What it does**:
#  - Creates one pull request per language after running a series of automated checks,
#    removing translations that are broken in any known way
# **Why we have it**:
#  - To deploy translations
# **Who does it impact**: It automates what would otherwise be manual work,
#   helping docs engineering focus on higher value work

on:
  workflow_dispatch:
  schedule:
    - cron: '02 17 * * *' # Once a day at 17:02 UTC / 9:02 PST

permissions:
  contents: write

jobs:
  create-translation-batch:
    name: Create translation batch
    if: github.repository == 'github/docs-internal'
    runs-on: ubuntu-latest
    # A sync's average run time is ~3.2 hours.
    # This sets a maximum execution time of 300 minutes (5 hours) to prevent the workflow from running longer than necessary.
    timeout-minutes: 300
    strategy:
      fail-fast: false
      max-parallel: 1
      matrix:
        include:
          - language: pt
            crowdin_language: pt-BR
            language_dir: translations/pt-BR

          - language: es
            crowdin_language: es-ES
            language_dir: translations/es-ES

          - language: cn
            crowdin_language: zh-CN
            language_dir: translations/zh-CN

          - language_dir: translations/ja-JP
            crowdin_language: ja
            language: ja

    steps:
      - name: Set branch name
        id: set-branch
        run: |
          echo "::set-output name=BRANCH_NAME::translation-batch-${{ matrix.language }}-$(date +%Y-%m-%d__%H-%M)"

      - run: git config --global user.name "docubot"
      - run: git config --global user.email "67483024+docubot@users.noreply.github.com"

      - name: Checkout
        uses: actions/checkout@dcd71f646680f2efd8db4afa5ad64fdcba30e748
        with:
          fetch-depth: 0
          lfs: true

      - run: git checkout -b ${{ steps.set-branch.outputs.BRANCH_NAME }}

      - name: Remove unwanted git hooks
        run: rm .git/hooks/post-checkout

      # https://support.crowdin.com/cli-tool/#debian
      - name: Download and install the public signing key
        run: wget -qO - https://artifacts.crowdin.com/repo/GPG-KEY-crowdin | sudo apt-key add -
      - name: Create the crowdin.list file in the /etc/apt/sources.list.d directory
        run: |
          sudo touch /etc/apt/sources.list.d/crowdin.list
          echo "deb https://artifacts.crowdin.com/repo/deb/ /" | sudo tee -a /etc/apt/sources.list.d/crowdin.list
      - name: Install the Crowdin CLI Debian Package
        run: sudo apt-get update && sudo apt-get install crowdin3

      # Delete empty source files that would be rejected by Crowdin breaking the workflow
      - name: Remove empty source files
        run: |
          find content -type f -empty -delete
          find data    -type f -empty -delete

      - name: Upload files to crowdin
        run: crowdin upload sources --delete-obsolete --no-progress --no-colors --verbose --debug '--branch=main' '--config=crowdin.yml'
        env:
          # This is a numeric id, not to be confused with Crowdin API v1 "project identifier" string
          # See "API v2" on https://crowdin.com/project/<your-project>/settings#api
          CROWDIN_PROJECT_ID: ${{ secrets.CROWDIN_PROJECT_ID }}

          # A personal access token, not to be confused with Crowdin API v1 "API key"
          # See https://crowdin.com/settings#api-key to generate a token
          # This token was created by logging into Crowdin with the octoglot user
          CROWDIN_PERSONAL_TOKEN: ${{ secrets.CROWDIN_PERSONAL_TOKEN }}

      - name: Remove all language translations
        run: |
          git rm -rf --quiet ${{ matrix.language_dir }}/content
          git rm -rf --quiet ${{ matrix.language_dir }}/data

      - name: Download crowdin translations
        run: crowdin download  --no-progress --no-colors --verbose --debug '--branch=main' '--config=crowdin.yml' --language=${{ matrix.crowdin_language }}
        env:
          # This is a numeric id, not to be confused with Crowdin API v1 "project identifier" string
          # See "API v2" on https://crowdin.com/project/<your-project>/settings#api
          CROWDIN_PROJECT_ID: ${{ secrets.CROWDIN_PROJECT_ID }}

          # A personal access token, not to be confused with Crowdin API v1 "API key"
          # See https://crowdin.com/settings#api-key to generate a token
          # This token was created by logging into Crowdin with the octoglot user
          CROWDIN_PERSONAL_TOKEN: ${{ secrets.CROWDIN_PERSONAL_TOKEN }}

      - name: Commit crowdin sync
        run: |
          git add ${{ matrix.language_dir }}
          git commit -m "Add crowdin translations" || echo "Nothing to commit"

      - name: 'Setup node'
        uses: actions/setup-node@1f8c6b94b26d0feae1e387ca63ccbdc44d27b561
        with:
          node-version: 16.14.x

      - run: npm ci

      # step 6 in docs-engineering/crowdin.md
      - name: Homogenize frontmatter
        run: |
          node script/i18n/homogenize-frontmatter.js
          git add ${{ matrix.language_dir }} && git commit -m "Run script/i18n/homogenize-frontmatter.js" || echo "Nothing to commit"

      # step 7 in docs-engineering/crowdin.md
      - name: Fix translation errors
        run: |
          node script/i18n/fix-translation-errors.js
          git add ${{ matrix.language_dir }} && git commit -m "Run script/i18n/fix-translation-errors.js" || echo "Nothing to commit"

      # step 8a in docs-engineering/crowdin.md
      - name: Check parsing
        run: |
          node script/i18n/lint-translation-files.js --check parsing | tee -a /tmp/batch.log | cat
          git add ${{ matrix.language_dir }} && git commit -m "Run script/i18n/lint-translation-files.js --check parsing" || echo "Nothing to commit"

      # step 8b in docs-engineering/crowdin.md
      - name: Check rendering
        run: |
          node script/i18n/lint-translation-files.js --check rendering | tee -a /tmp/batch.log | cat
          git add ${{ matrix.language_dir }} && git commit -m "Run script/i18n/lint-translation-files.js --check rendering" || echo "Nothing to commit"

      - name: Reset files with broken liquid tags
        run: |
          node script/i18n/reset-files-with-broken-liquid-tags.js --language=${{ matrix.language }} | tee -a /tmp/batch.log | cat
          git add ${{ matrix.language_dir }} && git commit -m "run script/i18n/reset-files-with-broken-liquid-tags.js --language=${{ matrix.language }}" || echo "Nothing to commit"

      # step 5 in docs-engineering/crowdin.md using script from docs-internal#22709
      - name: Reset known broken files
        run: |
          node script/i18n/reset-known-broken-translation-files.js | tee -a /tmp/batch.log | cat
          git add ${{ matrix.language_dir }} && git commit -m "run script/i18n/reset-known-broken-translation-files.js" || echo "Nothing to commit"
        env:
          GITHUB_TOKEN: ${{ secrets.DOCUBOT_REPO_PAT }}

      - name: Check in CSV report
        run: |
          mkdir -p translations/log
          csvFile=translations/log/${{ matrix.language }}-resets.csv
          script/i18n/report-reset-files.js --report-type=csv --language=${{ matrix.language }} --log-file=/tmp/batch.log > $csvFile
          git add -f $csvFile && git commit -m "Check in ${{ matrix.language }} CSV report" || echo "Nothing to commit"

      - name: Write the reported files that were reset to /tmp/pr-body.txt
        run: script/i18n/report-reset-files.js --report-type=pull-request-body --language=${{ matrix.language }} --log-file=/tmp/batch.log > /tmp/pr-body.txt

      - name: Push filtered translations
        run: git push origin ${{ steps.set-branch.outputs.BRANCH_NAME }}

      - name: Close existing stale batches
        uses: lee-dohm/close-matching-issues@e9e43aad2fa6f06a058cedfd8fb975fd93b56d8f
        with:
          token: ${{ secrets.OCTOMERGER_PAT_WITH_REPO_AND_WORKFLOW_SCOPE }}
          query: 'type:pr label:translation-batch-${{ matrix.language }}'

      - name: Create translation batch pull request
        env:
          GITHUB_TOKEN: ${{ secrets.DOCUBOT_REPO_PAT }}
          TITLE: 'New translation batch for ${{ matrix.language }}'
          BASE: 'main'
          HEAD: ${{ steps.set-branch.outputs.BRANCH_NAME }}
          LANGUAGE: ${{ matrix.language }}
          BODY_FILE: '/tmp/pr-body.txt'
        run: .github/actions-scripts/create-translation-batch-pr.js

      - name: Approve PR
        if: github.ref_name == 'main'
        env:
          GITHUB_TOKEN: ${{ secrets.OCTOMERGER_PAT_WITH_REPO_AND_WORKFLOW_SCOPE }}
        run: gh pr review --approve || echo "Nothing to approve"

      - name: Set auto-merge
        if: github.ref_name == 'main'
        env:
          GITHUB_TOKEN: ${{ secrets.OCTOMERGER_PAT_WITH_REPO_AND_WORKFLOW_SCOPE }}
        run: gh pr merge ${{ steps.set-branch.outputs.BRANCH_NAME }} --auto --squash || echo "Nothing to merge"

      # When the maximum execution time is reached for this job, Actions cancels the workflow run.
      # This emits a notification for the first responder to triage.
      - name: Send Slack notification if workflow is cancelled
        uses: someimportantcompany/github-actions-slack-message@f8d28715e7b8a4717047d23f48c39827cacad340
        if: cancelled()
        with:
          channel: ${{ secrets.DOCS_ALERTS_SLACK_CHANNEL_ID }}
          bot-token: ${{ secrets.SLACK_DOCS_BOT_TOKEN }}
          color: failure
          text: 'The new translation batch for ${{ matrix.language }} was cancelled.'

      # Emit a notification for the first responder to triage if the workflow failed.
      - name: Send Slack notification if workflow failed
        uses: someimportantcompany/github-actions-slack-message@f8d28715e7b8a4717047d23f48c39827cacad340
        if: failure()
        with:
          channel: ${{ secrets.DOCS_ALERTS_SLACK_CHANNEL_ID }}
          bot-token: ${{ secrets.SLACK_DOCS_BOT_TOKEN }}
          color: failure
          text: 'The new translation batch for ${{ matrix.language }} failed.'