doc_review.yml

name: AI Doc Review

on:
  workflow_dispatch:

  issue_comment:
    types:
      - created

permissions:
  contents: read
  pull-requests: write
  issues: write

jobs:
  prepare:
    runs-on: ubuntu-latest
    if: >
      github.event_name == 'workflow_dispatch' ||
      (
        github.event_name == 'issue_comment' &&
        contains(github.event.comment.body, '/bot-review') &&
        (github.event.comment.user.login == 'hfxsd' || github.event.comment.user.login == 'likidu' || github.event.comment.user.login == 'lilin90' || github.event.comment.user.login == 'Oreoxmt' || github.event.comment.user.login == 'qiancai')
      )
    outputs:
      zh_files_count: ${{ steps.diff.outputs.zh_files_count }}
      en_files_count: ${{ steps.diff.outputs.en_files_count }}
      review_mode: ${{ steps.extract.outputs.REVIEW_MODE }}
      commit_sha: ${{ steps.extract.outputs.COMMIT_SHA }}
      base_sha: ${{ steps.extract.outputs.BASE_SHA }}
      head_sha: ${{ steps.extract.outputs.HEAD_SHA }}
    steps:

      - name: Checkout Repo
        uses: actions/checkout@v3
        with:
          fetch-depth: 0

      - name: Extract review parameters
        id: extract
        if: github.event_name == 'issue_comment'
        run: |
          COMMENT="${{ github.event.comment.body }}"
          
          # Match commit range
          if [[ "$COMMENT" =~ \/bot-review:[[:space:]]*([a-f0-9]{7,40})[[:space:]]*\.\.[[:space:]]*([a-f0-9]{7,40}) ]]; then
            echo "BASE_SHA=${BASH_REMATCH[1]}" >> $GITHUB_OUTPUT
            echo "HEAD_SHA=${BASH_REMATCH[2]}" >> $GITHUB_OUTPUT
            echo "REVIEW_MODE=commit_range" >> $GITHUB_OUTPUT
            echo "Detected commit range"
          # Match a single commit
          elif [[ "$COMMENT" =~ \/bot-review:[[:space:]]*([a-f0-9]{7,40}) ]]; then
            echo "COMMIT_SHA=${BASH_REMATCH[1]}" >> $GITHUB_OUTPUT
            echo "REVIEW_MODE=single_commit" >> $GITHUB_OUTPUT
            printf "Detected single commit: %s\n" "${BASH_REMATCH[1]}"
          # Match "/bot-review" or "/bot-review "
          elif [[ "$COMMENT" =~ ^\/bot-review[[:space:]]*$ ]]; then
            echo "REVIEW_MODE=latest" >> $GITHUB_OUTPUT
            echo "Detected default review mode"
          # Invalid format
          else
            echo "REVIEW_MODE=invalid" >> $GITHUB_OUTPUT
            echo "Invalid bot-review command format"
          fi

      - name: Get PR diff
        id: diff
        run: |
          # Initialize variables
          PR_NUMBER=""
          FILES=""
          REVIEW_MODE="${{ steps.extract.outputs.REVIEW_MODE }}"
          printf "Review mode detected: %s\n" "$REVIEW_MODE"
          
          # Process single commit mode
          if [[ "$REVIEW_MODE" == "single_commit" ]]; then
            COMMIT_SHA="${{ steps.extract.outputs.COMMIT_SHA }}"
            printf "Processing single commit mode with SHA: %s\n" "$COMMIT_SHA"
            
            FILES=$(git diff-tree --no-commit-id --name-only -r "$COMMIT_SHA" 2>/dev/null || echo "")
            
            if [[ -z "$FILES" ]]; then
              printf "Fetching commit %s\n" "$COMMIT_SHA"
              git fetch --depth=1 origin "$COMMIT_SHA"
              FILES=$(git diff-tree --no-commit-id --name-only -r "$COMMIT_SHA" 2>/dev/null || git show --name-only --format="" "$COMMIT_SHA" 2>/dev/null || echo "")
              
              if [[ -z "$FILES" ]]; then
                printf "No files found for commit %s\n" "$COMMIT_SHA"
                echo "zh_files_count=0" >> "$GITHUB_OUTPUT"
                echo "en_files_count=0" >> "$GITHUB_OUTPUT"
                exit 0
              fi
            fi
            
            ZH_COUNT=0
            EN_COUNT=0
            
            while IFS= read -r file; do
              if [[ "$file" == zh/* ]]; then
                ZH_COUNT=$((ZH_COUNT + 1))
              else
                # Count any non-zh file as English
                EN_COUNT=$((EN_COUNT + 1))
              fi
            done <<< "$FILES"
            
            printf "Files from commit %s:\n" "$COMMIT_SHA"
            echo "$FILES"
            echo "Chinese files: $ZH_COUNT"
            echo "English files: $EN_COUNT"
            
            # Set output variables
            echo "zh_files_count=$ZH_COUNT" >> "$GITHUB_OUTPUT"
            echo "en_files_count=$EN_COUNT" >> "$GITHUB_OUTPUT"
            exit 0
          
          # Process commit range mode
          elif [[ "$REVIEW_MODE" == "commit_range" ]]; then
            BASE_SHA="${{ steps.extract.outputs.BASE_SHA }}"
            HEAD_SHA="${{ steps.extract.outputs.HEAD_SHA }}"
            printf "Processing commit range mode with SHA range: %s..%s\n" "$BASE_SHA" "$HEAD_SHA"
            
            # Try to get files for this commit range
            FILES=$(git diff --name-only "$BASE_SHA" "$HEAD_SHA" 2>/dev/null || echo "")
            
            if [[ -z "$FILES" ]]; then
              printf "Fetching commits %s and %s\n" "$BASE_SHA" "$HEAD_SHA"
              git fetch --depth=1 origin "$BASE_SHA" "$HEAD_SHA"
              FILES=$(git diff --name-only "$BASE_SHA" "$HEAD_SHA" 2>/dev/null || echo "")
              
              if [[ -z "$FILES" ]]; then
                printf "No files found in range %s..%s\n" "$BASE_SHA" "$HEAD_SHA"
                echo "zh_files_count=0" >> "$GITHUB_OUTPUT"
                echo "en_files_count=0" >> "$GITHUB_OUTPUT"
                exit 0
              fi
            fi
            
            ZH_COUNT=0
            EN_COUNT=0
            
            while IFS= read -r file; do
              if [[ "$file" == zh/* ]]; then
                ZH_COUNT=$((ZH_COUNT + 1))
              else
                # Count any non-zh file as English
                EN_COUNT=$((EN_COUNT + 1))
              fi
            done <<< "$FILES"
            
            printf "Files from commit range %s..%s:\n" "$BASE_SHA" "$HEAD_SHA"
            echo "$FILES"
            echo "Chinese files: $ZH_COUNT"
            echo "English files: $EN_COUNT"
            
            # Set output variables
            echo "zh_files_count=$ZH_COUNT" >> "$GITHUB_OUTPUT"
            echo "en_files_count=$EN_COUNT" >> "$GITHUB_OUTPUT"
            exit 0
          
          # Process default/latest mode (entire PR)
          elif [[ "$REVIEW_MODE" == "latest" ]]; then
            # Get PR number
            if [[ "${{ github.event_name }}" == "workflow_dispatch" ]]; then
              if [[ "${{ github.ref }}" =~ ^refs/pull/([0-9]+)/merge$ ]]; then
                PR_NUMBER="${BASH_REMATCH[1]}"
              else
                echo "zh_files_count=0" >> "$GITHUB_OUTPUT"
                echo "en_files_count=0" >> "$GITHUB_OUTPUT"
                exit 0
              fi
            else
              # For issue_comment event - get PR number from different sources
              PR_NUMBER=$(jq --raw-output .pull_request.number "$GITHUB_EVENT_PATH" || jq --raw-output .issue.number "$GITHUB_EVENT_PATH")
              
              # Extract from issue URL if null
              if [[ "$PR_NUMBER" == "null" ]]; then
                ISSUE_URL=$(jq --raw-output .issue.html_url "$GITHUB_EVENT_PATH")
                if [[ "$ISSUE_URL" =~ /pull/([0-9]+)$ ]]; then
                  PR_NUMBER="${BASH_REMATCH[1]}"
                fi
              fi
            fi
            
            if [[ -n "$PR_NUMBER" && "$PR_NUMBER" != "null" ]]; then
              # Default: get files for the entire PR
              FILES=$(gh pr view $PR_NUMBER --repo ${{ github.repository }} --json files --jq '.files[].path')
              if [[ -z "$FILES" ]]; then
                # Fallback to git
                PR_HEAD=$(gh pr view $PR_NUMBER --repo ${{ github.repository }} --json headRefName --jq '.headRefName')
                PR_BASE=$(gh pr view $PR_NUMBER --repo ${{ github.repository }} --json baseRefName --jq '.baseRefName')
                if [[ -n "$PR_HEAD" && -n "$PR_BASE" ]]; then
                  git fetch origin "$PR_HEAD" "$PR_BASE"
                  MERGE_BASE=$(git merge-base "origin/$PR_BASE" "origin/$PR_HEAD")
                  FILES=$(git diff --name-only "$MERGE_BASE" "origin/$PR_HEAD")
                fi
              fi
              
              ZH_COUNT=0
              EN_COUNT=0
              
              while IFS= read -r file; do
                if [[ "$file" == zh/* ]]; then
                  ZH_COUNT=$((ZH_COUNT + 1))
                else
                  # Count any non-zh file as English
                  EN_COUNT=$((EN_COUNT + 1))
                fi
              done <<< "$FILES"
              
              printf "Files from PR %s:\n" "$PR_NUMBER"
              echo "$FILES"
              echo "Chinese files: $ZH_COUNT"
              echo "English files: $EN_COUNT"
              
              echo "zh_files_count=$ZH_COUNT" >> "$GITHUB_OUTPUT"
              echo "en_files_count=$EN_COUNT" >> "$GITHUB_OUTPUT"
            else
              echo "Could not determine PR number"
              echo "zh_files_count=0" >> "$GITHUB_OUTPUT"
              echo "en_files_count=0" >> "$GITHUB_OUTPUT"
            fi
          else
            echo "Invalid review mode"
            echo "zh_files_count=0" >> "$GITHUB_OUTPUT"
            echo "en_files_count=0" >> "$GITHUB_OUTPUT"
          fi
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

  review-zh:
    needs: prepare
    if: needs.prepare.outputs.zh_files_count > 0
    runs-on: ubuntu-latest
    steps:
      - name: Checkout Repo
        uses: actions/checkout@v3
        with:
          fetch-depth: 0

      - name: AI Doc Reviewer for Chinese docs
        uses: qiancai/ai-codereviewer@main
        continue-on-error: false
        with:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
          API_PROVIDER: "deepseek"
          DEEPSEEK_API_KEY: ${{ secrets.DEEPSEEK_API_KEY }}
          DEEPSEEK_API_MODEL: "deepseek-chat"
          exclude: "**/*.json,!zh/**" # Exclude JSON files and non-Chinese docs
          REVIEW_MODE: ${{ needs.prepare.outputs.review_mode }}
          COMMIT_SHA: ${{ needs.prepare.outputs.commit_sha }}
          BASE_SHA: ${{ needs.prepare.outputs.base_sha }}
          HEAD_SHA: ${{ needs.prepare.outputs.head_sha }}
          PROMPT_PATH: "doc-review-prompt-zh.txt"

  review-en:
    needs: prepare
    if: needs.prepare.outputs.en_files_count > 0
    runs-on: ubuntu-latest
    steps:
      - name: Checkout Repo
        uses: actions/checkout@v3
        with:
          fetch-depth: 0

      - name: AI Doc Reviewer for English docs
        uses: qiancai/ai-codereviewer@main
        continue-on-error: false
        with:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
          API_PROVIDER: "openai"
          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
          OPENAI_API_MODEL: "gpt-4o"
          exclude: "**/*.json,zh/**" # Exclude JSON files and Chinese docs
          REVIEW_MODE: ${{ needs.prepare.outputs.review_mode }}
          COMMIT_SHA: ${{ needs.prepare.outputs.commit_sha }}
          BASE_SHA: ${{ needs.prepare.outputs.base_sha }}
          HEAD_SHA: ${{ needs.prepare.outputs.head_sha }}
          PROMPT_PATH: "doc-review-prompt-en.txt"