auto-assign-reviewers.yml

#
# Copyright (c) 2006-2025, RT-Thread Development Team
#
# SPDX-License-Identifier: Apache-2.0
#
# Change Logs:
# Date           Author       Notes
# 2025-01-21     kurisaW      Initial version
# 2025-03-14     hydevcode
# 2025-05-10     kurisaW      Fixed file existence, cache, and comment time issues
# 2025-05-11     kurisaW      Fixed missing unique files creation and cache logic

# Script Function Description: Assign PR reviews based on the MAINTAINERS list.

name: Auto Review Assistant

on:
  pull_request_target:
    branches: [ master ]
    types: [opened, synchronize, reopened]

jobs:
  assign-reviewers:
    runs-on: ubuntu-22.04
    if: github.repository_owner == 'RT-Thread'
    permissions:
      issues: read
      pull-requests: write
      contents: read
    steps:
      - name: Extract PR number
        id: extract-pr
        run: |
          PR_NUMBER=${{ github.event.pull_request.number }}
          echo "PR_NUMBER=${PR_NUMBER}" >> $GITHUB_OUTPUT

      - name: Checkout code
        uses: actions/checkout@v4
        with:
          ref: master
          sparse-checkout: MAINTAINERS
          persist-credentials: false

      - name: Get changed files
        id: changed_files
        run: |
          # 通过 GitHub API 获取 PR 的变更文件列表
          changed_files=$(curl -s \
            "https://api.github.com/repos/${{ github.repository }}/pulls/${{ steps.extract-pr.outputs.PR_NUMBER }}/files" | \
            jq -r '.[].filename')  # 使用 jq 提取文件名
          echo "$changed_files" > changed_files.txt
          
          existing_comment=$(curl -s \
            "https://api.github.com/repos/${{ github.repository }}/issues/${{ steps.extract-pr.outputs.PR_NUMBER }}/comments" | \
            jq -r '.[] | select(.user.login == "github-actions[bot]") | {body: .body} | @base64')

          echo "=== Changed Files ==="
          cat changed_files.txt
          echo "====================="

          comment_body=""
          if [[ ! -z "$existing_comment" ]]; then
            comment_body=$(echo "$existing_comment" | head -1 | base64 -d | jq -r .body | sed -nE 's/.*Last Updated: ([0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2} UTC).*/\1/p')
            comment_time=$(date -d "$comment_body" +%s)
            echo "${comment_body}"
            echo "COMMENT_TIME=${comment_time}" >> $GITHUB_OUTPUT
          else
            comment_time=""
            echo "COMMENT_TIME=${comment_time}" >> $GITHUB_OUTPUT
          fi
          echo "COMMENT_TIME=${comment_time}"

      - name: Parse MAINTAINERS file
        id: parse_maintainer
        run: |
          # 使用 AWK 解析 MAINTAINERS 文件格式：
          # 提取 tag(标签)、path(路径)和 owners(维护者 GitHub ID)
          awk '
            /^tag:/ { 
              tag = substr($0, index($0, $2))  # 提取标签内容
            }
            /^path:/ {
              # 提取 path 字段并去除前后空格
              path = substr($0, index($0, $2))
              gsub(/^[ \t]+|[ \t]+$/, "", path)  # 清理前后空格和制表符
            }
            /^owners:/ {
              owners = substr($0, index($0, $2))  # 提取维护者信息
              split(owners, parts, /[()]/)       # 拆分出 GitHub ID（括号内内容）
              github_ids = ""
              for (i=2; i<=length(parts); i+=2) {
                github_ids = github_ids "@" parts[i] " "  # 拼接为 @user 格式
              }
              print tag "|" path "|" github_ids
            }
          ' MAINTAINERS > tag_data.csv

      - name: Generate reviewers list
        id: generate_reviewers
        run: |
          rm -f triggered_reviewers.txt triggered_tags.txt unique_reviewers.txt unique_tags.txt
          touch triggered_reviewers.txt triggered_tags.txt unique_reviewers.txt unique_tags.txt

          while IFS='|' read -r tag path reviewers; do
            # 转义路径中的正则特殊字符
            escaped_path=$(sed 's/[.[\*^$]/\\&/g' <<< "$path")
            
            # 使用增强型正则匹配路径及其所有子目录
            if grep -qE "^$escaped_path(/.*)*" changed_files.txt; then
              echo "$reviewers" | tr -s ' ' '\n' | sed '/^$/d' >> triggered_reviewers.txt
              echo "$tag" >> triggered_tags.txt
              echo "Matched: $path → $tag"
            fi
          done < tag_data.csv

          # 生成去重的 unique_reviewers.txt 和 unique_tags.txt
          sort -u triggered_reviewers.txt > unique_reviewers.txt
          sort -u triggered_tags.txt > unique_tags.txt

          # 检查是否有匹配的 reviewers
          if [[ -s unique_reviewers.txt ]]; then
            echo "HAS_REVIEWERS=true" >> $GITHUB_OUTPUT
          else
            echo "HAS_REVIEWERS=false" >> $GITHUB_OUTPUT
          fi

          echo "=== Matched Paths ==="
          cat unique_tags.txt
          echo "=== Matched Reviewers ==="
          cat unique_reviewers.txt

      - name: Restore Reviewers Cache
        id: reviewers-cache-restore 
        if: ${{ steps.changed_files.outputs.COMMENT_TIME != '' }}
        uses: actions/cache/restore@v4
        with: 
          path: |
            unique_tags_bak.txt
            unique_reviewers_bak.txt
          key: ${{ runner.os }}-auto-assign-reviewers-${{ steps.extract-pr.outputs.PR_NUMBER }}-${{ steps.changed_files.outputs.COMMENT_TIME }}

      - name: Get approval status
        id: get_approval
        run: |
          current_time=$(date -u +"%Y-%m-%d %H:%M UTC")
          
          # 检查 unique_reviewers.txt 是否存在且非空
          if [[ ! -s unique_reviewers.txt ]]; then
            echo "No reviewers found, creating empty unique_reviewers.txt"
            touch unique_reviewers.txt
          fi
          
          reviewers=$(cat unique_reviewers.txt | tr '\n' '|' | sed 's/|$//')
          
          # 获取 PR 的所有评论
          comments=$(curl -s \
            "https://api.github.com/repos/${{ github.repository }}/issues/${{ steps.extract-pr.outputs.PR_NUMBER }}/comments")

          echo '#!/bin/bash' > approval_data.sh
          echo 'declare -A approvals=()' >> approval_data.sh
          
          # 使用 jq 解析包含 LGTM 的有效评论
          jq -r --arg reviewers "$reviewers" '
            .[] | 
            select(.user.login != "github-actions[bot]") |  # 排除 bot 的评论
            select(.body | test("^\\s*LGTM\\s*$"; "i")) |   # 匹配 LGTM 评论（不区分大小写）
            .user.login as $user |
            "@\($user)" as $mention |
            select($mention | inside($reviewers)) |         # 过滤有效审查者
            "approvals[\"\($mention)\"]=\"\(.created_at)\""  # 记录审批时间
          ' <<< "$comments" >> approval_data.sh
          
          # 加载审查数据并生成状态报告
          chmod +x approval_data.sh
          source ./approval_data.sh

          jq -r --arg reviewers "$reviewers" '
            .[] | 
            select(.user.login != "github-actions[bot]") |  # 排除 bot 的评论
            select(.body | test("^\\s*LGTM\\s*$"; "i")) |   # 匹配 LGTM 评论（不区分大小写）
            .user.login as $user |
            "@\($user)" as $mention |
            select($mention | inside($reviewers)) |         # 过滤有效审查者
            "\($mention) \(.created_at)"                   # 输出审查者和时间
          ' <<< "$comments" >> approval_data.txt

          notified_users=""
          if [[ -f unique_reviewers_bak.txt ]]; then
            notified_users=$(cat unique_reviewers_bak.txt | xargs)
          else
            notified_users=""
          fi

          {
            echo "---"
            echo "### 📊 Current Review Status (Last Updated: $current_time)"
            while read -r reviewer; do
              formatted_reviewers=""
              for r in $reviewers; do
                if [[ " ${notified_users[@]} " =~ " $reviewer " ]]; then
                    formatted_reviewers+="${reviewer#@}"
                else
                    formatted_reviewers+="$reviewer"
                fi
              done
            
              if [[ -n "${approvals[$reviewer]}" ]]; then
                timestamp=$(date -d "${approvals[$reviewer]}" -u +"%Y-%m-%d %H:%M UTC")
                echo "- ✅ **$formatted_reviewers** Reviewed On $timestamp"
              else
                echo "- ⌛ **$formatted_reviewers** Pending Review"
              fi
            done < unique_reviewers.txt
          } > review_status.md
          
          echo "CURRENT_TIME=${current_time}" >> $GITHUB_OUTPUT

      - name: Generate review data
        id: generate_review
        if: steps.generate_reviewers.outputs.HAS_REVIEWERS == 'true'
        run: |
          unique_tags=""
          if [[ -s unique_tags.txt ]]; then
            unique_tags=$(cat unique_tags.txt | xargs)
          fi
          unique_tags_bak=""
          if [[ -f unique_tags_bak.txt ]]; then
            unique_tags_bak=$(cat unique_tags_bak.txt | xargs)
          fi

          existing_tags=""
          for r in $unique_tags; do
            if [[ " ${unique_tags_bak[@]} " =~ " $r " ]]; then
              echo "$r 不存在于数组中"
            else
              existing_tags+="$r "
            fi
          done

          current_time=$(date -u +"%Y-%m-%d %H:%M UTC")
          {
            # 生成审查分配信息
            echo "## 📌 Code Review Assignment"
            echo ""

            while IFS='|' read -r tag path reviewers; do
              if grep -qE "^$path(/|$)" changed_files.txt; then
                echo "### 🏷️ Tag: $tag"
                echo "**Path:** \`$path\`  "

                if [[ " ${existing_tags[@]} " =~ " $tag " ]]; then
                  echo "**Reviewers:** $reviewers  "
                else
                  formatted_reviewers=""
                  for r in $reviewers; do
                    formatted_reviewers+="${r#@} "
                  done
                  echo "**Reviewers:** $formatted_reviewers  "
                fi

                echo "<details>"
                echo "<summary><b>Changed Files</b> (Click to expand)</summary>"
                echo ""
                grep -E "^$path(/|$)" changed_files.txt | sed 's/^/- /'  # 列出匹配的变更文件
                echo ""
                echo "</details>"
                echo ""
              fi
            done < tag_data.csv
            # 插入审查状态
            cat review_status.md

            echo "---"
            echo "### 📝 Review Instructions"
            echo ""
            echo "1. **维护者可以通过单击此处来刷新审查状态:** [🔄 刷新状态](${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }})"
            echo "   **Maintainers can refresh the review status by clicking here:** [🔄 Refresh Status](${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }})"
            echo ""
            echo "2. **确认审核通过后评论 \`LGTM/lgtm\`**"
            echo "   **Comment \`LGTM/lgtm\` after confirming approval**"
            echo ""
            echo "3. **PR合并前需至少一位维护者确认**"
            echo "   **PR must be confirmed by at least one maintainer before merging**"
            echo ""
            echo "> ℹ️ **刷新CI状态操作需要具备仓库写入权限。**"
            echo "> ℹ️ **Refresh CI status operation requires repository Write permission.**"
          } > review_data.md

      - name: Post/Update comment
        id: post_comment
        if: steps.generate_reviewers.outputs.HAS_REVIEWERS == 'true'
        run: |
          # 查找现有的 bot 评论
          existing_comment=$(curl -s \
            -H "Authorization: Bearer ${{ secrets.GITHUB_TOKEN }}" \
            "https://api.github.com/repos/${{ github.repository }}/issues/${{ steps.extract-pr.outputs.PR_NUMBER }}/comments" | \
            jq -r '.[] | select(.user.login == "github-actions[bot]") | {id: .id, body: .body} | @base64')
          
          if [[ -n "$existing_comment" ]]; then
            # 更新现有评论
            comment_id=$(echo "$existing_comment" | head -1 | base64 -d | jq -r .id)
            echo "Updating existing comment $comment_id"
            response=$(curl -s -X PATCH \
              -H "Authorization: Bearer ${{ secrets.GITHUB_TOKEN }}" \
              -d "$(jq -n --arg body "$(cat review_data.md)" '{body: $body}')" \
              "https://api.github.com/repos/${{ github.repository }}/issues/comments/$comment_id")
          else
            # 创建新评论
            echo "Creating new comment"
            response=$(curl -s -X POST \
              -H "Authorization: Bearer ${{ secrets.GITHUB_TOKEN }}" \
              -d "$(jq -n --arg body "$(cat review_data.md)" '{body: $body}')" \
              "https://api.github.com/repos/${{ github.repository }}/issues/${{ steps.extract-pr.outputs.PR_NUMBER }}/comments")
          fi

      - name: Get Comment Time
        id: get_comment_time
        if: steps.generate_reviewers.outputs.HAS_REVIEWERS == 'true'
        run: |
          existing_comment=$(curl -s \
            "https://api.github.com/repos/${{ github.repository }}/issues/${{ steps.extract-pr.outputs.PR_NUMBER }}/comments" | \
            jq -r '.[] | select(.user.login == "github-actions[bot]") | {body: .body} | @base64')
          comment_body="${{ steps.get_approval.outputs.CURRENT_TIME }}"
          comment_time=$(date -d "$comment_body" +%s)
          echo "CURRENT_TIME=${comment_time}" >> $GITHUB_OUTPUT
          cp unique_reviewers.txt unique_reviewers_bak.txt
          cp unique_tags.txt unique_tags_bak.txt

      - name: Save Reviewers Cache
        id: reviewers-cache-save
        if: steps.generate_reviewers.outputs.HAS_REVIEWERS == 'true'
        uses: actions/cache/save@v4
        with:
          path: |
            unique_tags_bak.txt
            unique_reviewers_bak.txt
          key: ${{ runner.os }}-auto-assign-reviewers-${{ steps.extract-pr.outputs.PR_NUMBER }}-${{ steps.get_comment_time.outputs.CURRENT_TIME }}