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
# 2025-07-14     kurisaW      Merge same tag with different paths, remove Path display from CI comment

# 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]

concurrency:
  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
  cancel-in-progress: true

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（带重试机制和错误处理）
          max_retries=3
          retry_count=0
          changed_files=""
          api_response=""
          
          echo "Fetching changed files for PR #${{ steps.extract-pr.outputs.PR_NUMBER }}..."
          
          while [ $retry_count -lt $max_retries ]; do
            api_response=$(curl -s \
              -H "Authorization: Bearer ${{ secrets.GITHUB_TOKEN }}" \
              -H "Accept: application/vnd.github.v3+json" \
              "https://api.github.com/repos/${{ github.repository }}/pulls/${{ steps.extract-pr.outputs.PR_NUMBER }}/files")
            
            # 验证响应是否为有效JSON且包含文件数组
            if jq -e 'if type=="array" then .[0].filename else empty end' <<<"$api_response" >/dev/null 2>&1; then
              changed_files=$(jq -r '.[].filename' <<<"$api_response")
              break
            else
              echo "Retry $((retry_count+1)): API response not ready or invalid format"
              echo "API Response: $api_response"
              sleep 5
              ((retry_count++))
            fi
          done
          
          if [ -z "$changed_files" ]; then
            echo "Error: Failed to get changed files after $max_retries attempts"
            echo "Final API Response: $api_response"
            exit 1
          fi
          
          echo "$changed_files" > changed_files.txt
          echo "Successfully fetched $(wc -l < changed_files.txt) changed files"
          
          # 以下是原有的评论处理逻辑（保持不变）
          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")
            
          # Check if response is valid JSON
          if jq -e . >/dev/null 2>&1 <<<"$existing_comment"; then
            existing_comment=$(jq -r '.[] | select(.user.login == "github-actions[bot]" and (.body | contains("<!-- Auto Review Assistant Comment -->"))) | {body: .body} | @base64' <<< "$existing_comment")
          else
            existing_comment=""
            echo "Warning: Invalid JSON response from GitHub API for comments"
            echo "Response: $existing_comment"
          fi

          comment_body=""
          if [[ ! -z "$existing_comment" ]]; then
            comment_body=$(echo "$existing_comment" | head -1 | base64 -d | jq -r .body | sed -nE 's/.*Last Updated: ([0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2} CST).*/\1/p')
            comment_time=$(TZ='Asia/Shanghai' date -d "$comment_body" +%s)
            echo "CACHE_TIMESTAMP=${comment_time}" >> $GITHUB_OUTPUT  # 统一使用这个变量名
            echo "COMMENT_TIME=${comment_time}" >> $GITHUB_OUTPUT
          else
            comment_time=""
            echo "CACHE_TIMESTAMP=" >> $GITHUB_OUTPUT
            echo "COMMENT_TIME=" >> $GITHUB_OUTPUT
          fi
          echo "Debug - CACHE_TIMESTAMP: $comment_time"

      - name: Parse MAINTAINERS file
        id: parse_maintainer
        run: |
          set -euo pipefail
          awk '
            BEGIN{ tag=""; paths=""; owners="" }
            /^tag:/ { 
                tag = substr($0, index($0, $2)); 
                paths=""; owners=""
            }
            /^path:/ {
                path = substr($0, index($0, $2))
                gsub(/^[ \t]+|[ \t]+$/, "", path)
                paths = (paths == "" ? path : paths "|" path)
            }
            /^owners:/ {
                owners = substr($0, index($0, $2))
                n = split(owners, parts, /[()]/)
                github_ids=""
                for (i=2; i<=n; i+=2) {
                  id=parts[i]
                  gsub(/^[ \t@]+|[ \t]+$/, "", id)
                  if(id != "") github_ids=github_ids "@" id " "
                }
                print tag "|" paths "|" github_ids
                tag=""; paths=""; owners=""
            }
          ' MAINTAINERS > tag_data.csv

      - name: Generate reviewers list and tag-file mapping
        id: generate_reviewers
        run: |
          rm -f triggered_reviewers.txt triggered_tags.txt unique_reviewers.txt unique_tags.txt tag_files_map.json tag_reviewers_map.txt
          touch triggered_reviewers.txt triggered_tags.txt unique_reviewers.txt unique_tags.txt

          # 1. 读取 tag_data.csv，建立 tag -> [paths], tag -> reviewers
          declare -A tag_paths_map
          declare -A tag_reviewers_map

          while IFS='|' read -r tag paths reviewers; do
            IFS='|' read -ra path_arr <<< "$paths"
            for p in "${path_arr[@]}"; do
              tag_paths_map["$tag"]+="$p;"
            done
            # 合并 reviewers，去重，只保留合法格式
            existing_reviewers="${tag_reviewers_map["$tag"]}"
            all_reviewers="$existing_reviewers $reviewers"
            # 只保留 @xxx 格式，去重
            all_reviewers=$(echo "$all_reviewers" | grep -o '@[A-Za-z0-9_-]\+' | sort -u | tr '\n' ' ')
            tag_reviewers_map["$tag"]="$all_reviewers"
          done < tag_data.csv

          # 2. 针对每个 tag，找出它所有 path 匹配的变更文件
          declare -A tag_changedfiles_map
          while IFS= read -r changed; do
            for tag in "${!tag_paths_map[@]}"; do
              IFS=';' read -ra tpaths <<< "${tag_paths_map[$tag]}"
              for tpath in "${tpaths[@]}"; do
                [[ -z "$tpath" ]] && continue
                if [[ -f "$tpath" ]]; then
                  # 精确文件名
                  [[ "$changed" == "$tpath" ]] && tag_changedfiles_map["$tag"]+="$changed;"
                else
                  # 目录前缀
                  [[ "$changed" == $tpath* ]] && tag_changedfiles_map["$tag"]+="$changed;"
                fi
              done
            done
          done < changed_files.txt

          # 3. 输出合并后的 tag reviewers、tag、并去重
          for tag in "${!tag_changedfiles_map[@]}"; do
            reviewers="${tag_reviewers_map[$tag]}"
            echo "$reviewers" | tr -s ' ' '\n' | sed '/^$/d' >> triggered_reviewers.txt
            echo "$tag" >> triggered_tags.txt
          done

          # 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

          # 4. 输出 tag_files_map.json，格式 { "tag1": ["file1","file2"], ... }
          {
            echo "{"
            first_tag=1
            for tag in "${!tag_changedfiles_map[@]}"; do
              [[ $first_tag -eq 0 ]] && echo ","
              echo -n "  \"${tag}\": ["
              IFS=';' read -ra files <<< "${tag_changedfiles_map[$tag]}"
              file_list=""
              for f in "${files[@]}"; do
                [[ -z "$f" ]] && continue
                [[ -n "$file_list" ]] && file_list+=", "
                file_list+="\"$f\""
              done
              echo -n "$file_list"
              echo -n "]"
              first_tag=0
            done
            echo ""
            echo "}"
          } > tag_files_map.json

          # 5. 保存聚合去重后的 reviewers 到 tag_reviewers_map.txt
          {
            for tag in "${!tag_reviewers_map[@]}"; do
              echo "$tag|${tag_reviewers_map[$tag]}"
            done
          } > tag_reviewers_map.txt

          # 6. 标记是否有 reviewer
          if [[ -s unique_reviewers.txt ]]; then
            echo "HAS_REVIEWERS=true" >> $GITHUB_OUTPUT
          else
            echo "HAS_REVIEWERS=false" >> $GITHUB_OUTPUT
          fi

          echo "=== Matched Tags ==="
          cat unique_tags.txt
          echo "=== Matched Reviewers ==="
          cat unique_reviewers.txt
          echo "=== Tag-ChangedFiles Map ==="
          cat tag_files_map.json

      - name: Restore Reviewers Cache
        id: reviewers-cache-restore
        if: ${{ steps.changed_files.outputs.CACHE_TIMESTAMP != '' }}
        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 }}-${{ steps.changed_files.outputs.CACHE_TIMESTAMP }}-${{ github.run_id }}
          restore-keys: |
            ${{ runner.os }}-auto-assign-reviewers-${{ steps.extract-pr.outputs.PR_NUMBER }}-${{ steps.changed_files.outputs.CACHE_TIMESTAMP }}-
            ${{ runner.os }}-auto-assign-reviewers-${{ steps.extract-pr.outputs.PR_NUMBER }}-

      - name: Get approval status
        id: get_approval
        run: |
          current_time=$(TZ='Asia/Shanghai' date +"%Y-%m-%d %H:%M CST")
          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[$reviewer]}" ]]; then
                timestamp=$(TZ='Asia/Shanghai' date -d "${approvals[$reviewer]}" +"%Y-%m-%d %H:%M CST")
                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 (tag merge, no path in comment, changed files summary per tag)
        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
          # 读取 tag->files 映射
          declare -A tag_files_map
          eval "$(jq -r 'to_entries[] | "tag_files_map[\"\(.key)\"]=\"\(.value | join(";"))\"" ' tag_files_map.json)"
          # 读取 tag->reviewers（只读聚合去重后的结果）
          declare -A tag_reviewers_map
          while IFS='|' read -r tag reviewers; do
            tag_reviewers_map["$tag"]="$reviewers"
          done < tag_reviewers_map.txt
          # 获取已通知的 reviewers
          notified_users=""
          if [[ -f unique_reviewers_bak.txt ]]; then
            notified_users=$(cat unique_reviewers_bak.txt | xargs)
          fi
          current_time=$(TZ='Asia/Shanghai' date +"%Y-%m-%d %H:%M CST")
          {
            echo "<!-- Auto Review Assistant Comment -->"
            echo "## 📌 Code Review Assignment"
            echo ""
            for tag in $unique_tags; do
              reviewers="${tag_reviewers_map[$tag]}"
              # 移除尾部空格并提取有效的@username格式
              reviewers=$(echo "$reviewers" | sed 's/[[:space:]]*$//' | grep -o '@[A-Za-z0-9_-]\+' | sort -u | tr '\n' ' ')
              
              # 格式化reviewers显示（仅对已通知用户去掉@）
              formatted_reviewers=""
              for reviewer in $reviewers; do
                if [[ " ${notified_users[@]} " =~ " $reviewer " ]]; then
                  formatted_reviewers+="${reviewer#@} "  # 已通知用户去掉@
                else
                  formatted_reviewers+="$reviewer "      # 未通知用户保留@
                fi
              done
              
              echo "### 🏷️ Tag: $tag"
              echo ""
              echo "**Reviewers:** $formatted_reviewers"  # 确保显示Reviewers
              echo "<details>"
              echo "<summary><b>Changed Files</b> (Click to expand)</summary>"
              echo ""
              IFS=';' read -ra files <<< "${tag_files_map[$tag]}"
              for file in "${files[@]}"; do
                [[ -z "$file" ]] && continue
                echo "- $file"
              done
              echo ""
              echo "</details>"
              echo ""
            done
            # 插入审查状态
            cat review_status.md
            echo "---"
            echo "### 📝 Review Instructions"
            echo ""
            echo "1. **维护者可以通过单击此处来刷新审查状态:** [🔄 刷新状态](${{ 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]" and (.body | contains("<!-- Auto Review Assistant Comment -->"))) | {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 \
            -H "Authorization: Bearer ${{ secrets.GITHUB_TOKEN }}" \
            "https://api.github.com/repos/${{ github.repository }}/issues/${{ steps.extract-pr.outputs.PR_NUMBER }}/comments")
            
          # Check if response is valid JSON
          if jq -e . >/dev/null 2>&1 <<<"$existing_comment"; then
            existing_comment=$(jq -r '.[] | select(.user.login == "github-actions[bot]" and (.body | contains("<!-- Auto Review Assistant Comment -->"))) | {body: .body} | @base64' <<< "$existing_comment")
          else
            existing_comment=""
            echo "Warning: Invalid JSON response from GitHub API"
            echo "Response: $existing_comment"
          fi
          comment_body="${{ steps.get_approval.outputs.CURRENT_TIME }}"
          comment_time=$(TZ='Asia/Shanghai' date -d "$comment_body" +%s)
          echo "CACHE_TIMESTAMP=${comment_time}" >> $GITHUB_OUTPUT  # 统一使用这个变量名
          echo "Debug - Saving cache with timestamp: $comment_time"
          
          mkdir -p $(dirname unique_reviewers_bak.txt)
          if [[ -s unique_reviewers.txt ]]; then
            cp unique_reviewers.txt unique_reviewers_bak.txt
          else
            touch unique_reviewers_bak.txt
          fi
          
          if [[ -s unique_tags.txt ]]; then
            cp unique_tags.txt unique_tags_bak.txt
          else
            touch unique_tags_bak.txt
          fi

      - name: Save Reviewers Cache
        id: reviewers-cache-save
        if: steps.generate_reviewers.outputs.HAS_REVIEWERS == 'true'
        continue-on-error: 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 }}-${{ steps.get_comment_time.outputs.CACHE_TIMESTAMP }}-${{ github.run_id }}