split-pr-safe-worktree.sh

split-pr-safe-worktree.sh

#!/bin/bash
set -euo pipefail

# ==============================================================================
# 功能：
# 1. 使用 git worktree 完全隔离环境，不污染原仓库
# 2. 保留完整 commit 历史、信息、作者、时间
# 3. 自动拆分为「代码分支」+「配置分支」
# 4. 自动验证文件不丢失、内容完全一致
# 5. 无论成功/失败/中断，都自动清理 worktree，不留垃圾
# 6. 最终输出：本地两个干净分支，可直接 push 开 PR
# ==============================================================================
# 使用示例：
# ./split-pr-safe-worktree-v3.sh \
#   --repo-dir      /path/to/your/repo \
#   --base-branch   develop \
#   --code-branch   pr-code-only \
#   --config-branch pr-config-only \
#   --worktree-dir  /tmp/pr-split-worktree
# ==============================================================================

# ====================== 全局善后清理函数（无论如何都执行）======================
cleanup() {
  if [[ -n "${WORKTREE_DIR:-}" && -d "${WORKTREE_DIR:-}" ]]; then
    echo -e "\n🧹 清理隔离工作区: $WORKTREE_DIR"
    rm -rf "$WORKTREE_DIR"
    git -C "$REPO_DIR" worktree prune 2>/dev/null || true
  fi
}

# 脚本结束（正常退出、报错、Ctrl+C）都执行清理
trap cleanup EXIT

# ====================== 参数解析 ======================
while [[ $# -gt 0 ]]; do
  case "$1" in
    --repo-dir)      REPO_DIR="$2"; shift 2 ;;
    --base-branch)   BASE_BRANCH="$2"; shift 2 ;;
    --code-branch)   CODE_BRANCH="$2"; shift 2 ;;
    --config-branch) CONFIG_BRANCH="$2"; shift 2 ;;
    --worktree-dir)  WORKTREE_DIR="$2"; shift 2 ;;
    *)
      echo "未知参数: $1"
      exit 1
      ;;
  esac
done

# ====================== 必传参数检查 ======================
required_vars=(REPO_DIR BASE_BRANCH CODE_BRANCH CONFIG_BRANCH WORKTREE_DIR)
for var in "${required_vars[@]}"; do
  if [[ -z "${!var:-}" ]]; then
    echo "错误：缺少必须参数 --${var,,}"
    exit 1
  fi
done

# ====================== 文件 类型配置 ======================
CODE_EXTS_REGEX="\.(java|kt)$"
CONFIG_EXTS_REGEX="\.(json|xml|txt|yml|yaml|properties|conf)$"

# ==========================================================
echo "=============================================="
echo "🔒 安全拆分 PR（git worktree 隔离模式）"
echo "原仓库:    $REPO_DIR"
echo "基准分支:  $BASE_BRANCH"
echo "代码分支:  $CODE_BRANCH"
echo "配置分支:  $CONFIG_BRANCH"
echo "工作目录:  $WORKTREE_DIR"
echo "=============================================="

# 进入原仓库
cd "$REPO_DIR"
git fetch

ORIGINAL_BRANCH=$(git branch --show-current)
echo -e "✅ 当前开发分支: $ORIGINAL_BRANCH"

# 创建隔离 worktree
echo -e "\n📂 创建隔离工作区..."
rm -rf "$WORKTREE_DIR"
git worktree add "$WORKTREE_DIR" "$BASE_BRANCH"

cd "$WORKTREE_DIR"

# 获取所有 commit（从旧到新）
COMMITS=$(git rev-list --reverse "$BASE_BRANCH..$ORIGINAL_BRANCH")
COMMIT_COUNT=$(echo "$COMMITS" | wc -l | tr -d ' ')
echo -e "🔍 共 $COMMIT_COUNT 个 commit 需要处理"

# --------------------------------------------------------------------
# 函数：创建过滤分支
# $1 分支名
# $2 保留文件正则
# --------------------------------------------------------------------
create_filtered_branch() {
  local branch_name="$1"
  local keep_pattern="$2"

  echo -e "\n=============================================="
  echo "🚀 构建分支: $branch_name"
  echo "=============================================="

  git checkout -B "$branch_name" "$BASE_BRANCH"

  for commit in $COMMITS; do
    echo -e "\n➤ 处理 commit: $commit"
    git cherry-pick --no-commit "$commit"

    git add -A
    # 不属于当前类型的文件全部恢复为基准版本
    git diff --name-only | grep -vE "$keep_pattern" | xargs -r git checkout HEAD --

    # 保留原 commit 的所有信息
    git commit -C "$commit"
  done
}

# 开始拆分
create_filtered_branch "$CODE_BRANCH"   "$CODE_EXTS_REGEX"
create_filtered_branch "$CONFIG_BRANCH" "$CONFIG_EXTS_REGEX"

cd "$REPO_DIR"

# ==============================================
# 🧪 自动验证：文件是否完整不丢失
# ==============================================
echo -e "\n=============================================="
echo "🧪 开始自动验证"
echo "=============================================="

CODE_LIST=$(mktemp)
CONFIG_LIST=$(mktemp)
ALL_LIST=$(mktemp)

git -C "$WORKTREE_DIR" diff --name-only "$BASE_BRANCH" "$CODE_BRANCH" | sort -u > "$CODE_LIST"
git -C "$WORKTREE_DIR" diff --name-only "$BASE_BRANCH" "$CONFIG_BRANCH" | sort -u > "$CONFIG_LIST"
git diff --name-only "$BASE_BRANCH" "$ORIGINAL_BRANCH" | sort -u > "$ALL_LIST"

COMBINED=$(mktemp)
cat "$CODE_LIST" "$CONFIG_LIST" | sort -u | uniq > "$COMBINED"

if diff -q "$COMBINED" "$ALL_LIST" >/dev/null; then
  echo -e "\n✅ 校验通过：代码 + 配置 = 原分支全部修改，无丢失！"
else
  echo -e "\n❌ 校验失败：存在文件丢失或不一致！"
  diff "$COMBINED" "$ALL_LIST"
  exit 1
fi

# 展示统计
CODE_CNT=$(wc -l < "$CODE_LIST")
CONFIG_CNT=$(wc -l < "$CONFIG_LIST")
ALL_CNT=$(wc -l < "$ALL_LIST")

echo -e "\n📊 修改文件统计："
echo "代码文件：$CODE_CNT 个"
echo "配置文件：$CONFIG_CNT 个"
echo "原分支总修改：$ALL_CNT 个"
echo "校验结果：$CODE_CNT + $CONFIG_CNT = $ALL_CNT"

rm -f "$CODE_LIST" "$CONFIG_LIST" "$ALL_LIST" "$COMBINED"

# ==============================================
# 🎉 最终提示
# ==============================================
echo -e "\n=============================================="
echo "🎉 全部完成！"
echo ""
echo "✅ 两个分支已创建在原仓库中："
echo "   - $CODE_BRANCH"
echo "   - $CONFIG_BRANCH"
echo ""
echo "✅ 可直接推送远程："
echo "  git push origin $CODE_BRANCH"
echo "  git push origin $CONFIG_BRANCH"
echo ""
echo "✅ worktree 已自动清理，环境干净无残留"
echo "=============================================="