TsDB2/cmd/mrepro/main.go

// mrepro — 迁移孤儿页复现工具。
//
//	mrepro -src <sqlite 源库> [-workdir 目录] [-vacuum] [-keep]
//
// 流程: 把源库 cp 进工作目录 (绝不碰原件) → 官方 migrate.Run(sqlite→VRTX) →
// (可选) VACUUM → orphancheck 取证 → 打印两套口径孤儿数。
// 退出码: 真实孤儿页 > 0 时 exit 1 (供 CI/脚本判定复现)。
package main

import (
	"flag"
	"fmt"
	"io"
	"os"
	"path/filepath"

	"git.ycl.pub/yl32940/TsDB2/DB/core"
	"git.ycl.pub/yl32940/TsDB2/internal/orphancheck"
	"git.ycl.pub/yl32940/TsDB2/migrate"
)

func main() {
	src := flag.String("src", "", "sqlite 源库路径 (会先复制再迁, 不直接打开原件)")
	workdir := flag.String("workdir", "", "工作目录 (默认自动创建临时目录)")
	vac := flag.Bool("vacuum", false, "迁移后跑一次 VACUUM")
	keep := flag.Bool("keep", false, "保留工作目录 (默认结束后删除)")
	flag.Parse()

	if *src == "" {
		fmt.Fprintln(os.Stderr, "用法: mrepro -src <sqlite 源库> [-workdir 目录] [-vacuum] [-keep]")
		os.Exit(2)
	}

	wd := *workdir
	var err error
	if wd == "" {
		wd, err = os.MkdirTemp("", "mrepro-")
		if err != nil {
			fatal(err)
		}
	} else if err := os.MkdirAll(wd, 0o755); err != nil {
		fatal(err)
	}
	if !*keep {
		defer os.RemoveAll(wd)
	}
	fmt.Println("工作目录:", wd)

	// 1. 复制源库 —— 只迁副本
	srcCopy := filepath.Join(wd, "src.db")
	if err := copyFile(srcCopy, *src); err != nil {
		fatal(err)
	}

	// 2. 官方 migrate: sqlite → VRTX
	sqliteSrc, err := migrate.OpenSQLite(srcCopy, nil)
	if err != nil {
		fatal(fmt.Errorf("打开 sqlite 副本: %w", err))
	}
	outDB := filepath.Join(wd, "out.db")
	vrtxDst, err := migrate.OpenVRTX(outDB, nil)
	if err != nil {
		fatal(fmt.Errorf("打开 VRTX 目标: %w", err))
	}
	rep, runErr := migrate.Run(sqliteSrc, vrtxDst, migrate.DefaultOptions())
	_ = vrtxDst.Close()
	_ = sqliteSrc.Close()
	if runErr != nil {
		fmt.Fprintln(os.Stderr, "迁移报错 (继续取证):", runErr)
	}
	for _, tr := range rep.Tables {
		note := ""
		if tr.Error != "" {
			note = "  ERROR: " + tr.Error
		}
		fmt.Printf("  表 %-32s %8d 行 %s\n", tr.Table, tr.Rows, note)
	}

	// 3. (可选) VACUUM
	if *vac {
		db, err := core.Open(outDB)
		if err != nil {
			fatal(fmt.Errorf("打开迁移产物: %w", err))
		}
		vres, err := db.VACUUM()
		_ = db.Close()
		if err != nil {
			fatal(fmt.Errorf("VACUUM: %w", err))
		}
		fmt.Printf("VACUUM: %d → %d 字节 (省 %d), 复制 %d 行\n",
			vres.OldFileSize, vres.NewFileSize, vres.SavedBytes, vres.RowsCopied)
	}

	// 4. 取证
	R, err := orphancheck.Run(outDB, orphancheck.Options{EngineBlobScan: true, SamplesPerType: 3})
	if err != nil {
		fatal(err)
	}
	fmt.Printf("\n== 取证: %s\n", outDB)
	fmt.Printf("总页数=%d  空闲链=%d  结构可达=%d  blob 链页=%d\n",
		R.TotalPages, R.FreeCount, R.ReachableNoBlob, R.BlobPages)
	fmt.Printf("analyzer 口径孤儿 = %d (按类型 %v)\n", len(R.AnalyzerOrphans), R.AnalyzerByType)
	fmt.Printf("blob 感知真实孤儿 = %d (按类型 %v)\n", len(R.Orphans), R.ByType)
	for _, n := range R.Notes {
		fmt.Println("备注:", n)
	}
	for _, tn := range orphancheck.SortedTypes(R.ByType) {
		for _, p := range R.Samples[tn] {
			fmt.Printf("-- 样例 页 %d 类型 %s UseSize=%d NextPage=%d\n", p.ID, tn, p.UseSize, p.Next)
			fmt.Print(orphancheck.SampleHex(p))
		}
	}

	if len(R.Orphans) > 0 {
		fmt.Printf("\n复现成功: %d 页真实孤儿。\n", len(R.Orphans))
		os.Exit(1)
	}
	extra := ""
	if d := len(R.AnalyzerOrphans) - len(R.Orphans); d > 0 {
		extra = fmt.Sprintf(" (analyzer 口径的 %d 页全部是活 blob 链页 — analyzer 误报)", d)
	}
	fmt.Printf("\n纯净: 真实孤儿页 = 0%s\n", extra)
}

func copyFile(dst, src string) error { //nolint:revive // 参数序与 io.Copy 对齐习惯
	in, err := os.Open(src)
	if err != nil {
		return err
	}
	defer in.Close()
	out, err := os.Create(dst)
	if err != nil {
		return err
	}
	if _, err := io.Copy(out, in); err != nil {
		_ = out.Close()
		return err
	}
	return out.Close()
}

func fatal(err error) {
	fmt.Fprintln(os.Stderr, "错误:", err)
	os.Exit(2)
}