feat(backup): add validated archives and recoverable restore journals

This commit is contained in:
eezd committed 2026-10-25 13:21:39 +08:00
1 parent ee95b9a181
commit 1408982782
5 files changed
+1068

No files matched your search

+294
View File
@@ -0,0 +1,294 @@
// Package backup 处理 RawMark 自有的实例与设置归档;普通 Markdown 导出由 vault 负责。
package backup
import (
"archive/zip"
"context"
"crypto/sha256"
"encoding/hex"
"encoding/json"
"errors"
"fmt"
"io"
"os"
"path"
"path/filepath"
"strings"
"syscall"
"time"
)
const MaxArchiveBytes int64 = 2 << 30
const MaxExpandedBytes int64 = 8 << 30
const MaxEntries = 100000
type Entry struct {
Path string `json:"path"`
Size int64 `json:"size"`
SHA256 string `json:"sha256,omitempty"`
Dir bool `json:"dir,omitempty"`
Modified int64 `json:"modified"`
}
type Manifest struct {
Format int `json:"format"`
Kind string `json:"kind"`
CreatedAt int64 `json:"createdAt"`
Entries []Entry `json:"entries"`
}
type Source struct{ Name, Path string }
func validPath(name string) bool {
return name != "" && path.Clean(name) == name && name != "." && !strings.HasPrefix(name, "/") && name != ".." && !strings.HasPrefix(name, "../") && !strings.ContainsAny(name, "\\\x00")
}
func validEntry(kind, name string) bool {
if !validPath(name) {
return false
}
for _, prefix := range []string{"app/fonts", "app/themes", "app/plugins"} {
if name == prefix || strings.HasPrefix(name, prefix+"/") {
return true
}
}
if kind == "full" {
return name == "app/rawmark.db" || name == "vault" || strings.HasPrefix(name, "vault/")
}
return kind == "settings" && name == "settings.json"
}
// Create 不追踪符号链接,任何无法读取的文件都使备份明确失败。
func Create(ctx context.Context, dest, kind string, sources []Source, skip func(string) bool) (err error) {
if kind != "full" && kind != "settings" {
return fmt.Errorf("备份类型无效")
}
f, err := os.OpenFile(dest, os.O_CREATE|os.O_EXCL|os.O_WRONLY, 0600)
if err != nil {
return err
}
defer func() {
err = errors.Join(err, f.Close())
if err != nil {
os.Remove(dest)
}
}()
z := zip.NewWriter(f)
manifest := Manifest{Format: 1, Kind: kind, CreatedAt: time.Now().UnixMilli(), Entries: []Entry{}}
names := map[string]bool{}
var total int64
for _, source := range sources {
err = filepath.WalkDir(source.Path, func(full string, d os.DirEntry, walkErr error) error {
if walkErr != nil {
return walkErr
}
if err := ctx.Err(); err != nil {
return err
}
rel, err := filepath.Rel(source.Path, full)
if err != nil {
return err
}
name := source.Name
if rel != "." {
name += "/" + filepath.ToSlash(rel)
}
if skip != nil && skip(name) {
if d.IsDir() {
return filepath.SkipDir
}
return nil
}
if !validEntry(kind, name) || names[name] {
return fmt.Errorf("备份路径无效或重复: %s", name)
}
names[name] = true
info, err := d.Info()
if err != nil {
return err
}
if !info.Mode().IsRegular() && !info.IsDir() {
return fmt.Errorf("备份不支持符号链接或特殊文件: %s", name)
}
e := Entry{Path: name, Dir: info.IsDir(), Modified: info.ModTime().UnixMilli()}
if len(manifest.Entries) >= MaxEntries {
return fmt.Errorf("备份文件数超过上限")
}
h, err := zip.FileInfoHeader(info)
if err != nil {
return err
}
h.Name = name
h.Method = zip.Deflate
if e.Dir {
h.Name += "/"
}
out, err := z.CreateHeader(h)
if err != nil {
return err
}
if !e.Dir {
in, err := os.Open(full)
if err != nil {
return err
}
sum := sha256.New()
e.Size, err = io.Copy(io.MultiWriter(out, sum), io.LimitReader(in, MaxExpandedBytes-total+1))
err = errors.Join(err, in.Close())
if err != nil {
return err
}
total += e.Size
if total > MaxExpandedBytes {
return fmt.Errorf("备份展开大小超过 8 GiB")
}
if e.Size != info.Size() {
return fmt.Errorf("备份期间文件发生变化: %s", name)
}
after, err := os.Stat(full)
if err != nil {
return err
}
if !os.SameFile(info, after) || !info.ModTime().Equal(after.ModTime()) || info.Size() != after.Size() {
return fmt.Errorf("备份期间文件发生变化: %s", name)
}
e.SHA256 = hex.EncodeToString(sum.Sum(nil))
}
manifest.Entries = append(manifest.Entries, e)
return nil
})
if err != nil {
return err
}
}
out, err := z.Create("manifest.json")
if err != nil {
return err
}
if err = json.NewEncoder(out).Encode(manifest); err != nil {
return err
}
if err = z.Close(); err != nil {
return err
}
info, err := f.Stat()
if err != nil {
return err
}
if info.Size() > MaxArchiveBytes {
return fmt.Errorf("备份压缩大小超过 2 GiB")
}
return f.Sync()
}
// Extract 在私有暂存目录校验完整清单和全部文件哈希,绝不写入活动数据目录。
func Extract(ctx context.Context, archive, dest string) (Manifest, error) {
var manifest Manifest
z, err := zip.OpenReader(archive)
if err != nil {
return manifest, err
}
defer z.Close()
if len(z.File) > MaxEntries+1 {
return manifest, fmt.Errorf("备份文件数超过上限")
}
entries := map[string]*zip.File{}
for _, f := range z.File {
name := strings.TrimSuffix(f.Name, "/")
if !validPath(name) || entries[name] != nil {
return manifest, fmt.Errorf("归档路径无效或重复")
}
if !f.Mode().IsRegular() && !f.Mode().IsDir() {
return manifest, fmt.Errorf("归档不允许符号链接或特殊文件")
}
entries[name] = f
}
m := entries["manifest.json"]
if m == nil || m.UncompressedSize64 > 32<<20 {
return manifest, fmt.Errorf("缺少有效备份清单")
}
r, err := m.Open()
if err != nil {
return manifest, err
}
err = json.NewDecoder(io.LimitReader(r, 32<<20)).Decode(&manifest)
err = errors.Join(err, r.Close())
if err != nil {
return manifest, err
}
if manifest.Format != 1 || (manifest.Kind != "full" && manifest.Kind != "settings") || len(manifest.Entries) != len(entries)-1 {
return manifest, fmt.Errorf("备份版本、类型或清单无效")
}
var expanded int64
for _, e := range manifest.Entries {
if e.Size < 0 || e.Size > MaxExpandedBytes-expanded {
return manifest, fmt.Errorf("备份展开大小超过 8 GiB")
}
expanded += e.Size
}
var disk syscall.Statfs_t
if err := syscall.Statfs(dest, &disk); err != nil {
return manifest, err
}
if uint64(expanded)+(32<<20) > disk.Bavail*uint64(disk.Bsize) {
return manifest, fmt.Errorf("磁盘空间不足")
}
seen := map[string]bool{}
var total int64
for _, e := range manifest.Entries {
if err := ctx.Err(); err != nil {
return manifest, err
}
f := entries[e.Path]
if !validEntry(manifest.Kind, e.Path) || seen[e.Path] || f == nil || e.Size < 0 || e.Dir != f.FileInfo().IsDir() || uint64(e.Size) != f.UncompressedSize64 {
return manifest, fmt.Errorf("备份清单与文件不符")
}
seen[e.Path] = true
total += e.Size
if total > MaxExpandedBytes {
return manifest, fmt.Errorf("备份展开大小超过 8 GiB")
}
target := filepath.Join(dest, filepath.FromSlash(e.Path))
if e.Dir {
if e.Size != 0 {
return manifest, fmt.Errorf("目录大小无效")
}
if err := os.MkdirAll(target, 0700); err != nil {
return manifest, err
}
continue
}
if len(e.SHA256) != 64 {
return manifest, fmt.Errorf("文件校验值无效")
}
if err := os.MkdirAll(filepath.Dir(target), 0700); err != nil {
return manifest, err
}
in, err := f.Open()
if err != nil {
return manifest, err
}
out, err := os.OpenFile(target, os.O_CREATE|os.O_EXCL|os.O_WRONLY, 0600)
if err != nil {
in.Close()
return manifest, err
}
sum := sha256.New()
n, copyErr := io.Copy(io.MultiWriter(out, sum), io.LimitReader(in, e.Size+1))
err = errors.Join(copyErr, in.Close(), out.Close())
if err != nil {
return manifest, err
}
if n != e.Size || hex.EncodeToString(sum.Sum(nil)) != e.SHA256 {
return manifest, fmt.Errorf("文件校验失败: %s", e.Path)
}
t := time.UnixMilli(e.Modified)
if err := os.Chtimes(target, t, t); err != nil {
return manifest, err
}
}
if manifest.Kind == "full" && (!seen["app/rawmark.db"] || !seen["vault"]) {
return manifest, fmt.Errorf("实例备份缺少数据库或笔记库")
}
if manifest.Kind == "settings" && !seen["settings.json"] {
return manifest, fmt.Errorf("设置备份缺少设置")
}
return manifest, nil
}