ThothII/tools/tht/internal/backup/preflight.go

package backup

import (
	"archive/tar"
	"archive/zip"
	"bytes"
	"context"
	"crypto/sha256"
	"encoding/binary"
	"encoding/hex"
	"encoding/json"
	"errors"
	"fmt"
	"hash"
	"io"
	"os"
	"path"
	"path/filepath"
	"regexp"
	"strings"

	"github.com/aritmolab/thothii/tools/tht/internal/config"
)

var archiveDrivePath = regexp.MustCompile(`^[A-Za-z]:/`)

// PreflightRequest identifies an archive and the explicit protections required to inspect a
// restore that contains external secret payloads. Preflight never writes to the installation.
type PreflightRequest struct {
	Archive              string
	Confirm              bool
	AllowExternalSecrets bool
	Limits               PreflightLimits
}

// PreflightLimits bounds hostile archive processing. Zero values select the conservative
// defaults; callers that need larger installation backups must opt in explicitly.
type PreflightLimits struct {
	MaxMembers           int
	MaxArchiveBytes      uint64
	MaxUncompressedBytes uint64
	MaxCompressionRatio  uint64
}

const (
	defaultPreflightMaxMembers           = 10_000
	defaultPreflightMaxArchiveBytes      = 32 << 30
	defaultPreflightMaxUncompressedBytes = 128 << 30
	defaultPreflightMaxCompressionRatio  = 100
)

// PreflightDependencies supplies checks that require knowledge of the current Docker targets.
// Part B can bind these callbacks to read-only Docker Compose and filesystem inspections before
// it starts its restore transaction.
type PreflightDependencies struct {
	FreeBytes                     func(target string) (uint64, error)
	CheckOwnershipPermissions     func(context.Context, config.Installation, Manifest) error
	CheckVolumeMapping            func(context.Context, config.Installation, Manifest) error
	CheckImageConfigCompatibility func(context.Context, config.Installation, Manifest) error
}

// ArchiveEntryMetadata is safe restore metadata. It intentionally contains no archive payload.
type ArchiveEntryMetadata struct {
	Path        string
	Kind        string
	Owner       string
	LogicalName string
	SourcePath  string
	Size        int64
	SHA256      string
	Mode        uint32
	Sensitive   bool
}

// PreflightResult is the validated, non-mutating input for a future restore transaction.
// Manifest and Entries contain checksums and ownership metadata only; no secret or other
// archive bytes are returned.
type PreflightResult struct {
	ArchivePath   string
	ArchiveSize   int64
	RequiredBytes uint64
	Manifest      Manifest
	Entries       []ArchiveEntryMetadata
	archive       *verifiedArchive
}

type verifiedArchive struct {
	file   *os.File
	info   os.FileInfo
	digest string
	limits PreflightLimits
}

type inspectedArchiveEntry struct {
	metadata ArchiveEntryMetadata
	member   *zip.File
}

// Preflight validates an archive completely before restore mutation. It streams each archive
// member once for checksum verification and never extracts a member to a destination.
func Preflight(ctx context.Context, installation config.Installation, request PreflightRequest, dependencies PreflightDependencies) (PreflightResult, error) {
	if dependencies.FreeBytes == nil || dependencies.CheckOwnershipPermissions == nil ||
		dependencies.CheckVolumeMapping == nil || dependencies.CheckImageConfigCompatibility == nil {
		return PreflightResult{}, errors.New("backup preflight dependencies are incomplete")
	}
	if err := contextError(ctx); err != nil {
		return PreflightResult{}, err
	}
	limits, err := normalizePreflightLimits(request.Limits)
	if err != nil {
		return PreflightResult{}, err
	}
	archivePath, archiveInfo, err := openableArchivePath(request.Archive)
	if err != nil {
		return PreflightResult{}, err
	}
	if uint64(archiveInfo.Size()) > limits.MaxArchiveBytes {
		return PreflightResult{}, errors.New("backup archive exceeds the configured archive-size limit")
	}
	archiveFile, err := os.Open(archivePath)
	if err != nil {
		return PreflightResult{}, fmt.Errorf("open backup archive: %w", err)
	}
	keepArchiveOpen := false
	defer func() {
		if !keepArchiveOpen {
			_ = archiveFile.Close()
		}
	}()
	openedInfo, err := archiveFile.Stat()
	if err != nil {
		return PreflightResult{}, fmt.Errorf("inspect opened backup archive: %w", err)
	}
	if !openedInfo.Mode().IsRegular() || openedInfo.Size() != archiveInfo.Size() {
		return PreflightResult{}, errors.New("backup archive changed while opening")
	}
	memberCount, err := zipMemberCount(archiveFile, openedInfo.Size())
	if err != nil {
		return PreflightResult{}, err
	}
	if memberCount > uint64(limits.MaxMembers) {
		return PreflightResult{}, errors.New("backup archive exceeds the configured member limit")
	}
	initialDigest, err := digestArchive(ctx, archiveFile, limits.MaxArchiveBytes)
	if err != nil {
		return PreflightResult{}, err
	}
	reader, err := zip.NewReader(archiveFile, openedInfo.Size())
	if err != nil {
		return PreflightResult{}, fmt.Errorf("read backup archive: %w", err)
	}

	manifestBytes, entries, err := inspectArchiveMembers(ctx, reader, limits)
	if err != nil {
		return PreflightResult{}, err
	}
	if err := validateRawManifestPaths(manifestBytes); err != nil {
		return PreflightResult{}, err
	}
	manifest, err := DecodeManifest(manifestBytes)
	if err != nil {
		return PreflightResult{}, fmt.Errorf("validate backup manifest: %w", err)
	}
	expectedID, err := backupInstallationID(installation)
	if err != nil {
		return PreflightResult{}, err
	}
	if manifest.InstallationID != expectedID {
		return PreflightResult{}, errors.New("backup archive belongs to a different installation")
	}
	if manifest.IncludesSecrets && (!request.Confirm || !request.AllowExternalSecrets) {
		return PreflightResult{}, errors.New("secret-bearing backup requires explicit confirmation and external-secret permission")
	}

	metadata, requiredBytes, err := reconcileArchiveEntries(ctx, manifest, entries)
	if err != nil {
		return PreflightResult{}, err
	}
	freeBytes, err := dependencies.FreeBytes(installation.ProjectDirectory)
	if err != nil {
		return PreflightResult{}, fmt.Errorf("check free disk space: %w", err)
	}
	if freeBytes < requiredBytes {
		return PreflightResult{}, fmt.Errorf("insufficient free disk space for restore: need %d bytes, have %d", requiredBytes, freeBytes)
	}
	if err := contextError(ctx); err != nil {
		return PreflightResult{}, err
	}
	if err := dependencies.CheckOwnershipPermissions(ctx, installation, manifest); err != nil {
		return PreflightResult{}, fmt.Errorf("validate target ownership and permissions: %w", err)
	}
	if err := dependencies.CheckVolumeMapping(ctx, installation, manifest); err != nil {
		return PreflightResult{}, fmt.Errorf("validate volume mapping: %w", err)
	}
	if err := dependencies.CheckImageConfigCompatibility(ctx, installation, manifest); err != nil {
		return PreflightResult{}, fmt.Errorf("validate image/config compatibility: %w", err)
	}
	finalDigest, err := digestArchive(ctx, archiveFile, limits.MaxArchiveBytes)
	if err != nil {
		return PreflightResult{}, err
	}
	if initialDigest != finalDigest {
		return PreflightResult{}, errors.New("backup archive changed during preflight")
	}
	keepArchiveOpen = true
	return PreflightResult{
		ArchivePath: archivePath, ArchiveSize: openedInfo.Size(), RequiredBytes: requiredBytes,
		Manifest: manifest, Entries: metadata,
		archive: &verifiedArchive{file: archiveFile, info: openedInfo, digest: finalDigest, limits: limits},
	}, nil
}

// RevalidateArchive binds a restore to the bytes inspected by Preflight. A caller must invoke
// it immediately before a restore transaction and use the returned retained handle, never reopen
// ArchivePath. It refuses a path replacement or in-place content change.
func (result PreflightResult) RevalidateArchive() (*os.File, error) {
	if result.archive == nil || result.archive.file == nil {
		return nil, errors.New("backup archive has not been retained by preflight")
	}
	pathInfo, err := os.Lstat(result.ArchivePath)
	if err != nil {
		return nil, errors.New("backup archive changed after preflight")
	}
	if !pathInfo.Mode().IsRegular() || !os.SameFile(result.archive.info, pathInfo) {
		return nil, errors.New("backup archive changed after preflight")
	}
	heldInfo, err := result.archive.file.Stat()
	if err != nil || !os.SameFile(result.archive.info, heldInfo) || heldInfo.Size() != result.ArchiveSize {
		return nil, errors.New("backup archive changed after preflight")
	}
	digest, err := digestArchive(context.Background(), result.archive.file, result.archive.limits.MaxArchiveBytes)
	if err != nil || digest != result.archive.digest {
		return nil, errors.New("backup archive changed after preflight")
	}
	return result.archive.file, nil
}

// CloseArchive releases the retained read-only archive handle after the caller finishes the
// restore transaction or decides not to proceed.
func (result PreflightResult) CloseArchive() error {
	if result.archive == nil || result.archive.file == nil {
		return nil
	}
	return result.archive.file.Close()
}

func normalizePreflightLimits(requested PreflightLimits) (PreflightLimits, error) {
	if requested.MaxMembers < 0 {
		return PreflightLimits{}, errors.New("backup preflight member limit must be positive")
	}
	if requested.MaxMembers == 0 {
		requested.MaxMembers = defaultPreflightMaxMembers
	}
	if requested.MaxArchiveBytes == 0 {
		requested.MaxArchiveBytes = defaultPreflightMaxArchiveBytes
	}
	if requested.MaxUncompressedBytes == 0 {
		requested.MaxUncompressedBytes = defaultPreflightMaxUncompressedBytes
	}
	if requested.MaxCompressionRatio == 0 {
		requested.MaxCompressionRatio = defaultPreflightMaxCompressionRatio
	}
	return requested, nil
}

func validateRawManifestPaths(value []byte) error {
	var raw struct {
		Entries []struct {
			Path string `json:"path"`
		} `json:"entries"`
	}
	decoder := json.NewDecoder(bytes.NewReader(value))
	if err := decoder.Decode(&raw); err != nil {
		return fmt.Errorf("validate backup manifest paths: %w", err)
	}
	for _, entry := range raw.Entries {
		if _, err := validateArchiveMemberPath(entry.Path); err != nil {
			return fmt.Errorf("validate backup manifest entry path: %w", err)
		}
	}
	return nil
}

func openableArchivePath(requested string) (string, os.FileInfo, error) {
	if strings.TrimSpace(requested) == "" {
		return "", nil, errors.New("backup archive path is required")
	}
	archivePath, err := filepath.Abs(requested)
	if err != nil {
		return "", nil, fmt.Errorf("resolve backup archive path: %w", err)
	}
	archivePath = filepath.Clean(archivePath)
	info, err := os.Lstat(archivePath)
	if err != nil {
		return "", nil, fmt.Errorf("inspect backup archive: %w", err)
	}
	if !info.Mode().IsRegular() {
		return "", nil, errors.New("backup archive must be a regular file")
	}
	return archivePath, info, nil
}

func inspectArchiveMembers(ctx context.Context, reader *zip.Reader, limits PreflightLimits) ([]byte, map[string]inspectedArchiveEntry, error) {
	if len(reader.File) > limits.MaxMembers {
		return nil, nil, errors.New("backup archive exceeds the configured member limit")
	}
	var manifestBytes []byte
	entries := make(map[string]inspectedArchiveEntry, len(reader.File))
	var totalUncompressed uint64
	for _, member := range reader.File {
		if err := contextError(ctx); err != nil {
			return nil, nil, err
		}
		if err := checkArchiveMemberLimits(member, limits, &totalUncompressed); err != nil {
			return nil, nil, err
		}
		name, err := validateArchiveMemberPath(member.Name)
		if err != nil {
			return nil, nil, err
		}
		if _, exists := entries[name]; exists || name == ManifestPath && manifestBytes != nil {
			return nil, nil, fmt.Errorf("backup archive contains duplicate entry %q", name)
		}
		if member.Mode()&os.ModeSymlink != 0 {
			return nil, nil, fmt.Errorf("backup archive contains a symlink entry %q", name)
		}
		if member.FileInfo().IsDir() {
			return nil, nil, fmt.Errorf("backup archive contains unsupported directory entry %q", name)
		}
		opened, err := member.Open()
		if err != nil {
			return nil, nil, fmt.Errorf("open backup archive entry: %w", err)
		}
		if name == ManifestPath {
			manifestBytes, err = io.ReadAll(io.LimitReader(opened, maxPreflightManifestBytes+1))
			closeErr := opened.Close()
			if err != nil {
				return nil, nil, fmt.Errorf("read backup manifest: %w", err)
			}
			if closeErr != nil {
				return nil, nil, fmt.Errorf("close backup manifest: %w", closeErr)
			}
			if int64(len(manifestBytes)) > maxPreflightManifestBytes {
				return nil, nil, errors.New("backup manifest is too large")
			}
			if uint64(len(manifestBytes)) != member.UncompressedSize64 {
				return nil, nil, errors.New("backup manifest size does not match its archive metadata")
			}
			continue
		}
		hashValue := sha256.New()
		size, copyErr := io.Copy(hashValue, io.LimitReader(opened, int64(member.UncompressedSize64)+1))
		closeErr := opened.Close()
		if copyErr != nil {
			return nil, nil, fmt.Errorf("read backup archive entry: %w", copyErr)
		}
		if closeErr != nil {
			return nil, nil, fmt.Errorf("close backup archive entry: %w", closeErr)
		}
		if size < 0 {
			return nil, nil, errors.New("backup archive entry size overflow")
		}
		if uint64(size) != member.UncompressedSize64 {
			return nil, nil, errors.New("backup archive entry size does not match its archive metadata")
		}
		entries[name] = inspectedArchiveEntry{
			metadata: ArchiveEntryMetadata{Path: name, Size: size, SHA256: digestForHash(hashValue), Mode: uint32(member.Mode().Perm())},
			member:   member,
		}
	}
	if manifestBytes == nil {
		return nil, nil, errors.New("backup archive is missing manifest.json")
	}
	return manifestBytes, entries, nil
}

func checkArchiveMemberLimits(member *zip.File, limits PreflightLimits, total *uint64) error {
	if member.UncompressedSize64 > limits.MaxUncompressedBytes || *total > limits.MaxUncompressedBytes-member.UncompressedSize64 {
		return errors.New("backup archive exceeds the configured uncompressed-size limit")
	}
	*total += member.UncompressedSize64
	if member.CompressedSize64 == 0 {
		if member.UncompressedSize64 != 0 {
			return errors.New("backup archive exceeds the configured compression-ratio limit")
		}
		return nil
	}
	if limits.MaxCompressionRatio <= ^uint64(0)/member.CompressedSize64 &&
		member.UncompressedSize64 > limits.MaxCompressionRatio*member.CompressedSize64 {
		return errors.New("backup archive exceeds the configured compression-ratio limit")
	}
	return nil
}

func digestArchive(ctx context.Context, file *os.File, maximum uint64) (string, error) {
	if _, err := file.Seek(0, io.SeekStart); err != nil {
		return "", fmt.Errorf("seek backup archive: %w", err)
	}
	digest := sha256.New()
	buffer := make([]byte, 128*1024)
	var total uint64
	for {
		if err := contextError(ctx); err != nil {
			return "", err
		}
		count, err := file.Read(buffer)
		if count > 0 {
			if uint64(count) > maximum-total {
				return "", errors.New("backup archive exceeds the configured archive-size limit")
			}
			total += uint64(count)
			if _, writeErr := digest.Write(buffer[:count]); writeErr != nil {
				return "", fmt.Errorf("hash backup archive: %w", writeErr)
			}
		}
		if errors.Is(err, io.EOF) {
			break
		}
		if err != nil {
			return "", fmt.Errorf("read backup archive: %w", err)
		}
	}
	if _, err := file.Seek(0, io.SeekStart); err != nil {
		return "", fmt.Errorf("rewind backup archive: %w", err)
	}
	return digestForHash(digest), nil
}

func zipMemberCount(file *os.File, size int64) (uint64, error) {
	const (
		endOfCentralDirectorySignature = 0x06054b50
		zip64LocatorSignature          = 0x07064b50
		zip64EndSignature              = 0x06064b50
		endOfCentralDirectorySize      = 22
		zipCommentMaximum              = 1<<16 - 1
		zip64LocatorSize               = 20
		zip64EndMinimumSize            = 56
	)
	if size < endOfCentralDirectorySize {
		return 0, errors.New("read backup archive directory: archive is too small")
	}
	tailSize := int64(endOfCentralDirectorySize + zipCommentMaximum)
	if size < tailSize {
		tailSize = size
	}
	tail := make([]byte, tailSize)
	if _, err := file.ReadAt(tail, size-tailSize); err != nil {
		return 0, fmt.Errorf("read backup archive directory: %w", err)
	}
	for index := len(tail) - endOfCentralDirectorySize; index >= 0; index-- {
		if binary.LittleEndian.Uint32(tail[index:index+4]) != endOfCentralDirectorySignature {
			continue
		}
		commentLength := int(binary.LittleEndian.Uint16(tail[index+20 : index+22]))
		if index+endOfCentralDirectorySize+commentLength != len(tail) {
			continue
		}
		count := uint64(binary.LittleEndian.Uint16(tail[index+10 : index+12]))
		if count != 0xffff {
			return count, nil
		}
		endOffset := size - tailSize + int64(index)
		if endOffset < zip64LocatorSize {
			return 0, errors.New("read backup archive directory: ZIP64 locator is missing")
		}
		locator := make([]byte, zip64LocatorSize)
		if _, err := file.ReadAt(locator, endOffset-zip64LocatorSize); err != nil {
			return 0, fmt.Errorf("read backup archive directory: %w", err)
		}
		if binary.LittleEndian.Uint32(locator[:4]) != zip64LocatorSignature {
			return 0, errors.New("read backup archive directory: ZIP64 locator is invalid")
		}
		zip64Offset := binary.LittleEndian.Uint64(locator[8:16])
		if size < zip64EndMinimumSize || zip64Offset > uint64(size-zip64EndMinimumSize) {
			return 0, errors.New("read backup archive directory: ZIP64 record is invalid")
		}
		zip64End := make([]byte, zip64EndMinimumSize)
		if _, err := file.ReadAt(zip64End, int64(zip64Offset)); err != nil {
			return 0, fmt.Errorf("read backup archive directory: %w", err)
		}
		if binary.LittleEndian.Uint32(zip64End[:4]) != zip64EndSignature || binary.LittleEndian.Uint64(zip64End[4:12]) < 44 {
			return 0, errors.New("read backup archive directory: ZIP64 record is invalid")
		}
		return binary.LittleEndian.Uint64(zip64End[32:40]), nil
	}
	return 0, errors.New("read backup archive directory: end record is missing")
}

const maxPreflightManifestBytes = 16 << 20

func validateArchiveMemberPath(value string) (string, error) {
	if value == "" || strings.ContainsRune(value, '\x00') {
		return "", errors.New("backup archive contains an invalid empty or NUL path")
	}
	if strings.ContainsRune(value, '\\') {
		return "", fmt.Errorf("backup archive entry path %q uses an unsafe separator", value)
	}
	if strings.HasPrefix(value, "/") || strings.HasPrefix(value, "//") || archiveDrivePath.MatchString(value) {
		return "", fmt.Errorf("backup archive entry path %q is absolute", value)
	}
	clean := path.Clean(value)
	if clean != value || clean == "." || clean == ".." || strings.HasPrefix(clean, "../") {
		return "", fmt.Errorf("backup archive entry path %q escapes the archive", value)
	}
	return clean, nil
}

func reconcileArchiveEntries(ctx context.Context, manifest Manifest, entries map[string]inspectedArchiveEntry) ([]ArchiveEntryMetadata, uint64, error) {
	metadata := make([]ArchiveEntryMetadata, 0, len(entries))
	var requiredBytes uint64
	for _, entry := range manifest.Entries {
		actual, present := entries[entry.Path]
		if entry.Archived && !present {
			return nil, 0, fmt.Errorf("backup archive is missing entry %q", entry.Path)
		}
		if !entry.Archived {
			if present {
				return nil, 0, fmt.Errorf("non-archived backup entry %q has a payload", entry.Path)
			}
			continue
		}
		if actual.metadata.Size != entry.Size || actual.metadata.SHA256 != entry.SHA256 {
			return nil, 0, fmt.Errorf("checksum or size mismatch for backup entry %q", entry.Path)
		}
		if actual.metadata.Mode != entry.Mode {
			return nil, 0, fmt.Errorf("mode mismatch for backup entry %q", entry.Path)
		}
		if entry.Kind == EntryVolume {
			if err := validateVolumeTar(ctx, actual.member); err != nil {
				return nil, 0, fmt.Errorf("validate volume archive %q: %w", entry.Path, err)
			}
		}
		if ^uint64(0)-requiredBytes < uint64(actual.metadata.Size) {
			return nil, 0, errors.New("backup archive size overflows free-space calculation")
		}
		requiredBytes += uint64(actual.metadata.Size)
		actual.metadata.Kind = entry.Kind
		actual.metadata.Owner = entry.Owner
		actual.metadata.LogicalName = entry.LogicalName
		actual.metadata.SourcePath = entry.SourcePath
		actual.metadata.Sensitive = entry.Sensitive
		metadata = append(metadata, actual.metadata)
		delete(entries, entry.Path)
	}
	if len(entries) != 0 {
		for name := range entries {
			return nil, 0, fmt.Errorf("backup archive contains unmanifested entry %q", name)
		}
	}
	return metadata, requiredBytes, nil
}

func validateVolumeTar(ctx context.Context, member *zip.File) error {
	if member == nil {
		return errors.New("volume archive member is unavailable")
	}
	opened, err := member.Open()
	if err != nil {
		return fmt.Errorf("open volume archive: %w", err)
	}
	defer opened.Close()
	reader := tar.NewReader(opened)
	for {
		if err := contextError(ctx); err != nil {
			return err
		}
		header, err := reader.Next()
		if errors.Is(err, io.EOF) {
			return nil
		}
		if err != nil {
			return fmt.Errorf("read volume archive: %w", err)
		}
		name := strings.TrimSuffix(header.Name, "/")
		if _, err := validateArchiveMemberPath(name); err != nil {
			return fmt.Errorf("volume archive path is unsafe: %w", err)
		}
		switch header.Typeflag {
		case tar.TypeSymlink, tar.TypeLink:
			return fmt.Errorf("volume archive contains a link entry %q", name)
		case tar.TypeChar, tar.TypeBlock, tar.TypeFifo:
			return fmt.Errorf("volume archive contains a special entry %q", name)
		}
	}
}

func digestForHash(value hash.Hash) string {
	return "sha256:" + hex.EncodeToString(value.Sum(nil))
}

func contextError(ctx context.Context) error {
	select {
	case <-ctx.Done():
		return ctx.Err()
	default:
		return nil
	}
}