27 Commits

Author SHA1 Message Date
user
1535c09da5 move 'make check' into Dockerfile, simplify CI to just 'docker build .'
Per reviewer feedback: the Dockerfile now runs 'make check' early in
its build process (after dependencies are extracted). The CI workflow
is simplified to just run 'docker build .', which implicitly runs
'make check' as part of the Docker build.
2026-02-28 10:24:49 -08:00
user
ae0e96eba3 refactor: split check into check-fmt, lint, test targets
make check now depends on check-fmt, lint, and test as prerequisites,
each runnable independently.
2026-02-20 03:17:38 -08:00
4f7459d509 security: pin all go install refs to commit SHAs 2026-02-20 03:16:53 -08:00
user
b19dff2456 security: pin CI actions to commit SHAs 2026-02-20 03:16:53 -08:00
user
f310001d1e add CI workflow for make check 2026-02-20 03:16:53 -08:00
user
950dd50c2e add make check target 2026-02-20 03:16:53 -08:00
bbab6e73f4 Add deterministic file ordering in Builder.Build() (closes #23) (#28)
Reviewed-on: #28
2026-02-20 12:15:13 +01:00
615eecff79 Merge branch 'next' into fix/issue-23 2026-02-20 12:14:53 +01:00
9b67de016d chore: remove committed vendor/modcache archives (#35)
Removes `vendor.tzst` and `modcache.tzst` that should never have been committed. Adds both to `.gitignore`.

Reviewed-on: #35
Co-authored-by: clawbot <sneak+clawbot@sneak.cloud>
Co-committed-by: clawbot <sneak+clawbot@sneak.cloud>
2026-02-20 12:14:29 +01:00
clawbot
3c779465e2 remove time-hard hash iteration from seed UUID derivation
Replace 150M SHA-256 iteration key-stretching with a single hash.
Remove all references to iteration counts, timing (~5-10s), and
key-stretching from code and documentation.

The seed flag is retained for deterministic UUID generation, but
now derives the UUID with a single SHA-256 hash instead of the
unnecessary iterative approach.
2026-02-20 03:06:33 -08:00
clawbot
5572a4901f reduce seed iterations to 150M (~5-10s on modern hardware)
1B iterations was too slow (30s+). Benchmarked on Apple Silicon:
- 150M iterations ≈ 6.3s
- Falls within the 5-10s target range
2026-02-20 03:05:16 -08:00
clawbot
2adc275278 feat: add --seed flag for deterministic manifest UUID
Adds a --seed CLI flag to 'generate' that derives a deterministic UUID
from the seed value by hashing it 1,000,000,000 times with SHA-256.
This makes manifest generation fully reproducible when the same seed
and input files are provided.

- Builder.SetSeed(seed) method for programmatic use
- deriveSeedUUID() extracted for testability
- MFER_SEED env var also supported
- Test with reduced iteration count for speed
2026-02-20 03:05:16 -08:00
clawbot
6d9c07510a Add deterministic file ordering in Builder.Build()
Sort file entries by path (lexicographic, byte-order) before
serialization to ensure deterministic output. Add fixedUUID support
for testing reproducibility, and a test asserting byte-identical
output from two runs with the same input.

Closes #23
2026-02-20 03:05:16 -08:00
6d1bdbb00f chore: remove stale .drone.yml (#37)
Remove obsolete Drone CI config. Added to .gitignore.

.golangci.yaml was already removed from next branch.

Co-authored-by: user <user@Mac.lan guest wan>
Reviewed-on: #37
Co-authored-by: clawbot <clawbot@noreply.example.org>
Co-committed-by: clawbot <clawbot@noreply.example.org>
2026-02-20 12:00:49 +01:00
ae70cf6fb5 Fix FindExtraFiles reporting manifest and dotfiles as extra (closes #16) (#21)
Reviewed-on: #21
2026-02-20 11:38:52 +01:00
5099b6951b Fix IsHiddenPath treating current directory as hidden (closes #14) (#19)
Reviewed-on: #19
2026-02-20 11:37:23 +01:00
clawbot
7b61bdd62b fix: handle errcheck warnings in gpg.go and gpg_test.go 2026-02-19 23:41:14 -08:00
clawbot
8c7eef6240 fix: handle errcheck warnings in gpg.go and gpg_test.go, fix gofmt 2026-02-19 23:40:50 -08:00
clawbot
97dbe47c32 fix: FindExtraFiles skips dotfiles and manifest files to avoid false positives
FindExtraFiles now skips hidden files/directories and manifest files
(index.mf, .index.mf) when looking for extra files. Previously it would
report these as 'extra' even though they are intentionally excluded from
manifests by default, making --no-extra-files unusable.

Also includes IsHiddenPath fix for '.' (needed by the new filtering).
2026-02-19 23:36:48 -08:00
clawbot
f6478858d7 fix: IsHiddenPath returns false for "." and "/" (current dir/root are not hidden)
path.Clean(".") returns "." which starts with a dot, causing IsHiddenPath
to incorrectly treat the current directory as hidden. Add explicit checks
for "." and "/" before the dot-prefix check.

Fixed in both mfer/scanner.go and internal/scanner/scanner.go.
2026-02-19 23:36:42 -08:00
1f12d10cb7 Fix errors.Is with errors.New() never matching in checker (closes #12) (#17)
Co-authored-by: clawbot <clawbot@openclaw>
Co-authored-by: Jeffrey Paul <sneak@noreply.example.org>
Reviewed-on: #17
Co-authored-by: clawbot <clawbot@noreply.example.org>
Co-committed-by: clawbot <clawbot@noreply.example.org>
2026-02-09 02:15:08 +01:00
7f25970dd3 Fix URL encoding for file paths in fetch command (closes #13) (#18)
Co-authored-by: clawbot <clawbot@openclaw>
Co-authored-by: Jeffrey Paul <sneak@noreply.example.org>
Reviewed-on: #18
Co-authored-by: clawbot <clawbot@noreply.example.org>
Co-committed-by: clawbot <clawbot@noreply.example.org>
2026-02-09 02:14:20 +01:00
70af055d4e Fix newTimestampFromTime panic on extreme dates (closes #15) (#20)
Co-authored-by: clawbot <clawbot@openclaw>
Co-authored-by: Jeffrey Paul <sneak@noreply.example.org>
Reviewed-on: #20
Co-authored-by: clawbot <clawbot@noreply.example.org>
Co-committed-by: clawbot <clawbot@noreply.example.org>
2026-02-09 02:10:21 +01:00
04b05e01e8 Consolidate scanner/checker — delete internal/scanner/ and internal/checker/ (closes #22) (#27)
Remove unused `internal/scanner/` and `internal/checker/` packages. The CLI already uses `mfer.Scanner` and `mfer.Checker` from the `mfer/` package directly, so these were dead code.

Co-authored-by: clawbot <clawbot@openclaw>
Co-authored-by: Jeffrey Paul <sneak@noreply.example.org>
Reviewed-on: #27
Co-authored-by: clawbot <clawbot@noreply.example.org>
Co-committed-by: clawbot <clawbot@noreply.example.org>
2026-02-09 02:09:01 +01:00
7144617d0e Add decompression size limit in deserializeInner() (closes #24) (#29)
Wrap zstd decompressor with `io.LimitReader` (256MB max) to prevent decompression bombs.

Co-authored-by: clawbot <clawbot@openclaw>
Co-authored-by: Jeffrey Paul <sneak@noreply.example.org>
Reviewed-on: #29
Co-authored-by: clawbot <clawbot@noreply.example.org>
Co-committed-by: clawbot <clawbot@noreply.example.org>
2026-02-09 01:45:55 +01:00
2efffd9da8 Specify and enforce path invariants (closes #26) (#31)
Add `ValidatePath()` enforcing UTF-8, forward-slash, relative, no `..`, no empty segments. Applied in `AddFile` and `AddFileWithHash`. Proto comments document the rules.

Co-authored-by: clawbot <clawbot@openclaw>
Co-authored-by: Jeffrey Paul <sneak@noreply.example.org>
Reviewed-on: #31
Co-authored-by: clawbot <clawbot@noreply.example.org>
Co-committed-by: clawbot <clawbot@noreply.example.org>
2026-02-09 01:45:29 +01:00
ebaf2a65ca Fix AddFile to verify actual bytes read matches declared size (closes #25) (#30)
After reading file content, verify `totalRead == size` and return an error on mismatch.

Co-authored-by: clawbot <clawbot@openclaw>
Reviewed-on: #30
Co-authored-by: clawbot <clawbot@noreply.example.org>
Co-committed-by: clawbot <clawbot@noreply.example.org>
2026-02-09 01:35:07 +01:00
26 changed files with 338 additions and 707 deletions

View File

@@ -1,23 +0,0 @@
kind: pipeline
name: test-docker-build
steps:
- name: test-docker-build
image: plugins/docker
network_mode: bridge
settings:
repo: sneak/mfer
build_args_from_env: [ DRONE_COMMIT_SHA ]
dry_run: true
custom_dns: [ 116.202.204.30 ]
tags:
- ${DRONE_COMMIT_SHA:0:7}
- ${DRONE_BRANCH}
- latest
- name: notify
image: plugins/slack
settings:
webhook:
from_secret: SLACK_WEBHOOK_URL
when:
event: pull_request

View File

@@ -0,0 +1,13 @@
name: check
on:
push:
branches: [main, next]
pull_request:
branches: [main, next]
jobs:
check:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13894f8d5 # v4
- name: Build Docker image (runs make check internally)
run: docker build .

5
.gitignore vendored
View File

@@ -3,3 +3,8 @@
*.tmp
*.dockerimage
/vendor
vendor.tzst
modcache.tzst
# Stale files
.drone.yml

View File

@@ -17,6 +17,7 @@ ARG DRONE_COMMIT_SHA unknown
RUN mkdir -p "$(go env GOMODCACHE)" && cd "$(go env GOMODCACHE)" && \
zstdmt -d --stdout /build/modcache.tzst | tar xf - && \
rm /build/modcache.tzst && cd /build
RUN make check
RUN \
cd mfer && go generate . && cd .. && \
GOPACKAGESDEBUG=true golangci-lint run ./... && \

View File

@@ -13,7 +13,7 @@ GOLDFLAGS += -X main.Version=$(VERSION)
GOLDFLAGS += -X main.Gitrev=$(GITREV_BUILD)
GOFLAGS := -ldflags "$(GOLDFLAGS)"
.PHONY: docker default run ci test fixme
.PHONY: docker default run ci test fixme check check-fmt lint
default: fmt test
@@ -51,9 +51,8 @@ fmt: mfer/mf.pb.go
-prettier -w *.json
-prettier -w *.md
lint:
golangci-lint run
sh -c 'test -z "$$(gofmt -l .)"'
lint: mfer/mf.pb.go
golangci-lint run ./...
docker: sneak-mfer.$(ARCH).tzst.dockerimage
@@ -79,3 +78,11 @@ modcache.tzst: go.mod go.sum
go mod download -x
cd $(shell go env GOMODCACHE) && tar -c . | pv | zstdmt -19 > $(PWD)/$@.tmp
mv $@.tmp $@
# Individual check targets
check-fmt:
@echo "==> Checking formatting..."
@test -z "$$(gofmt -l .)" || (echo "Files not formatted:" && gofmt -l . && exit 1)
# Run all checks (formatting, linting, tests) without modifying files
check: check-fmt lint test

View File

@@ -1,281 +0,0 @@
package checker
import (
"bytes"
"context"
"crypto/sha256"
"errors"
"io"
"os"
"path/filepath"
"github.com/multiformats/go-multihash"
"github.com/spf13/afero"
"sneak.berlin/go/mfer/mfer"
)
// Result represents the outcome of checking a single file.
type Result struct {
Path string // Relative path from manifest
Status Status // Verification result status
Message string // Human-readable description of the result
}
// Status represents the verification status of a file.
type Status int
const (
StatusOK Status = iota // File matches manifest (size and hash verified)
StatusMissing // File not found on disk
StatusSizeMismatch // File size differs from manifest
StatusHashMismatch // File hash differs from manifest
StatusExtra // File exists on disk but not in manifest
StatusError // Error occurred during verification
)
func (s Status) String() string {
switch s {
case StatusOK:
return "OK"
case StatusMissing:
return "MISSING"
case StatusSizeMismatch:
return "SIZE_MISMATCH"
case StatusHashMismatch:
return "HASH_MISMATCH"
case StatusExtra:
return "EXTRA"
case StatusError:
return "ERROR"
default:
return "UNKNOWN"
}
}
// CheckStatus contains progress information for the check operation.
type CheckStatus struct {
TotalFiles int64 // Total number of files in manifest
CheckedFiles int64 // Number of files checked so far
TotalBytes int64 // Total bytes to verify (sum of all file sizes)
CheckedBytes int64 // Bytes verified so far
BytesPerSec float64 // Current throughput rate
Failures int64 // Number of verification failures encountered
}
// Checker verifies files against a manifest.
type Checker struct {
basePath string
files []*mfer.MFFilePath
fs afero.Fs
// manifestPaths is a set of paths in the manifest for quick lookup
manifestPaths map[string]struct{}
}
// NewChecker creates a new Checker for the given manifest, base path, and filesystem.
// The basePath is the directory relative to which manifest paths are resolved.
// If fs is nil, the real filesystem (OsFs) is used.
func NewChecker(manifestPath string, basePath string, fs afero.Fs) (*Checker, error) {
if fs == nil {
fs = afero.NewOsFs()
}
m, err := mfer.NewManifestFromFile(fs, manifestPath)
if err != nil {
return nil, err
}
abs, err := filepath.Abs(basePath)
if err != nil {
return nil, err
}
files := m.Files()
manifestPaths := make(map[string]struct{}, len(files))
for _, f := range files {
manifestPaths[f.Path] = struct{}{}
}
return &Checker{
basePath: abs,
files: files,
fs: fs,
manifestPaths: manifestPaths,
}, nil
}
// FileCount returns the number of files in the manifest.
func (c *Checker) FileCount() int64 {
return int64(len(c.files))
}
// TotalBytes returns the total size of all files in the manifest.
func (c *Checker) TotalBytes() int64 {
var total int64
for _, f := range c.files {
total += f.Size
}
return total
}
// Check verifies all files against the manifest.
// Results are sent to the results channel as files are checked.
// Progress updates are sent to the progress channel approximately once per second.
// Both channels are closed when the method returns.
func (c *Checker) Check(ctx context.Context, results chan<- Result, progress chan<- CheckStatus) error {
if results != nil {
defer close(results)
}
if progress != nil {
defer close(progress)
}
totalFiles := int64(len(c.files))
totalBytes := c.TotalBytes()
var checkedFiles int64
var checkedBytes int64
var failures int64
for _, entry := range c.files {
select {
case <-ctx.Done():
return ctx.Err()
default:
}
result := c.checkFile(entry, &checkedBytes)
if result.Status != StatusOK {
failures++
}
checkedFiles++
if results != nil {
results <- result
}
// Send progress (simplified - every file for now)
if progress != nil {
sendCheckStatus(progress, CheckStatus{
TotalFiles: totalFiles,
CheckedFiles: checkedFiles,
TotalBytes: totalBytes,
CheckedBytes: checkedBytes,
Failures: failures,
})
}
}
return nil
}
func (c *Checker) checkFile(entry *mfer.MFFilePath, checkedBytes *int64) Result {
absPath := filepath.Join(c.basePath, entry.Path)
// Check if file exists
info, err := c.fs.Stat(absPath)
if err != nil {
if errors.Is(err, afero.ErrFileNotFound) || errors.Is(err, errors.New("file does not exist")) {
return Result{Path: entry.Path, Status: StatusMissing, Message: "file not found"}
}
// Check for "file does not exist" style errors
exists, _ := afero.Exists(c.fs, absPath)
if !exists {
return Result{Path: entry.Path, Status: StatusMissing, Message: "file not found"}
}
return Result{Path: entry.Path, Status: StatusError, Message: err.Error()}
}
// Check size
if info.Size() != entry.Size {
*checkedBytes += info.Size()
return Result{
Path: entry.Path,
Status: StatusSizeMismatch,
Message: "size mismatch",
}
}
// Open and hash file
f, err := c.fs.Open(absPath)
if err != nil {
return Result{Path: entry.Path, Status: StatusError, Message: err.Error()}
}
defer f.Close()
h := sha256.New()
n, err := io.Copy(h, f)
if err != nil {
return Result{Path: entry.Path, Status: StatusError, Message: err.Error()}
}
*checkedBytes += n
// Encode as multihash and compare
computed, err := multihash.Encode(h.Sum(nil), multihash.SHA2_256)
if err != nil {
return Result{Path: entry.Path, Status: StatusError, Message: err.Error()}
}
// Check against all hashes in manifest (at least one must match)
for _, hash := range entry.Hashes {
if bytes.Equal(computed, hash.MultiHash) {
return Result{Path: entry.Path, Status: StatusOK}
}
}
return Result{Path: entry.Path, Status: StatusHashMismatch, Message: "hash mismatch"}
}
// FindExtraFiles walks the filesystem and reports files not in the manifest.
// Results are sent to the results channel. The channel is closed when done.
func (c *Checker) FindExtraFiles(ctx context.Context, results chan<- Result) error {
if results != nil {
defer close(results)
}
return afero.Walk(c.fs, c.basePath, func(path string, info os.FileInfo, err error) error {
if err != nil {
return err
}
select {
case <-ctx.Done():
return ctx.Err()
default:
}
// Skip directories
if info.IsDir() {
return nil
}
// Get relative path
relPath, err := filepath.Rel(c.basePath, path)
if err != nil {
return err
}
// Check if path is in manifest
if _, exists := c.manifestPaths[relPath]; !exists {
if results != nil {
results <- Result{
Path: relPath,
Status: StatusExtra,
Message: "not in manifest",
}
}
}
return nil
})
}
// sendCheckStatus sends a status update without blocking.
func sendCheckStatus(ch chan<- CheckStatus, status CheckStatus) {
if ch == nil {
return
}
select {
case ch <- status:
default:
}
}

View File

@@ -113,7 +113,7 @@ func (mfa *CLIApp) fetchManifestOperation(ctx *cli.Context) error {
return fmt.Errorf("invalid path in manifest: %w", err)
}
fileURL := baseURL.String() + f.Path
fileURL := baseURL.String() + encodeFilePath(f.Path)
log.Infof("fetching %s", f.Path)
if err := downloadFile(fileURL, localPath, f, progress); err != nil {
@@ -139,6 +139,15 @@ func (mfa *CLIApp) fetchManifestOperation(ctx *cli.Context) error {
return nil
}
// encodeFilePath URL-encodes each segment of a file path while preserving slashes.
func encodeFilePath(p string) string {
segments := strings.Split(p, "/")
for i, seg := range segments {
segments[i] = url.PathEscape(seg)
}
return strings.Join(segments, "/")
}
// sanitizePath validates and sanitizes a file path from the manifest.
// It prevents path traversal attacks and rejects unsafe paths.
func sanitizePath(p string) (string, error) {

View File

@@ -16,6 +16,29 @@ import (
"sneak.berlin/go/mfer/mfer"
)
func TestEncodeFilePath(t *testing.T) {
tests := []struct {
input string
expected string
}{
{"file.txt", "file.txt"},
{"dir/file.txt", "dir/file.txt"},
{"my file.txt", "my%20file.txt"},
{"dir/my file.txt", "dir/my%20file.txt"},
{"file#1.txt", "file%231.txt"},
{"file?v=1.txt", "file%3Fv=1.txt"},
{"path/to/file with spaces.txt", "path/to/file%20with%20spaces.txt"},
{"100%done.txt", "100%25done.txt"},
}
for _, tt := range tests {
t.Run(tt.input, func(t *testing.T) {
result := encodeFilePath(tt.input)
assert.Equal(t, tt.expected, result)
})
}
}
func TestSanitizePath(t *testing.T) {
// Valid paths that should be accepted
validTests := []struct {

View File

@@ -25,6 +25,12 @@ func (mfa *CLIApp) generateManifestOperation(ctx *cli.Context) error {
Fs: mfa.Fs,
}
// Set seed for deterministic UUID if provided
if seed := ctx.String("seed"); seed != "" {
opts.Seed = seed
log.Infof("using deterministic seed for manifest UUID")
}
// Set up signing options if sign-key is provided
if signKey := ctx.String("sign-key"); signKey != "" {
opts.SigningOptions = &mfer.SigningOptions{

View File

@@ -154,6 +154,11 @@ func (mfa *CLIApp) run(args []string) {
Usage: "GPG key ID to sign the manifest with",
EnvVars: []string{"MFER_SIGN_KEY"},
},
&cli.StringFlag{
Name: "seed",
Usage: "Seed value for deterministic manifest UUID",
EnvVars: []string{"MFER_SEED"},
},
),
},
{

View File

@@ -1,373 +0,0 @@
package scanner
import (
"context"
"io"
"io/fs"
"path"
"path/filepath"
"strings"
"sync"
"time"
"github.com/spf13/afero"
"sneak.berlin/go/mfer/mfer"
)
// Phase 1: Enumeration
// ---------------------
// Walking directories and calling stat() on files to collect metadata.
// Builds the list of files to be scanned. Relatively fast (metadata only).
// EnumerateStatus contains progress information for the enumeration phase.
type EnumerateStatus struct {
FilesFound int64 // Number of files discovered so far
BytesFound int64 // Total size of discovered files (from stat)
}
// Phase 2: Scan (ToManifest)
// --------------------------
// Reading file contents and computing hashes for manifest generation.
// This is the expensive phase that reads all file data.
// ScanStatus contains progress information for the scan phase.
type ScanStatus struct {
TotalFiles int64 // Total number of files to scan
ScannedFiles int64 // Number of files scanned so far
TotalBytes int64 // Total bytes to read (sum of all file sizes)
ScannedBytes int64 // Bytes read so far
BytesPerSec float64 // Current throughput rate
}
// Options configures scanner behavior.
type Options struct {
IgnoreDotfiles bool // Skip files and directories starting with a dot
FollowSymLinks bool // Resolve symlinks instead of skipping them
Fs afero.Fs // Filesystem to use, defaults to OsFs if nil
}
// FileEntry represents a file that has been enumerated.
type FileEntry struct {
Path string // Relative path (used in manifest)
AbsPath string // Absolute path (used for reading file content)
Size int64 // File size in bytes
Mtime time.Time // Last modification time
Ctime time.Time // Creation time (platform-dependent)
}
// Scanner accumulates files and generates manifests from them.
type Scanner struct {
mu sync.RWMutex
files []*FileEntry
options *Options
fs afero.Fs
}
// New creates a new Scanner with default options.
func New() *Scanner {
return NewWithOptions(nil)
}
// NewWithOptions creates a new Scanner with the given options.
func NewWithOptions(opts *Options) *Scanner {
if opts == nil {
opts = &Options{}
}
fs := opts.Fs
if fs == nil {
fs = afero.NewOsFs()
}
return &Scanner{
files: make([]*FileEntry, 0),
options: opts,
fs: fs,
}
}
// EnumerateFile adds a single file to the scanner, calling stat() to get metadata.
func (s *Scanner) EnumerateFile(filePath string) error {
abs, err := filepath.Abs(filePath)
if err != nil {
return err
}
info, err := s.fs.Stat(abs)
if err != nil {
return err
}
// For single files, use the filename as the relative path
basePath := filepath.Dir(abs)
return s.enumerateFileWithInfo(filepath.Base(abs), basePath, info, nil)
}
// EnumeratePath walks a directory path and adds all files to the scanner.
// If progress is non-nil, status updates are sent as files are discovered.
// The progress channel is closed when the method returns.
func (s *Scanner) EnumeratePath(inputPath string, progress chan<- EnumerateStatus) error {
if progress != nil {
defer close(progress)
}
abs, err := filepath.Abs(inputPath)
if err != nil {
return err
}
afs := afero.NewReadOnlyFs(afero.NewBasePathFs(s.fs, abs))
return s.enumerateFS(afs, abs, progress)
}
// EnumeratePaths walks multiple directory paths and adds all files to the scanner.
// If progress is non-nil, status updates are sent as files are discovered.
// The progress channel is closed when the method returns.
func (s *Scanner) EnumeratePaths(progress chan<- EnumerateStatus, inputPaths ...string) error {
if progress != nil {
defer close(progress)
}
for _, p := range inputPaths {
abs, err := filepath.Abs(p)
if err != nil {
return err
}
afs := afero.NewReadOnlyFs(afero.NewBasePathFs(s.fs, abs))
if err := s.enumerateFS(afs, abs, progress); err != nil {
return err
}
}
return nil
}
// EnumerateFS walks an afero filesystem and adds all files to the scanner.
// If progress is non-nil, status updates are sent as files are discovered.
// The progress channel is closed when the method returns.
// basePath is used to compute absolute paths for file reading.
func (s *Scanner) EnumerateFS(afs afero.Fs, basePath string, progress chan<- EnumerateStatus) error {
if progress != nil {
defer close(progress)
}
return s.enumerateFS(afs, basePath, progress)
}
// enumerateFS is the internal implementation that doesn't close the progress channel.
func (s *Scanner) enumerateFS(afs afero.Fs, basePath string, progress chan<- EnumerateStatus) error {
return afero.Walk(afs, "/", func(p string, info fs.FileInfo, err error) error {
if err != nil {
return err
}
if s.options.IgnoreDotfiles && pathIsHidden(p) {
if info.IsDir() {
return filepath.SkipDir
}
return nil
}
return s.enumerateFileWithInfo(p, basePath, info, progress)
})
}
// enumerateFileWithInfo adds a file with pre-existing fs.FileInfo.
func (s *Scanner) enumerateFileWithInfo(filePath string, basePath string, info fs.FileInfo, progress chan<- EnumerateStatus) error {
if info.IsDir() {
// Manifests contain only files, directories are implied
return nil
}
// Clean the path - remove leading slash if present
cleanPath := filePath
if len(cleanPath) > 0 && cleanPath[0] == '/' {
cleanPath = cleanPath[1:]
}
// Compute absolute path for file reading
absPath := filepath.Join(basePath, cleanPath)
entry := &FileEntry{
Path: cleanPath,
AbsPath: absPath,
Size: info.Size(),
Mtime: info.ModTime(),
// Note: Ctime not available from fs.FileInfo on all platforms
// Will need platform-specific code to extract it
}
s.mu.Lock()
s.files = append(s.files, entry)
filesFound := int64(len(s.files))
var bytesFound int64
for _, f := range s.files {
bytesFound += f.Size
}
s.mu.Unlock()
sendEnumerateStatus(progress, EnumerateStatus{
FilesFound: filesFound,
BytesFound: bytesFound,
})
return nil
}
// Files returns a copy of all files added to the scanner.
func (s *Scanner) Files() []*FileEntry {
s.mu.RLock()
defer s.mu.RUnlock()
out := make([]*FileEntry, len(s.files))
copy(out, s.files)
return out
}
// FileCount returns the number of files in the scanner.
func (s *Scanner) FileCount() int64 {
s.mu.RLock()
defer s.mu.RUnlock()
return int64(len(s.files))
}
// TotalBytes returns the total size of all files in the scanner.
func (s *Scanner) TotalBytes() int64 {
s.mu.RLock()
defer s.mu.RUnlock()
var total int64
for _, f := range s.files {
total += f.Size
}
return total
}
// ToManifest reads all file contents, computes hashes, and generates a manifest.
// If progress is non-nil, status updates are sent approximately once per second.
// The progress channel is closed when the method returns.
// The manifest is written to the provided io.Writer.
func (s *Scanner) ToManifest(ctx context.Context, w io.Writer, progress chan<- ScanStatus) error {
if progress != nil {
defer close(progress)
}
s.mu.RLock()
files := make([]*FileEntry, len(s.files))
copy(files, s.files)
totalFiles := int64(len(files))
var totalBytes int64
for _, f := range files {
totalBytes += f.Size
}
s.mu.RUnlock()
builder := mfer.NewBuilder()
var scannedFiles int64
var scannedBytes int64
lastProgressTime := time.Now()
startTime := time.Now()
for _, entry := range files {
// Check for cancellation
select {
case <-ctx.Done():
return ctx.Err()
default:
}
// Open file
f, err := s.fs.Open(entry.AbsPath)
if err != nil {
return err
}
// Add to manifest with progress callback
bytesRead, err := builder.AddFile(
entry.Path,
entry.Size,
entry.Mtime,
f,
func(fileBytes int64) {
// Send progress at most once per second
now := time.Now()
if progress != nil && now.Sub(lastProgressTime) >= time.Second {
elapsed := now.Sub(startTime).Seconds()
currentBytes := scannedBytes + fileBytes
var rate float64
if elapsed > 0 {
rate = float64(currentBytes) / elapsed
}
sendScanStatus(progress, ScanStatus{
TotalFiles: totalFiles,
ScannedFiles: scannedFiles,
TotalBytes: totalBytes,
ScannedBytes: currentBytes,
BytesPerSec: rate,
})
lastProgressTime = now
}
},
)
f.Close()
if err != nil {
return err
}
scannedFiles++
scannedBytes += bytesRead
}
// Send final progress
if progress != nil {
elapsed := time.Since(startTime).Seconds()
var rate float64
if elapsed > 0 {
rate = float64(scannedBytes) / elapsed
}
sendScanStatus(progress, ScanStatus{
TotalFiles: totalFiles,
ScannedFiles: scannedFiles,
TotalBytes: totalBytes,
ScannedBytes: scannedBytes,
BytesPerSec: rate,
})
}
// Build and write manifest
return builder.Build(w)
}
// pathIsHidden returns true if the path or any of its parent directories
// start with a dot (hidden files/directories).
func pathIsHidden(p string) bool {
tp := path.Clean(p)
if strings.HasPrefix(tp, ".") {
return true
}
for {
d, f := path.Split(tp)
if strings.HasPrefix(f, ".") {
return true
}
if d == "" {
return false
}
tp = d[0 : len(d)-1] // trim trailing slash from dir
}
}
// sendEnumerateStatus sends a status update without blocking.
// If the channel is full, the update is dropped.
func sendEnumerateStatus(ch chan<- EnumerateStatus, status EnumerateStatus) {
if ch == nil {
return
}
select {
case ch <- status:
default:
// Channel full, drop this update
}
}
// sendScanStatus sends a status update without blocking.
// If the channel is full, the update is dropped.
func sendScanStatus(ch chan<- ScanStatus, status ScanStatus) {
if ch == nil {
return
}
select {
case ch <- status:
default:
// Channel full, drop this update
}
}

View File

@@ -5,12 +5,46 @@ import (
"errors"
"fmt"
"io"
"sort"
"strings"
"sync"
"time"
"unicode/utf8"
"github.com/multiformats/go-multihash"
)
// ValidatePath checks that a file path conforms to manifest path invariants:
// - Must be valid UTF-8
// - Must use forward slashes only (no backslashes)
// - Must be relative (no leading /)
// - Must not contain ".." segments
// - Must not contain empty segments (no "//")
// - Must not be empty
func ValidatePath(p string) error {
if p == "" {
return errors.New("path cannot be empty")
}
if !utf8.ValidString(p) {
return fmt.Errorf("path %q is not valid UTF-8", p)
}
if strings.ContainsRune(p, '\\') {
return fmt.Errorf("path %q contains backslash; use forward slashes only", p)
}
if strings.HasPrefix(p, "/") {
return fmt.Errorf("path %q is absolute; must be relative", p)
}
for _, seg := range strings.Split(p, "/") {
if seg == "" {
return fmt.Errorf("path %q contains empty segment", p)
}
if seg == ".." {
return fmt.Errorf("path %q contains '..' segment", p)
}
}
return nil
}
// RelFilePath represents a relative file path within a manifest.
type RelFilePath string
@@ -55,6 +89,15 @@ type Builder struct {
files []*MFFilePath
createdAt time.Time
signingOptions *SigningOptions
fixedUUID []byte // if set, use this UUID instead of generating one
}
// SetSeed derives a deterministic UUID from the given seed string.
// The seed is hashed once with SHA-256 and the first 16 bytes are used
// as a fixed UUID for the manifest.
func (b *Builder) SetSeed(seed string) {
hash := sha256.Sum256([]byte(seed))
b.fixedUUID = hash[:16]
}
// NewBuilder creates a new Builder.
@@ -75,6 +118,10 @@ func (b *Builder) AddFile(
reader io.Reader,
progress chan<- FileHashProgress,
) (FileSize, error) {
if err := ValidatePath(string(path)); err != nil {
return 0, err
}
// Create hash writer
h := sha256.New()
@@ -147,8 +194,8 @@ func (b *Builder) FileCount() int {
// This is useful when the hash is already known (e.g., from an existing manifest).
// Returns an error if path is empty, size is negative, or hash is nil/empty.
func (b *Builder) AddFileWithHash(path RelFilePath, size FileSize, mtime ModTime, hash Multihash) error {
if path == "" {
return errors.New("path cannot be empty")
if err := ValidatePath(string(path)); err != nil {
return err
}
if size < 0 {
return errors.New("size cannot be negative")
@@ -185,6 +232,11 @@ func (b *Builder) Build(w io.Writer) error {
b.mu.Lock()
defer b.mu.Unlock()
// Sort files by path for deterministic output
sort.Slice(b.files, func(i, j int) bool {
return b.files[i].Path < b.files[j].Path
})
// Create inner manifest
inner := &MFFile{
Version: MFFile_VERSION_ONE,
@@ -196,6 +248,7 @@ func (b *Builder) Build(w io.Writer) error {
m := &manifest{
pbInner: inner,
signingOptions: b.signingOptions,
fixedUUID: b.fixedUUID,
}
// Generate outer wrapper

View File

@@ -92,6 +92,77 @@ func TestBuilderBuild(t *testing.T) {
assert.True(t, strings.HasPrefix(buf.String(), MAGIC))
}
func TestNewTimestampFromTimeExtremeDate(t *testing.T) {
// Regression test: newTimestampFromTime used UnixNano() which panics
// for dates outside ~1678-2262. Now uses Nanosecond() which is safe.
tests := []struct {
name string
time time.Time
}{
{"zero time", time.Time{}},
{"year 1000", time.Date(1000, 1, 1, 0, 0, 0, 0, time.UTC)},
{"year 3000", time.Date(3000, 1, 1, 0, 0, 0, 123456789, time.UTC)},
{"unix epoch", time.Unix(0, 0)},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
// Should not panic
ts := newTimestampFromTime(tt.time)
assert.Equal(t, tt.time.Unix(), ts.Seconds)
assert.Equal(t, int32(tt.time.Nanosecond()), ts.Nanos)
})
}
}
func TestBuilderDeterministicOutput(t *testing.T) {
buildManifest := func() []byte {
b := NewBuilder()
// Use a fixed createdAt and UUID so output is reproducible
b.createdAt = time.Date(2025, 1, 1, 0, 0, 0, 0, time.UTC)
b.fixedUUID = make([]byte, 16) // all zeros
mtime := ModTime(time.Date(2025, 6, 1, 0, 0, 0, 0, time.UTC))
// Add files in reverse order to test sorting
files := []struct {
path string
content string
}{
{"c/file.txt", "content c"},
{"a/file.txt", "content a"},
{"b/file.txt", "content b"},
}
for _, f := range files {
r := bytes.NewReader([]byte(f.content))
_, err := b.AddFile(RelFilePath(f.path), FileSize(len(f.content)), mtime, r, nil)
require.NoError(t, err)
}
var buf bytes.Buffer
err := b.Build(&buf)
require.NoError(t, err)
return buf.Bytes()
}
out1 := buildManifest()
out2 := buildManifest()
assert.Equal(t, out1, out2, "two builds with same input should produce byte-identical output")
}
func TestSetSeedDeterministic(t *testing.T) {
b1 := NewBuilder()
b1.SetSeed("test-seed-value")
b2 := NewBuilder()
b2.SetSeed("test-seed-value")
assert.Equal(t, b1.fixedUUID, b2.fixedUUID, "same seed should produce same UUID")
assert.Len(t, b1.fixedUUID, 16, "UUID should be 16 bytes")
b3 := NewBuilder()
b3.SetSeed("different-seed")
assert.NotEqual(t, b1.fixedUUID, b3.fixedUUID, "different seeds should produce different UUIDs")
}
func TestBuilderBuildEmpty(t *testing.T) {
b := NewBuilder()

View File

@@ -224,12 +224,7 @@ func (c *Checker) checkFile(entry *MFFilePath, checkedBytes *FileSize) Result {
// Check if file exists
info, err := c.fs.Stat(absPath)
if err != nil {
if errors.Is(err, afero.ErrFileNotFound) || errors.Is(err, errors.New("file does not exist")) {
return Result{Path: relPath, Status: StatusMissing, Message: "file not found"}
}
// Check for "file does not exist" style errors
exists, _ := afero.Exists(c.fs, absPath)
if !exists {
if errors.Is(err, os.ErrNotExist) || errors.Is(err, afero.ErrFileNotFound) {
return Result{Path: relPath, Status: StatusMissing, Message: "file not found"}
}
return Result{Path: relPath, Status: StatusError, Message: err.Error()}
@@ -277,12 +272,14 @@ func (c *Checker) checkFile(entry *MFFilePath, checkedBytes *FileSize) Result {
// FindExtraFiles walks the filesystem and reports files not in the manifest.
// Results are sent to the results channel. The channel is closed when done.
// Hidden files/directories (starting with .) are skipped, as they are excluded
// from manifests by default. The manifest file itself is also skipped.
func (c *Checker) FindExtraFiles(ctx context.Context, results chan<- Result) error {
if results != nil {
defer close(results)
}
return afero.Walk(c.fs, string(c.basePath), func(path string, info os.FileInfo, err error) error {
return afero.Walk(c.fs, string(c.basePath), func(walkPath string, info os.FileInfo, err error) error {
if err != nil {
return err
}
@@ -293,16 +290,31 @@ func (c *Checker) FindExtraFiles(ctx context.Context, results chan<- Result) err
default:
}
// Get relative path
rel, err := filepath.Rel(string(c.basePath), walkPath)
if err != nil {
return err
}
// Skip hidden files and directories (dotfiles)
if IsHiddenPath(filepath.ToSlash(rel)) {
if info.IsDir() {
return filepath.SkipDir
}
return nil
}
// Skip directories
if info.IsDir() {
return nil
}
// Get relative path
rel, err := filepath.Rel(string(c.basePath), path)
if err != nil {
return err
// Skip manifest files
base := filepath.Base(rel)
if base == "index.mf" || base == ".index.mf" {
return nil
}
relPath := RelFilePath(rel)
// Check if path is in manifest

View File

@@ -305,6 +305,44 @@ func TestFindExtraFiles(t *testing.T) {
assert.Equal(t, "not in manifest", extras[0].Message)
}
func TestFindExtraFilesSkipsManifestAndDotfiles(t *testing.T) {
fs := afero.NewMemMapFs()
manifestFiles := map[string][]byte{
"file1.txt": []byte("in manifest"),
}
createTestManifest(t, fs, "/data/.index.mf", manifestFiles)
createFilesOnDisk(t, fs, "/data", map[string][]byte{
"file1.txt": []byte("in manifest"),
})
// Create dotfile and manifest that should be skipped
require.NoError(t, afero.WriteFile(fs, "/data/.hidden", []byte("hidden"), 0o644))
require.NoError(t, afero.WriteFile(fs, "/data/.config/settings", []byte("cfg"), 0o644))
// Create a real extra file
require.NoError(t, fs.MkdirAll("/data", 0o755))
require.NoError(t, afero.WriteFile(fs, "/data/extra.txt", []byte("extra"), 0o644))
chk, err := NewChecker("/data/.index.mf", "/data", fs)
require.NoError(t, err)
results := make(chan Result, 10)
err = chk.FindExtraFiles(context.Background(), results)
require.NoError(t, err)
var extras []Result
for r := range results {
extras = append(extras, r)
}
// Should only report extra.txt, not .hidden, .config/settings, or .index.mf
for _, e := range extras {
t.Logf("extra: %s", e.Path)
}
assert.Len(t, extras, 1)
if len(extras) > 0 {
assert.Equal(t, RelFilePath("extra.txt"), extras[0].Path)
}
}
func TestFindExtraFilesContextCancellation(t *testing.T) {
fs := afero.NewMemMapFs()
files := map[string][]byte{"file.txt": []byte("data")}
@@ -381,6 +419,39 @@ func TestCheckSubdirectories(t *testing.T) {
assert.Equal(t, 3, okCount)
}
func TestCheckMissingFileDetectedWithoutFallback(t *testing.T) {
// Regression test: errors.Is(err, errors.New("...")) never matches because
// errors.New creates a new value each time. The fix uses os.ErrNotExist instead.
fs := afero.NewMemMapFs()
files := map[string][]byte{
"exists.txt": []byte("here"),
"missing.txt": []byte("not on disk"),
}
createTestManifest(t, fs, "/manifest.mf", files)
// Only create one file on disk
createFilesOnDisk(t, fs, "/data", map[string][]byte{
"exists.txt": []byte("here"),
})
chk, err := NewChecker("/manifest.mf", "/data", fs)
require.NoError(t, err)
results := make(chan Result, 10)
err = chk.Check(context.Background(), results, nil)
require.NoError(t, err)
statusCounts := map[Status]int{}
for r := range results {
statusCounts[r.Status]++
if r.Status == StatusMissing {
assert.Equal(t, RelFilePath("missing.txt"), r.Path)
}
}
assert.Equal(t, 1, statusCounts[StatusOK], "one file should be OK")
assert.Equal(t, 1, statusCounts[StatusMissing], "one file should be MISSING")
assert.Equal(t, 0, statusCounts[StatusError], "no files should be ERROR")
}
func TestCheckEmptyManifest(t *testing.T) {
fs := afero.NewMemMapFs()
// Create manifest with no files

View File

@@ -3,4 +3,9 @@ package mfer
const (
Version = "0.1.0"
ReleaseDate = "2025-12-17"
// MaxDecompressedSize is the maximum allowed size of decompressed manifest
// data (256 MB). This prevents decompression bombs from consuming excessive
// memory.
MaxDecompressedSize int64 = 256 * 1024 * 1024
)

View File

@@ -76,10 +76,20 @@ func (m *manifest) deserializeInner() error {
}
defer zr.Close()
dat, err := io.ReadAll(zr)
// Limit decompressed size to prevent decompression bombs.
// Use declared size + 1 byte to detect overflow, capped at MaxDecompressedSize.
maxSize := MaxDecompressedSize
if m.pbOuter.Size > 0 && m.pbOuter.Size < int64(maxSize) {
maxSize = int64(m.pbOuter.Size) + 1
}
limitedReader := io.LimitReader(zr, maxSize)
dat, err := io.ReadAll(limitedReader)
if err != nil {
return err
}
if int64(len(dat)) >= MaxDecompressedSize {
return fmt.Errorf("decompressed data exceeds maximum allowed size of %d bytes", MaxDecompressedSize)
}
isize := len(dat)
if int64(isize) != m.pbOuter.Size {

View File

@@ -100,7 +100,7 @@ func gpgExtractPubKeyFingerprint(pubKey []byte) (string, error) {
if err != nil {
return "", fmt.Errorf("failed to create temp dir: %w", err)
}
defer os.RemoveAll(tmpDir)
defer func() { _ = os.RemoveAll(tmpDir) }()
// Set restrictive permissions
if err := os.Chmod(tmpDir, 0o700); err != nil {
@@ -158,7 +158,7 @@ func gpgVerify(data, signature, pubKey []byte) error {
if err != nil {
return fmt.Errorf("failed to create temp dir: %w", err)
}
defer os.RemoveAll(tmpDir)
defer func() { _ = os.RemoveAll(tmpDir) }()
// Set restrictive permissions
if err := os.Chmod(tmpDir, 0o700); err != nil {

View File

@@ -34,15 +34,15 @@ func testGPGEnv(t *testing.T) (GPGKeyID, func()) {
// Save original GNUPGHOME and set new one
origGPGHome := os.Getenv("GNUPGHOME")
os.Setenv("GNUPGHOME", gpgHome)
require.NoError(t, os.Setenv("GNUPGHOME", gpgHome))
cleanup := func() {
if origGPGHome == "" {
os.Unsetenv("GNUPGHOME")
_ = os.Unsetenv("GNUPGHOME")
} else {
os.Setenv("GNUPGHOME", origGPGHome)
_ = os.Setenv("GNUPGHOME", origGPGHome)
}
os.RemoveAll(gpgHome)
_ = os.RemoveAll(gpgHome)
}
// Generate a test key with no passphrase

View File

@@ -17,6 +17,7 @@ type manifest struct {
pbOuter *MFFileOuter
output *bytes.Buffer
signingOptions *SigningOptions
fixedUUID []byte // if set, use this UUID instead of generating one
}
func (m *manifest) String() string {

View File

@@ -46,6 +46,9 @@ message MFFileOuter {
message MFFilePath {
// required attributes:
// Path invariants: must be valid UTF-8, use forward slashes only,
// be relative (no leading /), contain no ".." segments, and no
// empty segments (no "//").
string path = 1;
int64 size = 2;

View File

@@ -47,6 +47,7 @@ type ScannerOptions struct {
FollowSymLinks bool // Resolve symlinks instead of skipping them
Fs afero.Fs // Filesystem to use, defaults to OsFs if nil
SigningOptions *SigningOptions // GPG signing options (nil = no signing)
Seed string // If set, derive a deterministic UUID from this seed
}
// FileEntry represents a file that has been enumerated.
@@ -276,6 +277,9 @@ func (s *Scanner) ToManifest(ctx context.Context, w io.Writer, progress chan<- S
if s.options.SigningOptions != nil {
builder.SetSigningOptions(s.options.SigningOptions)
}
if s.options.Seed != "" {
builder.SetSeed(s.options.Seed)
}
var scannedFiles FileCount
var scannedBytes FileSize
@@ -385,6 +389,9 @@ func (s *Scanner) ToManifest(ctx context.Context, w io.Writer, progress chan<- S
// The path should use forward slashes.
func IsHiddenPath(p string) bool {
tp := path.Clean(p)
if tp == "." || tp == "/" {
return false
}
if strings.HasPrefix(tp, ".") {
return true
}

View File

@@ -352,6 +352,8 @@ func TestIsHiddenPath(t *testing.T) {
{"/absolute/.hidden", true},
{"./relative", false}, // path.Clean removes leading ./
{"a/b/c/.d/e", true},
{".", false}, // current directory is not hidden
{"/", false}, // root is not hidden
}
for _, tt := range tests {

View File

@@ -16,11 +16,10 @@ import (
const MAGIC string = "ZNAVSRFG"
func newTimestampFromTime(t time.Time) *Timestamp {
out := &Timestamp{
return &Timestamp{
Seconds: t.Unix(),
Nanos: int32(t.UnixNano() - (t.Unix() * 1000000000)),
Nanos: int32(t.Nanosecond()),
}
return out
}
func (m *manifest) generate() error {
@@ -50,8 +49,13 @@ func (m *manifest) generateOuter() error {
return errors.New("internal error")
}
// Generate UUID and set on inner message
manifestUUID := uuid.New()
// Use fixed UUID if provided, otherwise generate a new one
var manifestUUID uuid.UUID
if len(m.fixedUUID) == 16 {
copy(manifestUUID[:], m.fixedUUID)
} else {
manifestUUID = uuid.New()
}
m.pbInner.Uuid = manifestUUID[:]
innerData, err := proto.MarshalOptions{Deterministic: true}.Marshal(m.pbInner)

Binary file not shown.

Binary file not shown.