Default the manifest to index.mf and keep it out of its own listing (closes #100)

mfer gen now writes index.mf instead of .index.mf, the name fetch
requests and the README calls the standard filename. Given a
directory, check, freshen, list and export look only for index.mf;
.index.mf is no longer recognized. Because index.mf is not hidden, gen
and freshen now leave the manifest they write out of its own listing by
file identity (os.SameFile), so it is recognized however its path is
spelled, through a symlink included, while an ordinary file of the same
name elsewhere in the tree is still listed.

Model: opus-5-5

.gitignore

@@ -8,7 +8,7 @@ vendor.tzst
 modcache.tzst
 
 # Generated manifest files
-.index.mf
+/index.mf
 
 # Secrets
 .env

FORMAT.md

@@ -49,7 +49,9 @@ The `innerMessage` field is compressed with
 enforce a decompression size limit to prevent decompression bombs. The reference
 implementation limits decompressed size to 256 MB. It writes zstd frames with a
 window of at most 8 MiB, the largest window the zstd format recommends decoders
-support, and refuses frames that ask for a larger one.
+support, and refuses frames that ask for a larger one. It also refuses an inner
+message whose file entries, hashes, timestamps and MIME types, counted at 160,
+112, 64 and 16 bytes each, add up to more than 8 times its size.
 
 ## Inner Message (`MFFile`)
 

README.md

@@ -36,12 +36,12 @@ Generate a manifest for a directory tree, verify it later, and fetch a published
 tree by URL:
 
 ```sh
-# Write .index.mf, a manifest of the files under the current directory.
+# Write index.mf, a manifest of the files under the current directory.
 bin/mfer gen .
 
 # Verify the files on disk against the manifest. Exits nonzero if any file
 # is missing or corrupted.
-bin/mfer check .index.mf
+bin/mfer check index.mf
 
 # Download and cryptographically verify a tree published over HTTP: mfer
 # fetches <url>/index.mf, then downloads every file it lists.

internal/cli/check.go

@@ -75,25 +75,22 @@ func safeRateUint64(rate float64) uint64 {
 	return uint64(rate)
 }
 
-// findManifest looks for a manifest file in the given directory.
-// It checks for index.mf and .index.mf, returning the first one found.
+// findManifest returns the path of the manifest with the default name in
+// dir, or an error if there is none.
 func findManifest(fs afero.Fs, dir string) (string, error) {
-	candidates := []string{"index.mf", ".index.mf"}
-	for _, name := range candidates {
-		path := filepath.Join(dir, name)
+	path := filepath.Join(dir, defaultManifestName)
 
-		exists, err := afero.Exists(fs, path)
-		if err != nil {
-			return "", err
-		}
-
-		if exists {
-			return path, nil
-		}
+	exists, err := afero.Exists(fs, path)
+	if err != nil {
+		return "", err
 	}
 
-	return "", fmt.Errorf(
-		"%w in %s (looked for index.mf and .index.mf)", errNoManifestFound, dir)
+	if !exists {
+		return "", fmt.Errorf("%w in %s (looked for %s)",
+			errNoManifestFound, dir, defaultManifestName)
+	}
+
+	return path, nil
 }
 
 // fetchManifestToTemp downloads a manifest URL to a temporary file and

internal/cli/entry_test.go

@@ -8,6 +8,7 @@ import (
 	"io"
 	"math/rand"
 	"os"
+	"path/filepath"
 	"slices"
 	"strings"
 	"sync"
@@ -737,6 +738,80 @@ func TestGenerateFailsWithoutForceWhenOutputExists(t *testing.T) {
 	assert.Equal(t, "existing", string(content), "original file should be preserved")
 }
 
+// manifestPaths returns the file paths listed by the manifest at path.
+func manifestPaths(t *testing.T, fs afero.Fs, path string) []string {
+	t.Helper()
+
+	manifest, err := mfer.NewManifestFromFile(fs, path)
+	require.NoError(t, err)
+
+	paths := make([]string, 0, len(manifest.Files()))
+	for _, f := range manifest.Files() {
+		paths = append(paths, f.GetPath())
+	}
+
+	return paths
+}
+
+// TestGenerateLeavesOutputOutOfListing overwrites an output file inside the
+// scanned tree with --force: the old file is not listed, even when the tree
+// is named through a symlink, while a file named index.mf in a subdirectory
+// still is. The output file is recognized by file identity, which needs
+// the real filesystem.
+func TestGenerateLeavesOutputOutOfListing(t *testing.T) {
+	t.Parallel()
+
+	// Paths are relative to a temp dir holding data/tree and link, a
+	// symlink to data.
+	for name, tc := range map[string]struct{ input, output string }{
+		"default name":                 {"data/tree", "data/tree/index.mf"},
+		"other name in a subdirectory": {"data/tree", "data/tree/sub/listing.mf"},
+		"tree named through a symlink": {"link/tree", "data/tree/index.mf"},
+	} {
+		t.Run(name, func(t *testing.T) {
+			t.Parallel()
+
+			root := t.TempDir()
+			tree := filepath.Join(root, "data", "tree")
+			output := filepath.Join(root, tc.output)
+
+			fs := afero.NewOsFs()
+			require.NoError(t, fs.MkdirAll(filepath.Join(tree, "sub"), 0o750))
+			require.NoError(t,
+				os.Symlink(filepath.Join(root, "data"), filepath.Join(root, "link")))
+			writeTestFile(t, fs, filepath.Join(tree, testFileTxt), "hello")
+			writeTestFile(t, fs, filepath.Join(tree, "sub", "index.mf"), "an ordinary file")
+			writeTestFile(t, fs, output, "previous manifest")
+
+			opts := testOpts([]string{
+				testApp, cmdGenerate, "-q", "--force", "-o", output, filepath.Join(root, tc.input),
+			}, fs)
+			require.Equal(t, 0, runCLI(opts), "stderr: %s", testStderr(t, opts))
+
+			assert.ElementsMatch(t, []string{testFileTxt, "sub/index.mf"},
+				manifestPaths(t, fs, output))
+		})
+	}
+}
+
+// TestGenerateDefaultOutputLeftOutOfListing runs gen with --force and no
+// other arguments, so it scans the current directory and writes the
+// relative path index.mf: the index.mf already there is not listed.
+//
+//nolint:paralleltest // changes the process-global working directory
+func TestGenerateDefaultOutputLeftOutOfListing(t *testing.T) {
+	chdirTemp(t)
+
+	fs := afero.NewOsFs()
+	writeTestFile(t, fs, testFileTxt, "hello")
+	writeTestFile(t, fs, "index.mf", "previous manifest")
+
+	opts := testOpts([]string{testApp, cmdGenerate, "-q", "--force"}, fs)
+	require.Equal(t, 0, runCLI(opts), "stderr: %s", testStderr(t, opts))
+
+	assert.Equal(t, []string{testFileTxt}, manifestPaths(t, fs, "index.mf"))
+}
+
 func TestGenerateAtomicWriteUsesTemp(t *testing.T) {
 	t.Parallel()
 

internal/cli/errmsg_test.go

@@ -74,7 +74,7 @@ func TestNoManifestFoundMessage(t *testing.T) {
 	_, err := findManifest(afero.NewMemMapFs(), "/tmp/x")
 	require.ErrorIs(t, err, errNoManifestFound)
 	assert.EqualError(t, err,
-		"no manifest found in /tmp/x (looked for index.mf and .index.mf)")
+		"no manifest found in /tmp/x (looked for index.mf)")
 }
 
 func TestVerifyRequiredSignerMessages(t *testing.T) {

internal/cli/fetch.go

@@ -313,7 +313,7 @@ func checkNoSymlinks(p string) error {
 
 // resolveManifestURL takes a URL and returns the manifest URL.
 // If the URL already ends with .mf, it's returned as-is.
-// Otherwise, index.mf is appended.
+// Otherwise, the default manifest name is appended.
 func resolveManifestURL(inputURL string) (string, error) {
 	parsed, err := url.Parse(inputURL)
 	if err != nil {
@@ -330,8 +330,7 @@ func resolveManifestURL(inputURL string) (string, error) {
 		parsed.Path += "/"
 	}
 
-	// Append index.mf
-	parsed.Path += "index.mf"
+	parsed.Path += defaultManifestName
 
 	return parsed.String(), nil
 }

internal/cli/freshen.go

@@ -7,6 +7,7 @@ import (
 	"fmt"
 	"io"
 	"io/fs"
+	"os"
 	"path/filepath"
 	"time"
 
@@ -56,7 +57,7 @@ type freshenEntry struct {
 type freshenScanner struct {
 	fs              afero.Fs
 	absBase         string
-	manifestBase    string
+	manifestInfo    fs.FileInfo // the manifest being freshened
 	includeDotfiles bool
 	followSymlinks  bool
 	showProgress    bool
@@ -156,11 +157,6 @@ func (s *freshenScanner) walk(path string, info fs.FileInfo, walkErr error) erro
 			"freshen: failed to compute relative path for %s: %w", path, err)
 	}
 
-	// Skip the manifest file itself
-	if relPath == s.manifestBase || relPath == "."+s.manifestBase {
-		return nil
-	}
-
 	// Handle dotfiles
 	if !s.includeDotfiles && mfer.IsHiddenPath(filepath.ToSlash(relPath)) {
 		if info.IsDir() {
@@ -185,6 +181,12 @@ func (s *freshenScanner) walk(path string, info fs.FileInfo, walkErr error) erro
 		info = realInfo
 	}
 
+	// Skip the manifest file itself, however its path is spelled; gen
+	// leaves out its output file the same way.
+	if os.SameFile(info, s.manifestInfo) {
+		return nil
+	}
+
 	s.scanCount++
 
 	// Check against existing manifest
@@ -364,17 +366,22 @@ func (mfa *CLIApp) freshenScan(
 	startScan := time.Now()
 	showProgress := ctx.Bool("progress")
 
+	manifestInfo, err := mfa.Fs.Stat(manifestPath)
+	if err != nil {
+		return nil, 0, fmt.Errorf("freshen: %w", err)
+	}
+
 	scanner := &freshenScanner{
 		fs:              mfa.Fs,
 		absBase:         absBase,
-		manifestBase:    filepath.Base(manifestPath),
+		manifestInfo:    manifestInfo,
 		includeDotfiles: ctx.Bool("include-dotfiles"),
 		followSymlinks:  ctx.Bool("follow-symlinks"),
 		showProgress:    showProgress,
 		existingByPath:  existingByPath,
 	}
 
-	err := afero.Walk(mfa.Fs, absBase, scanner.walk)
+	err = afero.Walk(mfa.Fs, absBase, scanner.walk)
 
 	if showProgress {
 		log.ProgressDone()

internal/cli/freshen_test.go

@@ -5,6 +5,7 @@ import (
 	"bytes"
 	"context"
 	"os"
+	"path/filepath"
 	"testing"
 	"time"
 
@@ -29,7 +30,7 @@ func (s stubFileInfo) IsDir() bool        { return false }
 func (s stubFileInfo) Sys() any           { return nil }
 
 // setupFreshenDir populates /testdir with two files, scans it, and
-// writes the resulting manifest to /testdir/.index.mf.
+// writes the resulting manifest to /testdir/index.mf.
 func setupFreshenDir(t *testing.T, fs afero.Fs) {
 	t.Helper()
 
@@ -48,7 +49,7 @@ func setupFreshenDir(t *testing.T, fs afero.Fs) {
 
 	// Write manifest to filesystem
 	require.NoError(t,
-		afero.WriteFile(fs, "/testdir/.index.mf", manifestBuf.Bytes(), 0o644))
+		afero.WriteFile(fs, "/testdir/index.mf", manifestBuf.Bytes(), 0o644))
 }
 
 func TestFreshenUnchanged(t *testing.T) {
@@ -58,7 +59,7 @@ func TestFreshenUnchanged(t *testing.T) {
 	setupFreshenDir(t, fs)
 
 	// Parse manifest to verify
-	manifest, err := mfer.NewManifestFromFile(fs, "/testdir/.index.mf")
+	manifest, err := mfer.NewManifestFromFile(fs, "/testdir/index.mf")
 	require.NoError(t, err)
 	assert.Len(t, manifest.Files(), 2)
 }
@@ -70,7 +71,7 @@ func TestFreshenWithChanges(t *testing.T) {
 	setupFreshenDir(t, fs)
 
 	// Verify initial manifest has 2 files
-	manifest, err := mfer.NewManifestFromFile(fs, "/testdir/.index.mf")
+	manifest, err := mfer.NewManifestFromFile(fs, "/testdir/index.mf")
 	require.NoError(t, err)
 	assert.Len(t, manifest.Files(), 2)
 
@@ -95,6 +96,36 @@ func TestFreshenWithChanges(t *testing.T) {
 	assert.Equal(t, "modified content2", string(content))
 }
 
+// TestFreshenLeavesManifestOutOfListing freshens a manifest kept in a
+// subdirectory of the tree it lists: the manifest is not listed, while an
+// ordinary file of the same name at the top of the tree is. The manifest
+// is recognized by file identity, which needs the real filesystem.
+func TestFreshenLeavesManifestOutOfListing(t *testing.T) {
+	t.Parallel()
+
+	root := t.TempDir()
+	manifestPath := filepath.Join(root, "sub", "listing.mf")
+
+	fs := afero.NewOsFs()
+	require.NoError(t, fs.MkdirAll(filepath.Join(root, "sub"), 0o750))
+	writeTestFile(t, fs, filepath.Join(root, testFileTxt), "hello")
+	writeTestFile(t, fs, filepath.Join(root, "listing.mf"), "an ordinary file")
+
+	opts := testOpts([]string{testApp, cmdGenerate, "-q", "-o", manifestPath, root}, fs)
+	require.Equal(t, 0, runCLI(opts), "stderr: %s", testStderr(t, opts))
+
+	// A new file gives freshen something to write.
+	writeTestFile(t, fs, filepath.Join(root, "added.txt"), "added")
+
+	opts = testOpts([]string{
+		testApp, "freshen", "-q", testFlagBase, root, manifestPath,
+	}, fs)
+	require.Equal(t, 0, runCLI(opts), "stderr: %s", testStderr(t, opts))
+
+	assert.ElementsMatch(t, []string{testFileTxt, "listing.mf", "added.txt"},
+		manifestPaths(t, fs, manifestPath))
+}
+
 // TestFreshenRecordEntryMtimePresence pins the behavior of recordEntry
 // with respect to MFFilePath.Mtime, which is a message pointer with
 // proto3 field presence and may legitimately be absent.

internal/cli/gen.go

@@ -94,6 +94,7 @@ func (mfa *CLIApp) buildScannerOptions(ctx *cli.Context) *mfer.ScannerOptions {
 		FollowSymLinks:    ctx.Bool("follow-symlinks"),
 		IncludeTimestamps: ctx.Bool("include-timestamps"),
 		Fs:                mfa.Fs,
+		OutputPath:        ctx.String("output"),
 	}
 
 	// Set seed for deterministic UUID if provided

internal/cli/manifest_loader.go

@@ -65,7 +65,7 @@ func (mfa *CLIApp) openManifestReader(pathOrURL string) (io.ReadCloser, error) {
 }
 
 // resolveManifestArg resolves the manifest path from CLI arguments.
-// HTTP(S) URLs are returned as-is. Directories are searched for index.mf/.index.mf.
+// HTTP(S) URLs are returned as-is. Directories are searched for index.mf.
 // If no argument is given, the current directory is searched.
 func (mfa *CLIApp) resolveManifestArg(ctx *cli.Context) (string, error) {
 	if ctx.Args().Len() > 0 {

internal/cli/mfer.go

@@ -25,6 +25,11 @@ const (
 	flagProgress = "progress"
 
 	manifestArgsUsage = "[manifest file]"
+
+	// defaultManifestName is the filename gen writes by default, the one
+	// looked for when a command is given a directory, and the one fetch
+	// appends to a directory URL.
+	defaultManifestName = "index.mf"
 )
 
 // errUnknownCommand indicates an unrecognized command argument.
@@ -156,7 +161,7 @@ func (mfa *CLIApp) generateCommand() *cli.Command {
 			},
 			&cli.StringFlag{
 				Name:    "output",
-				Value:   "./.index.mf",
+				Value:   defaultManifestName,
 				Aliases: []string{"o"},
 				Usage:   "Specify output filename",
 			},

mfer/checker_test.go

@@ -360,7 +360,7 @@ func TestFindExtraFilesSkipsManifestAndDotfiles(t *testing.T) {
 	manifestFiles := map[string][]byte{
 		testFile1: []byte("in manifest"),
 	}
-	createTestManifest(t, fs, "/data/.index.mf", manifestFiles)
+	createTestManifest(t, fs, "/data/index.mf", manifestFiles)
 	createFilesOnDisk(t, fs, map[string][]byte{
 		testFile1: []byte("in manifest"),
 	})
@@ -371,7 +371,7 @@ func TestFindExtraFilesSkipsManifestAndDotfiles(t *testing.T) {
 	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)
+	chk, err := NewChecker("/data/index.mf", "/data", fs)
 	require.NoError(t, err)
 
 	results := make(chan Result, 10)
@@ -383,7 +383,7 @@ func TestFindExtraFilesSkipsManifestAndDotfiles(t *testing.T) {
 		extras = append(extras, r)
 	}
 
-	// Should only report extra.txt, not .hidden, .config/settings, or .index.mf
+	// Should only report extra.txt, not .hidden, .config/settings, or index.mf
 	for _, e := range extras {
 		t.Logf("extra: %s", e.Path)
 	}
@@ -528,7 +528,7 @@ func TestFindExtraFilesSkipsDotfiles(t *testing.T) {
 	files := map[string][]byte{
 		testFile1: []byte("in manifest"),
 	}
-	createTestManifest(t, fs, "/data/.index.mf", files)
+	createTestManifest(t, fs, "/data/index.mf", files)
 	createFilesOnDisk(t, fs, files)
 
 	// Add dotfiles and manifest file on disk
@@ -537,7 +537,7 @@ func TestFindExtraFilesSkipsDotfiles(t *testing.T) {
 	require.NoError(t,
 		afero.WriteFile(fs, "/data/.git/config", []byte("git config"), 0o644))
 
-	chk, err := NewChecker("/data/.index.mf", "/data", fs)
+	chk, err := NewChecker("/data/index.mf", "/data", fs)
 	require.NoError(t, err)
 
 	results := make(chan Result, 10)

mfer/constants.go

@@ -17,4 +17,27 @@ const (
 
 	// uuidLength is the length in bytes of a binary UUID.
 	uuidLength = 16
+
+	// Numbers in mf.proto of MFFile.files and of the MFFilePath fields
+	// that decoding sets aside a fixed amount of memory for.
+	filesFieldNumber    = 101
+	hashesFieldNumber   = 3
+	mimeTypeFieldNumber = 301
+	mtimeFieldNumber    = 302
+	ctimeFieldNumber    = 303
+
+	// Bytes decoding sets aside for each file entry, hash, timestamp and
+	// MIME type, however short its encoding. checkDecodedSize refuses an
+	// inner message for which these add up to more than maxDecodedGrowth
+	// times its size.
+	decodedFileEntrySize = 160
+	decodedHashSize      = 112
+	decodedTimestampSize = 64
+	decodedMIMETypeSize  = 16
+
+	// Each file entry mfer writes holds a path of at least one byte, a
+	// 34-byte SHA-256 multihash and a modification time: at least 47 bytes,
+	// counted at 336. So its manifests add up to at most about 7.15 times
+	// their size, and this limit is about 12% above that.
+	maxDecodedGrowth = 8
 )

mfer/deserialize.go

@@ -11,6 +11,7 @@ import (
 	"github.com/google/uuid"
 	"github.com/klauspost/compress/zstd"
 	"github.com/spf13/afero"
+	"google.golang.org/protobuf/encoding/protowire"
 	"google.golang.org/protobuf/proto"
 	"sneak.berlin/go/mfer/internal/bork"
 	"sneak.berlin/go/mfer/internal/log"
@@ -27,6 +28,8 @@ var (
 	errUUIDMismatch         = errors.New("outer and inner UUID mismatch")
 	errInvalidFileFormat    = errors.New("invalid file format")
 	errInvalidManifestPath  = errors.New("manifest contains invalid path")
+	errDecodedTooLarge      = errors.New(
+		"manifest would take too much memory to decode")
 )
 
 // validateUUID checks that the byte slice is a valid UUID (16 bytes, parseable).
@@ -154,6 +157,85 @@ func (m *manifest) decompressInner() ([]byte, error) {
 	return dat, nil
 }
 
+// checkDecodedSize refuses an encoded inner message whose file entries,
+// hashes, timestamps and MIME types would take more than maxDecodedGrowth
+// times its size to decode. Decoding sets aside a fixed amount for each,
+// however short its encoding, so a message of empty ones would take about
+// 50 times its size.
+func checkDecodedSize(inner []byte) error {
+	limit := maxDecodedGrowth * int64(len(inner))
+
+	var decoded int64
+
+	add := func(size int64) error {
+		decoded += size
+		if decoded > limit {
+			return errDecodedTooLarge
+		}
+
+		return nil
+	}
+
+	return forEachBytesField(inner, func(num protowire.Number, entry []byte) error {
+		if num != filesFieldNumber {
+			return nil
+		}
+
+		err := add(decodedFileEntrySize)
+		if err != nil {
+			return err
+		}
+
+		return forEachBytesField(entry, func(num protowire.Number, _ []byte) error {
+			if num == hashesFieldNumber {
+				return add(decodedHashSize)
+			}
+
+			if num == mtimeFieldNumber || num == ctimeFieldNumber {
+				return add(decodedTimestampSize)
+			}
+
+			if num == mimeTypeFieldNumber {
+				return add(decodedMIMETypeSize)
+			}
+
+			return nil
+		})
+	})
+}
+
+// forEachBytesField calls fn with the number and value of each
+// length-delimited field in the encoded message msg, and fails if msg is
+// malformed.
+func forEachBytesField(
+	msg []byte, fn func(num protowire.Number, value []byte) error,
+) error {
+	for len(msg) > 0 {
+		num, wireType, tagLen := protowire.ConsumeTag(msg)
+		if tagLen < 0 {
+			return protowire.ParseError(tagLen)
+		}
+
+		valueLen := protowire.ConsumeFieldValue(num, wireType, msg[tagLen:])
+		if valueLen < 0 {
+			return protowire.ParseError(valueLen)
+		}
+
+		if wireType == protowire.BytesType {
+			value, _ := protowire.ConsumeBytes(msg[tagLen:])
+
+			err := fn(num, value)
+			if err != nil {
+				return err
+			}
+		}
+
+		msg = msg[tagLen+valueLen:]
+	}
+
+	return nil
+}
+
 func (m *manifest) deserializeInner() error {
 	err := m.validateOuterHeader()
 	if err != nil {
@@ -177,10 +259,16 @@ func (m *manifest) deserializeInner() error {
 		return bork.ErrFileTruncated
 	}
 
+	err = checkDecodedSize(dat)
+	if err != nil {
+		return fmt.Errorf("deserialize: unmarshal inner: %w", err)
+	}
+
 	// Deserialize inner message
 	m.pbInner = new(MFFile)
 
-	err = proto.Unmarshal(dat, m.pbInner)
+	// Unknown fields would cost memory; mfer never writes a loaded manifest out.
+	err = proto.UnmarshalOptions{DiscardUnknown: true}.Unmarshal(dat, m.pbInner)
 	if err != nil {
 		return fmt.Errorf("deserialize: unmarshal inner: %w", err)
 	}
@@ -249,7 +337,8 @@ func NewManifestFromReader(input io.Reader) (*manifest, error) {
 	// deserialize outer:
 	m.pbOuter = new(MFFileOuter)
 
-	err = proto.Unmarshal(dat, m.pbOuter)
+	// Unknown fields would cost memory; mfer never writes a loaded manifest out.
+	err = proto.UnmarshalOptions{DiscardUnknown: true}.Unmarshal(dat, m.pbOuter)
 	if err != nil {
 		return nil, err
 	}

mfer/deserialize_fuzz_test.go

@@ -54,9 +54,15 @@ func FuzzNewManifestFromReader(f *testing.F) {
 		}
 
 		// It also keeps a few copies of its input. Buffers grow by
-		// copying, so reaching those sizes allocates a few times them in
-		// total: sixteen times the input and the decompressed data leaves
-		// room for that.
+		// copying, so reaching those sizes allocates up to about six times
+		// them in total. Decoding the decompressed data takes up to
+		// maxDecodedGrowth times its size for file entries, hashes,
+		// timestamps and MIME types, and drops fields it does not know.
+		// The strings and bytes it copies out of it, such as many one-byte
+		// values in one hash, take up to about five times more under the
+		// race detector, which pads every small copy to 16 bytes, and about
+		// half that without it. Twenty times the input and the
+		// decompressed data leaves room for all of that.
 		//
 		// The decoder also sets aside a new buffer of one to two times the
 		// window for each frame that asks for a larger window than the
@@ -71,8 +77,9 @@ func FuzzNewManifestFromReader(f *testing.F) {
 		// fails if the decoder accepts windows of twice zstdWindowSize; the
 		// seed whose two frames together exceed MaxDecompressedSize fails
 		// if the decoder decodes them in full instead of stopping at the
-		// declared size.
-		limit := 16*(uint64(len(data))+decompressed) + 24*zstdWindowSize
+		// declared size; the seeds of empty file entries and of a file
+		// entry of empty hashes fail if the parser decodes them.
+		limit := 20*(uint64(len(data))+decompressed) + 24*zstdWindowSize
 
 		allocated := after.TotalAlloc - before.TotalAlloc
 		if allocated > limit {

mfer/deserialize_path_test.go

@@ -6,7 +6,10 @@ import (
 	"context"
 	"crypto/sha256"
 	"fmt"
+	"strconv"
+	"strings"
 	"testing"
+	"time"
 
 	"github.com/google/uuid"
 	"github.com/klauspost/compress/zstd"
@@ -114,6 +117,113 @@ func TestDeserializeRejectsInvalidEntryPaths(t *testing.T) {
 	}
 }
 
+// Entries of a path, an empty hash, an empty MIME type and empty modification
+// and change times are counted at 416 bytes each (160 + 112 + 16 + 64 + 64)
+// and take 16 bytes plus the path to encode. A 35-character path makes that
+// 51 bytes, about 8.2 times: refused, and leaving any one of the five
+// uncounted, even the MIME type, brings it under 8. A 37-character path makes
+// it 53 bytes, about 7.8 times: loaded.
+func TestDeserializeRefusesEntriesThatDecodeTooLarge(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		pathLen int
+		refused bool
+	}{
+		{35, true},
+		{37, false},
+	}
+
+	for _, tt := range tests {
+		t.Run(strconv.Itoa(tt.pathLen), func(t *testing.T) {
+			t.Parallel()
+
+			entry := protowire.AppendTag(nil, 1, protowire.BytesType) // MFFilePath.path
+			entry = protowire.AppendString(entry, strings.Repeat("a", tt.pathLen))
+			entry = protowire.AppendTag(entry, 3, protowire.BytesType) // MFFilePath.hashes
+			entry = protowire.AppendBytes(entry, nil)
+			entry = protowire.AppendTag(entry, 301, protowire.BytesType) // MFFilePath.mimeType
+			entry = protowire.AppendBytes(entry, nil)
+			entry = protowire.AppendTag(entry, 302, protowire.BytesType) // MFFilePath.mtime
+			entry = protowire.AppendBytes(entry, nil)
+			entry = protowire.AppendTag(entry, 303, protowire.BytesType) // MFFilePath.ctime
+			entry = protowire.AppendBytes(entry, nil)
+
+			id := uuid.New()
+			inner := protowire.AppendTag(nil, 102, protowire.BytesType) // MFFile.uuid
+			inner = protowire.AppendBytes(inner, id[:])
+
+			for range 1000 {
+				inner = protowire.AppendTag(inner, 101, protowire.BytesType) // MFFile.files
+				inner = protowire.AppendBytes(inner, entry)
+			}
+
+			_, err := NewManifestFromReader(bytes.NewReader(wrapInner(t, id, inner)))
+			if tt.refused {
+				require.ErrorIs(t, err, errDecodedTooLarge)
+			} else {
+				require.NoError(t, err)
+			}
+		})
+	}
+}
+
+// Fields the decoder does not know are dropped, in the outer message, the inner
+// message and a file entry, so that they take no memory once loaded.
+func TestDeserializeDropsUnknownFields(t *testing.T) {
+	t.Parallel()
+
+	unknown := protowire.AppendTag(nil, 99, protowire.BytesType) // in no message
+	unknown = protowire.AppendBytes(unknown, []byte("not known"))
+
+	entry := protowire.AppendTag(nil, 1, protowire.BytesType) // MFFilePath.path
+	entry = protowire.AppendString(entry, "a")
+	entry = append(entry, unknown...)
+
+	id := uuid.New()
+	inner := protowire.AppendTag(nil, 101, protowire.BytesType) // MFFile.files
+	inner = protowire.AppendBytes(inner, entry)
+	inner = protowire.AppendTag(inner, 102, protowire.BytesType) // MFFile.uuid
+	inner = protowire.AppendBytes(inner, id[:])
+	inner = append(inner, unknown...)
+
+	data := wrapInner(t, id, inner)
+	data = append(data, unknown...) // the outer message ends the file
+
+	m, err := NewManifestFromReader(bytes.NewReader(data))
+	require.NoError(t, err)
+	require.Len(t, m.Files(), 1)
+	assert.Empty(t, m.pbOuter.ProtoReflect().GetUnknown())
+	assert.Empty(t, m.pbInner.ProtoReflect().GetUnknown())
+	assert.Empty(t, m.Files()[0].ProtoReflect().GetUnknown())
+}
+
+// Many empty files with names of at most three characters and modification
+// times at the epoch make about the densest manifest mfer writes: it takes
+// about 7 times its size to decode, and still loads. A signature would not
+// change the inner message, so none is added.
+func TestDeserializeLoadsDensestManifest(t *testing.T) {
+	t.Parallel()
+
+	hash := make([]byte, 34) // multihash: 2-byte prefix + 32-byte SHA-256
+
+	b := NewBuilder()
+	b.SetIncludeTimestamps(true)
+
+	const files = 10000
+	for i := range files {
+		name := RelFilePath(strconv.FormatInt(int64(i), 36))
+		require.NoError(t, b.AddFileWithHash(name, 0, ModTime(time.Unix(0, 0)), hash))
+	}
+
+	var buf bytes.Buffer
+	require.NoError(t, b.Build(context.Background(), &buf))
+
+	m, err := NewManifestFromReader(&buf)
+	require.NoError(t, err)
+	assert.Len(t, m.Files(), files)
+}
+
 func TestDeserializeValidManifestRoundTrips(t *testing.T) {
 	t.Parallel()
 

mfer/scanner.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"io"
 	"io/fs"
+	"os"
 	"path"
 	"path/filepath"
 	"strings"
@@ -57,6 +58,14 @@ type ScannerOptions struct {
 	SigningOptions *SigningOptions
 	// Seed, if set, derives a deterministic UUID from this seed.
 	Seed string
+	// OutputPath, if set, is the file the manifest will be written to.
+	// If that file exists when the scanner is created, enumeration leaves
+	// it out however its path is spelled (through a symlink, a hard link,
+	// or in different letter case), so a manifest being replaced is never
+	// listed.
+	// Files are matched with os.SameFile, which only recognizes files on
+	// the operating system's filesystem.
+	OutputPath string
 }
 
 // FileEntry represents a file that has been enumerated.
@@ -75,6 +84,7 @@ type Scanner struct {
 	totalBytes FileSize // cached sum of all file sizes
 	options    *ScannerOptions
 	fs         afero.Fs
+	outputInfo fs.FileInfo // the existing output file; nil when there is none
 }
 
 // NewScanner creates a new Scanner with default options.
@@ -93,11 +103,22 @@ func NewScannerWithOptions(opts *ScannerOptions) *Scanner {
 		fs = afero.NewOsFs()
 	}
 
-	return &Scanner{
+	s := &Scanner{
 		files:   make([]*FileEntry, 0),
 		options: opts,
 		fs:      fs,
 	}
+
+	if opts.OutputPath != "" {
+		// If the output file cannot be stat'd, normally because it does
+		// not exist yet, there is nothing to leave out.
+		info, err := s.fs.Stat(opts.OutputPath)
+		if err == nil {
+			s.outputInfo = info
+		}
+	}
+
+	return s
 }
 
 // EnumerateFile adds a single file to the scanner, calling stat() to get metadata.
@@ -435,6 +456,11 @@ func (s *Scanner) enumerateFileWithInfo(
 		info = realInfo
 	}
 
+	// The manifest being written is not one of the files it lists.
+	if s.outputInfo != nil && os.SameFile(info, s.outputInfo) {
+		return nil
+	}
+
 	entry := &FileEntry{
 		Path:    RelFilePath(cleanPath),
 		AbsPath: AbsFilePath(absPath),

mfer/testdata/fuzz/FuzzNewManifestFromReader/empty-file-entries

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("ZNAVSRFG\xa8\x06\x01\xb0\x06\x01\xb8\x06\xff\xff\xff\x03\xc2\x06 {\x16\xbdu\xa0\xa2\x11\xfcH\xef*\x1b7\r\x99\xefb\x04\x02g\n\xa9\xf3B5\xe5p\x96\x8c\x8c\xac\x0e\xca\x06\x10\x03Q\xb2\xd0\x19`F\xc1\xb1\xc0Z\xf4x\xf4g^\xba\f\xa1\x06(\xb5/\xfd\x04h\x04\x01\x00d\x01\xb2\x06\x10\x03Q\xb2\xd0\x19`F\xc1\xb1\xc0Z\xf4x\xf4g^\xaa\x06\x00\x01T\x13\x024\xce\xff\rL\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15M\x00\x00\x00\x01T\x00\x044\xfc\xff\x153\xea\a\xb4")

mfer/testdata/fuzz/FuzzNewManifestFromReader/file-entry-of-empty-hashes

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("ZNAVSRFG\xa8\x06\x01\xb0\x06\x01\xb8\x06\xff\xff\xff\x03\xc2\x06 .\xcd\x11|0\xfcP\xe5\x1b\xe3\xc6Ӡ\xcdڤx\xcd\x169t\x1a9~ǽB\xc9\xe8G`\x05\xca\x06\x10\x11*!\x0e\x95EF\xb8\xbd\x9f\xde\x12MF\r\x99\xba\f\xa6\x06(\xb5/\xfd\x04h,\x01\x00\xb4\x01\xb2\x06\x10\x11*!\x0e\x95EF\xb8\xbd\x9f\xde\x12MF\r\x99\xaa\x06\xe6\xff\xff\x03\x1a\x00\x01T\x14\x024\x8b\xff\x17L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15M\x00\x00\x00\x01T\x00\x044\xfc\xff\x15\x02\xd1.\xe3")