internal/pkg: encode entry in custom format

The fact that Gob serialisation is deterministic is an implementation detail. This change replaces Gob with a simple custom format.

Signed-off-by: Ophestra <cat@gensokyo.uk>

3 files changed, 194 insertions, 28 deletions

diff --git a/internal/pkg/dir.go b/internal/pkg/dir.go
index 794fdf5e..780b54e5 100644
--- a/internal/pkg/dir.go
+++ b/internal/pkg/dir.go
@@ -2,27 +2,149 @@ package pkg
 
 import (
 	"crypto/sha512"
-	"encoding/gob"
+	"encoding/binary"
+	"errors"
 	"io"
 	"io/fs"
+	"math"
 	"os"
+	"syscall"
 
 	"hakurei.app/container/check"
 )
 
-// FlatEntry is the representation of a directory entry via [Flatten].
+// FlatEntry is a directory entry to be encoded for [Flatten].
 type FlatEntry struct {
-	Name string      // base name of the file
 	Mode fs.FileMode // file mode bits
+	Path string      // pathname of the file
 	Data []byte      // file content or symlink destination
 }
 
+/*
+| mode uint32 | path_sz uint32 |
+|        data_sz uint64        |
+|         path string          |
+|         data []byte          |
+*/
+
+// wordSize is the boundary which binary segments are always aligned to.
+const wordSize = 8
+
+// alignSize returns the padded size for aligning sz.
+func alignSize(sz int) int {
+	return sz + (wordSize-(sz)%wordSize)%wordSize
+}
+
+// Encode encodes the entry for transmission or hashing.
+func (ent *FlatEntry) Encode(w io.Writer) (n int, err error) {
+	pPathSize := alignSize(len(ent.Path))
+	if pPathSize > math.MaxUint32 {
+		return 0, syscall.E2BIG
+	}
+	pDataSize := alignSize(len(ent.Data))
+
+	payload := make([]byte, wordSize*2+pPathSize+pDataSize)
+	binary.LittleEndian.PutUint32(payload, uint32(ent.Mode))
+	binary.LittleEndian.PutUint32(payload[wordSize/2:], uint32(len(ent.Path)))
+	binary.LittleEndian.PutUint64(payload[wordSize:], uint64(len(ent.Data)))
+	copy(payload[wordSize*2:], ent.Path)
+	copy(payload[wordSize*2+pPathSize:], ent.Data)
+	return w.Write(payload)
+}
+
+// Decode decodes the entry from its representation produced by Encode.
+func (ent *FlatEntry) Decode(r io.Reader) (n int, err error) {
+	var nr int
+
+	header := make([]byte, wordSize*2)
+	nr, err = r.Read(header)
+	n += nr
+	if err != nil {
+		if errors.Is(err, io.EOF) && n != 0 {
+			err = io.ErrUnexpectedEOF
+		}
+		return
+	}
+
+	ent.Mode = fs.FileMode(binary.LittleEndian.Uint32(header))
+	pathSize := int(binary.LittleEndian.Uint32(header[wordSize/2:]))
+	pPathSize := alignSize(pathSize)
+	dataSize := int(binary.LittleEndian.Uint64(header[wordSize:]))
+	pDataSize := alignSize(dataSize)
+
+	buf := make([]byte, pPathSize+pDataSize)
+	nr, err = r.Read(buf)
+	n += nr
+	if err != nil {
+		if errors.Is(err, io.EOF) {
+			if nr != len(buf) {
+				err = io.ErrUnexpectedEOF
+				return
+			}
+		} else {
+			return
+		}
+	}
+
+	ent.Path = string(buf[:pathSize])
+	if ent.Mode.IsDir() {
+		ent.Data = nil
+	} else {
+		ent.Data = buf[pPathSize : pPathSize+dataSize]
+	}
+	return
+}
+
+// DirScanner provides an efficient interface for reading a stream of encoded
+// [FlatEntry]. Successive calls to the Scan method will step through the
+// entries in the stream.
+type DirScanner struct {
+	// Underlying reader to scan [FlatEntry] representations from.
+	r io.Reader
+
+	// First non-EOF I/O error, returned by the Err method.
+	err error
+
+	// Entry to store results in. Its address is returned by the Entry method
+	// and is updated on every call to Scan.
+	ent FlatEntry
+}
+
+// NewDirScanner returns the address of a new instance of [DirScanner] reading
+// from r. The caller must no longer read from r after this function returns.
+func NewDirScanner(r io.Reader) *DirScanner { return &DirScanner{r: r} }
+
+// Err returns the first non-EOF I/O error.
+func (s *DirScanner) Err() error {
+	if errors.Is(s.err, io.EOF) {
+		return nil
+	}
+	return s.err
+}
+
+// Entry returns the address to the [FlatEntry] value storing the last result.
+func (s *DirScanner) Entry() *FlatEntry { return &s.ent }
+
+// Scan advances to the next [FlatEntry].
+func (s *DirScanner) Scan() bool {
+	if s.err != nil {
+		return false
+	}
+
+	var n int
+	n, s.err = s.ent.Decode(s.r)
+	if errors.Is(s.err, io.EOF) {
+		return n != 0
+	}
+	return s.err == nil
+}
+
 // Flatten writes a deterministic representation of the contents of fsys to w.
 // The resulting data can be hashed to produce a deterministic checksum for the
 // directory.
-func Flatten(fsys fs.FS, root string, w io.Writer) error {
-	e := gob.NewEncoder(w)
-	return fs.WalkDir(fsys, root, func(path string, d fs.DirEntry, err error) error {
+func Flatten(fsys fs.FS, root string, w io.Writer) (n int, err error) {
+	var nr int
+	err = fs.WalkDir(fsys, root, func(path string, d fs.DirEntry, err error) error {
 		if err != nil {
 			return err
 		}
@@ -34,7 +156,7 @@ func Flatten(fsys fs.FS, root string, w io.Writer) error {
 		}
 
 		ent := FlatEntry{
-			Name: fi.Name(),
+			Path: path,
 			Mode: fi.Mode(),
 		}
 		if ent.Mode.IsRegular() {
@@ -49,14 +171,17 @@ func Flatten(fsys fs.FS, root string, w io.Writer) error {
 			ent.Data = []byte(newpath)
 		}
 
-		return e.Encode(&ent)
+		nr, err = ent.Encode(w)
+		n += nr
+		return err
 	})
+	return
 }
 
 // HashFS returns a checksum produced by hashing the result of [Flatten].
 func HashFS(fsys fs.FS, root string) (Checksum, error) {
 	h := sha512.New384()
-	if err := Flatten(fsys, root, h); err != nil {
+	if _, err := Flatten(fsys, root, h); err != nil {
 		return Checksum{}, err
 	}
 	return (Checksum)(h.Sum(nil)), nil
diff --git a/internal/pkg/dir_test.go b/internal/pkg/dir_test.go
index 2e3b9079..7bc877d1 100644
--- a/internal/pkg/dir_test.go
+++ b/internal/pkg/dir_test.go
@@ -1,7 +1,9 @@
 package pkg_test
 
 import (
+	"bytes"
 	"io/fs"
+	"reflect"
 	"testing"
 	"testing/fstest"
 
@@ -12,36 +14,75 @@ func TestFlatten(t *testing.T) {
 	t.Parallel()
 
 	testCases := []struct {
-		name string
-		fsys fs.FS
-		want pkg.Checksum
+		name    string
+		fsys    fs.FS
+		entries []pkg.FlatEntry
+		sum     pkg.Checksum
 	}{
 		{"sample cache file", fstest.MapFS{
 			".": {Mode: 020000000700},
 
 			"checksum": {Mode: 020000000700},
-			"checksum/vsAhtPNo4waRNOASwrQwcIPTqb3SBuJOXw2G4T1mNmVZM-wrQTRllmgXqcIIoRcX": {Mode: 0400, Data: []byte{0x0}},
-			"checksum/0bSFPu5Tnd-2Jj0Mv6co23PW2t3BmHc7eLFj9TgY3eIBg8zislo7xZYNBqovVLcq": {Mode: 0400, Data: []byte{0x0, 0x0, 0x0, 0x0, 0xad, 0xb, 0x0, 0x4, 0xfe, 0xfe, 0x0, 0x0, 0xfe, 0xca, 0x0, 0x0}},
+			"checksum/vsAhtPNo4waRNOASwrQwcIPTqb3SBuJOXw2G4T1mNmVZM-wrQTRllmgXqcIIoRcX": {Mode: 0400, Data: []byte{0}},
+			"checksum/0bSFPu5Tnd-2Jj0Mv6co23PW2t3BmHc7eLFj9TgY3eIBg8zislo7xZYNBqovVLcq": {Mode: 0400, Data: []byte{0, 0, 0, 0, 0xad, 0xb, 0, 4, 0xfe, 0xfe, 0, 0, 0xfe, 0xca, 0, 0}},
 
 			"identifier": {Mode: 020000000700},
-			"identifier/vsAhtPNo4waRNOASwrQwcIPTqb3SBuJOXw2G4T1mNmVZM-wrQTRllmgXqcIIoRcX": {Mode: 0400, Data: []byte{0x0}},
-			"identifier/0bSFPu5Tnd-2Jj0Mv6co23PW2t3BmHc7eLFj9TgY3eIBg8zislo7xZYNBqovVLcq": {Mode: 0400, Data: []byte{0x0, 0x0, 0x0, 0x0, 0xad, 0xb, 0x0, 0x4, 0xfe, 0xfe, 0x0, 0x0, 0xfe, 0xca, 0x0, 0x0}},
-			"identifier/cafebabecafebabecafebabecafebabecafebabecafebabecafebabecafebabe": {Mode: 0400, Data: []byte{0x0, 0x0, 0x0, 0x0, 0xad, 0xb, 0x0, 0x4, 0xfe, 0xfe, 0x0, 0x0, 0xfe, 0xca, 0x0, 0x0}},
-			"identifier/deadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeef": {Mode: 0400, Data: []byte{0x0, 0x0, 0x0, 0x0, 0xad, 0xb, 0x0, 0x4, 0xfe, 0xfe, 0x0, 0x0, 0xfe, 0xca, 0x0, 0x0}},
-		}, pkg.MustDecode("lvK4lY9bQUFscHpxqHmiPvptjUwOgn3BFhzCXZMeupkY1n22WUPSuh7pswEvVZrx")},
+			"identifier/vsAhtPNo4waRNOASwrQwcIPTqb3SBuJOXw2G4T1mNmVZM-wrQTRllmgXqcIIoRcX": {Mode: 0400, Data: []byte{0}},
+			"identifier/0bSFPu5Tnd-2Jj0Mv6co23PW2t3BmHc7eLFj9TgY3eIBg8zislo7xZYNBqovVLcq": {Mode: 0400, Data: []byte{0, 0, 0, 0, 0xad, 0xb, 0, 4, 0xfe, 0xfe, 0, 0, 0xfe, 0xca, 0, 0}},
+			"identifier/cafebabecafebabecafebabecafebabecafebabecafebabecafebabecafebabe": {Mode: 0400, Data: []byte{0, 0, 0, 0, 0xad, 0xb, 0, 4, 0xfe, 0xfe, 0, 0, 0xfe, 0xca, 0, 0}},
+			"identifier/deadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeef": {Mode: 0400, Data: []byte{0, 0, 0, 0, 0xad, 0xb, 0, 4, 0xfe, 0xfe, 0, 0, 0xfe, 0xca, 0, 0}},
+		}, []pkg.FlatEntry{
+			{Mode: 020000000700, Path: "."},
+
+			{Mode: 020000000700, Path: "checksum"},
+			{Mode: 0400, Path: "checksum/0bSFPu5Tnd-2Jj0Mv6co23PW2t3BmHc7eLFj9TgY3eIBg8zislo7xZYNBqovVLcq", Data: []byte{0, 0, 0, 0, 0xad, 0xb, 0, 4, 0xfe, 0xfe, 0, 0, 0xfe, 0xca, 0, 0}},
+			{Mode: 0400, Path: "checksum/vsAhtPNo4waRNOASwrQwcIPTqb3SBuJOXw2G4T1mNmVZM-wrQTRllmgXqcIIoRcX", Data: []byte{0}},
+
+			{Mode: 020000000700, Path: "identifier"},
+			{Mode: 0400, Path: "identifier/0bSFPu5Tnd-2Jj0Mv6co23PW2t3BmHc7eLFj9TgY3eIBg8zislo7xZYNBqovVLcq", Data: []byte{0, 0, 0, 0, 0xad, 0xb, 0, 4, 0xfe, 0xfe, 0, 0, 0xfe, 0xca, 0, 0}},
+			{Mode: 0400, Path: "identifier/cafebabecafebabecafebabecafebabecafebabecafebabecafebabecafebabe", Data: []byte{0, 0, 0, 0, 0xad, 0xb, 0, 4, 0xfe, 0xfe, 0, 0, 0xfe, 0xca, 0, 0}},
+			{Mode: 0400, Path: "identifier/deadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeef", Data: []byte{0, 0, 0, 0, 0xad, 0xb, 0, 4, 0xfe, 0xfe, 0, 0, 0xfe, 0xca, 0, 0}},
+			{Mode: 0400, Path: "identifier/vsAhtPNo4waRNOASwrQwcIPTqb3SBuJOXw2G4T1mNmVZM-wrQTRllmgXqcIIoRcX", Data: []byte{0}},
+		}, pkg.MustDecode("ZNSQH-mjhtIbFvi51lQ0UjatjoS8_5ILrBPNWlO2LWTq9P6MJEnekYzP0esUJnVr")},
 	}
 	for _, tc := range testCases {
 		t.Run(tc.name, func(t *testing.T) {
 			t.Parallel()
 
-			if got, err := pkg.HashFS(tc.fsys, "."); err != nil {
-				t.Fatalf("HashFS: error = %v", err)
-			} else if got != tc.want {
-				t.Fatalf("HashFS: %v", &pkg.ChecksumMismatchError{
-					Got:  got,
-					Want: tc.want,
-				})
-			}
+			t.Run("roundtrip", func(t *testing.T) {
+				t.Parallel()
+
+				var buf bytes.Buffer
+				if _, err := pkg.Flatten(tc.fsys, ".", &buf); err != nil {
+					t.Fatalf("Flatten: error = %v", err)
+				}
+
+				s := pkg.NewDirScanner(bytes.NewReader(buf.Bytes()))
+				var got []pkg.FlatEntry
+				for s.Scan() {
+					got = append(got, *s.Entry())
+				}
+				if err := s.Err(); err != nil {
+					t.Fatalf("Err: error = %v", err)
+				}
+
+				if !reflect.DeepEqual(got, tc.entries) {
+					t.Fatalf("Scan: %#v, want %#v", got, tc.entries)
+				}
+			})
+
+			t.Run("hash", func(t *testing.T) {
+				t.Parallel()
+
+				if got, err := pkg.HashFS(tc.fsys, "."); err != nil {
+					t.Fatalf("HashFS: error = %v", err)
+				} else if got != tc.sum {
+					t.Fatalf("HashFS: %v", &pkg.ChecksumMismatchError{
+						Got:  got,
+						Want: tc.sum,
+					})
+				}
+			})
 		})
 	}
 }
diff --git a/internal/pkg/pkg_test.go b/internal/pkg/pkg_test.go
index c73049e5..bd2096d9 100644
--- a/internal/pkg/pkg_test.go
+++ b/internal/pkg/pkg_test.go
@@ -211,7 +211,7 @@ func TestCache(t *testing.T) {
 			}
 		}, func(t *testing.T, base *check.Absolute) {
 			wantChecksum := pkg.MustDecode(
-				"lvK4lY9bQUFscHpxqHmiPvptjUwOgn3BFhzCXZMeupkY1n22WUPSuh7pswEvVZrx",
+				"ZNSQH-mjhtIbFvi51lQ0UjatjoS8_5ILrBPNWlO2LWTq9P6MJEnekYzP0esUJnVr",
 			)
 			if checksum, err := pkg.HashDir(base); err != nil {
 				t.Fatalf("HashDir: error = %v", err)