Initial attempt to make a compressor with less overhead

io.Reader actually has massive overhead...

2 files changed, 153 insertions, 0 deletions

diff --git a/internal/zlib/decompress.go b/internal/zlib/decompress.go
new file mode 100644
index 00000000..55f6d3e2
--- /dev/null
+++ b/internal/zlib/decompress.go
@@ -0,0 +1,69 @@
+package zlib
+
+import (
+	"encoding/binary"
+	"io"
+
+	"git.sr.ht/~runxiyu/furgit/internal/adler32"
+	"git.sr.ht/~runxiyu/furgit/internal/bufpool"
+	"git.sr.ht/~runxiyu/furgit/internal/flate"
+)
+
+// Decompress inflates the provided zlib wrapped stream and returns the
+// uncompressed data inside a pooled bufpool.Buffer.
+func Decompress(src []byte) (bufpool.Buffer, error) {
+	return DecompressDict(src, nil)
+}
+
+// DecompressDict is like Decompress but accepts a preset dictionary. The
+// dictionary must match the checksum embedded in the stream if the dictionary
+// flag is present.
+func DecompressDict(src []byte, dict []byte) (bufpool.Buffer, error) {
+	if len(src) < 6 {
+		return bufpool.Buffer{}, io.ErrUnexpectedEOF
+	}
+
+	cmf := src[0]
+	flg := src[1]
+	if (cmf&0x0f != zlibDeflate) || (cmf>>4 > zlibMaxWindow) || (binary.BigEndian.Uint16(src[:2])%31 != 0) {
+		return bufpool.Buffer{}, ErrHeader
+	}
+
+	offset := 2
+	haveDict := flg&0x20 != 0
+	if haveDict {
+		if len(src) < offset+4 {
+			return bufpool.Buffer{}, io.ErrUnexpectedEOF
+		}
+		if dict == nil {
+			return bufpool.Buffer{}, ErrDictionary
+		}
+		checksum := binary.BigEndian.Uint32(src[offset : offset+4])
+		if checksum != adler32.Checksum(dict) {
+			return bufpool.Buffer{}, ErrDictionary
+		}
+		offset += 4
+	}
+
+	if len(src[offset:]) < 4 {
+		return bufpool.Buffer{}, io.ErrUnexpectedEOF
+	}
+
+	deflateData := src[offset:]
+	out, consumed, err := flate.DecompressDict(deflateData, dict)
+	if err != nil {
+		return bufpool.Buffer{}, err
+	}
+
+	checksumPos := offset + consumed
+	if checksumPos+4 > len(src) {
+		out.Release()
+		return bufpool.Buffer{}, io.ErrUnexpectedEOF
+	}
+	expected := binary.BigEndian.Uint32(src[checksumPos : checksumPos+4])
+	if expected != adler32.Checksum(out.Bytes()) {
+		out.Release()
+		return bufpool.Buffer{}, ErrChecksum
+	}
+	return out, nil
+}
diff --git a/internal/zlib/decompress_test.go b/internal/zlib/decompress_test.go
new file mode 100644
index 00000000..bb517ae6
--- /dev/null
+++ b/internal/zlib/decompress_test.go
@@ -0,0 +1,84 @@
+package zlib
+
+import (
+	"bytes"
+	stdzlib "compress/zlib"
+	"testing"
+)
+
+func compressZlib(t *testing.T, payload, dict []byte) []byte {
+	t.Helper()
+	var buf bytes.Buffer
+	var (
+		w   *stdzlib.Writer
+		err error
+	)
+	if dict != nil {
+		w, err = stdzlib.NewWriterLevelDict(&buf, stdzlib.DefaultCompression, dict)
+	} else {
+		w = stdzlib.NewWriter(&buf)
+	}
+	if err != nil {
+		t.Fatalf("NewWriter: %v", err)
+	}
+	if _, err := w.Write(payload); err != nil {
+		t.Fatalf("Write: %v", err)
+	}
+	if err := w.Close(); err != nil {
+		t.Fatalf("Close: %v", err)
+	}
+	return buf.Bytes()
+}
+
+func TestDecompress(t *testing.T) {
+	payload := []byte("hello, zlib world!")
+	compressed := compressZlib(t, payload, nil)
+
+	out, err := Decompress(compressed)
+	if err != nil {
+		t.Fatalf("Decompress: %v", err)
+	}
+	defer out.Release()
+
+	if !bytes.Equal(out.Bytes(), payload) {
+		t.Fatalf("unexpected payload %q", out.Bytes())
+	}
+}
+
+func TestDecompressDict(t *testing.T) {
+	dict := []byte("git dictionary for zlib")
+	payload := append([]byte(nil), dict...)
+	payload = append(payload, []byte(" -- extended body -- extended body")...)
+	compressed := compressZlib(t, payload, dict)
+
+	out, err := DecompressDict(compressed, dict)
+	if err != nil {
+		t.Fatalf("DecompressDict: %v", err)
+	}
+	defer out.Release()
+
+	if !bytes.Equal(out.Bytes(), payload) {
+		t.Fatalf("unexpected payload %q", out.Bytes())
+	}
+}
+
+func TestDecompressDictMissing(t *testing.T) {
+	dict := []byte("preset dictionary")
+	payload := append([]byte(nil), dict...)
+	payload = append(payload, []byte(" .. more data ..")...)
+	compressed := compressZlib(t, payload, dict)
+
+	if _, err := Decompress(compressed); err != ErrDictionary {
+		t.Fatalf("expected ErrDictionary, got %v", err)
+	}
+}
+
+func TestDecompressChecksumError(t *testing.T) {
+	payload := []byte("checksum check")
+	compressed := compressZlib(t, payload, nil)
+	compressed[len(compressed)-1] ^= 0xff
+
+	if _, err := Decompress(compressed); err != ErrChecksum {
+		t.Fatalf("expected ErrChecksum, got %v", err)
+	}
+}