Tweak encoding

klauspost · Dec 11, 2023 · 617aeec · 617aeec
1 parent fb2efcb
commit 617aeec
Show file tree

Hide file tree

Showing 3 changed files with 42 additions and 9 deletions.
diff --git a/s2/encode_best.go b/s2/encode_best.go
@@ -721,9 +721,14 @@ func emitCopySize(offset, length int) int {
 	}
 
 	// Offset no more than 2 bytes.
-	if offset < 2048 && length < 12 {
-		// Emit up to 11 bytes with short offset.
-		return 2
+	if offset < 1024 {
+		if length < 11+8 {
+			// Emit up to 18 bytes with short offset.
+			return 2
+		}
+		if length < 18+256 {
+			return 3
+		}
 	}
 	// 2 byte offset + Variable length (base length 4).
 	return emitCopy2Size(length)

diff --git a/s2/encode_go.go b/s2/encode_go.go
@@ -197,6 +197,7 @@ func encodeLength60(dst []byte, tag uint8, length int) int {
 //	1 <= offset && offset <= math.MaxUint32
 //	4 <= length && length <= 1 << 24
 func emitCopy(dst []byte, offset, length int) int {
+	offset--
 	if offset >= 65536 {
 		// Encode tag+length as up to 4 bytes.
 		n := encodeLength(dst, tagCopy4, length-3)
@@ -208,12 +209,20 @@ func emitCopy(dst []byte, offset, length int) int {
 	}
 
 	// Offset no more than 2 bytes and length less than 12.
-	if offset < 2048 && length < 12 {
-		// emit 12 bytes as tagCopy1, rest as repeats.
-		dst[1] = uint8(offset)
-		dst[0] = uint8(offset>>8)<<5 | uint8(7)<<2 | tagCopy1
-		length -= 12
-		return 2
+	if offset < 1024 {
+		if length < 12+8 {
+			// FIXME: Incorrect encoding
+			dst[1] = uint8(offset)
+			dst[0] = uint8(offset>>8)<<5 | uint8(length-4)<<2 | tagCopy1
+			return 2
+		}
+		if length < 18+256 {
+			// FIXME: Incorrect encoding
+			dst[2] = uint8(length - 18)
+			dst[1] = uint8(offset)
+			dst[0] = uint8(offset>>8)<<5 | uint8(11+8)<<2 | tagCopy1
+			return 3
+		}
 	}
 
 	// 2 byte offset, with variable length.

diff --git a/s2/encode_test.go b/s2/encode_test.go
@@ -6,6 +6,7 @@ package s2
 
 import (
 	"bytes"
+	"encoding/binary"
 	"fmt"
 	"math"
 	"testing"
@@ -67,3 +68,21 @@ func TestEncodeHuge(t *testing.T) {
 	}
 	test(t, make([]byte, MaxBlockSize))
 }
+
+func TestSizes(t *testing.T) {
+	var src [2]byte
+	src[0] = 123
+	src[1] = 57
+	s := 2
+
+	want := int(uint32(src[s-2])&0xe0<<3 | uint32(src[s-1]))
+	//got := bits.RotateLeft16(binary.LittleEndian.Uint16(src[:]), 16-5) & 2047
+	got := binary.LittleEndian.Uint16(src[:])
+	t.Logf("w:%012b G:%016b", want, got)
+	for i := 4; i < 100; i++ {
+		if i == 99 {
+			i = (1 << 24) - 1
+		}
+		t.Logf("%d: short:%d medium: %d long: %d repeat: %d", i, emitCopySize(10, i), emitCopySize(4000, i), emitCopySize(70000, i), emitRepeatSize(0, i))
+	}
+}