适配字节序问题

base.go

@@ -7,15 +7,15 @@ import (
 // Base has an encoding buffer thus should not be copied.
 //    total size: 8 bytes
 type Base struct {
-	off uint16  // starting offset
+	off uint16  // starting offset (0 is not permitted)
 	til uint16  // remianing indicator starting offset
 	bit uint8   // 2^bit, max is 15 (32768)
-	pos uint8   // bitwise buffer position
-	buf [2]byte // en/decoding buffer
+	_   [3]byte // always 0, indicates the byte order
 }
 
 var (
 	ErrInvalidBitSize   = errors.New("bit size >= 16 or == 0")
+	ErrZeroOffsetStart  = errors.New("zero offset start")
 	ErrOffsetOverflow   = errors.New("offset overflow")
 	ErrTailOverflow     = errors.New("tail overflow")
 	ErrTailInCodingArea = errors.New("tail in coding area")
@@ -23,6 +23,9 @@ var (
 
 // NewBase generates a new base2n config
 func NewBase(off, til uint16, bit uint8) (*Base, error) {
+	if off == 0 {
+		return nil, ErrZeroOffsetStart
+	}
 	if bit >= 16 || bit == 0 {
 		return nil, ErrInvalidBitSize
 	}
@@ -31,7 +34,7 @@ func NewBase(off, til uint16, bit uint8) (*Base, error) {
 		return nil, ErrOffsetOverflow
 	}
 	tile := uint32(til) // [til, tile)
-	if bit%2 == 0 {
+	if bit > 8 && bit%2 == 0 {
 		tile += uint32(bit / 2)
 	} else {
 		tile += uint32(bit)

code.go

@@ -0,0 +1,9 @@
+package unibase2n
+
+func (bs *Base) Encode(data []byte) []byte {
+	return nil
+}
+
+func (bs *Base) Decode(data []byte) []byte {
+	return nil
+}

conv.go

@@ -0,0 +1,57 @@
+package unibase2n
+
+import (
+	"golang.org/x/text/encoding/unicode"
+)
+
+var format = unicode.UTF16(unicode.BigEndian, unicode.IgnoreBOM)
+
+// UTF16BE2UTF8 to display the result as string
+func UTF16BE2UTF8(b []byte) ([]byte, error) {
+	return format.NewDecoder().Bytes(b)
+}
+
+// UTF82UTF16BE to decode from string
+func UTF82UTF16BE(b []byte) ([]byte, error) {
+	return format.NewEncoder().Bytes(b)
+}
+
+func (bs *Base) EncodeToString(b []byte) string {
+	out, err := UTF16BE2UTF8(bs.Encode(b))
+	if err != nil {
+		return ""
+	}
+	return BytesToString(out)
+}
+
+func (bs *Base) EncodeFromString(s string) []byte {
+	return bs.Encode(StringToBytes(s))
+}
+
+func (bs *Base) EncodeString(s string) string {
+	out, err := UTF16BE2UTF8(bs.Encode(StringToBytes(s)))
+	if err != nil {
+		return ""
+	}
+	return BytesToString(out)
+}
+
+func (bs *Base) DecodeToString(d []byte) string {
+	return BytesToString(bs.Decode(d))
+}
+
+func (bs *Base) DecodeFromString(s string) []byte {
+	d, err := UTF82UTF16BE(StringToBytes(s))
+	if err != nil {
+		return nil
+	}
+	return bs.Decode(d)
+}
+
+func (bs *Base) DecodeString(s string) string {
+	d, err := UTF82UTF16BE(StringToBytes(s))
+	if err != nil {
+		return ""
+	}
+	return BytesToString(bs.Decode(d))
+}

go.mod

@@ -1,3 +1,5 @@
 module github.com/fumiama/unibase2n
 
 go 1.18
+
+require golang.org/x/text v0.3.7

go.sum

@@ -0,0 +1,2 @@
+golang.org/x/text v0.3.7 h1:olpwvP2KacW1ZWvsR7uQhoyTYvKAupfQrRGBFM352Gk=
+golang.org/x/text v0.3.7/go.mod h1:u+2+/6zg+i71rQMx5EYifcz6MCKuco9NR6JIITiCfzQ=

pack.go

@@ -1,13 +1,35 @@
 package unibase2n
 
-import "unsafe"
+import (
+	"encoding/binary"
+	"math/bits"
+	"unsafe"
+)
 
 type Pack uint64
 
 var (
+	// Base16384 CJK Unified Ideographs
+	//    see https://github.com/fumiama/base16384
 	Base16384 = newbasepack(0x4e00, 0x3d00, 14)
-	Base8192  = newbasepack(0xac00, 0xcc00, 13)
-	Base256   = newbasepack(0x1100, 0x0000, 4)
+	// Base8192 谚文音節 Hangul Syllables
+	Base8192 = newbasepack(0xac00, 0xcc00, 13)
+	// Base256 谚文字母 Hangul Jamo
+	Base256 = newbasepack(0x1100, 0, 8)
+	// BaseMath (256) 數學運算符 Mathematical Operators
+	BaseMath = newbasepack(0x2200, 0, 8)
+	// Base128 帶圈或括號的字母數字 Enclosed Alphanumerics
+	Base128 = newbasepack(0x2460, 0x24f4, 7)
+	// Base64 箭頭 Arrows
+	Base64 = newbasepack(0x2190, 0x21d0, 6)
+	// Base64Gua 六十四卦 YiJing Hexagram Symbols
+	Base64Gua = newbasepack(0x3400, 0x262f, 6)
+	// Base32 方塊元素 Block Elements
+	Base32 = newbasepack(0x2580, 0x259f, 5)
+	// Base16 漢文訓讀點 Kanbun Kundoku Den
+	Base16 = newbasepack(0x3190, 0, 4)
+	// Base8 八卦 YiJing Hexagram Symbols
+	Base8 = newbasepack(0x2630, 0x2689, 3)
 )
 
 func newbasepack(off, til uint16, bit uint8) Pack {
@@ -18,12 +40,36 @@ func newbasepack(off, til uint16, bit uint8) Pack {
 	return b.Pack()
 }
 
+// New base2n from a packed config
 func New(pack Pack) *Base {
 	b := &Base{}
-	*(*Pack)(unsafe.Pointer(b)) = pack
+	ismele := isLittleEndian()
+	isitle := pack&0xffffff != 0
+	if ismele == isitle { // same endian
+		*(*Pack)(unsafe.Pointer(b)) = pack
+		return b
+	}
+	// change to native endian
+	n := bits.Reverse64(uint64(pack))
+	field := (*[8]byte)(unsafe.Pointer(&n))
+	if isitle { // packed in little endian but I am big
+		b.off = binary.BigEndian.Uint16(field[6:8])
+		b.til = binary.BigEndian.Uint16(field[4:6])
+	} else { // packed in big endian but I am little
+		b.off = binary.LittleEndian.Uint16(field[6:8])
+		b.til = binary.LittleEndian.Uint16(field[4:6])
+	}
+	b.bit = field[3]
 	return b
 }
 
-func (b *Base) Pack() Pack {
-	return *(*Pack)(unsafe.Pointer(b))
+// Pack this config into an integer
+func (bs *Base) Pack() Pack {
+	return *(*Pack)(unsafe.Pointer(bs))
+}
+
+func isLittleEndian() bool {
+	n := 0x1234
+	f := *((*byte)(unsafe.Pointer(&n)))
+	return (f ^ 0x34) == 0
 }