encoding/charmap/charmap.go

// Copyright 2013 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

// Package charmap provides simple character encodings such as IBM Code Page 437
// and Windows 1252.
package charmap // import "golang.org/x/text/encoding/charmap"

import (
	"unicode/utf8"

	"golang.org/x/text/transform"
)

// utf8Enc holds a rune's UTF-8 encoding in data[:len].
type utf8Enc struct {
	len  uint8
	data [3]byte
}

// charmap describes an 8-bit character set encoding.
type charmap struct {
	// name is the encoding's name.
	name string
	// asciiSuperset states whether the encoding is a superset of ASCII.
	asciiSuperset bool
	// low is the lower bound of the encoded byte for a non-ASCII rune. If
	// charmap.asciiSuperset is true then this will be 0x80, otherwise 0x00.
	low uint8
	// replacement is the encoded replacement character.
	replacement byte
	// decode is the map from encoded byte to UTF-8.
	decode [256]utf8Enc
	// encoding is the map from runes to encoded bytes. Each entry is a
	// uint32: the high 8 bits are the encoded byte and the low 24 bits are
	// the rune. The table entries are sorted by ascending rune.
	encode [256]uint32
}

func (m *charmap) NewDecoder() transform.Transformer {
	return charmapDecoder{charmap: m}
}

func (m *charmap) NewEncoder() transform.Transformer {
	return charmapEncoder{charmap: m}
}

func (m *charmap) String() string {
	return m.name
}

// charmapDecoder implements transform.Transformer by decoding to UTF-8.
type charmapDecoder struct {
	transform.NopResetter
	charmap *charmap
}

func (m charmapDecoder) Transform(dst, src []byte, atEOF bool) (nDst, nSrc int, err error) {
	for i, c := range src {
		if m.charmap.asciiSuperset && c < utf8.RuneSelf {
			if nDst >= len(dst) {
				err = transform.ErrShortDst
				break
			}
			dst[nDst] = c
			nDst++
			nSrc = i + 1
			continue
		}

		decode := &m.charmap.decode[c]
		n := int(decode.len)
		if nDst+n > len(dst) {
			err = transform.ErrShortDst
			break
		}
		// It's 15% faster to avoid calling copy for these tiny slices.
		for j := 0; j < n; j++ {
			dst[nDst] = decode.data[j]
			nDst++
		}
		nSrc = i + 1
	}
	return nDst, nSrc, err
}

// charmapEncoder implements transform.Transformer by encoding from UTF-8.
type charmapEncoder struct {
	transform.NopResetter
	charmap *charmap
}

func (m charmapEncoder) Transform(dst, src []byte, atEOF bool) (nDst, nSrc int, err error) {
	r, size := rune(0), 0
	for nSrc < len(src) {
		if nDst >= len(dst) {
			err = transform.ErrShortDst
			break
		}
		r = rune(src[nSrc])

		// Decode a 1-byte rune.
		if r < utf8.RuneSelf {
			nSrc++
			if m.charmap.asciiSuperset {
				dst[nDst] = uint8(r)
				nDst++
				continue
			}

		} else {
			// Decode a multi-byte rune.
			r, size = utf8.DecodeRune(src[nSrc:])
			if size == 1 {
				// All valid runes of size 1 (those below utf8.RuneSelf) were
				// handled above. We have invalid UTF-8 or we haven't seen the
				// full character yet.
				if !atEOF && !utf8.FullRune(src[nSrc:]) {
					err = transform.ErrShortSrc
					break
				}
			}
			nSrc += size
			if r == utf8.RuneError {
				dst[nDst] = m.charmap.replacement
				nDst++
				continue
			}
		}

		// Binary search in [low, high) for that rune in the m.charmap.encode table.
		for low, high := int(m.charmap.low), 0x100; ; {
			if low >= high {
				dst[nDst] = m.charmap.replacement
				nDst++
				break
			}
			mid := (low + high) / 2
			got := m.charmap.encode[mid]
			gotRune := rune(got & (1<<24 - 1))
			if gotRune < r {
				low = mid + 1
			} else if gotRune > r {
				high = mid
			} else {
				dst[nDst] = byte(got >> 24)
				nDst++
				break
			}
		}
	}
	return nDst, nSrc, err
}