1// Copyright 2013 The Go Authors. All rights reserved.
  2// Use of this source code is governed by a BSD-style
  3// license that can be found in the LICENSE file.
  4
  5package simplifiedchinese
  6
  7import (
  8	"unicode/utf8"
  9
 10	"golang.org/x/text/encoding"
 11	"golang.org/x/text/encoding/internal"
 12	"golang.org/x/text/encoding/internal/identifier"
 13	"golang.org/x/text/transform"
 14)
 15
 16var (
 17	// GB18030 is the GB18030 encoding.
 18	GB18030 encoding.Encoding = &gbk18030
 19	// GBK is the GBK encoding. It encodes an extension of the GB2312 character set
 20	// and is also known as Code Page 936.
 21	GBK encoding.Encoding = &gbk
 22)
 23
 24var gbk = internal.Encoding{
 25	&internal.SimpleEncoding{
 26		gbkDecoder{gb18030: false},
 27		gbkEncoder{gb18030: false},
 28	},
 29	"GBK",
 30	identifier.GBK,
 31}
 32
 33var gbk18030 = internal.Encoding{
 34	&internal.SimpleEncoding{
 35		gbkDecoder{gb18030: true},
 36		gbkEncoder{gb18030: true},
 37	},
 38	"GB18030",
 39	identifier.GB18030,
 40}
 41
 42type gbkDecoder struct {
 43	transform.NopResetter
 44	gb18030 bool
 45}
 46
 47func (d gbkDecoder) Transform(dst, src []byte, atEOF bool) (nDst, nSrc int, err error) {
 48	r, size := rune(0), 0
 49loop:
 50	for ; nSrc < len(src); nSrc += size {
 51		switch c0 := src[nSrc]; {
 52		case c0 < utf8.RuneSelf:
 53			r, size = rune(c0), 1
 54
 55		// Microsoft's Code Page 936 extends GBK 1.0 to encode the euro sign U+20AC
 56		// as 0x80. The HTML5 specification at http://encoding.spec.whatwg.org/#gbk
 57		// says to treat "gbk" as Code Page 936.
 58		case c0 == 0x80:
 59			r, size = '€', 1
 60
 61		case c0 < 0xff:
 62			if nSrc+1 >= len(src) {
 63				if !atEOF {
 64					err = transform.ErrShortSrc
 65					break loop
 66				}
 67				r, size = utf8.RuneError, 1
 68				goto write
 69			}
 70			c1 := src[nSrc+1]
 71			switch {
 72			case 0x40 <= c1 && c1 < 0x7f:
 73				c1 -= 0x40
 74			case 0x80 <= c1 && c1 < 0xff:
 75				c1 -= 0x41
 76			case d.gb18030 && 0x30 <= c1 && c1 < 0x40:
 77				if nSrc+3 >= len(src) {
 78					if !atEOF {
 79						err = transform.ErrShortSrc
 80						break loop
 81					}
 82					// The second byte here is always ASCII, so we can set size
 83					// to 1 in all cases.
 84					r, size = utf8.RuneError, 1
 85					goto write
 86				}
 87				c2 := src[nSrc+2]
 88				if c2 < 0x81 || 0xff <= c2 {
 89					r, size = utf8.RuneError, 1
 90					goto write
 91				}
 92				c3 := src[nSrc+3]
 93				if c3 < 0x30 || 0x3a <= c3 {
 94					r, size = utf8.RuneError, 1
 95					goto write
 96				}
 97				size = 4
 98				r = ((rune(c0-0x81)*10+rune(c1-0x30))*126+rune(c2-0x81))*10 + rune(c3-0x30)
 99				if r < 39420 {
100					i, j := 0, len(gb18030)
101					for i < j {
102						h := i + (j-i)/2
103						if r >= rune(gb18030[h][0]) {
104							i = h + 1
105						} else {
106							j = h
107						}
108					}
109					dec := &gb18030[i-1]
110					r += rune(dec[1]) - rune(dec[0])
111					goto write
112				}
113				r -= 189000
114				if 0 <= r && r < 0x100000 {
115					r += 0x10000
116				} else {
117					r, size = utf8.RuneError, 1
118				}
119				goto write
120			default:
121				r, size = utf8.RuneError, 1
122				goto write
123			}
124			r, size = '\ufffd', 2
125			if i := int(c0-0x81)*190 + int(c1); i < len(decode) {
126				r = rune(decode[i])
127				if r == 0 {
128					r = '\ufffd'
129				}
130			}
131
132		default:
133			r, size = utf8.RuneError, 1
134		}
135
136	write:
137		if nDst+utf8.RuneLen(r) > len(dst) {
138			err = transform.ErrShortDst
139			break loop
140		}
141		nDst += utf8.EncodeRune(dst[nDst:], r)
142	}
143	return nDst, nSrc, err
144}
145
146type gbkEncoder struct {
147	transform.NopResetter
148	gb18030 bool
149}
150
151func (e gbkEncoder) Transform(dst, src []byte, atEOF bool) (nDst, nSrc int, err error) {
152	r, r2, size := rune(0), rune(0), 0
153	for ; nSrc < len(src); nSrc += size {
154		r = rune(src[nSrc])
155
156		// Decode a 1-byte rune.
157		if r < utf8.RuneSelf {
158			size = 1
159
160		} else {
161			// Decode a multi-byte rune.
162			r, size = utf8.DecodeRune(src[nSrc:])
163			if size == 1 {
164				// All valid runes of size 1 (those below utf8.RuneSelf) were
165				// handled above. We have invalid UTF-8 or we haven't seen the
166				// full character yet.
167				if !atEOF && !utf8.FullRune(src[nSrc:]) {
168					err = transform.ErrShortSrc
169					break
170				}
171			}
172
173			// func init checks that the switch covers all tables.
174			switch {
175			case encode0Low <= r && r < encode0High:
176				if r2 = rune(encode0[r-encode0Low]); r2 != 0 {
177					goto write2
178				}
179			case encode1Low <= r && r < encode1High:
180				// Microsoft's Code Page 936 extends GBK 1.0 to encode the euro sign U+20AC
181				// as 0x80. The HTML5 specification at http://encoding.spec.whatwg.org/#gbk
182				// says to treat "gbk" as Code Page 936.
183				if r == '€' {
184					r = 0x80
185					goto write1
186				}
187				if r2 = rune(encode1[r-encode1Low]); r2 != 0 {
188					goto write2
189				}
190			case encode2Low <= r && r < encode2High:
191				if r2 = rune(encode2[r-encode2Low]); r2 != 0 {
192					goto write2
193				}
194			case encode3Low <= r && r < encode3High:
195				if r2 = rune(encode3[r-encode3Low]); r2 != 0 {
196					goto write2
197				}
198			case encode4Low <= r && r < encode4High:
199				if r2 = rune(encode4[r-encode4Low]); r2 != 0 {
200					goto write2
201				}
202			}
203
204			if e.gb18030 {
205				if r < 0x10000 {
206					i, j := 0, len(gb18030)
207					for i < j {
208						h := i + (j-i)/2
209						if r >= rune(gb18030[h][1]) {
210							i = h + 1
211						} else {
212							j = h
213						}
214					}
215					dec := &gb18030[i-1]
216					r += rune(dec[0]) - rune(dec[1])
217					goto write4
218				} else if r < 0x110000 {
219					r += 189000 - 0x10000
220					goto write4
221				}
222			}
223			err = internal.ErrASCIIReplacement
224			break
225		}
226
227	write1:
228		if nDst >= len(dst) {
229			err = transform.ErrShortDst
230			break
231		}
232		dst[nDst] = uint8(r)
233		nDst++
234		continue
235
236	write2:
237		if nDst+2 > len(dst) {
238			err = transform.ErrShortDst
239			break
240		}
241		dst[nDst+0] = uint8(r2 >> 8)
242		dst[nDst+1] = uint8(r2)
243		nDst += 2
244		continue
245
246	write4:
247		if nDst+4 > len(dst) {
248			err = transform.ErrShortDst
249			break
250		}
251		dst[nDst+3] = uint8(r%10 + 0x30)
252		r /= 10
253		dst[nDst+2] = uint8(r%126 + 0x81)
254		r /= 126
255		dst[nDst+1] = uint8(r%10 + 0x30)
256		r /= 10
257		dst[nDst+0] = uint8(r + 0x81)
258		nDst += 4
259		continue
260	}
261	return nDst, nSrc, err
262}
263
264func init() {
265	// Check that the hard-coded encode switch covers all tables.
266	if numEncodeTables != 5 {
267		panic("bad numEncodeTables")
268	}
269}