Skip to content

Commit f39cc11

Browse files
saintfishsaintfish
committed
Add multi-byte charset detectors
1 parent e19ac93 commit f39cc11

File tree

2 files changed

+350
-5
lines changed

2 files changed

+350
-5
lines changed

‎detector.go

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -46,11 +46,11 @@ var recognizers = []recognizer{
4646
newRecognizer_KOI8_R(),
4747
newRecognizer_8859_9_tr(),
4848

49-
// newRecognizer_sjis(),
50-
// newRecognizer_gb_18030(),
51-
// newRecognizer_euc_jp(),
52-
// newRecognizer_euc_kr(),
53-
// newRecognizer_big5(),
49+
newRecognizer_sjis(),
50+
newRecognizer_gb_18030(),
51+
newRecognizer_euc_jp(),
52+
newRecognizer_euc_kr(),
53+
newRecognizer_big5(),
5454

5555
// newRecognizer_2022JP(),
5656
// newRecognizer_2022KR(),

‎multi_byte.go

Lines changed: 345 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,345 @@
1+
package chardet
2+
3+
import (
4+
"errors"
5+
"math"
6+
)
7+
8+
type recognizerMultiByte struct {
9+
charset string
10+
language string
11+
decoder charDecoder
12+
commonChars []uint16
13+
}
14+
15+
type charDecoder interface {
16+
DecodeOneChar([]byte) (c uint16, remain []byte, err error)
17+
}
18+
19+
func (r *recognizerMultiByte) Match(input *recognizerInput) (output recognizerOutput) {
20+
return recognizerOutput{
21+
Charset: r.charset,
22+
Language: r.language,
23+
Confidence: r.matchConfidence(input),
24+
}
25+
}
26+
27+
func (r *recognizerMultiByte) matchConfidence(input *recognizerInput) int {
28+
raw := input.raw
29+
var c uint16
30+
var err error
31+
var totalCharCount, badCharCount, singleByteCharCount, doubleByteCharCount, commonCharCount int
32+
for c, raw, err = r.decoder.DecodeOneChar(raw); len(raw) > 0; c, raw, err = r.decoder.DecodeOneChar(raw) {
33+
totalCharCount++
34+
if err != nil {
35+
badCharCount++
36+
} else if c <= 0xFF {
37+
singleByteCharCount++
38+
} else {
39+
doubleByteCharCount++
40+
if r.commonChars != nil && binarySearch(r.commonChars, c) {
41+
commonCharCount++
42+
}
43+
}
44+
if badCharCount >= 2 && badCharCount*5 >= doubleByteCharCount {
45+
return 0
46+
}
47+
}
48+
49+
if doubleByteCharCount <= 10 && badCharCount == 0 {
50+
if doubleByteCharCount == 0 && totalCharCount < 10 {
51+
return 0
52+
} else {
53+
return 10
54+
}
55+
}
56+
57+
if doubleByteCharCount < 20*badCharCount {
58+
return 0
59+
}
60+
if r.commonChars == nil {
61+
confidence := 30 + doubleByteCharCount - 20*badCharCount
62+
if confidence > 100 {
63+
confidence = 100
64+
}
65+
return confidence
66+
}
67+
maxVal := math.Log(float64(doubleByteCharCount) / 4)
68+
scaleFactor := 90 / maxVal
69+
confidence := int(math.Log(float64(commonCharCount)+1)*scaleFactor + 10)
70+
if confidence > 100 {
71+
confidence = 100
72+
}
73+
if confidence < 0 {
74+
confidence = 0
75+
}
76+
return confidence
77+
}
78+
79+
func binarySearch(l []uint16, c uint16) bool {
80+
start := 0
81+
end := len(l)
82+
for start <= end {
83+
mid := (start + end) / 2
84+
if c == l[mid] {
85+
return true
86+
} else if c < l[mid] {
87+
end = mid - 1
88+
} else {
89+
start = mid + 1
90+
}
91+
}
92+
return false
93+
}
94+
95+
var eobError = errors.New("End of input buffer")
96+
var badCharError = errors.New("Decode a bad char")
97+
98+
type charDecoder_sjis struct {
99+
}
100+
101+
func (charDecoder_sjis) DecodeOneChar(input []byte) (c uint16, remain []byte, err error) {
102+
if len(input) == 0 {
103+
return 0, nil, eobError
104+
}
105+
first := input[0]
106+
c = uint16(first)
107+
remain = input[1:]
108+
if first <= 0x7F || (first > 0xA0 && first <= 0xDF) {
109+
return
110+
}
111+
if len(remain) == 0 {
112+
return c, remain, badCharError
113+
}
114+
second := remain[0]
115+
remain = remain[1:]
116+
c = c<<8 | uint16(second)
117+
if (second >= 0x40 && second <= 0x7F) || (second >= 0x80 && second <= 0xFE) {
118+
} else {
119+
err = badCharError
120+
}
121+
return
122+
}
123+
124+
var commonChars_sjis = []uint16{
125+
0x8140, 0x8141, 0x8142, 0x8145, 0x815b, 0x8169, 0x816a, 0x8175, 0x8176, 0x82a0,
126+
0x82a2, 0x82a4, 0x82a9, 0x82aa, 0x82ab, 0x82ad, 0x82af, 0x82b1, 0x82b3, 0x82b5,
127+
0x82b7, 0x82bd, 0x82be, 0x82c1, 0x82c4, 0x82c5, 0x82c6, 0x82c8, 0x82c9, 0x82cc,
128+
0x82cd, 0x82dc, 0x82e0, 0x82e7, 0x82e8, 0x82e9, 0x82ea, 0x82f0, 0x82f1, 0x8341,
129+
0x8343, 0x834e, 0x834f, 0x8358, 0x835e, 0x8362, 0x8367, 0x8375, 0x8376, 0x8389,
130+
0x838a, 0x838b, 0x838d, 0x8393, 0x8e96, 0x93fa, 0x95aa,
131+
}
132+
133+
func newRecognizer_sjis() *recognizerMultiByte {
134+
return &recognizerMultiByte{
135+
"Shift_JIS",
136+
"ja",
137+
charDecoder_sjis{},
138+
commonChars_sjis,
139+
}
140+
}
141+
142+
type charDecoder_euc struct {
143+
}
144+
145+
func (charDecoder_euc) DecodeOneChar(input []byte) (c uint16, remain []byte, err error) {
146+
if len(input) == 0 {
147+
return 0, nil, eobError
148+
}
149+
first := input[0]
150+
remain = input[1:]
151+
c = uint16(first)
152+
if first <= 0x8D {
153+
return uint16(first), remain, nil
154+
}
155+
if len(remain) == 0 {
156+
return 0, nil, eobError
157+
}
158+
second := remain[0]
159+
remain = remain[1:]
160+
c = c<<8 | uint16(second)
161+
if first >= 0xA1 && first <= 0xFE {
162+
if second < 0xA1 {
163+
err = badCharError
164+
}
165+
return
166+
}
167+
if first == 0x8E {
168+
if second < 0xA1 {
169+
err = badCharError
170+
}
171+
return
172+
}
173+
if first == 0x8F {
174+
if len(remain) == 0 {
175+
return 0, nil, eobError
176+
}
177+
third := remain[0]
178+
remain = remain[1:]
179+
c = c<<0 | uint16(third)
180+
if third < 0xa1 {
181+
err = badCharError
182+
}
183+
}
184+
return
185+
}
186+
187+
var commonChars_euc_jp = []uint16{
188+
0xa1a1, 0xa1a2, 0xa1a3, 0xa1a6, 0xa1bc, 0xa1ca, 0xa1cb, 0xa1d6, 0xa1d7, 0xa4a2,
189+
0xa4a4, 0xa4a6, 0xa4a8, 0xa4aa, 0xa4ab, 0xa4ac, 0xa4ad, 0xa4af, 0xa4b1, 0xa4b3,
190+
0xa4b5, 0xa4b7, 0xa4b9, 0xa4bb, 0xa4bd, 0xa4bf, 0xa4c0, 0xa4c1, 0xa4c3, 0xa4c4,
191+
0xa4c6, 0xa4c7, 0xa4c8, 0xa4c9, 0xa4ca, 0xa4cb, 0xa4ce, 0xa4cf, 0xa4d0, 0xa4de,
192+
0xa4df, 0xa4e1, 0xa4e2, 0xa4e4, 0xa4e8, 0xa4e9, 0xa4ea, 0xa4eb, 0xa4ec, 0xa4ef,
193+
0xa4f2, 0xa4f3, 0xa5a2, 0xa5a3, 0xa5a4, 0xa5a6, 0xa5a7, 0xa5aa, 0xa5ad, 0xa5af,
194+
0xa5b0, 0xa5b3, 0xa5b5, 0xa5b7, 0xa5b8, 0xa5b9, 0xa5bf, 0xa5c3, 0xa5c6, 0xa5c7,
195+
0xa5c8, 0xa5c9, 0xa5cb, 0xa5d0, 0xa5d5, 0xa5d6, 0xa5d7, 0xa5de, 0xa5e0, 0xa5e1,
196+
0xa5e5, 0xa5e9, 0xa5ea, 0xa5eb, 0xa5ec, 0xa5ed, 0xa5f3, 0xb8a9, 0xb9d4, 0xbaee,
197+
0xbbc8, 0xbef0, 0xbfb7, 0xc4ea, 0xc6fc, 0xc7bd, 0xcab8, 0xcaf3, 0xcbdc, 0xcdd1,
198+
}
199+
200+
var commonChars_euc_kr = []uint16{
201+
0xb0a1, 0xb0b3, 0xb0c5, 0xb0cd, 0xb0d4, 0xb0e6, 0xb0ed, 0xb0f8, 0xb0fa, 0xb0fc,
202+
0xb1b8, 0xb1b9, 0xb1c7, 0xb1d7, 0xb1e2, 0xb3aa, 0xb3bb, 0xb4c2, 0xb4cf, 0xb4d9,
203+
0xb4eb, 0xb5a5, 0xb5b5, 0xb5bf, 0xb5c7, 0xb5e9, 0xb6f3, 0xb7af, 0xb7c2, 0xb7ce,
204+
0xb8a6, 0xb8ae, 0xb8b6, 0xb8b8, 0xb8bb, 0xb8e9, 0xb9ab, 0xb9ae, 0xb9cc, 0xb9ce,
205+
0xb9fd, 0xbab8, 0xbace, 0xbad0, 0xbaf1, 0xbbe7, 0xbbf3, 0xbbfd, 0xbcad, 0xbcba,
206+
0xbcd2, 0xbcf6, 0xbdba, 0xbdc0, 0xbdc3, 0xbdc5, 0xbec6, 0xbec8, 0xbedf, 0xbeee,
207+
0xbef8, 0xbefa, 0xbfa1, 0xbfa9, 0xbfc0, 0xbfe4, 0xbfeb, 0xbfec, 0xbff8, 0xc0a7,
208+
0xc0af, 0xc0b8, 0xc0ba, 0xc0bb, 0xc0bd, 0xc0c7, 0xc0cc, 0xc0ce, 0xc0cf, 0xc0d6,
209+
0xc0da, 0xc0e5, 0xc0fb, 0xc0fc, 0xc1a4, 0xc1a6, 0xc1b6, 0xc1d6, 0xc1df, 0xc1f6,
210+
0xc1f8, 0xc4a1, 0xc5cd, 0xc6ae, 0xc7cf, 0xc7d1, 0xc7d2, 0xc7d8, 0xc7e5, 0xc8ad,
211+
}
212+
213+
func newRecognizer_euc_jp() *recognizerMultiByte {
214+
return &recognizerMultiByte{
215+
"EUC-JP",
216+
"ja",
217+
charDecoder_euc{},
218+
commonChars_euc_jp,
219+
}
220+
}
221+
222+
func newRecognizer_euc_kr() *recognizerMultiByte {
223+
return &recognizerMultiByte{
224+
"EUC-KR",
225+
"ko",
226+
charDecoder_euc{},
227+
commonChars_euc_kr,
228+
}
229+
}
230+
231+
type charDecoder_big5 struct {
232+
}
233+
234+
func (charDecoder_big5) DecodeOneChar(input []byte) (c uint16, remain []byte, err error) {
235+
if len(input) == 0 {
236+
return 0, nil, eobError
237+
}
238+
first := input[0]
239+
remain = input[1:]
240+
c = uint16(first)
241+
if first <= 0x7F || first == 0xFF {
242+
return
243+
}
244+
if len(remain) == 0 {
245+
return c, nil, eobError
246+
}
247+
second := remain[0]
248+
remain = remain[1:]
249+
c = c<<8 | uint16(second)
250+
if second < 0x40 || second == 0x7F || second == 0xFF {
251+
err = badCharError
252+
}
253+
return
254+
}
255+
256+
var commonChars_big5 = []uint16{
257+
0xa140, 0xa141, 0xa142, 0xa143, 0xa147, 0xa149, 0xa175, 0xa176, 0xa440, 0xa446,
258+
0xa447, 0xa448, 0xa451, 0xa454, 0xa457, 0xa464, 0xa46a, 0xa46c, 0xa477, 0xa4a3,
259+
0xa4a4, 0xa4a7, 0xa4c1, 0xa4ce, 0xa4d1, 0xa4df, 0xa4e8, 0xa4fd, 0xa540, 0xa548,
260+
0xa558, 0xa569, 0xa5cd, 0xa5e7, 0xa657, 0xa661, 0xa662, 0xa668, 0xa670, 0xa6a8,
261+
0xa6b3, 0xa6b9, 0xa6d3, 0xa6db, 0xa6e6, 0xa6f2, 0xa740, 0xa751, 0xa759, 0xa7da,
262+
0xa8a3, 0xa8a5, 0xa8ad, 0xa8d1, 0xa8d3, 0xa8e4, 0xa8fc, 0xa9c0, 0xa9d2, 0xa9f3,
263+
0xaa6b, 0xaaba, 0xaabe, 0xaacc, 0xaafc, 0xac47, 0xac4f, 0xacb0, 0xacd2, 0xad59,
264+
0xaec9, 0xafe0, 0xb0ea, 0xb16f, 0xb2b3, 0xb2c4, 0xb36f, 0xb44c, 0xb44e, 0xb54c,
265+
0xb5a5, 0xb5bd, 0xb5d0, 0xb5d8, 0xb671, 0xb7ed, 0xb867, 0xb944, 0xbad8, 0xbb44,
266+
0xbba1, 0xbdd1, 0xc2c4, 0xc3b9, 0xc440, 0xc45f,
267+
}
268+
269+
func newRecognizer_big5() *recognizerMultiByte {
270+
return &recognizerMultiByte{
271+
"Big5",
272+
"zh",
273+
charDecoder_big5{},
274+
commonChars_big5,
275+
}
276+
}
277+
278+
type charDecoder_gb_18030 struct {
279+
}
280+
281+
func (charDecoder_gb_18030) DecodeOneChar(input []byte) (c uint16, remain []byte, err error) {
282+
if len(input) == 0 {
283+
return 0, nil, eobError
284+
}
285+
first := input[0]
286+
remain = input[1:]
287+
c = uint16(first)
288+
if first <= 0x80 {
289+
return
290+
}
291+
if len(remain) == 0 {
292+
return 0, nil, eobError
293+
}
294+
second := remain[0]
295+
remain = remain[1:]
296+
c = c<<8 | uint16(second)
297+
if first >= 0x81 && first <= 0xFE {
298+
if (second >= 0x40 && second <= 0x7E) || (second >= 0x80 && second <= 0xFE) {
299+
return
300+
}
301+
302+
if second >= 0x30 && second <= 0x39 {
303+
if len(remain) == 0 {
304+
return 0, nil, eobError
305+
}
306+
third := remain[0]
307+
remain = remain[1:]
308+
if third >= 0x81 && third <= 0xFE {
309+
if len(remain) == 0 {
310+
return 0, nil, eobError
311+
}
312+
fourth := remain[0]
313+
remain = remain[1:]
314+
if fourth >= 0x30 && fourth <= 0x39 {
315+
c = c<<16 | uint16(third)<<8 | uint16(fourth)
316+
return
317+
}
318+
}
319+
}
320+
err = badCharError
321+
}
322+
return
323+
}
324+
325+
var commonChars_gb_18030 = []uint16{
326+
0xa1a1, 0xa1a2, 0xa1a3, 0xa1a4, 0xa1b0, 0xa1b1, 0xa1f1, 0xa1f3, 0xa3a1, 0xa3ac,
327+
0xa3ba, 0xb1a8, 0xb1b8, 0xb1be, 0xb2bb, 0xb3c9, 0xb3f6, 0xb4f3, 0xb5bd, 0xb5c4,
328+
0xb5e3, 0xb6af, 0xb6d4, 0xb6e0, 0xb7a2, 0xb7a8, 0xb7bd, 0xb7d6, 0xb7dd, 0xb8b4,
329+
0xb8df, 0xb8f6, 0xb9ab, 0xb9c9, 0xb9d8, 0xb9fa, 0xb9fd, 0xbacd, 0xbba7, 0xbbd6,
330+
0xbbe1, 0xbbfa, 0xbcbc, 0xbcdb, 0xbcfe, 0xbdcc, 0xbecd, 0xbedd, 0xbfb4, 0xbfc6,
331+
0xbfc9, 0xc0b4, 0xc0ed, 0xc1cb, 0xc2db, 0xc3c7, 0xc4dc, 0xc4ea, 0xc5cc, 0xc6f7,
332+
0xc7f8, 0xc8ab, 0xc8cb, 0xc8d5, 0xc8e7, 0xc9cf, 0xc9fa, 0xcab1, 0xcab5, 0xcac7,
333+
0xcad0, 0xcad6, 0xcaf5, 0xcafd, 0xccec, 0xcdf8, 0xceaa, 0xcec4, 0xced2, 0xcee5,
334+
0xcfb5, 0xcfc2, 0xcfd6, 0xd0c2, 0xd0c5, 0xd0d0, 0xd0d4, 0xd1a7, 0xd2aa, 0xd2b2,
335+
0xd2b5, 0xd2bb, 0xd2d4, 0xd3c3, 0xd3d0, 0xd3fd, 0xd4c2, 0xd4da, 0xd5e2, 0xd6d0,
336+
}
337+
338+
func newRecognizer_gb_18030() *recognizerMultiByte {
339+
return &recognizerMultiByte{
340+
"GB-18030",
341+
"zh",
342+
charDecoder_gb_18030{},
343+
commonChars_gb_18030,
344+
}
345+
}

0 commit comments

Comments
 (0)