Mercurial > libguess
comparison turkish_impl.c @ 3:70e2c306231e
- implemented dfa utility functions.
- added dfa.c.
- rewrote guess functions for ar, gr, hw and tr scripts with dfa utilities.
- guess functions for cjk scripts too.
author | Yoshiki Yazawa <yaz@cc.rim.or.jp> |
---|---|
date | Thu, 12 Jun 2008 20:20:43 +0900 |
parents | 754a4550c64e |
children |
comparison
equal
deleted
inserted
replaced
2:754a4550c64e | 3:70e2c306231e |
---|---|
1 #include "libguess.h" | 1 #include "libguess.h" |
2 #include "dfa.h" | |
3 #include "guess_tab.c" | |
2 | 4 |
3 static const char *_guess_tr(const unsigned char *ptr, int size) | 5 /* precedence order */ |
6 #define ORDER &utf8, &iso8859_9, &cp1254 | |
7 | |
8 /* encodings */ | |
9 static guess_dfa cp1254 = DFA_INIT(guess_cp1253_st, guess_cp1253_ar, "CP1254"); | |
10 static guess_dfa iso8859_9 = DFA_INIT(guess_iso8859_9_st, guess_iso8859_9_ar, "ISO-8859-9"); | |
11 static guess_dfa utf8 = DFA_INIT(guess_utf8_st, guess_utf8_ar, "UTF-8"); | |
12 | |
13 /* common */ | |
14 const char *guess_tr(const char *buf, int buflen) | |
4 { | 15 { |
5 int i; | 16 int i; |
17 const char *rv = NULL; | |
18 guess_dfa *top = NULL; | |
19 guess_dfa *order[] = { ORDER, NULL }; | |
6 | 20 |
7 for (i = 0; i < size; i++) | 21 for (i = 0; i < buflen; i++) { |
8 { | 22 int c = (unsigned char) buf[i]; |
9 if (ptr[i] == 0x80 || | 23 |
10 (ptr[i] >= 0x82 && ptr[i] <= 0x8C) || | 24 /* special treatment of BOM */ |
11 (ptr[i] >= 0x91 && ptr[i] <= 0x9C) || | 25 if (i == 0 && c == 0xff) { |
12 ptr[ i ] == 0x9F) | 26 if (i < buflen - 1) { |
13 return "CP1254"; | 27 c = (unsigned char) buf[i + 1]; |
28 if (c == 0xfe) | |
29 return UCS_2LE; | |
30 } | |
31 } | |
32 if (i == 0 && c == 0xfe) { | |
33 if (i < buflen - 1) { | |
34 c = (unsigned char) buf[i + 1]; | |
35 if (c == 0xff) | |
36 return UCS_2BE; | |
37 } | |
38 } | |
39 | |
40 rv = dfa_process(order, c); | |
41 if(rv) | |
42 return rv; | |
43 | |
44 if (dfa_none(order)) { | |
45 /* we ran out the possibilities */ | |
46 return NULL; | |
47 } | |
14 } | 48 } |
15 | 49 |
16 return "ISO-8859-9"; | 50 top = dfa_top(order); |
51 if (top) | |
52 return top->name; | |
53 else | |
54 return NULL; | |
17 } | 55 } |
18 | |
19 const char *guess_tr(const char *ptr, int size) | |
20 { | |
21 if (dfa_validate_utf8(ptr, size)) | |
22 return "UTF-8"; | |
23 | |
24 return _guess_tr((const unsigned char *)ptr, size); | |
25 } |