libguess_d.guess.common source code

1 /**
2  * 
3  *
4  * License: BSD 3-clause
5  */
6 module libguess_d.guess.common;
7 
8 
9 private static import libguess_d.dfa;
10 private static import libguess_d.encoding;
11 
12 pure nothrow @safe @nogc
13 private void common_first_check(const char[] input, ref libguess_d.encoding.libguess_encoding rv)
14 
15 	do
16 	{
17 		static import libguess_d.dfa;
18 
19 		rv = libguess_d.dfa.check_UTF16_BOM(input);
20 	}
21 
22 /**
23  * 
24  *
25  * Params:
26  *      input = input string
27  *
28  * Returns: Character encoding or null
29  */
30 pure nothrow @trusted @nogc
31 libguess_d.encoding.libguess_encoding guess_common(libguess_d.encoding.supported_lang lang)(const char[] input)
32 
33 	do
34 	{
35 		static import libguess_d.dfa;
36 		static import libguess_d.encoding;
37 
38 		if (input.length == 0) {
39 			return libguess_d.encoding.libguess_encoding.undefined;
40 		}
41 
42 		static if (lang == libguess_d.encoding.supported_lang.ar) {
43 			libguess_d.dfa.guess_dfa[3] order = [libguess_d.encoding.utf8, libguess_d.encoding.iso8859_6, libguess_d.encoding.cp1256];
44 		} else static if (lang == libguess_d.encoding.supported_lang.gr) {
45 			libguess_d.dfa.guess_dfa[3] order = [libguess_d.encoding.utf8, libguess_d.encoding.iso8859_7, libguess_d.encoding.cp1253];
46 		} else static if (lang == libguess_d.encoding.supported_lang.hw) {
47 			libguess_d.dfa.guess_dfa[3] order = [libguess_d.encoding.utf8, libguess_d.encoding.iso8859_8, libguess_d.encoding.cp1255];
48 		} else static if (lang == libguess_d.encoding.supported_lang.tr) {
49 			libguess_d.dfa.guess_dfa[3] order = [libguess_d.encoding.utf8, libguess_d.encoding.iso8859_9, libguess_d.encoding.cp1254];
50 		} else {
51 			static assert(0);
52 		}
53 
54 		libguess_d.encoding.libguess_encoding rv = libguess_d.encoding.libguess_encoding.undefined;
55 
56 		.common_first_check(input, rv);
57 
58 		if (rv != libguess_d.encoding.libguess_encoding.undefined) {
59 			return rv;
60 		}
61 
62 		for (size_t i = 0; i < input.length; i++) {
63 			ubyte c = cast(ubyte)(input[i]);
64 
65 			rv = libguess_d.dfa.dfa_process(order, c);
66 
67 			if (rv != libguess_d.encoding.libguess_encoding.undefined) {
68 				return rv;
69 			}
70 
71 			if (libguess_d.dfa.dfa_none(order)) {
72 				/* we ran out the possibilities */
73 				return libguess_d.encoding.libguess_encoding.undefined;
74 			}
75 		}
76 
77 		rv = libguess_d.dfa.dfa_top(order);
78 
79 		return (rv != libguess_d.encoding.libguess_encoding.undefined) ? (rv) : (libguess_d.encoding.libguess_encoding.undefined);
80 	}
81 
82 alias guess_ar = .guess_common!(libguess_d.encoding.supported_lang.ar);
83 alias guess_gr = .guess_common!(libguess_d.encoding.supported_lang.gr);
84 alias guess_hw = .guess_common!(libguess_d.encoding.supported_lang.hw);
85 alias guess_tr = .guess_common!(libguess_d.encoding.supported_lang.tr);