c6d25d8d |
1 | /* |
2 | * local.c - translate our internal character set codes to and from |
3 | * our own set of plausibly legible character-set names. Also |
4 | * provides a canonical name for each encoding (useful for software |
5 | * announcing what character set it will be using), and a set of |
6 | * enumeration functions which return a list of supported |
7 | * encodings one by one. |
8 | * |
9 | * charset_from_localenc will attempt all other text translations |
10 | * as well as this table, to maximise the number of different ways |
11 | * you can select a supported charset. |
12 | */ |
13 | |
14 | #include <ctype.h> |
15 | #include "charset.h" |
16 | #include "internal.h" |
17 | |
18 | static const struct { |
19 | const char *name; |
20 | int charset; |
21 | int return_in_enum; /* enumeration misses some charsets */ |
22 | } localencs[] = { |
23 | { "<UNKNOWN>", CS_NONE, 0 }, |
24 | { "ASCII", CS_ASCII, 1 }, |
25 | { "ISO-8859-1", CS_ISO8859_1, 1 }, |
26 | { "ISO-8859-1 with X11 line drawing", CS_ISO8859_1_X11, 0 }, |
27 | { "ISO-8859-2", CS_ISO8859_2, 1 }, |
28 | { "ISO-8859-3", CS_ISO8859_3, 1 }, |
29 | { "ISO-8859-4", CS_ISO8859_4, 1 }, |
30 | { "ISO-8859-5", CS_ISO8859_5, 1 }, |
31 | { "ISO-8859-6", CS_ISO8859_6, 1 }, |
32 | { "ISO-8859-7", CS_ISO8859_7, 1 }, |
33 | { "ISO-8859-8", CS_ISO8859_8, 1 }, |
34 | { "ISO-8859-9", CS_ISO8859_9, 1 }, |
35 | { "ISO-8859-10", CS_ISO8859_10, 1 }, |
36 | { "ISO-8859-11", CS_ISO8859_11, 1 }, |
37 | { "ISO-8859-13", CS_ISO8859_13, 1 }, |
38 | { "ISO-8859-14", CS_ISO8859_14, 1 }, |
39 | { "ISO-8859-15", CS_ISO8859_15, 1 }, |
40 | { "ISO-8859-16", CS_ISO8859_16, 1 }, |
41 | { "CP437", CS_CP437, 1 }, |
42 | { "CP850", CS_CP850, 1 }, |
43 | { "CP1250", CS_CP1250, 1 }, |
44 | { "CP1251", CS_CP1251, 1 }, |
45 | { "CP1252", CS_CP1252, 1 }, |
46 | { "CP1253", CS_CP1253, 1 }, |
47 | { "CP1254", CS_CP1254, 1 }, |
48 | { "CP1255", CS_CP1255, 1 }, |
49 | { "CP1256", CS_CP1256, 1 }, |
50 | { "CP1257", CS_CP1257, 1 }, |
51 | { "CP1258", CS_CP1258, 1 }, |
52 | { "KOI8-R", CS_KOI8_R, 1 }, |
53 | { "KOI8-U", CS_KOI8_U, 1 }, |
54 | { "KOI8-RU", CS_KOI8_RU, 1 }, |
55 | { "Mac Roman", CS_MAC_ROMAN, 1 }, |
56 | { "Mac Turkish", CS_MAC_TURKISH, 1 }, |
57 | { "Mac Croatian", CS_MAC_CROATIAN, 1 }, |
58 | { "Mac Iceland", CS_MAC_ICELAND, 1 }, |
59 | { "Mac Romanian", CS_MAC_ROMANIAN, 1 }, |
60 | { "Mac Greek", CS_MAC_GREEK, 1 }, |
61 | { "Mac Cyrillic", CS_MAC_CYRILLIC, 1 }, |
62 | { "Mac Thai", CS_MAC_THAI, 1 }, |
63 | { "Mac Centeuro", CS_MAC_CENTEURO, 1 }, |
64 | { "Mac Symbol", CS_MAC_SYMBOL, 1 }, |
65 | { "Mac Dingbats", CS_MAC_DINGBATS, 1 }, |
66 | { "Mac Roman (old)", CS_MAC_ROMAN_OLD, 0 }, |
67 | { "Mac Croatian (old)", CS_MAC_CROATIAN_OLD, 0 }, |
68 | { "Mac Iceland (old)", CS_MAC_ICELAND_OLD, 0 }, |
69 | { "Mac Romanian (old)", CS_MAC_ROMANIAN_OLD, 0 }, |
70 | { "Mac Greek (old)", CS_MAC_GREEK_OLD, 0 }, |
71 | { "Mac Cyrillic (old)", CS_MAC_CYRILLIC_OLD, 0 }, |
72 | { "Mac Ukraine", CS_MAC_UKRAINE, 1 }, |
73 | { "Mac VT100", CS_MAC_VT100, 1 }, |
74 | { "Mac VT100 (old)", CS_MAC_VT100_OLD, 0 }, |
75 | { "VISCII", CS_VISCII, 1 }, |
76 | { "HP ROMAN8", CS_HP_ROMAN8, 1 }, |
77 | { "DEC MCS", CS_DEC_MCS, 1 }, |
78 | { "UTF-8", CS_UTF8, 1 }, |
79 | { "UTF-7", CS_UTF7, 1 }, |
80 | { "UTF-7-conservative", CS_UTF7_CONSERVATIVE, 0 }, |
81 | { "EUC-CN", CS_EUC_CN, 1 }, |
82 | { "EUC-KR", CS_EUC_KR, 1 }, |
83 | { "EUC-JP", CS_EUC_JP, 1 }, |
84 | { "ISO-2022-JP", CS_ISO2022_JP, 1 }, |
85 | { "ISO-2022-KR", CS_ISO2022_KR, 1 }, |
86 | { "Big5", CS_BIG5, 1 }, |
87 | { "Shift-JIS", CS_SHIFT_JIS, 1 }, |
88 | { "HZ", CS_HZ, 1 }, |
89 | { "UTF-16BE", CS_UTF16BE, 1 }, |
90 | { "UTF-16LE", CS_UTF16LE, 1 }, |
91 | { "UTF-16", CS_UTF16, 1 }, |
92 | { "CP949", CS_CP949, 1 }, |
cdb08fdc |
93 | { "PDFDocEncoding", CS_PDF, 1 }, |
c6d25d8d |
94 | }; |
95 | |
96 | const char *charset_to_localenc(int charset) |
97 | { |
98 | int i; |
99 | |
100 | for (i = 0; i < (int)lenof(localencs); i++) |
101 | if (charset == localencs[i].charset) |
102 | return localencs[i].name; |
103 | |
104 | return NULL; /* not found */ |
105 | } |
106 | |
107 | int charset_from_localenc(const char *name) |
108 | { |
109 | int i; |
110 | |
111 | if ( (i = charset_from_mimeenc(name)) != CS_NONE) |
112 | return i; |
113 | if ( (i = charset_from_xenc(name)) != CS_NONE) |
114 | return i; |
115 | |
116 | for (i = 0; i < (int)lenof(localencs); i++) { |
117 | const char *p, *q; |
118 | p = name; |
119 | q = localencs[i].name; |
120 | while (*p || *q) { |
121 | if (tolower(*p) != tolower(*q)) |
122 | break; |
123 | p++; q++; |
124 | } |
125 | if (!*p && !*q) |
126 | return localencs[i].charset; |
127 | } |
128 | |
129 | return CS_NONE; /* not found */ |
130 | } |
131 | |
132 | int charset_localenc_nth(int n) |
133 | { |
134 | int i; |
135 | |
136 | for (i = 0; i < (int)lenof(localencs); i++) |
137 | if (localencs[i].return_in_enum && !n--) |
138 | return localencs[i].charset; |
139 | |
140 | return CS_NONE; /* end of list */ |
141 | } |