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 | * |
51e2b7c7 |
9 | * Also in this table are other ways people might plausibly refer |
10 | * to a charset (for example, Win1252 as well as CP1252). Where |
11 | * more than one string is given for a particular character set, |
12 | * the first one is the canonical one returned by |
13 | * charset_to_localenc. |
14 | * |
c6d25d8d |
15 | * charset_from_localenc will attempt all other text translations |
16 | * as well as this table, to maximise the number of different ways |
17 | * you can select a supported charset. |
18 | */ |
19 | |
20 | #include <ctype.h> |
21 | #include "charset.h" |
22 | #include "internal.h" |
23 | |
24 | static const struct { |
25 | const char *name; |
26 | int charset; |
27 | int return_in_enum; /* enumeration misses some charsets */ |
28 | } localencs[] = { |
29 | { "<UNKNOWN>", CS_NONE, 0 }, |
30 | { "ASCII", CS_ASCII, 1 }, |
294941fa |
31 | { "BS 4730", CS_BS4730, 1 }, |
c6d25d8d |
32 | { "ISO-8859-1", CS_ISO8859_1, 1 }, |
33 | { "ISO-8859-1 with X11 line drawing", CS_ISO8859_1_X11, 0 }, |
34 | { "ISO-8859-2", CS_ISO8859_2, 1 }, |
35 | { "ISO-8859-3", CS_ISO8859_3, 1 }, |
36 | { "ISO-8859-4", CS_ISO8859_4, 1 }, |
37 | { "ISO-8859-5", CS_ISO8859_5, 1 }, |
38 | { "ISO-8859-6", CS_ISO8859_6, 1 }, |
39 | { "ISO-8859-7", CS_ISO8859_7, 1 }, |
40 | { "ISO-8859-8", CS_ISO8859_8, 1 }, |
41 | { "ISO-8859-9", CS_ISO8859_9, 1 }, |
42 | { "ISO-8859-10", CS_ISO8859_10, 1 }, |
43 | { "ISO-8859-11", CS_ISO8859_11, 1 }, |
44 | { "ISO-8859-13", CS_ISO8859_13, 1 }, |
45 | { "ISO-8859-14", CS_ISO8859_14, 1 }, |
46 | { "ISO-8859-15", CS_ISO8859_15, 1 }, |
47 | { "ISO-8859-16", CS_ISO8859_16, 1 }, |
48 | { "CP437", CS_CP437, 1 }, |
49 | { "CP850", CS_CP850, 1 }, |
9b7e7a92 |
50 | { "CP866", CS_CP866, 1 }, |
36eb7564 |
51 | { "CP874", CS_CP874, 1 }, |
52 | { "Win874", CS_CP874, 0 }, |
53 | { "Win-874", CS_CP874, 0 }, |
c6d25d8d |
54 | { "CP1250", CS_CP1250, 1 }, |
5cd6dcd4 |
55 | { "Win1250", CS_CP1250, 0 }, |
c6d25d8d |
56 | { "CP1251", CS_CP1251, 1 }, |
5cd6dcd4 |
57 | { "Win1251", CS_CP1251, 0 }, |
c6d25d8d |
58 | { "CP1252", CS_CP1252, 1 }, |
5cd6dcd4 |
59 | { "Win1252", CS_CP1252, 0 }, |
c6d25d8d |
60 | { "CP1253", CS_CP1253, 1 }, |
5cd6dcd4 |
61 | { "Win1253", CS_CP1253, 0 }, |
c6d25d8d |
62 | { "CP1254", CS_CP1254, 1 }, |
5cd6dcd4 |
63 | { "Win1254", CS_CP1254, 0 }, |
c6d25d8d |
64 | { "CP1255", CS_CP1255, 1 }, |
5cd6dcd4 |
65 | { "Win1255", CS_CP1255, 0 }, |
c6d25d8d |
66 | { "CP1256", CS_CP1256, 1 }, |
5cd6dcd4 |
67 | { "Win1256", CS_CP1256, 0 }, |
c6d25d8d |
68 | { "CP1257", CS_CP1257, 1 }, |
5cd6dcd4 |
69 | { "Win1257", CS_CP1257, 0 }, |
c6d25d8d |
70 | { "CP1258", CS_CP1258, 1 }, |
5cd6dcd4 |
71 | { "Win1258", CS_CP1258, 0 }, |
c6d25d8d |
72 | { "KOI8-R", CS_KOI8_R, 1 }, |
73 | { "KOI8-U", CS_KOI8_U, 1 }, |
74 | { "KOI8-RU", CS_KOI8_RU, 1 }, |
01081d4e |
75 | { "JIS X 0201", CS_JISX0201, 1 }, |
5cd6dcd4 |
76 | { "JIS-X-0201", CS_JISX0201, 0 }, |
77 | { "JIS_X_0201", CS_JISX0201, 0 }, |
78 | { "JISX0201", CS_JISX0201, 0 }, |
c6d25d8d |
79 | { "Mac Roman", CS_MAC_ROMAN, 1 }, |
80 | { "Mac Turkish", CS_MAC_TURKISH, 1 }, |
81 | { "Mac Croatian", CS_MAC_CROATIAN, 1 }, |
82 | { "Mac Iceland", CS_MAC_ICELAND, 1 }, |
83 | { "Mac Romanian", CS_MAC_ROMANIAN, 1 }, |
84 | { "Mac Greek", CS_MAC_GREEK, 1 }, |
85 | { "Mac Cyrillic", CS_MAC_CYRILLIC, 1 }, |
86 | { "Mac Thai", CS_MAC_THAI, 1 }, |
87 | { "Mac Centeuro", CS_MAC_CENTEURO, 1 }, |
88 | { "Mac Symbol", CS_MAC_SYMBOL, 1 }, |
89 | { "Mac Dingbats", CS_MAC_DINGBATS, 1 }, |
90 | { "Mac Roman (old)", CS_MAC_ROMAN_OLD, 0 }, |
91 | { "Mac Croatian (old)", CS_MAC_CROATIAN_OLD, 0 }, |
92 | { "Mac Iceland (old)", CS_MAC_ICELAND_OLD, 0 }, |
93 | { "Mac Romanian (old)", CS_MAC_ROMANIAN_OLD, 0 }, |
94 | { "Mac Greek (old)", CS_MAC_GREEK_OLD, 0 }, |
95 | { "Mac Cyrillic (old)", CS_MAC_CYRILLIC_OLD, 0 }, |
96 | { "Mac Ukraine", CS_MAC_UKRAINE, 1 }, |
97 | { "Mac VT100", CS_MAC_VT100, 1 }, |
98 | { "Mac VT100 (old)", CS_MAC_VT100_OLD, 0 }, |
99 | { "VISCII", CS_VISCII, 1 }, |
100 | { "HP ROMAN8", CS_HP_ROMAN8, 1 }, |
101 | { "DEC MCS", CS_DEC_MCS, 1 }, |
294941fa |
102 | { "DEC graphics", CS_DEC_GRAPHICS, 1 }, |
103 | { "DEC-graphics", CS_DEC_GRAPHICS, 0 }, |
104 | { "DECgraphics", CS_DEC_GRAPHICS, 0 }, |
c6d25d8d |
105 | { "UTF-8", CS_UTF8, 1 }, |
106 | { "UTF-7", CS_UTF7, 1 }, |
107 | { "UTF-7-conservative", CS_UTF7_CONSERVATIVE, 0 }, |
108 | { "EUC-CN", CS_EUC_CN, 1 }, |
109 | { "EUC-KR", CS_EUC_KR, 1 }, |
110 | { "EUC-JP", CS_EUC_JP, 1 }, |
2d50ebfa |
111 | { "EUC-TW", CS_EUC_TW, 1 }, |
c6d25d8d |
112 | { "ISO-2022-JP", CS_ISO2022_JP, 1 }, |
113 | { "ISO-2022-KR", CS_ISO2022_KR, 1 }, |
114 | { "Big5", CS_BIG5, 1 }, |
115 | { "Shift-JIS", CS_SHIFT_JIS, 1 }, |
116 | { "HZ", CS_HZ, 1 }, |
117 | { "UTF-16BE", CS_UTF16BE, 1 }, |
118 | { "UTF-16LE", CS_UTF16LE, 1 }, |
119 | { "UTF-16", CS_UTF16, 1 }, |
120 | { "CP949", CS_CP949, 1 }, |
cdb08fdc |
121 | { "PDFDocEncoding", CS_PDF, 1 }, |
032fbecf |
122 | { "StandardEncoding", CS_PSSTD, 1 }, |
01081d4e |
123 | { "COMPOUND_TEXT", CS_CTEXT, 1 }, |
5cd6dcd4 |
124 | { "COMPOUND-TEXT", CS_CTEXT, 0 }, |
125 | { "COMPOUND TEXT", CS_CTEXT, 0 }, |
126 | { "COMPOUNDTEXT", CS_CTEXT, 0 }, |
127 | { "CTEXT", CS_CTEXT, 0 }, |
b97e5427 |
128 | { "ISO-2022", CS_ISO2022, 1 }, |
129 | { "ISO2022", CS_ISO2022, 0 }, |
c6d25d8d |
130 | }; |
131 | |
132 | const char *charset_to_localenc(int charset) |
133 | { |
134 | int i; |
135 | |
136 | for (i = 0; i < (int)lenof(localencs); i++) |
137 | if (charset == localencs[i].charset) |
138 | return localencs[i].name; |
139 | |
140 | return NULL; /* not found */ |
141 | } |
142 | |
143 | int charset_from_localenc(const char *name) |
144 | { |
145 | int i; |
146 | |
147 | if ( (i = charset_from_mimeenc(name)) != CS_NONE) |
148 | return i; |
149 | if ( (i = charset_from_xenc(name)) != CS_NONE) |
150 | return i; |
32361bda |
151 | if ( (i = charset_from_emacsenc(name)) != CS_NONE) |
152 | return i; |
c6d25d8d |
153 | |
154 | for (i = 0; i < (int)lenof(localencs); i++) { |
155 | const char *p, *q; |
156 | p = name; |
157 | q = localencs[i].name; |
158 | while (*p || *q) { |
273ef1a1 |
159 | if (tolower((unsigned char)*p) != tolower((unsigned char)*q)) |
c6d25d8d |
160 | break; |
161 | p++; q++; |
162 | } |
163 | if (!*p && !*q) |
164 | return localencs[i].charset; |
165 | } |
166 | |
167 | return CS_NONE; /* not found */ |
168 | } |
169 | |
170 | int charset_localenc_nth(int n) |
171 | { |
172 | int i; |
173 | |
174 | for (i = 0; i < (int)lenof(localencs); i++) |
175 | if (localencs[i].return_in_enum && !n--) |
176 | return localencs[i].charset; |
177 | |
178 | return CS_NONE; /* end of list */ |
179 | } |