2 * local.c - translate our internal character set codes to and from
\r
3 * our own set of plausibly legible character-set names. Also
\r
4 * provides a canonical name for each encoding (useful for software
\r
5 * announcing what character set it will be using), and a set of
\r
6 * enumeration functions which return a list of supported
\r
7 * encodings one by one.
\r
9 * charset_from_localenc will attempt all other text translations
\r
10 * as well as this table, to maximise the number of different ways
\r
11 * you can select a supported charset.
\r
15 #include "charset.h"
\r
16 #include "internal.h"
\r
18 static const struct {
\r
21 int return_in_enum; /* enumeration misses some charsets */
\r
23 { "<UNKNOWN>", CS_NONE, 0 },
\r
24 { "ISO-8859-1", CS_ISO8859_1, 1 },
\r
25 { "ISO-8859-1 with X11 line drawing", CS_ISO8859_1_X11, 0 },
\r
26 { "ISO-8859-2", CS_ISO8859_2, 1 },
\r
27 { "ISO-8859-3", CS_ISO8859_3, 1 },
\r
28 { "ISO-8859-4", CS_ISO8859_4, 1 },
\r
29 { "ISO-8859-5", CS_ISO8859_5, 1 },
\r
30 { "ISO-8859-6", CS_ISO8859_6, 1 },
\r
31 { "ISO-8859-7", CS_ISO8859_7, 1 },
\r
32 { "ISO-8859-8", CS_ISO8859_8, 1 },
\r
33 { "ISO-8859-9", CS_ISO8859_9, 1 },
\r
34 { "ISO-8859-10", CS_ISO8859_10, 1 },
\r
35 { "ISO-8859-11", CS_ISO8859_11, 1 },
\r
36 { "ISO-8859-13", CS_ISO8859_13, 1 },
\r
37 { "ISO-8859-14", CS_ISO8859_14, 1 },
\r
38 { "ISO-8859-15", CS_ISO8859_15, 1 },
\r
39 { "ISO-8859-16", CS_ISO8859_16, 1 },
\r
40 { "CP437", CS_CP437, 1 },
\r
41 { "CP850", CS_CP850, 1 },
\r
42 { "CP866", CS_CP866, 1 },
\r
43 { "CP1250", CS_CP1250, 1 },
\r
44 { "CP1251", CS_CP1251, 1 },
\r
45 { "CP1252", CS_CP1252, 1 },
\r
46 { "CP1253", CS_CP1253, 1 },
\r
47 { "CP1254", CS_CP1254, 1 },
\r
48 { "CP1255", CS_CP1255, 1 },
\r
49 { "CP1256", CS_CP1256, 1 },
\r
50 { "CP1257", CS_CP1257, 1 },
\r
51 { "CP1258", CS_CP1258, 1 },
\r
52 { "KOI8-R", CS_KOI8_R, 1 },
\r
53 { "KOI8-U", CS_KOI8_U, 1 },
\r
54 { "Mac Roman", CS_MAC_ROMAN, 1 },
\r
55 { "Mac Turkish", CS_MAC_TURKISH, 1 },
\r
56 { "Mac Croatian", CS_MAC_CROATIAN, 1 },
\r
57 { "Mac Iceland", CS_MAC_ICELAND, 1 },
\r
58 { "Mac Romanian", CS_MAC_ROMANIAN, 1 },
\r
59 { "Mac Greek", CS_MAC_GREEK, 1 },
\r
60 { "Mac Cyrillic", CS_MAC_CYRILLIC, 1 },
\r
61 { "Mac Thai", CS_MAC_THAI, 1 },
\r
62 { "Mac Centeuro", CS_MAC_CENTEURO, 1 },
\r
63 { "Mac Symbol", CS_MAC_SYMBOL, 1 },
\r
64 { "Mac Dingbats", CS_MAC_DINGBATS, 1 },
\r
65 { "Mac Roman (old)", CS_MAC_ROMAN_OLD, 0 },
\r
66 { "Mac Croatian (old)", CS_MAC_CROATIAN_OLD, 0 },
\r
67 { "Mac Iceland (old)", CS_MAC_ICELAND_OLD, 0 },
\r
68 { "Mac Romanian (old)", CS_MAC_ROMANIAN_OLD, 0 },
\r
69 { "Mac Greek (old)", CS_MAC_GREEK_OLD, 0 },
\r
70 { "Mac Cyrillic (old)", CS_MAC_CYRILLIC_OLD, 0 },
\r
71 { "Mac Ukraine", CS_MAC_UKRAINE, 1 },
\r
72 { "Mac VT100", CS_MAC_VT100, 1 },
\r
73 { "Mac VT100 (old)", CS_MAC_VT100_OLD, 0 },
\r
74 { "VISCII", CS_VISCII, 1 },
\r
75 { "HP ROMAN8", CS_HP_ROMAN8, 1 },
\r
76 { "DEC MCS", CS_DEC_MCS, 1 },
\r
77 { "UTF-8", CS_UTF8, 1 },
\r
80 const char *charset_to_localenc(int charset)
\r
84 for (i = 0; i < (int)lenof(localencs); i++)
\r
85 if (charset == localencs[i].charset)
\r
86 return localencs[i].name;
\r
88 return NULL; /* not found */
\r
91 int charset_from_localenc(const char *name)
\r
95 if ( (i = charset_from_mimeenc(name)) != CS_NONE)
\r
97 if ( (i = charset_from_xenc(name)) != CS_NONE)
\r
100 for (i = 0; i < (int)lenof(localencs); i++) {
\r
103 q = localencs[i].name;
\r
105 if (tolower((unsigned char)*p) != tolower((unsigned char)*q))
\r
110 return localencs[i].charset;
\r
113 return CS_NONE; /* not found */
\r
116 int charset_localenc_nth(int n)
\r
120 for (i = 0; i < (int)lenof(localencs); i++)
\r
121 if (localencs[i].return_in_enum && !n--)
\r
122 return localencs[i].charset;
\r
124 return CS_NONE; /* end of list */
\r