15 * Unix Unicode-handling routines.
18 int is_dbcs_leadbyte(int codepage, char byte)
20 return 0; /* we don't do DBCS */
23 int mb_to_wc(int codepage, int flags, char *mbstr, int mblen,
24 wchar_t *wcstr, int wclen)
26 if (codepage == DEFAULT_CODEPAGE) {
28 mbstate_t state = { 0 };
30 setlocale(LC_CTYPE, "");
33 size_t i = mbrtowc(wcstr+n, mbstr, (size_t)mblen, &state);
34 if (i == (size_t)-1 || i == (size_t)-2)
41 setlocale(LC_CTYPE, "C");
44 } else if (codepage == CS_NONE) {
48 wcstr[n] = 0xD800 | (mbstr[0] & 0xFF);
56 return charset_to_unicode(&mbstr, &mblen, wcstr, wclen, codepage,
60 int wc_to_mb(int codepage, int flags, wchar_t *wcstr, int wclen,
61 char *mbstr, int mblen, char *defchr, int *defused)
63 /* FIXME: we should remove the defused param completely... */
67 if (codepage == DEFAULT_CODEPAGE) {
68 char output[MB_LEN_MAX];
69 mbstate_t state = { 0 };
72 setlocale(LC_CTYPE, "");
75 int i = wcrtomb(output, wcstr[0], &state);
76 if (i == (size_t)-1 || i > n - mblen)
78 memcpy(mbstr+n, output, i);
84 setlocale(LC_CTYPE, "C");
87 } else if (codepage == CS_NONE) {
89 while (wclen > 0 && n < mblen) {
90 if (*wcstr >= 0xD800 && *wcstr < 0xD900)
91 mbstr[n++] = (*wcstr & 0xFF);
99 return charset_from_unicode(&wcstr, &wclen, mbstr, mblen, codepage,
104 void init_ucs(int font_charset)
109 * In the platform-independent parts of the code, font_codepage
110 * is used only for system DBCS support - which we don't
111 * support at all. So we set this to something which will never
117 * line_codepage should be decoded from the specification in
120 line_codepage = charset_from_mimeenc(cfg.line_codepage);
121 if (line_codepage == CS_NONE)
122 line_codepage = charset_from_xenc(cfg.line_codepage);
125 * If line_codepage is _still_ CS_NONE, we assume we're using
126 * the font's own encoding. This has been passed in to us, so
127 * we use that. If it's still CS_NONE after _that_ - i.e. the
128 * font we were given had an incomprehensible charset - then we
129 * fall back to using the D800 page.
131 if (line_codepage == CS_NONE)
132 line_codepage = font_charset;
135 * Set up unitab_line, by translating each individual character
136 * in the line codepage into Unicode.
138 for (i = 0; i < 256; i++) {
145 if (line_codepage == CS_NONE)
146 unitab_line[i] = 0xD800 | i;
147 else if (1 == charset_to_unicode(&p, &len, wc, 1, line_codepage,
149 unitab_line[i] = wc[0];
151 unitab_line[i] = 0xFFFD;
155 * Set up unitab_xterm. This is the same as unitab_line except
156 * in the line-drawing regions, where it follows the Unicode
159 * (Note that the strange X encoding of line-drawing characters
160 * in the bottom 32 glyphs of ISO8859-1 fonts is taken care of
161 * by the font encoding, which will spot such a font and act as
162 * if it were in a variant encoding of ISO8859-1.)
164 for (i = 0; i < 256; i++) {
165 static const wchar_t unitab_xterm_std[32] = {
166 0x2666, 0x2592, 0x2409, 0x240c, 0x240d, 0x240a, 0x00b0, 0x00b1,
167 0x2424, 0x240b, 0x2518, 0x2510, 0x250c, 0x2514, 0x253c, 0x23ba,
168 0x23bb, 0x2500, 0x23bc, 0x23bd, 0x251c, 0x2524, 0x2534, 0x252c,
169 0x2502, 0x2264, 0x2265, 0x03c0, 0x2260, 0x00a3, 0x00b7, 0x0020
171 if (i >= 0x5F && i < 0x7F)
172 unitab_xterm[i] = unitab_xterm_std[i & 0x1F];
174 unitab_xterm[i] = unitab_line[i];
178 * Set up unitab_scoacs. The SCO Alternate Character Set is
181 for (i = 0; i < 256; i++) {
188 if (1 == charset_to_unicode(&p, &len, wc, 1, CS_CP437, NULL, L"", 0))
189 unitab_scoacs[i] = wc[0];
191 unitab_scoacs[i] = 0xFFFD;
195 * Find the control characters in the line codepage. For
196 * direct-to-font mode using the D800 hack, we assume 00-1F and
197 * 7F are controls, but allow 80-9F through. (It's as good a
198 * guess as anything; and my bet is that half the weird fonts
199 * used in this way will be IBM or MS code pages anyway.)
201 for (i = 0; i < 256; i++) {
202 int lineval = unitab_line[i];
203 if (lineval < ' ' || (lineval >= 0x7F && lineval < 0xA0) ||
204 (lineval >= 0xD800 && lineval < 0xD820) || (lineval == 0xD87F))
207 unitab_ctrl[i] = 0xFF;