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,
100 NULL, defchr?defchr:NULL, defchr?1:0);
105 * Return value is TRUE if pterm is to run in direct-to-font mode.
107 int init_ucs(int font_charset)
112 * In the platform-independent parts of the code, font_codepage
113 * is used only for system DBCS support - which we don't
114 * support at all. So we set this to something which will never
120 * line_codepage should be decoded from the specification in
123 line_codepage = charset_from_mimeenc(cfg.line_codepage);
124 if (line_codepage == CS_NONE)
125 line_codepage = charset_from_xenc(cfg.line_codepage);
128 * If line_codepage is _still_ CS_NONE, we assume we're using
129 * the font's own encoding. This has been passed in to us, so
130 * we use that. If it's still CS_NONE after _that_ - i.e. the
131 * font we were given had an incomprehensible charset - then we
132 * fall back to using the D800 page.
134 if (line_codepage == CS_NONE)
135 line_codepage = font_charset;
137 if (line_codepage == CS_NONE)
141 * Set up unitab_line, by translating each individual character
142 * in the line codepage into Unicode.
144 for (i = 0; i < 256; i++) {
151 if (line_codepage == CS_NONE)
152 unitab_line[i] = 0xD800 | i;
153 else if (1 == charset_to_unicode(&p, &len, wc, 1, line_codepage,
155 unitab_line[i] = wc[0];
157 unitab_line[i] = 0xFFFD;
161 * Set up unitab_xterm. This is the same as unitab_line except
162 * in the line-drawing regions, where it follows the Unicode
165 * (Note that the strange X encoding of line-drawing characters
166 * in the bottom 32 glyphs of ISO8859-1 fonts is taken care of
167 * by the font encoding, which will spot such a font and act as
168 * if it were in a variant encoding of ISO8859-1.)
170 for (i = 0; i < 256; i++) {
171 static const wchar_t unitab_xterm_std[32] = {
172 0x2666, 0x2592, 0x2409, 0x240c, 0x240d, 0x240a, 0x00b0, 0x00b1,
173 0x2424, 0x240b, 0x2518, 0x2510, 0x250c, 0x2514, 0x253c, 0x23ba,
174 0x23bb, 0x2500, 0x23bc, 0x23bd, 0x251c, 0x2524, 0x2534, 0x252c,
175 0x2502, 0x2264, 0x2265, 0x03c0, 0x2260, 0x00a3, 0x00b7, 0x0020
177 if (i >= 0x5F && i < 0x7F)
178 unitab_xterm[i] = unitab_xterm_std[i & 0x1F];
180 unitab_xterm[i] = unitab_line[i];
184 * Set up unitab_scoacs. The SCO Alternate Character Set is
187 for (i = 0; i < 256; i++) {
194 if (1 == charset_to_unicode(&p, &len, wc, 1, CS_CP437, NULL, L"", 0))
195 unitab_scoacs[i] = wc[0];
197 unitab_scoacs[i] = 0xFFFD;
201 * Find the control characters in the line codepage. For
202 * direct-to-font mode using the D800 hack, we assume 00-1F and
203 * 7F are controls, but allow 80-9F through. (It's as good a
204 * guess as anything; and my bet is that half the weird fonts
205 * used in this way will be IBM or MS code pages anyway.)
207 for (i = 0; i < 256; i++) {
208 int lineval = unitab_line[i];
209 if (lineval < ' ' || (lineval >= 0x7F && lineval < 0xA0) ||
210 (lineval >= 0xD800 && lineval < 0xD820) || (lineval == 0xD87F))
213 unitab_ctrl[i] = 0xFF;