mdw@git.distorted.org.uk Git - sgt/putty/blob - unix/uxucs.c

   1 #include <stdio.h>
   2 #include <stdlib.h>
   3 #include <ctype.h>
   4 #include <locale.h>
   5 #include <limits.h>
   6 #include <wchar.h>
   7
   8 #include <time.h>
   9
  10 #include "putty.h"
  11 #include "charset.h"
  12 #include "terminal.h"
  13 #include "misc.h"
  14
  15 /*
  16  * Unix Unicode-handling routines.
  17  */
  18
  19 int is_dbcs_leadbyte(int codepage, char byte)
  20 {
  21     return 0;                          /* we don't do DBCS */
  22 }
  23
  24 int mb_to_wc(int codepage, int flags, char *mbstr, int mblen,
  25              wchar_t *wcstr, int wclen)
  26 {
  27     if (codepage == DEFAULT_CODEPAGE) {
  28         int n = 0;
  29         mbstate_t state = { 0 };
  30
  31         setlocale(LC_CTYPE, "");
  32
  33         while (mblen > 0) {
  34             size_t i = mbrtowc(wcstr+n, mbstr, (size_t)mblen, &state);
  35             if (i == (size_t)-1 || i == (size_t)-2)
  36                 break;
  37             n++;
  38             mbstr += i;
  39             mblen -= i;
  40         }
  41
  42         setlocale(LC_CTYPE, "C");
  43
  44         return n;
  45     } else if (codepage == CS_NONE) {
  46         int n = 0;
  47
  48         while (mblen > 0) {
  49             wcstr[n] = 0xD800 | (mbstr[0] & 0xFF);
  50             n++;
  51             mbstr++;
  52             mblen--;
  53         }
  54
  55         return n;
  56     } else
  57         return charset_to_unicode(&mbstr, &mblen, wcstr, wclen, codepage,
  58                                   NULL, NULL, 0);
  59 }
  60
  61 int wc_to_mb(int codepage, int flags, wchar_t *wcstr, int wclen,
  62              char *mbstr, int mblen, char *defchr, int *defused,
  63              struct unicode_data *ucsdata)
  64 {
  65     /* FIXME: we should remove the defused param completely... */
  66     if (defused)
  67         *defused = 0;
  68
  69     if (codepage == DEFAULT_CODEPAGE) {
  70         char output[MB_LEN_MAX];
  71         mbstate_t state = { 0 };
  72         int n = 0;
  73
  74         setlocale(LC_CTYPE, "");
  75
  76         while (wclen > 0) {
  77             int i = wcrtomb(output, wcstr[0], &state);
  78             if (i == (size_t)-1 || i > n - mblen)
  79                 break;
  80             memcpy(mbstr+n, output, i);
  81             n += i;
  82             wcstr++;
  83             wclen--;
  84         }
  85
  86         setlocale(LC_CTYPE, "C");
  87
  88         return n;
  89     } else if (codepage == CS_NONE) {
  90         int n = 0;
  91         while (wclen > 0 && n < mblen) {
  92             if (*wcstr >= 0xD800 && *wcstr < 0xD900)
  93                 mbstr[n++] = (*wcstr & 0xFF);
  94             else if (defchr)
  95                 mbstr[n++] = *defchr;
  96             wcstr++;
  97             wclen--;
  98         }
  99         return n;
 100     } else {
 101         return charset_from_unicode(&wcstr, &wclen, mbstr, mblen, codepage,
 102                                     NULL, defchr?defchr:NULL, defchr?1:0);
 103     }
 104 }
 105
 106 /*
 107  * Return value is TRUE if pterm is to run in direct-to-font mode.
 108  */
 109 int init_ucs(struct unicode_data *ucsdata,
 110              char *linecharset, int font_charset, int vtmode)
 111 {
 112     int i, ret = 0;
 113
 114     /*
 115      * In the platform-independent parts of the code, font_codepage
 116      * is used only for system DBCS support - which we don't
 117      * support at all. So we set this to something which will never
 118      * be used.
 119      */
 120     ucsdata->font_codepage = -1;
 121
 122     /*
 123      * line_codepage should be decoded from the specification in
 124      * cfg.
 125      */
 126     ucsdata->line_codepage = decode_codepage(linecharset);
 127
 128     /*
 129      * If line_codepage is _still_ CS_NONE, we assume we're using
 130      * the font's own encoding. This has been passed in to us, so
 131      * we use that. If it's still CS_NONE after _that_ - i.e. the
 132      * font we were given had an incomprehensible charset - then we
 133      * fall back to using the D800 page.
 134      */
 135     if (ucsdata->line_codepage == CS_NONE)
 136         ucsdata->line_codepage = font_charset;
 137
 138     if (ucsdata->line_codepage == CS_NONE)
 139         ret = 1;
 140
 141     /*
 142      * Set up unitab_line, by translating each individual character
 143      * in the line codepage into Unicode.
 144      */
 145     for (i = 0; i < 256; i++) {
 146         char c[1], *p;
 147         wchar_t wc[1];
 148         int len;
 149         c[0] = i;
 150         p = c;
 151         len = 1;
 152         if (ucsdata->line_codepage == CS_NONE)
 153             ucsdata->unitab_line[i] = 0xD800 | i;
 154         else if (1 == charset_to_unicode(&p, &len, wc, 1,
 155                                          ucsdata->line_codepage,
 156                                          NULL, L"", 0))
 157             ucsdata->unitab_line[i] = wc[0];
 158         else
 159             ucsdata->unitab_line[i] = 0xFFFD;
 160     }
 161
 162     /*
 163      * Set up unitab_xterm. This is the same as unitab_line except
 164      * in the line-drawing regions, where it follows the Unicode
 165      * encoding.
 166      *
 167      * (Note that the strange X encoding of line-drawing characters
 168      * in the bottom 32 glyphs of ISO8859-1 fonts is taken care of
 169      * by the font encoding, which will spot such a font and act as
 170      * if it were in a variant encoding of ISO8859-1.)
 171      */
 172     for (i = 0; i < 256; i++) {
 173         static const wchar_t unitab_xterm_std[32] = {
 174             0x2666, 0x2592, 0x2409, 0x240c, 0x240d, 0x240a, 0x00b0, 0x00b1,
 175             0x2424, 0x240b, 0x2518, 0x2510, 0x250c, 0x2514, 0x253c, 0x23ba,
 176             0x23bb, 0x2500, 0x23bc, 0x23bd, 0x251c, 0x2524, 0x2534, 0x252c,
 177             0x2502, 0x2264, 0x2265, 0x03c0, 0x2260, 0x00a3, 0x00b7, 0x0020
 178         };
 179         static const wchar_t unitab_xterm_poorman[32] =
 180             L"*#****o~**+++++-----++++|****L. ";
 181
 182         const wchar_t *ptr;
 183
 184         if (vtmode == VT_POORMAN)
 185             ptr = unitab_xterm_poorman;
 186         else
 187             ptr = unitab_xterm_std;
 188
 189         if (i >= 0x5F && i < 0x7F)
 190             ucsdata->unitab_xterm[i] = ptr[i & 0x1F];
 191         else
 192             ucsdata->unitab_xterm[i] = ucsdata->unitab_line[i];
 193     }
 194
 195     /*
 196      * Set up unitab_scoacs. The SCO Alternate Character Set is
 197      * simply CP437.
 198      */
 199     for (i = 0; i < 256; i++) {
 200         char c[1], *p;
 201         wchar_t wc[1];
 202         int len;
 203         c[0] = i;
 204         p = c;
 205         len = 1;
 206         if (1 == charset_to_unicode(&p, &len, wc, 1, CS_CP437, NULL, L"", 0))
 207             ucsdata->unitab_scoacs[i] = wc[0];
 208         else
 209             ucsdata->unitab_scoacs[i] = 0xFFFD;
 210     }
 211
 212     /*
 213      * Find the control characters in the line codepage. For
 214      * direct-to-font mode using the D800 hack, we assume 00-1F and
 215      * 7F are controls, but allow 80-9F through. (It's as good a
 216      * guess as anything; and my bet is that half the weird fonts
 217      * used in this way will be IBM or MS code pages anyway.)
 218      */
 219     for (i = 0; i < 256; i++) {
 220         int lineval = ucsdata->unitab_line[i];
 221         if (lineval < ' ' || (lineval >= 0x7F && lineval < 0xA0) ||
 222             (lineval >= 0xD800 && lineval < 0xD820) || (lineval == 0xD87F))
 223             ucsdata->unitab_ctrl[i] = i;
 224         else
 225             ucsdata->unitab_ctrl[i] = 0xFF;
 226     }
 227
 228     return ret;
 229 }
 230
 231 const char *cp_name(int codepage)
 232 {
 233     if (codepage == CS_NONE)
 234         return "Use font encoding";
 235     return charset_to_localenc(codepage);
 236 }
 237
 238 const char *cp_enumerate(int index)
 239 {
 240     int charset;
 241     if (index == 0)
 242         return "Use font encoding";
 243     charset = charset_localenc_nth(index-1);
 244     if (charset == CS_NONE)
 245         return NULL;
 246     return charset_to_localenc(charset);
 247 }
 248
 249 int decode_codepage(char *cp_name)
 250 {
 251     if (!*cp_name)
 252         return CS_NONE;                /* use font encoding */
 253     return charset_from_localenc(cp_name);
 254 }