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