common/conv.c

   1 /*-
   2  * Copyright (c) 1993, 1994
   3  *      The Regents of the University of California.  All rights reserved.
   4  * Copyright (c) 1993, 1994, 1995, 1996
   5  *      Keith Bostic.  All rights reserved.
   6  *
   7  * See the LICENSE file for redistribution information.
   8  */
   9
  10 #include "config.h"
  11
  12 #ifndef lint
  13 static const char sccsid[] = "$Id: conv.c,v 1.7 2000/08/27 17:15:04 skimo Exp $ (Berkeley) $Date: 2000/08/27 17:15:04 $";
  14 #endif /* not lint */
  15
  16 #include <sys/types.h>
  17 #include <sys/queue.h>
  18 #include <sys/time.h>
  19
  20 #include <bitstring.h>
  21 #include <errno.h>
  22 #include <limits.h>
  23 #include <stdio.h>
  24 #include <stdlib.h>
  25 #include <string.h>
  26 #include <unistd.h>
  27
  28 #include "common.h"
  29
  30 int
  31 default_char2int(CONV *conv, const char * str, ssize_t len, CHAR_T **tostr, size_t *tolen, size_t *blen)
  32 {
  33     int i;
  34
  35     BINC_RETW(NULL, *tostr, *blen, len);
  36
  37     *tolen = len;
  38     for (i = 0; i < len; ++i)
  39         (*tostr)[i] = (u_char) str[i];
  40
  41     return 0;
  42 }
  43
  44 int
  45 default_int2char(CONV *conv, const CHAR_T * str, ssize_t len, char **tostr, size_t *tolen, size_t *blen)
  46 {
  47     int i;
  48
  49     BINC_RET(NULL, *tostr, *blen, len);
  50
  51     *tolen = len;
  52     for (i = 0; i < len; ++i)
  53         (*tostr)[i] = str[i];
  54
  55     return 0;
  56 }
  57
  58 int
  59 default_int2disp(CONV *conv, const CHAR_T * str, ssize_t len, char **tostr, size_t *tolen, size_t *blen)
  60 {
  61     int i, j;
  62
  63     BINC_RET(NULL, *tostr, *blen, len * 2);
  64
  65     for (i = 0, j = 0; i < len; ++i)
  66         if (CHAR_WIDTH(NULL, str[i]) > 1) {
  67             (*tostr)[j++] = '[';
  68             (*tostr)[j++] = ']';
  69         } else
  70             (*tostr)[j++] = str[i];
  71     *tolen = j;
  72
  73     return 0;
  74 }
  75
  76 int
  77 gb2int(CONV *conv, const char * str, ssize_t len, CHAR_T **tostr, size_t *tolen, size_t *blen)
  78 {
  79     int i, j;
  80
  81     BINC_RETW(NULL, *tostr, *blen, len);
  82
  83     for (i = 0, j = 0; i < len; ++i) {
  84         if (str[i] & 0x80) {
  85             if (i+1 < len && str[i+1] & 0x80) {
  86                 (*tostr)[j++] = INT9494(F_GB,str[i]&0x7F,str[i+1]&0x7F);
  87                 ++i;
  88             } else {
  89                 (*tostr)[j++] = INTILL(str[i]);
  90             }
  91         } else
  92             (*tostr)[j++] = str[i];
  93     }
  94     *tolen = j;
  95
  96     return 0;
  97 }
  98
  99 int
 100 int2gb(CONV *conv, const CHAR_T * str, ssize_t len, char **tostr, size_t *tolen, size_t *blen)
 101 {
 102     int i, j;
 103
 104     BINC_RET(NULL, *tostr, *blen, len * 2);
 105
 106     for (i = 0, j = 0; i < len; ++i) {
 107         if (INTIS9494(str[i])) {
 108             (*tostr)[j++] = INT9494R(str[i]) | 0x80;
 109             (*tostr)[j++] = INT9494C(str[i]) | 0x80;
 110         } else {
 111             (*tostr)[j++] = str[i] & 0xFF;
 112         }
 113     }
 114     *tolen = j;
 115
 116     return 0;
 117 }
 118
 119 int
 120 utf82int(CONV *conv, const char * str, ssize_t len, CHAR_T **tostr, size_t *tolen, size_t *blen)
 121 {
 122     int i, j;
 123     CHAR_T c;
 124
 125     BINC_RETW(NULL, *tostr, *blen, len);
 126
 127     for (i = 0, j = 0; i < len; ++i) {
 128         if (str[i] & 0x80) {
 129             if ((str[i] & 0xe0) == 0xc0 && i+1 < len && str[i+1] & 0x80) {
 130                 c = (str[i] & 0x1f) << 6;
 131                 c |= (str[i+1] & 0x3f);
 132                 (*tostr)[j++] = c;
 133                 ++i;
 134             } else if ((str[i] & 0xf0) == 0xe0 && i+2 < len &&
 135                         str[i+1] & 0x80 && str[i+2] & 0x80) {
 136                 c = (str[i] & 0xf) << 12;
 137                 c |= (str[i+1] & 0x3f) << 6;
 138                 c |= (str[i+2] & 0x3f);
 139                 (*tostr)[j++] = c;
 140                 i += 2;
 141             } else {
 142                 (*tostr)[j++] = INTILL(str[i]);
 143             }
 144         } else
 145             (*tostr)[j++] = str[i];
 146     }
 147     *tolen = j;
 148
 149     return 0;
 150 }
 151
 152 int
 153 int2utf8(CONV *conv, const CHAR_T * str, ssize_t len, char **tostr, size_t *tolen, size_t *blen)
 154 {
 155     BINC_RET(NULL, *tostr, *blen, len * 3);
 156
 157     *tolen = ucs2utf8(str, len, *tostr);
 158
 159     return 0;
 160 }
 161
 162
 163 CONV default_conv = { 0, 0, default_char2int, default_int2char,
 164                       default_char2int, default_int2char, default_int2disp };
 165 CONV gb_conv = { 0, 0, default_char2int, default_int2char,
 166                       gb2int, int2gb, default_int2disp };
 167 CONV utf8_conv = { 0, 0, default_char2int, default_int2char,
 168                       utf82int, int2utf8, default_int2disp };
 169
 170 void
 171 conv_init (SCR *orig, SCR *sp)
 172 {
 173     if (orig != NULL)
 174         sp->conv = orig->conv;
 175     else
 176         sp->conv = &default_conv;
 177 }
 178
 179 int
 180 conv_enc (SCR *sp, char *enc)
 181 {
 182     if (!*enc) {
 183         sp->conv = &default_conv;
 184         return 0;
 185     }
 186     if (!strcmp(enc,"GB")) {
 187         sp->conv = &gb_conv;
 188         return 0;
 189     }
 190     if (!strcmp(enc,"UTF-8")) {
 191         sp->conv = &utf8_conv;
 192         return 0;
 193     }
 194     return 1;
 195 }
 196