src/charsets.c

   1 #include <config.h>
   2
   3 #ifdef HAVE_CHARSET
   4 #include <stdlib.h>
   5 #include <stdio.h>
   6 #include <string.h>
   7 #include <iconv.h>
   8
   9 #include "global.h"
  10 #include "charsets.h"
  11
  12 int n_codepages = 0;
  13
  14 struct codepage_desc *codepages;
  15
  16 uchar conv_displ[256];
  17 uchar conv_input[256];
  18 uchar printable[256];
  19
  20 int load_codepages_list(void)
  21 {
  22     int result = -1;
  23     FILE *f;
  24     char *fname;
  25     char buf[256];
  26     extern char* mc_home;
  27     extern int display_codepage;
  28     char * default_codepage = NULL;
  29
  30     fname = concat_dir_and_file (mc_home, CHARSETS_INDEX);
  31     if ( !( f = fopen( fname, "r" ) ) ) {
  32         fprintf (stderr, _("Warning: file %s not found\n"), fname);
  33         g_free (fname);
  34         return -1;
  35     }
  36     g_free (fname);
  37
  38     for ( n_codepages=0; fgets( buf, sizeof (buf), f ); )
  39         if ( buf[0] != '\n' && buf[0] != '\0' && buf [0] != '#' )
  40             ++n_codepages;
  41     rewind( f );
  42
  43     codepages = calloc( n_codepages + 1, sizeof(struct codepage_desc) );
  44
  45     for( n_codepages = 0; fgets( buf, sizeof buf, f ); ) {
  46         /* split string into id and cpname */
  47         char *p = buf;
  48         int buflen = strlen( buf );
  49
  50         if ( *p == '\n' || *p == '\0' || *p == '#')
  51             continue;
  52
  53         if ( buflen > 0 && buf[ buflen - 1 ] == '\n' )
  54             buf[ buflen - 1 ] = '\0';
  55         while ( *p != '\t' && *p != ' ' && *p != '\0' )
  56             ++p;
  57         if ( *p == '\0' )
  58             goto fail;
  59
  60         *p++ = 0;
  61
  62         while ( *p == '\t' || *p == ' ' )
  63             ++p;
  64         if ( *p == '\0' )
  65             goto fail;
  66
  67         if (strcmp (buf, "default") == 0) {
  68             default_codepage = strdup (p);
  69             continue;
  70         }
  71
  72         codepages[n_codepages].id = strdup( buf  );
  73         codepages[n_codepages].name = strdup( p );
  74         ++n_codepages;
  75     }
  76
  77     if (default_codepage) {
  78         display_codepage = get_codepage_index (default_codepage);
  79         free (default_codepage);
  80     }
  81
  82     result = n_codepages;
  83 fail:
  84     fclose( f );
  85     return result;
  86 }
  87
  88 #define OTHER_8BIT "Other_8_bit"
  89
  90 char *get_codepage_id( int n )
  91 {
  92     return (n < 0) ? OTHER_8BIT : codepages[ n ].id;
  93 }
  94
  95 int get_codepage_index( const char *id )
  96 {
  97     int i;
  98     if (strcmp( id, OTHER_8BIT ) == 0)
  99         return -1;
 100     for ( i=0; codepages[ i ].id; ++i )
 101         if (strcmp( id, codepages[ i ].id ) == 0)
 102             return i;
 103     return -1;
 104 }
 105
 106 static char translate_character( iconv_t cd, char c )
 107 {
 108     char outbuf[4], *obuf;
 109     size_t ibuflen, obuflen, count;
 110
 111     char *ibuf = &c;
 112     obuf = outbuf;
 113     ibuflen = 1; obuflen = 4;
 114
 115     count = iconv(cd, &ibuf, &ibuflen, &obuf, &obuflen);
 116     if (count >= 0 && ibuflen == 0)
 117         return outbuf[0];
 118
 119     return UNKNCHAR;
 120 }
 121
 122 char errbuf[255];
 123
 124 /*
 125  * FIXME: This assumes that ASCII is always the first encoding
 126  * in mc.charsets
 127  */
 128 #define CP_ASCII 0
 129
 130 char* init_printable_table( int cpdisplay )
 131 {
 132     int i;
 133
 134     /* Fill printable characters table */
 135     for (i=0; i<=127; ++i)
 136         printable[i] = (i > 31 && i != 127);
 137
 138     for (i=128; i<=255; ++i) {
 139         printable[i] = 1;
 140     }
 141
 142     return NULL;
 143 }
 144
 145 char* init_translation_table( int cpsource, int cpdisplay )
 146 {
 147     int i;
 148     iconv_t cd;
 149     char *cpsour, *cpdisp;
 150
 151     /* Fill inpit <-> display tables */
 152
 153     if (cpsource < 0 || cpdisplay < 0 || cpsource == cpdisplay) {
 154         for (i=0; i<=255; ++i) {
 155             conv_displ[i] = i;
 156             conv_input[i] = i;
 157         }
 158         return NULL;
 159     }
 160
 161     for (i=0; i<=127; ++i) {
 162         conv_displ[i] = i;
 163         conv_input[i] = i;
 164     }
 165
 166     cpsour = codepages[ cpsource ].id;
 167     cpdisp = codepages[ cpdisplay ].id;
 168
 169     /* display <- inpit table */
 170
 171     cd = iconv_open( cpdisp, cpsour );
 172     if (cd == (iconv_t) -1) {
 173         sprintf( errbuf, _("Cannot translate from %s to %s"), cpsour, cpdisp );
 174         return errbuf;
 175     }
 176
 177     for (i=128; i<=255; ++i)
 178         conv_displ[i] = translate_character( cd, i );
 179
 180     iconv_close( cd );
 181
 182     /* inpit <- display table */
 183
 184     cd = iconv_open( cpsour, cpdisp );
 185     if (cd == (iconv_t) -1) {
 186         sprintf( errbuf, _("Cannot translate from %s to %s"), cpdisp, cpsour );
 187         return errbuf;
 188     }
 189
 190     for (i=128; i<=255; ++i) {
 191         uchar ch;
 192         ch = translate_character( cd, i );
 193         conv_input[i] = (ch == UNKNCHAR) ? i : ch;
 194     }
 195
 196     iconv_close( cd );
 197
 198     return NULL;
 199 }
 200
 201 void convert_to_display( char *str )
 202 {
 203     while ((*str = conv_displ[ (uchar) *str ]))
 204         str++;
 205 }
 206
 207 void convert_from_input( char *str )
 208 {
 209     while ((*str = conv_input[ (uchar) *str ]))
 210         str++;
 211 }
 212 #endif /* HAVE_CHARSET */