citadel/html.c

   1 /*
   2  * html.c -- Functions which handle translation between HTML and plain text
   3  * $Id$
   4  */
   5
   6 #include "sysdep.h"
   7 #include <stdlib.h>
   8 #include <unistd.h>
   9 #include <stdio.h>
  10 #include <fcntl.h>
  11 #include <signal.h>
  12 #include <time.h>
  13 #include <ctype.h>
  14 #include <string.h>
  15 #include <errno.h>
  16 #include <limits.h>
  17 #include <syslog.h>
  18 #include "citadel.h"
  19 #include "server.h"
  20 #include "control.h"
  21 #include "sysdep_decls.h"
  22 #include "support.h"
  23 #include "config.h"
  24 #include "msgbase.h"
  25 #include "tools.h"
  26 #include "room_ops.h"
  27 #include "html.h"
  28
  29
  30 /*
  31  * Convert HTML to plain text.
  32  *
  33  * inputmsg      = pointer to raw HTML message
  34  * screenwidth   = desired output screenwidth
  35  * do_citaformat = set to 1 to indent newlines with spaces
  36  */
  37 char *html_to_ascii(char *inputmsg, int screenwidth, int do_citaformat) {
  38         char inbuf[256];
  39         char outbuf[256];
  40         char tag[1024];
  41         int done_reading = 0;
  42         char *inptr;
  43         char *outptr;
  44         size_t outlen;
  45         int i, j, ch, did_out, rb;
  46         int nest = 0;           /* Bracket nesting level */
  47
  48         inptr = inputmsg;
  49         strcpy(inbuf, "");
  50         strcpy(outbuf, "");
  51
  52         outptr = mallok(strlen(inptr) + 256);
  53         if (outptr == NULL) return NULL;
  54         strcpy(outptr, "");
  55         outlen = 0;
  56
  57         do {
  58                 /* Fill the input buffer */
  59                 if ( (done_reading == 0) && (strlen(inbuf) < 128) ) {
  60
  61                         /* FIX ... genericize this */
  62                         ch = *inputmsg++;
  63                         if (ch > 0) {
  64                                 inbuf[strlen(inbuf)+1] = 0;
  65                                 inbuf[strlen(inbuf)] = ch;
  66                         }
  67                         else {
  68                                 done_reading = 1;
  69                         }
  70
  71                 }
  72
  73                 /* Do some parsing */
  74                 if (strlen(inbuf)>0) {
  75
  76                     /* Fold in all the spacing */
  77                     for (i=0; i<strlen(inbuf); ++i) {
  78                         if (inbuf[i]==10) inbuf[i]=32;
  79                         if (inbuf[i]==13) inbuf[i]=32;
  80                         if (inbuf[i]==9) inbuf[i]=32;
  81                         if ((inbuf[i]<32) || (inbuf[i]>126))
  82                                 strcpy(&inbuf[i], &inbuf[i+1]);
  83                         while ((inbuf[i]==32)&&(inbuf[i+1]==32))
  84                                 strcpy(&inbuf[i], &inbuf[i+1]);
  85                     }
  86
  87                     for (i=0; i<strlen(inbuf); ++i) {
  88
  89                         ch = inbuf[i];
  90
  91                         if (ch == '<') {
  92                                 ++nest;
  93                                 strcpy(tag, "");
  94                         }
  95
  96                         else if (ch == '>') {
  97                                 if (nest > 0) --nest;
  98
  99                                 if (!strcasecmp(tag, "P")) {
 100                                         strcat(outbuf, "\n\n");
 101                                 }
 102
 103                                 if (!strcasecmp(tag, "/DIV")) {
 104                                         strcat(outbuf, "\n\n");
 105                                 }
 106
 107                                 else if (!strcasecmp(tag, "H1")) {
 108                                         strcat(outbuf, "\n\n");
 109                                 }
 110
 111                                 else if (!strcasecmp(tag, "H2")) {
 112                                         strcat(outbuf, "\n\n");
 113                                 }
 114
 115                                 else if (!strcasecmp(tag, "H3")) {
 116                                         strcat(outbuf, "\n\n");
 117                                 }
 118
 119                                 else if (!strcasecmp(tag, "H4")) {
 120                                         strcat(outbuf, "\n\n");
 121                                 }
 122
 123                                 else if (!strcasecmp(tag, "/H1")) {
 124                                         strcat(outbuf, "\n");
 125                                 }
 126
 127                                 else if (!strcasecmp(tag, "/H2")) {
 128                                         strcat(outbuf, "\n");
 129                                 }
 130
 131                                 else if (!strcasecmp(tag, "/H3")) {
 132                                         strcat(outbuf, "\n");
 133                                 }
 134
 135                                 else if (!strcasecmp(tag, "/H4")) {
 136                                         strcat(outbuf, "\n");
 137                                 }
 138
 139                                 else if (!strcasecmp(tag, "HR")) {
 140                                         strcat(outbuf, "\n ");
 141                                         for (j=0; j<screenwidth-2; ++j)
 142                                                 strcat(outbuf, "-");
 143                                         strcat(outbuf, "\n");
 144                                 }
 145
 146                                 else if (!strcasecmp(tag, "BR")) {
 147                                         strcat(outbuf, "\n");
 148                                 }
 149
 150                                 else if (!strcasecmp(tag, "TR")) {
 151                                         strcat(outbuf, "\n");
 152                                 }
 153
 154                                 else if (!strcasecmp(tag, "/TABLE")) {
 155                                         strcat(outbuf, "\n");
 156                                 }
 157
 158                         }
 159
 160                         else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
 161                                 tag[strlen(tag)+1] = 0;
 162                                 tag[strlen(tag)] = ch;
 163                         }
 164
 165                         else if (!nest) {
 166                                 outbuf[strlen(outbuf)+1] = 0;
 167                                 outbuf[strlen(outbuf)] = ch;
 168                         }
 169                     }
 170                     strcpy(inbuf, &inbuf[i]);
 171                 }
 172
 173                 /* Convert &; tags to the forbidden characters */
 174                 if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
 175
 176                         if (!strncasecmp(&outbuf[i], "&nbsp;", 6)) {
 177                                 outbuf[i] = ' ';
 178                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 179                         }
 180
 181                         else if (!strncasecmp(&outbuf[i], "&lt;", 4)) {
 182                                 outbuf[i] = '<';
 183                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 184                         }
 185
 186                         else if (!strncasecmp(&outbuf[i], "&gt;", 4)) {
 187                                 outbuf[i] = '>';
 188                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 189                         }
 190
 191                         else if (!strncasecmp(&outbuf[i], "&amp;", 5)) {
 192                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 193                         }
 194
 195                         else if (!strncasecmp(&outbuf[i], "&quot;", 6)) {
 196                                 outbuf[i] = '\"';
 197                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 198                         }
 199
 200                         else if (!strncasecmp(&outbuf[i], "&copy;", 6)) {
 201                                 outbuf[i] = '(';
 202                                 outbuf[i+1] = 'c';
 203                                 outbuf[i+2] = ')';
 204                                 strcpy(&outbuf[i+3], &outbuf[i+6]);
 205                         }
 206
 207                         else if (!strncasecmp(&outbuf[i], "&reg;", 5)) {
 208                                 outbuf[i] = '(';
 209                                 outbuf[i+1] = 'r';
 210                                 outbuf[i+2] = ')';
 211                                 strcpy(&outbuf[i+3], &outbuf[i+5]);
 212                         }
 213
 214                 }
 215
 216                 /* Make sure the output buffer is big enough */
 217                 if ((strlen(outptr) + strlen(outbuf) + 128) > outlen) {
 218                         outlen += 128;
 219                         outptr = realloc(outptr, outlen);
 220                 }
 221
 222                 /* Output any lines terminated with hard line breaks */
 223                 do {
 224                         did_out = 0;
 225                         if (strlen(outbuf)>0)
 226                             for (i = 0; i<strlen(outbuf); ++i) {
 227                                 if ( (i<(screenwidth-2)) && (outbuf[i]=='\n')) {
 228                                         strncat(outptr, outbuf, i+1);
 229                                         strcat(outptr, "\n");
 230                                         if (do_citaformat)
 231                                                 strcat(outptr, " ");
 232                                         strcpy(outbuf, &outbuf[i+1]);
 233                                         i = 0;
 234                                         did_out = 1;
 235                                 }
 236                         }
 237                 } while (did_out);
 238
 239                 /* Add soft line breaks */
 240                 if (strlen(outbuf) > (screenwidth - 2)) {
 241                         rb = (-1);
 242                         for (i=0; i<(screenwidth-2); ++i) {
 243                                 if (outbuf[i]==32) rb = i;
 244                         }
 245                         if (rb>=0) {
 246                                 strncat(outptr, outbuf, rb);
 247                                 strcat(outptr, "\n");
 248                                 if (do_citaformat)
 249                                         strcat(outptr, " ");
 250                                 strcpy(outbuf, &outbuf[rb+1]);
 251                         } else {
 252
 253                                 strncat(outptr, outbuf, screenwidth-2);
 254                                 strcat(outptr, "\n");
 255                                 if (do_citaformat)
 256                                         strcat(outptr, " ");
 257                                 strcpy(outbuf, &outbuf[screenwidth-2]);
 258                         }
 259                 }
 260
 261         } while (done_reading == 0);
 262
 263         strcat(outptr, outbuf);
 264         strcat(outptr, "\n");
 265
 266         return outptr;
 267
 268 }
 269