citadel/html.c

   1 /*
   2  * $Id$
   3  *
   4  * Functions which handle translation between HTML and plain text
   5  * Copyright (c) 2000-2001 by Art Cancro and others.   This program is
   6  * released under the terms of the GNU General Public License.
   7  */
   8
   9 #ifdef DLL_EXPORT
  10 #define IN_LIBCIT
  11 #endif
  12
  13 #include "sysdep.h"
  14 #include <stdlib.h>
  15 #include <unistd.h>
  16 #include <stdio.h>
  17 #include <fcntl.h>
  18 #include <signal.h>
  19
  20 #if TIME_WITH_SYS_TIME
  21 # include <sys/time.h>
  22 # include <time.h>
  23 #else
  24 # if HAVE_SYS_TIME_H
  25 #  include <sys/time.h>
  26 # else
  27 #  include <time.h>
  28 # endif
  29 #endif
  30
  31 #include <ctype.h>
  32 #include <string.h>
  33 #include <errno.h>
  34 #include <limits.h>
  35 #include <syslog.h>
  36 #include "citadel.h"
  37 #include "server.h"
  38 #include "dynloader.h"
  39 #include "control.h"
  40 #include "sysdep_decls.h"
  41 #include "support.h"
  42 #include "config.h"
  43 #include "msgbase.h"
  44 #include "tools.h"
  45 #include "room_ops.h"
  46 #include "html.h"
  47
  48
  49 /*
  50  * Convert HTML to plain text.
  51  *
  52  * inputmsg      = pointer to raw HTML message
  53  * screenwidth   = desired output screenwidth
  54  * do_citaformat = set to 1 to indent newlines with spaces
  55  */
  56 char *html_to_ascii(char *inputmsg, int screenwidth, int do_citaformat) {
  57         char inbuf[SIZ];
  58         char outbuf[SIZ];
  59         char tag[1024];
  60         int done_reading = 0;
  61         char *inptr;
  62         char *outptr;
  63         size_t outlen;
  64         int i, j, ch, did_out, rb;
  65         int nest = 0;           /* Bracket nesting level */
  66
  67         inptr = inputmsg;
  68         strcpy(inbuf, "");
  69         strcpy(outbuf, "");
  70
  71         outptr = mallok(strlen(inptr) + SIZ);
  72         if (outptr == NULL) return NULL;
  73         strcpy(outptr, "");
  74         outlen = 0;
  75
  76         do {
  77                 /* Fill the input buffer */
  78                 if ( (done_reading == 0) && (strlen(inbuf) < 128) ) {
  79
  80                         ch = *inputmsg++;
  81                         if (ch > 0) {
  82                                 inbuf[strlen(inbuf)+1] = 0;
  83                                 inbuf[strlen(inbuf)] = ch;
  84                         }
  85                         else {
  86                                 done_reading = 1;
  87                         }
  88
  89                 }
  90
  91                 /* Do some parsing */
  92                 if (strlen(inbuf)>0) {
  93
  94                     /* Fold in all the spacing */
  95                     for (i=0; i<strlen(inbuf); ++i) {
  96                         if (inbuf[i]==10) inbuf[i]=32;
  97                         if (inbuf[i]==13) inbuf[i]=32;
  98                         if (inbuf[i]==9) inbuf[i]=32;
  99                         if ((inbuf[i]<32) || (inbuf[i]>126))
 100                                 strcpy(&inbuf[i], &inbuf[i+1]);
 101                     }
 102                     for (i=0; i<strlen(inbuf); ++i) {
 103                         while ((inbuf[i]==32)&&(inbuf[i+1]==32))
 104                                 strcpy(&inbuf[i], &inbuf[i+1]);
 105                     }
 106
 107                     for (i=0; i<strlen(inbuf); ++i) {
 108
 109                         ch = inbuf[i];
 110
 111                         if (ch == '<') {
 112                                 ++nest;
 113                                 strcpy(tag, "");
 114                         }
 115
 116                         else if (ch == '>') {
 117                                 if (nest > 0) --nest;
 118
 119                                 if (!strcasecmp(tag, "P")) {
 120                                         strcat(outbuf, "\n\n");
 121                                 }
 122
 123                                 if (!strcasecmp(tag, "/DIV")) {
 124                                         strcat(outbuf, "\n\n");
 125                                 }
 126
 127                                 else if (!strcasecmp(tag, "H1")) {
 128                                         strcat(outbuf, "\n\n");
 129                                 }
 130
 131                                 else if (!strcasecmp(tag, "H2")) {
 132                                         strcat(outbuf, "\n\n");
 133                                 }
 134
 135                                 else if (!strcasecmp(tag, "H3")) {
 136                                         strcat(outbuf, "\n\n");
 137                                 }
 138
 139                                 else if (!strcasecmp(tag, "H4")) {
 140                                         strcat(outbuf, "\n\n");
 141                                 }
 142
 143                                 else if (!strcasecmp(tag, "/H1")) {
 144                                         strcat(outbuf, "\n");
 145                                 }
 146
 147                                 else if (!strcasecmp(tag, "/H2")) {
 148                                         strcat(outbuf, "\n");
 149                                 }
 150
 151                                 else if (!strcasecmp(tag, "/H3")) {
 152                                         strcat(outbuf, "\n");
 153                                 }
 154
 155                                 else if (!strcasecmp(tag, "/H4")) {
 156                                         strcat(outbuf, "\n");
 157                                 }
 158
 159                                 else if (!strcasecmp(tag, "HR")) {
 160                                         strcat(outbuf, "\n ");
 161                                         for (j=0; j<screenwidth-2; ++j)
 162                                                 strcat(outbuf, "-");
 163                                         strcat(outbuf, "\n");
 164                                 }
 165
 166                                 else if (!strcasecmp(tag, "BR")) {
 167                                         strcat(outbuf, "\n");
 168                                 }
 169
 170                                 else if (!strcasecmp(tag, "TR")) {
 171                                         strcat(outbuf, "\n");
 172                                 }
 173
 174                                 else if (!strcasecmp(tag, "/TABLE")) {
 175                                         strcat(outbuf, "\n");
 176                                 }
 177
 178                         }
 179
 180                         else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
 181                                 tag[strlen(tag)+1] = 0;
 182                                 tag[strlen(tag)] = ch;
 183                         }
 184
 185                         else if (!nest) {
 186                                 outbuf[strlen(outbuf)+1] = 0;
 187                                 outbuf[strlen(outbuf)] = ch;
 188                         }
 189                     }
 190                     strcpy(inbuf, &inbuf[i]);
 191                 }
 192
 193                 /* Convert &; tags to the forbidden characters */
 194                 if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
 195
 196                         if (!strncasecmp(&outbuf[i], "&nbsp;", 6)) {
 197                                 outbuf[i] = ' ';
 198                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 199                         }
 200
 201                         else if (!strncasecmp(&outbuf[i], "&lt;", 4)) {
 202                                 outbuf[i] = '<';
 203                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 204                         }
 205
 206                         else if (!strncasecmp(&outbuf[i], "&gt;", 4)) {
 207                                 outbuf[i] = '>';
 208                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 209                         }
 210
 211                         else if (!strncasecmp(&outbuf[i], "&amp;", 5)) {
 212                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 213                         }
 214
 215                         else if (!strncasecmp(&outbuf[i], "&quot;", 6)) {
 216                                 outbuf[i] = '\"';
 217                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 218                         }
 219
 220                         else if (!strncasecmp(&outbuf[i], "&copy;", 6)) {
 221                                 outbuf[i] = '(';
 222                                 outbuf[i+1] = 'c';
 223                                 outbuf[i+2] = ')';
 224                                 strcpy(&outbuf[i+3], &outbuf[i+6]);
 225                         }
 226
 227                         else if (!strncasecmp(&outbuf[i], "&reg;", 5)) {
 228                                 outbuf[i] = '(';
 229                                 outbuf[i+1] = 'r';
 230                                 outbuf[i+2] = ')';
 231                                 strcpy(&outbuf[i+3], &outbuf[i+5]);
 232                         }
 233
 234                 }
 235
 236                 /* Make sure the output buffer is big enough */
 237                 if ((strlen(outptr) + strlen(outbuf) + 128) > outlen) {
 238                         outlen += 128;
 239                         outptr = realloc(outptr, outlen);
 240                 }
 241
 242                 /* Output any lines terminated with hard line breaks */
 243                 do {
 244                         did_out = 0;
 245                         if (strlen(outbuf)>0)
 246                             for (i = 0; i<strlen(outbuf); ++i) {
 247                                 if ( (i<(screenwidth-2)) && (outbuf[i]=='\n')) {
 248                                         strncat(outptr, outbuf, i+1);
 249                                         /* strcat(outptr, "\n"); */
 250                                         if (do_citaformat)
 251                                                 strcat(outptr, " ");
 252                                         strcpy(outbuf, &outbuf[i+1]);
 253                                         i = 0;
 254                                         did_out = 1;
 255                                 }
 256                         }
 257                 } while (did_out);
 258
 259                 /* Add soft line breaks */
 260                 if (strlen(outbuf) > (screenwidth - 2)) {
 261                         rb = (-1);
 262                         for (i=0; i<(screenwidth-2); ++i) {
 263                                 if (outbuf[i]==32) rb = i;
 264                         }
 265                         if (rb>=0) {
 266                                 strncat(outptr, outbuf, rb);
 267                                 strcat(outptr, "\n");
 268                                 if (do_citaformat)
 269                                         strcat(outptr, " ");
 270                                 strcpy(outbuf, &outbuf[rb+1]);
 271                         } else {
 272
 273                                 strncat(outptr, outbuf, screenwidth-2);
 274                                 strcat(outptr, "\n");
 275                                 if (do_citaformat)
 276                                         strcat(outptr, " ");
 277                                 strcpy(outbuf, &outbuf[screenwidth-2]);
 278                         }
 279                 }
 280
 281         } while (done_reading == 0);
 282
 283         strcat(outptr, outbuf);
 284         striplt(outptr);
 285         if (outptr[strlen(outptr)-1] != '\n') strcat(outptr, "\n");
 286
 287         return outptr;
 288
 289 }