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 outptr_buffer_size;
  64         size_t output_len = 0;
  65         int i, j, ch, did_out, rb;
  66         int nest = 0;           /* Bracket nesting level */
  67
  68         inptr = inputmsg;
  69         strcpy(inbuf, "");
  70         strcpy(outbuf, "");
  71
  72         outptr_buffer_size = strlen(inptr) + SIZ;
  73         outptr = mallok(outptr_buffer_size);
  74         if (outptr == NULL) return NULL;
  75         strcpy(outptr, "");
  76         output_len = 0;
  77
  78         do {
  79                 /* Fill the input buffer */
  80                 if ( (done_reading == 0) && (strlen(inbuf) < 128) ) {
  81
  82                         ch = *inputmsg++;
  83                         if (ch > 0) {
  84                                 inbuf[strlen(inbuf)+1] = 0;
  85                                 inbuf[strlen(inbuf)] = ch;
  86                         }
  87                         else {
  88                                 done_reading = 1;
  89                         }
  90
  91                 }
  92
  93                 /* Do some parsing */
  94                 if (strlen(inbuf)>0) {
  95
  96                     /* Fold in all the spacing */
  97                     for (i=0; i<strlen(inbuf); ++i) {
  98                         if (inbuf[i]==10) inbuf[i]=32;
  99                         if (inbuf[i]==13) inbuf[i]=32;
 100                         if (inbuf[i]==9) inbuf[i]=32;
 101                         if ((inbuf[i]<32) || (inbuf[i]>126))
 102                                 strcpy(&inbuf[i], &inbuf[i+1]);
 103                     }
 104                     for (i=0; i<strlen(inbuf); ++i) {
 105                         while ((inbuf[i]==32)&&(inbuf[i+1]==32))
 106                                 strcpy(&inbuf[i], &inbuf[i+1]);
 107                     }
 108
 109                     for (i=0; i<strlen(inbuf); ++i) {
 110
 111                         ch = inbuf[i];
 112
 113                         if (ch == '<') {
 114                                 ++nest;
 115                                 strcpy(tag, "");
 116                         }
 117
 118                         else if (ch == '>') {
 119                                 if (nest > 0) --nest;
 120
 121                                 if (!strcasecmp(tag, "P")) {
 122                                         strcat(outbuf, "\n\n");
 123                                 }
 124
 125                                 if (!strcasecmp(tag, "/DIV")) {
 126                                         strcat(outbuf, "\n\n");
 127                                 }
 128
 129                                 else if (!strcasecmp(tag, "H1")) {
 130                                         strcat(outbuf, "\n\n");
 131                                 }
 132
 133                                 else if (!strcasecmp(tag, "H2")) {
 134                                         strcat(outbuf, "\n\n");
 135                                 }
 136
 137                                 else if (!strcasecmp(tag, "H3")) {
 138                                         strcat(outbuf, "\n\n");
 139                                 }
 140
 141                                 else if (!strcasecmp(tag, "H4")) {
 142                                         strcat(outbuf, "\n\n");
 143                                 }
 144
 145                                 else if (!strcasecmp(tag, "/H1")) {
 146                                         strcat(outbuf, "\n");
 147                                 }
 148
 149                                 else if (!strcasecmp(tag, "/H2")) {
 150                                         strcat(outbuf, "\n");
 151                                 }
 152
 153                                 else if (!strcasecmp(tag, "/H3")) {
 154                                         strcat(outbuf, "\n");
 155                                 }
 156
 157                                 else if (!strcasecmp(tag, "/H4")) {
 158                                         strcat(outbuf, "\n");
 159                                 }
 160
 161                                 else if (!strcasecmp(tag, "HR")) {
 162                                         strcat(outbuf, "\n ");
 163                                         for (j=0; j<screenwidth-2; ++j)
 164                                                 strcat(outbuf, "-");
 165                                         strcat(outbuf, "\n");
 166                                 }
 167
 168                                 else if (!strcasecmp(tag, "BR")) {
 169                                         strcat(outbuf, "\n");
 170                                 }
 171
 172                                 else if (!strcasecmp(tag, "TR")) {
 173                                         strcat(outbuf, "\n");
 174                                 }
 175
 176                                 else if (!strcasecmp(tag, "/TABLE")) {
 177                                         strcat(outbuf, "\n");
 178                                 }
 179
 180                         }
 181
 182                         else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
 183                                 tag[strlen(tag)+1] = 0;
 184                                 tag[strlen(tag)] = ch;
 185                         }
 186
 187                         else if (!nest) {
 188                                 outbuf[strlen(outbuf)+1] = 0;
 189                                 outbuf[strlen(outbuf)] = ch;
 190                         }
 191                     }
 192                     strcpy(inbuf, &inbuf[i]);
 193                 }
 194
 195                 /* Convert &; tags to the forbidden characters */
 196                 if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
 197
 198                         if (!strncasecmp(&outbuf[i], "&nbsp;", 6)) {
 199                                 outbuf[i] = ' ';
 200                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 201                         }
 202
 203                         else if (!strncasecmp(&outbuf[i], "&lt;", 4)) {
 204                                 outbuf[i] = '<';
 205                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 206                         }
 207
 208                         else if (!strncasecmp(&outbuf[i], "&gt;", 4)) {
 209                                 outbuf[i] = '>';
 210                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 211                         }
 212
 213                         else if (!strncasecmp(&outbuf[i], "&amp;", 5)) {
 214                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 215                         }
 216
 217                         else if (!strncasecmp(&outbuf[i], "&quot;", 6)) {
 218                                 outbuf[i] = '\"';
 219                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 220                         }
 221
 222                         else if (!strncasecmp(&outbuf[i], "&copy;", 6)) {
 223                                 outbuf[i] = '(';
 224                                 outbuf[i+1] = 'c';
 225                                 outbuf[i+2] = ')';
 226                                 strcpy(&outbuf[i+3], &outbuf[i+6]);
 227                         }
 228
 229                         else if (!strncasecmp(&outbuf[i], "&reg;", 5)) {
 230                                 outbuf[i] = '(';
 231                                 outbuf[i+1] = 'r';
 232                                 outbuf[i+2] = ')';
 233                                 strcpy(&outbuf[i+3], &outbuf[i+5]);
 234                         }
 235
 236                 }
 237
 238                 /* Make sure the output buffer is big enough */
 239                 if ((output_len + strlen(outbuf) + 128)
 240                    > outptr_buffer_size) {
 241                         outptr_buffer_size += 128;
 242                         outptr = realloc(outptr, outptr_buffer_size);
 243                 }
 244
 245                 /* Output any lines terminated with hard line breaks */
 246                 do {
 247                         did_out = 0;
 248                         if (strlen(outbuf)>0)
 249                             for (i = 0; i<strlen(outbuf); ++i) {
 250                                 if ( (i<(screenwidth-2)) && (outbuf[i]=='\n')) {
 251
 252                                         strncpy(&outptr[output_len],
 253                                                 outbuf, i+1);
 254                                         output_len += (i+1);
 255
 256                                         if (do_citaformat) {
 257                                                 strcpy(&outptr[output_len],
 258                                                         " ");
 259                                                 ++output_len;
 260                                         }
 261
 262                                         strcpy(outbuf, &outbuf[i+1]);
 263                                         i = 0;
 264                                         did_out = 1;
 265                                 }
 266                         }
 267                 } while (did_out);
 268
 269                 /* Add soft line breaks */
 270                 if (strlen(outbuf) > (screenwidth - 2)) {
 271                         rb = (-1);
 272                         for (i=0; i<(screenwidth-2); ++i) {
 273                                 if (outbuf[i]==32) rb = i;
 274                         }
 275                         if (rb>=0) {
 276                                 strncpy(&outptr[output_len], outbuf, rb);
 277                                 output_len += rb;
 278                                 strcpy(&outptr[output_len], "\n");
 279                                 output_len += 1;
 280                                 if (do_citaformat) {
 281                                         strcpy(&outptr[output_len], " ");
 282                                         ++output_len;
 283                                 }
 284                                 strcpy(outbuf, &outbuf[rb+1]);
 285                         } else {
 286                                 strncpy(&outptr[output_len], outbuf,
 287                                         screenwidth-2);
 288                                 output_len += (screenwidth-2);
 289                                 strcpy(&outptr[output_len], "\n");
 290                                 output_len += 1;
 291                                 if (do_citaformat) {
 292                                         strcpy(&outptr[output_len], " ");
 293                                         ++output_len;
 294                                 }
 295                                 strcpy(outbuf, &outbuf[screenwidth-2]);
 296                         }
 297                 }
 298
 299         } while (done_reading == 0);
 300
 301         strcpy(&outptr[output_len], outbuf);
 302         output_len += strlen(outbuf);
 303
 304         /* Strip leading/trailing whitespace.  We can't do this with
 305          * striplt() because it uses too many strlen()'s
 306          */
 307         while ((output_len > 0) && (isspace(outptr[0]))) {
 308                 strcpy(outptr, &outptr[1]);
 309                 --output_len;
 310         }
 311         while ((output_len > 0) && (isspace(outptr[output_len-1]))) {
 312                 outptr[output_len-1] = 0;
 313                 --output_len;
 314         }
 315
 316         if (outptr[output_len-1] != '\n') {
 317                 strcat(outptr, "\n");
 318                 ++output_len;
 319         }
 320
 321         return outptr;
 322
 323 }