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 "citadel.h"
  36 #include "server.h"
  37 #include "serv_extensions.h"
  38 #include "control.h"
  39 #include "sysdep_decls.h"
  40 #include "support.h"
  41 #include "config.h"
  42 #include "msgbase.h"
  43 #include "tools.h"
  44 #include "room_ops.h"
  45 #include "html.h"
  46
  47
  48 /*
  49  * Convert HTML to plain text.
  50  *
  51  * inputmsg      = pointer to raw HTML message
  52  * screenwidth   = desired output screenwidth
  53  * do_citaformat = set to 1 to indent newlines with spaces
  54  */
  55 char *html_to_ascii(char *inputmsg, int screenwidth, int do_citaformat) {
  56         char inbuf[SIZ];
  57         char outbuf[SIZ];
  58         char tag[1024];
  59         int done_reading = 0;
  60         char *inptr;
  61         char *outptr;
  62         size_t outptr_buffer_size;
  63         size_t output_len = 0;
  64         int i, j, ch, did_out, rb, scanch;
  65         int nest = 0;           /* Bracket nesting level */
  66
  67         inptr = inputmsg;
  68         strcpy(inbuf, "");
  69         strcpy(outbuf, "");
  70
  71         outptr_buffer_size = strlen(inptr) + SIZ;
  72         outptr = malloc(outptr_buffer_size);
  73         if (outptr == NULL) return NULL;
  74         strcpy(outptr, "");
  75         output_len = 0;
  76
  77         do {
  78                 /* Fill the input buffer */
  79                 if ( (done_reading == 0) && (strlen(inbuf) < (SIZ-128)) ) {
  80
  81                         ch = *inptr++;
  82                         if (ch != 0) {
  83                                 inbuf[strlen(inbuf)+1] = 0;
  84                                 inbuf[strlen(inbuf)] = ch;
  85                         }
  86                         else {
  87                                 done_reading = 1;
  88                         }
  89
  90                 }
  91
  92                 /* Do some parsing */
  93                 if (strlen(inbuf)>0) {
  94
  95                     /* Fold in all the spacing */
  96                     for (i=0; i<strlen(inbuf); ++i) {
  97                         if (inbuf[i]==10) inbuf[i]=32;
  98                         if (inbuf[i]==13) inbuf[i]=32;
  99                         if (inbuf[i]==9) inbuf[i]=32;
 100                         if ((inbuf[i]<32) || (inbuf[i]>126)) {
 101                                 inbuf[i] = '?';
 102                                 /* strcpy(&inbuf[i], &inbuf[i+1]); */
 103                         }
 104                     }
 105                     for (i=0; i<strlen(inbuf); ++i) {
 106                         while ((inbuf[i]==32)&&(inbuf[i+1]==32))
 107                                 strcpy(&inbuf[i], &inbuf[i+1]);
 108                     }
 109
 110                     for (i=0; i<strlen(inbuf); ++i) {
 111
 112                         ch = inbuf[i];
 113
 114                         if (ch == '<') {
 115                                 ++nest;
 116                                 strcpy(tag, "");
 117                         }
 118
 119                         else if (ch == '>') {   /* We have a tag. */
 120                                 if (nest > 0) --nest;
 121
 122                                 /* Unqualify the tag (truncate at first space) */
 123                                 if (strchr(tag, ' ') != NULL) {
 124                                         strcpy(strchr(tag, ' '), "");
 125                                 }
 126
 127                                 if (!strcasecmp(tag, "P")) {
 128                                         strcat(outbuf, "\n\n");
 129                                 }
 130
 131                                 if (!strcasecmp(tag, "/DIV")) {
 132                                         strcat(outbuf, "\n\n");
 133                                 }
 134
 135                                 else if (!strcasecmp(tag, "H1")) {
 136                                         strcat(outbuf, "\n\n");
 137                                 }
 138
 139                                 else if (!strcasecmp(tag, "H2")) {
 140                                         strcat(outbuf, "\n\n");
 141                                 }
 142
 143                                 else if (!strcasecmp(tag, "H3")) {
 144                                         strcat(outbuf, "\n\n");
 145                                 }
 146
 147                                 else if (!strcasecmp(tag, "H4")) {
 148                                         strcat(outbuf, "\n\n");
 149                                 }
 150
 151                                 else if (!strcasecmp(tag, "/H1")) {
 152                                         strcat(outbuf, "\n");
 153                                 }
 154
 155                                 else if (!strcasecmp(tag, "/H2")) {
 156                                         strcat(outbuf, "\n");
 157                                 }
 158
 159                                 else if (!strcasecmp(tag, "/H3")) {
 160                                         strcat(outbuf, "\n");
 161                                 }
 162
 163                                 else if (!strcasecmp(tag, "/H4")) {
 164                                         strcat(outbuf, "\n");
 165                                 }
 166
 167                                 else if (!strcasecmp(tag, "HR")) {
 168                                         strcat(outbuf, "\n ");
 169                                         for (j=0; j<screenwidth-2; ++j)
 170                                                 strcat(outbuf, "-");
 171                                         strcat(outbuf, "\n");
 172                                 }
 173
 174                                 else if (!strcasecmp(tag, "BR")) {
 175                                         strcat(outbuf, "\n");
 176                                 }
 177
 178                                 else if (!strcasecmp(tag, "TR")) {
 179                                         strcat(outbuf, "\n");
 180                                 }
 181
 182                                 else if (!strcasecmp(tag, "/TABLE")) {
 183                                         strcat(outbuf, "\n");
 184                                 }
 185
 186                         }
 187
 188                         else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
 189                                 tag[strlen(tag)+1] = 0;
 190                                 tag[strlen(tag)] = ch;
 191                         }
 192
 193                         else if (!nest) {
 194                                 outbuf[strlen(outbuf)+1] = 0;
 195                                 outbuf[strlen(outbuf)] = ch;
 196                         }
 197                     }
 198                     strcpy(inbuf, &inbuf[i]);
 199                 }
 200
 201                 /* Convert &; tags to the forbidden characters */
 202                 if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
 203
 204                         if (!strncasecmp(&outbuf[i], "&nbsp;", 6)) {
 205                                 outbuf[i] = ' ';
 206                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 207                         }
 208
 209                         else if (!strncasecmp(&outbuf[i], "&lt;", 4)) {
 210                                 outbuf[i] = '<';
 211                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 212                         }
 213
 214                         else if (!strncasecmp(&outbuf[i], "&gt;", 4)) {
 215                                 outbuf[i] = '>';
 216                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 217                         }
 218
 219                         else if (!strncasecmp(&outbuf[i], "&amp;", 5)) {
 220                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 221                         }
 222
 223                         else if (!strncasecmp(&outbuf[i], "&quot;", 6)) {
 224                                 outbuf[i] = '\"';
 225                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 226                         }
 227
 228                         else if (!strncasecmp(&outbuf[i], "&copy;", 6)) {
 229                                 outbuf[i] = '(';
 230                                 outbuf[i+1] = 'c';
 231                                 outbuf[i+2] = ')';
 232                                 strcpy(&outbuf[i+3], &outbuf[i+6]);
 233                         }
 234
 235                         else if (!strncasecmp(&outbuf[i], "&reg;", 5)) {
 236                                 outbuf[i] = '(';
 237                                 outbuf[i+1] = 'r';
 238                                 outbuf[i+2] = ')';
 239                                 strcpy(&outbuf[i+3], &outbuf[i+5]);
 240                         }
 241
 242                         /* two-digit decimal equivalents */
 243                         else if ((!strncmp(&outbuf[i], "&#", 2))
 244                               && (outbuf[i+4] == ';') ) {
 245                                 scanch = 0;
 246                                 sscanf(&outbuf[i+2], "%02d", &scanch);
 247                                 outbuf[i] = scanch;
 248                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 249                         }
 250
 251                         /* three-digit decimal equivalents */
 252                         else if ((!strncmp(&outbuf[i], "&#", 2))
 253                               && (outbuf[i+5] == ';') ) {
 254                                 scanch = 0;
 255                                 sscanf(&outbuf[i+2], "%03d", &scanch);
 256                                 outbuf[i] = scanch;
 257                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 258                         }
 259
 260                 }
 261
 262                 /* Make sure the output buffer is big enough */
 263                 if ((output_len + strlen(outbuf) + SIZ)
 264                    > outptr_buffer_size) {
 265                         outptr_buffer_size += SIZ;
 266                         outptr = realloc(outptr, outptr_buffer_size);
 267                 }
 268
 269                 /* Output any lines terminated with hard line breaks */
 270                 do {
 271                         did_out = 0;
 272                         if (strlen(outbuf)>0)
 273                             for (i = 0; i<strlen(outbuf); ++i) {
 274                                 if ( (i<(screenwidth-2)) && (outbuf[i]=='\n')) {
 275
 276                                         strncpy(&outptr[output_len],
 277                                                 outbuf, i+1);
 278                                         output_len += (i+1);
 279
 280                                         if (do_citaformat) {
 281                                                 strcpy(&outptr[output_len],
 282                                                         " ");
 283                                                 ++output_len;
 284                                         }
 285
 286                                         strcpy(outbuf, &outbuf[i+1]);
 287                                         i = 0;
 288                                         did_out = 1;
 289                                 }
 290                         }
 291                 } while (did_out);
 292
 293                 /* Add soft line breaks */
 294                 if (strlen(outbuf) > (screenwidth - 2)) {
 295                         rb = (-1);
 296                         for (i=0; i<(screenwidth-2); ++i) {
 297                                 if (outbuf[i]==32) rb = i;
 298                         }
 299                         if (rb>=0) {
 300                                 strncpy(&outptr[output_len], outbuf, rb);
 301                                 output_len += rb;
 302                                 strcpy(&outptr[output_len], "\n");
 303                                 output_len += 1;
 304                                 if (do_citaformat) {
 305                                         strcpy(&outptr[output_len], " ");
 306                                         ++output_len;
 307                                 }
 308                                 strcpy(outbuf, &outbuf[rb+1]);
 309                         } else {
 310                                 strncpy(&outptr[output_len], outbuf,
 311                                         screenwidth-2);
 312                                 output_len += (screenwidth-2);
 313                                 strcpy(&outptr[output_len], "\n");
 314                                 output_len += 1;
 315                                 if (do_citaformat) {
 316                                         strcpy(&outptr[output_len], " ");
 317                                         ++output_len;
 318                                 }
 319                                 strcpy(outbuf, &outbuf[screenwidth-2]);
 320                         }
 321                 }
 322
 323         } while (done_reading == 0);
 324
 325         strcpy(&outptr[output_len], outbuf);
 326         output_len += strlen(outbuf);
 327
 328         /* Strip leading/trailing whitespace.  We can't do this with
 329          * striplt() because it uses too many strlen()'s
 330          */
 331         while ((output_len > 0) && (isspace(outptr[0]))) {
 332                 strcpy(outptr, &outptr[1]);
 333                 --output_len;
 334         }
 335         while ((output_len > 0) && (isspace(outptr[output_len-1]))) {
 336                 outptr[output_len-1] = 0;
 337                 --output_len;
 338         }
 339
 340         if (outptr[output_len-1] != '\n') {
 341                 strcat(outptr, "\n");
 342                 ++output_len;
 343         }
 344
 345         return outptr;
 346
 347 }