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         int blockquote = 0;     /* BLOCKQUOTE nesting level */
  67
  68         inptr = inputmsg;
  69         strcpy(inbuf, "");
  70         strcpy(outbuf, "");
  71
  72         outptr_buffer_size = strlen(inptr) + SIZ;
  73         outptr = malloc(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) < (SIZ-128)) ) {
  81
  82                         ch = *inptr++;
  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                                 inbuf[i] = '?';
 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                                 if (!strcasecmp(tag, "LI")) {
 136                                         strcat(outbuf, "\n * ");
 137                                 }
 138
 139                                 else if (!strcasecmp(tag, "/UL")) {
 140                                         strcat(outbuf, "\n\n");
 141                                 }
 142
 143                                 else if (!strcasecmp(tag, "H1")) {
 144                                         strcat(outbuf, "\n\n");
 145                                 }
 146
 147                                 else if (!strcasecmp(tag, "H2")) {
 148                                         strcat(outbuf, "\n\n");
 149                                 }
 150
 151                                 else if (!strcasecmp(tag, "H3")) {
 152                                         strcat(outbuf, "\n\n");
 153                                 }
 154
 155                                 else if (!strcasecmp(tag, "H4")) {
 156                                         strcat(outbuf, "\n\n");
 157                                 }
 158
 159                                 else if (!strcasecmp(tag, "/H1")) {
 160                                         strcat(outbuf, "\n");
 161                                 }
 162
 163                                 else if (!strcasecmp(tag, "/H2")) {
 164                                         strcat(outbuf, "\n");
 165                                 }
 166
 167                                 else if (!strcasecmp(tag, "/H3")) {
 168                                         strcat(outbuf, "\n");
 169                                 }
 170
 171                                 else if (!strcasecmp(tag, "/H4")) {
 172                                         strcat(outbuf, "\n");
 173                                 }
 174
 175                                 else if (!strcasecmp(tag, "HR")) {
 176                                         strcat(outbuf, "\n ");
 177                                         for (j=0; j<screenwidth-2; ++j)
 178                                                 strcat(outbuf, "-");
 179                                         strcat(outbuf, "\n");
 180                                 }
 181
 182                                 else if (!strcasecmp(tag, "BR")) {
 183                                         strcat(outbuf, "\n");
 184                                 }
 185
 186                                 else if (!strcasecmp(tag, "TR")) {
 187                                         strcat(outbuf, "\n");
 188                                 }
 189
 190                                 else if (!strcasecmp(tag, "/TABLE")) {
 191                                         strcat(outbuf, "\n");
 192                                 }
 193
 194                                 else if (!strcasecmp(tag, "BLOCKQUOTE")) {
 195                                         strcat(outbuf, "\n\n <<\n");
 196                                         ++blockquote;
 197                                 }
 198
 199                                 else if (!strcasecmp(tag, "/BLOCKQUOTE")) {
 200                                         strcat(outbuf, "\n >>\n\n");
 201                                         --blockquote;
 202                                 }
 203
 204                         }
 205
 206                         else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
 207                                 tag[strlen(tag)+1] = 0;
 208                                 tag[strlen(tag)] = ch;
 209                         }
 210
 211                         else if (!nest) {
 212                                 outbuf[strlen(outbuf)+1] = 0;
 213                                 outbuf[strlen(outbuf)] = ch;
 214                         }
 215                     }
 216                     strcpy(inbuf, &inbuf[i]);
 217                 }
 218
 219                 /* Convert &; tags to the forbidden characters */
 220                 if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
 221
 222                         if (!strncasecmp(&outbuf[i], "&nbsp;", 6)) {
 223                                 outbuf[i] = ' ';
 224                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 225                         }
 226
 227                         else if (!strncasecmp(&outbuf[i], "&lt;", 4)) {
 228                                 outbuf[i] = '<';
 229                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 230                         }
 231
 232                         else if (!strncasecmp(&outbuf[i], "&gt;", 4)) {
 233                                 outbuf[i] = '>';
 234                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 235                         }
 236
 237                         else if (!strncasecmp(&outbuf[i], "&amp;", 5)) {
 238                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 239                         }
 240
 241                         else if (!strncasecmp(&outbuf[i], "&quot;", 6)) {
 242                                 outbuf[i] = '\"';
 243                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 244                         }
 245
 246                         else if (!strncasecmp(&outbuf[i], "&copy;", 6)) {
 247                                 outbuf[i] = '(';
 248                                 outbuf[i+1] = 'c';
 249                                 outbuf[i+2] = ')';
 250                                 strcpy(&outbuf[i+3], &outbuf[i+6]);
 251                         }
 252
 253                         else if (!strncasecmp(&outbuf[i], "&reg;", 5)) {
 254                                 outbuf[i] = '(';
 255                                 outbuf[i+1] = 'r';
 256                                 outbuf[i+2] = ')';
 257                                 strcpy(&outbuf[i+3], &outbuf[i+5]);
 258                         }
 259
 260                         /* two-digit decimal equivalents */
 261                         else if ((!strncmp(&outbuf[i], "&#", 2))
 262                               && (outbuf[i+4] == ';') ) {
 263                                 scanch = 0;
 264                                 sscanf(&outbuf[i+2], "%02d", &scanch);
 265                                 outbuf[i] = scanch;
 266                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 267                         }
 268
 269                         /* three-digit decimal equivalents */
 270                         else if ((!strncmp(&outbuf[i], "&#", 2))
 271                               && (outbuf[i+5] == ';') ) {
 272                                 scanch = 0;
 273                                 sscanf(&outbuf[i+2], "%03d", &scanch);
 274                                 outbuf[i] = scanch;
 275                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 276                         }
 277
 278                 }
 279
 280                 /* Make sure the output buffer is big enough */
 281                 if ((output_len + strlen(outbuf) + SIZ)
 282                    > outptr_buffer_size) {
 283                         outptr_buffer_size += SIZ;
 284                         outptr = realloc(outptr, outptr_buffer_size);
 285                 }
 286
 287                 /* Output any lines terminated with hard line breaks */
 288                 do {
 289                         did_out = 0;
 290                         if (strlen(outbuf)>0) {
 291                             for (i = 0; i<strlen(outbuf); ++i) {
 292                                 if ( (i<(screenwidth-2)) && (outbuf[i]=='\n')) {
 293
 294                                         strncpy(&outptr[output_len],
 295                                                 outbuf, i+1);
 296                                         output_len += (i+1);
 297
 298                                         if (do_citaformat) {
 299                                                 strcpy(&outptr[output_len],
 300                                                         " ");
 301                                                 ++output_len;
 302                                         }
 303
 304                                         strcpy(outbuf, &outbuf[i+1]);
 305                                         i = 0;
 306                                         did_out = 1;
 307                                 }
 308                         }
 309                     }
 310                 } while (did_out);
 311
 312                 /* Add soft line breaks */
 313                 if (strlen(outbuf) > (screenwidth - 2 )) {
 314                         rb = (-1);
 315                         for (i=0; i<(screenwidth-2); ++i) {
 316                                 if (outbuf[i]==32) rb = i;
 317                         }
 318                         if (rb>=0) {
 319                                 strncpy(&outptr[output_len], outbuf, rb);
 320                                 output_len += rb;
 321                                 strcpy(&outptr[output_len], "\n");
 322                                 output_len += 1;
 323                                 if (do_citaformat) {
 324                                         strcpy(&outptr[output_len], " ");
 325                                         ++output_len;
 326                                 }
 327                                 strcpy(outbuf, &outbuf[rb+1]);
 328                         } else {
 329                                 strncpy(&outptr[output_len], outbuf,
 330                                         screenwidth-2);
 331                                 output_len += (screenwidth-2);
 332                                 strcpy(&outptr[output_len], "\n");
 333                                 output_len += 1;
 334                                 if (do_citaformat) {
 335                                         strcpy(&outptr[output_len], " ");
 336                                         ++output_len;
 337                                 }
 338                                 strcpy(outbuf, &outbuf[screenwidth-2]);
 339                         }
 340                 }
 341
 342         } while (done_reading == 0);
 343
 344         strcpy(&outptr[output_len], outbuf);
 345         output_len += strlen(outbuf);
 346
 347         /* Strip leading/trailing whitespace.  We can't do this with
 348          * striplt() because it uses too many strlen()'s
 349          */
 350         while ((output_len > 0) && (isspace(outptr[0]))) {
 351                 strcpy(outptr, &outptr[1]);
 352                 --output_len;
 353         }
 354         while ((output_len > 0) && (isspace(outptr[output_len-1]))) {
 355                 outptr[output_len-1] = 0;
 356                 --output_len;
 357         }
 358
 359         if (outptr[output_len-1] != '\n') {
 360                 strcat(outptr, "\n");
 361                 ++output_len;
 362         }
 363
 364         return outptr;
 365
 366 }