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 "serv_extensions.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) < (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                                 /* strcpy(&inbuf[i], &inbuf[i+1]); */
 104                         }
 105                     }
 106                     for (i=0; i<strlen(inbuf); ++i) {
 107                         while ((inbuf[i]==32)&&(inbuf[i+1]==32))
 108                                 strcpy(&inbuf[i], &inbuf[i+1]);
 109                     }
 110
 111                     for (i=0; i<strlen(inbuf); ++i) {
 112
 113                         ch = inbuf[i];
 114
 115                         if (ch == '<') {
 116                                 ++nest;
 117                                 strcpy(tag, "");
 118                         }
 119
 120                         else if (ch == '>') {
 121                                 if (nest > 0) --nest;
 122
 123                                 if (!strcasecmp(tag, "P")) {
 124                                         strcat(outbuf, "\n\n");
 125                                 }
 126
 127                                 if (!strcasecmp(tag, "/DIV")) {
 128                                         strcat(outbuf, "\n\n");
 129                                 }
 130
 131                                 else if (!strcasecmp(tag, "H1")) {
 132                                         strcat(outbuf, "\n\n");
 133                                 }
 134
 135                                 else if (!strcasecmp(tag, "H2")) {
 136                                         strcat(outbuf, "\n\n");
 137                                 }
 138
 139                                 else if (!strcasecmp(tag, "H3")) {
 140                                         strcat(outbuf, "\n\n");
 141                                 }
 142
 143                                 else if (!strcasecmp(tag, "H4")) {
 144                                         strcat(outbuf, "\n\n");
 145                                 }
 146
 147                                 else if (!strcasecmp(tag, "/H1")) {
 148                                         strcat(outbuf, "\n");
 149                                 }
 150
 151                                 else if (!strcasecmp(tag, "/H2")) {
 152                                         strcat(outbuf, "\n");
 153                                 }
 154
 155                                 else if (!strcasecmp(tag, "/H3")) {
 156                                         strcat(outbuf, "\n");
 157                                 }
 158
 159                                 else if (!strcasecmp(tag, "/H4")) {
 160                                         strcat(outbuf, "\n");
 161                                 }
 162
 163                                 else if (!strcasecmp(tag, "HR")) {
 164                                         strcat(outbuf, "\n ");
 165                                         for (j=0; j<screenwidth-2; ++j)
 166                                                 strcat(outbuf, "-");
 167                                         strcat(outbuf, "\n");
 168                                 }
 169
 170                                 else if (!strcasecmp(tag, "BR")) {
 171                                         strcat(outbuf, "\n");
 172                                 }
 173
 174                                 else if (!strcasecmp(tag, "TR")) {
 175                                         strcat(outbuf, "\n");
 176                                 }
 177
 178                                 else if (!strcasecmp(tag, "/TABLE")) {
 179                                         strcat(outbuf, "\n");
 180                                 }
 181
 182                         }
 183
 184                         else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
 185                                 tag[strlen(tag)+1] = 0;
 186                                 tag[strlen(tag)] = ch;
 187                         }
 188
 189                         else if (!nest) {
 190                                 outbuf[strlen(outbuf)+1] = 0;
 191                                 outbuf[strlen(outbuf)] = ch;
 192                         }
 193                     }
 194                     strcpy(inbuf, &inbuf[i]);
 195                 }
 196
 197                 /* Convert &; tags to the forbidden characters */
 198                 if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
 199
 200                         if (!strncasecmp(&outbuf[i], "&nbsp;", 6)) {
 201                                 outbuf[i] = ' ';
 202                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 203                         }
 204
 205                         else if (!strncasecmp(&outbuf[i], "&lt;", 4)) {
 206                                 outbuf[i] = '<';
 207                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 208                         }
 209
 210                         else if (!strncasecmp(&outbuf[i], "&gt;", 4)) {
 211                                 outbuf[i] = '>';
 212                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 213                         }
 214
 215                         else if (!strncasecmp(&outbuf[i], "&amp;", 5)) {
 216                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 217                         }
 218
 219                         else if (!strncasecmp(&outbuf[i], "&quot;", 6)) {
 220                                 outbuf[i] = '\"';
 221                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 222                         }
 223
 224                         else if (!strncasecmp(&outbuf[i], "&copy;", 6)) {
 225                                 outbuf[i] = '(';
 226                                 outbuf[i+1] = 'c';
 227                                 outbuf[i+2] = ')';
 228                                 strcpy(&outbuf[i+3], &outbuf[i+6]);
 229                         }
 230
 231                         else if (!strncasecmp(&outbuf[i], "&reg;", 5)) {
 232                                 outbuf[i] = '(';
 233                                 outbuf[i+1] = 'r';
 234                                 outbuf[i+2] = ')';
 235                                 strcpy(&outbuf[i+3], &outbuf[i+5]);
 236                         }
 237
 238                 }
 239
 240                 /* Make sure the output buffer is big enough */
 241                 if ((output_len + strlen(outbuf) + SIZ)
 242                    > outptr_buffer_size) {
 243                         outptr_buffer_size += SIZ;
 244                         outptr = realloc(outptr, outptr_buffer_size);
 245                 }
 246
 247                 /* Output any lines terminated with hard line breaks */
 248                 do {
 249                         did_out = 0;
 250                         if (strlen(outbuf)>0)
 251                             for (i = 0; i<strlen(outbuf); ++i) {
 252                                 if ( (i<(screenwidth-2)) && (outbuf[i]=='\n')) {
 253
 254                                         strncpy(&outptr[output_len],
 255                                                 outbuf, i+1);
 256                                         output_len += (i+1);
 257
 258                                         if (do_citaformat) {
 259                                                 strcpy(&outptr[output_len],
 260                                                         " ");
 261                                                 ++output_len;
 262                                         }
 263
 264                                         strcpy(outbuf, &outbuf[i+1]);
 265                                         i = 0;
 266                                         did_out = 1;
 267                                 }
 268                         }
 269                 } while (did_out);
 270
 271                 /* Add soft line breaks */
 272                 if (strlen(outbuf) > (screenwidth - 2)) {
 273                         rb = (-1);
 274                         for (i=0; i<(screenwidth-2); ++i) {
 275                                 if (outbuf[i]==32) rb = i;
 276                         }
 277                         if (rb>=0) {
 278                                 strncpy(&outptr[output_len], outbuf, rb);
 279                                 output_len += rb;
 280                                 strcpy(&outptr[output_len], "\n");
 281                                 output_len += 1;
 282                                 if (do_citaformat) {
 283                                         strcpy(&outptr[output_len], " ");
 284                                         ++output_len;
 285                                 }
 286                                 strcpy(outbuf, &outbuf[rb+1]);
 287                         } else {
 288                                 strncpy(&outptr[output_len], outbuf,
 289                                         screenwidth-2);
 290                                 output_len += (screenwidth-2);
 291                                 strcpy(&outptr[output_len], "\n");
 292                                 output_len += 1;
 293                                 if (do_citaformat) {
 294                                         strcpy(&outptr[output_len], " ");
 295                                         ++output_len;
 296                                 }
 297                                 strcpy(outbuf, &outbuf[screenwidth-2]);
 298                         }
 299                 }
 300
 301         } while (done_reading == 0);
 302
 303         strcpy(&outptr[output_len], outbuf);
 304         output_len += strlen(outbuf);
 305
 306         /* Strip leading/trailing whitespace.  We can't do this with
 307          * striplt() because it uses too many strlen()'s
 308          */
 309         while ((output_len > 0) && (isspace(outptr[0]))) {
 310                 strcpy(outptr, &outptr[1]);
 311                 --output_len;
 312         }
 313         while ((output_len > 0) && (isspace(outptr[output_len-1]))) {
 314                 outptr[output_len-1] = 0;
 315                 --output_len;
 316         }
 317
 318         if (outptr[output_len-1] != '\n') {
 319                 strcat(outptr, "\n");
 320                 ++output_len;
 321         }
 322
 323         return outptr;
 324
 325 }