citadel/html.c

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