citadel/html.c

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