citadel/html.c

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