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