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 char *html_to_ascii(char *inputmsg, int screenwidth) {
  37         char inbuf[256];
  38         char outbuf[256];
  39         char tag[1024];
  40         int done_reading = 0;
  41         char *inptr;
  42         char *outptr;
  43         size_t outlen;
  44         int i, j, ch, did_out, rb;
  45         int nest = 0;           /* Bracket nesting level */
  46
  47         inptr = inputmsg;
  48         strcpy(inbuf, "");
  49         strcpy(outbuf, "");
  50
  51         outptr = mallok(strlen(inptr) + 256);
  52         if (outptr == NULL) return NULL;
  53         strcpy(outptr, "");
  54         outlen = 0;
  55
  56         do {
  57                 /* Fill the input buffer */
  58                 if ( (done_reading == 0) && (strlen(inbuf) < 128) ) {
  59
  60                         /* FIX ... genericize this */
  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                                 else if (!strcasecmp(tag, "H1")) {
 103                                         strcat(outbuf, "\n\n");
 104                                 }
 105
 106                                 else if (!strcasecmp(tag, "H2")) {
 107                                         strcat(outbuf, "\n\n");
 108                                 }
 109
 110                                 else if (!strcasecmp(tag, "H3")) {
 111                                         strcat(outbuf, "\n\n");
 112                                 }
 113
 114                                 else if (!strcasecmp(tag, "H4")) {
 115                                         strcat(outbuf, "\n\n");
 116                                 }
 117
 118                                 else if (!strcasecmp(tag, "/H1")) {
 119                                         strcat(outbuf, "\n");
 120                                 }
 121
 122                                 else if (!strcasecmp(tag, "/H2")) {
 123                                         strcat(outbuf, "\n");
 124                                 }
 125
 126                                 else if (!strcasecmp(tag, "/H3")) {
 127                                         strcat(outbuf, "\n");
 128                                 }
 129
 130                                 else if (!strcasecmp(tag, "/H4")) {
 131                                         strcat(outbuf, "\n");
 132                                 }
 133
 134                                 else if (!strcasecmp(tag, "HR")) {
 135                                         strcat(outbuf, "\n ");
 136                                         for (j=0; j<screenwidth-2; ++j)
 137                                                 strcat(outbuf, "-");
 138                                         strcat(outbuf, "\n");
 139                                 }
 140
 141                                 else if (!strcasecmp(tag, "BR")) {
 142                                         strcat(outbuf, "\n");
 143                                 }
 144
 145                                 else if (!strcasecmp(tag, "TR")) {
 146                                         strcat(outbuf, "\n");
 147                                 }
 148
 149                                 else if (!strcasecmp(tag, "/TABLE")) {
 150                                         strcat(outbuf, "\n");
 151                                 }
 152
 153                         }
 154
 155                         else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
 156                                 tag[strlen(tag)+1] = 0;
 157                                 tag[strlen(tag)] = ch;
 158                         }
 159
 160                         else if (!nest) {
 161                                 outbuf[strlen(outbuf)+1] = 0;
 162                                 outbuf[strlen(outbuf)] = ch;
 163                         }
 164                     }
 165                     strcpy(inbuf, &inbuf[i]);
 166                 }
 167
 168                 /* Convert &; tags to the forbidden characters */
 169                 if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
 170
 171                         if (!strncasecmp(&outbuf[i], "&nbsp;", 6)) {
 172                                 outbuf[i] = ' ';
 173                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 174                         }
 175
 176                         else if (!strncasecmp(&outbuf[i], "&lt;", 4)) {
 177                                 outbuf[i] = '<';
 178                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 179                         }
 180
 181                         else if (!strncasecmp(&outbuf[i], "&gt;", 4)) {
 182                                 outbuf[i] = '>';
 183                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 184                         }
 185
 186                         else if (!strncasecmp(&outbuf[i], "&amp;", 5)) {
 187                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 188                         }
 189
 190                         else if (!strncasecmp(&outbuf[i], "&quot;", 6)) {
 191                                 outbuf[i] = '\"';
 192                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 193                         }
 194
 195                         else if (!strncasecmp(&outbuf[i], "&copy;", 6)) {
 196                                 outbuf[i] = '(';
 197                                 outbuf[i+1] = 'c';
 198                                 outbuf[i+2] = ')';
 199                                 strcpy(&outbuf[i+3], &outbuf[i+6]);
 200                         }
 201
 202                         else if (!strncasecmp(&outbuf[i], "&reg;", 5)) {
 203                                 outbuf[i] = '(';
 204                                 outbuf[i+1] = 'r';
 205                                 outbuf[i+2] = ')';
 206                                 strcpy(&outbuf[i+3], &outbuf[i+5]);
 207                         }
 208
 209                 }
 210
 211                 /* Make sure the output buffer is big enough */
 212                 if ((strlen(outptr) + strlen(outbuf) + 128) > outlen) {
 213                         outlen += 128;
 214                         outptr = realloc(outptr, outlen);
 215                 }
 216
 217                 /* Output any lines terminated with hard line breaks */
 218                 do {
 219                         did_out = 0;
 220                         if (strlen(outbuf)>0)
 221                             for (i = 0; i<strlen(outbuf); ++i) {
 222                                 if ( (i<(screenwidth-2)) && (outbuf[i]=='\n')) {
 223
 224                                         strncat(outptr, outbuf, i+1);
 225
 226                                         strcpy(outbuf, &outbuf[i+1]);
 227                                         i = 0;
 228                                         did_out = 1;
 229                                 }
 230                         }
 231                 } while (did_out);
 232
 233                 /* Add soft line breaks */
 234                 if (strlen(outbuf) > (screenwidth - 2)) {
 235                         rb = (-1);
 236                         for (i=0; i<(screenwidth-2); ++i) {
 237                                 if (outbuf[i]==32) rb = i;
 238                         }
 239                         if (rb>=0) {
 240
 241                                 strncat(outptr, outbuf, rb);
 242                                 strcat(outptr, "\n");
 243
 244                                 strcpy(outbuf, &outbuf[rb+1]);
 245                         } else {
 246
 247                                 strncat(outptr, outbuf, screenwidth-2);
 248                                 strcat(outptr, "\n");
 249
 250                                 strcpy(outbuf, &outbuf[screenwidth-2]);
 251                         }
 252                 }
 253
 254         } while (done_reading == 0);
 255
 256         strcat(outptr, outbuf);
 257         strcat(outptr, "\n");
 258
 259         return outptr;
 260
 261 }
 262