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         int i, j, ch, did_out, rb;
  43         int nest = 0;           /* Bracket nesting level */
  44
  45         inptr = inputmsg;
  46         strcpy(inbuf, "");
  47         strcpy(outbuf, "");
  48
  49         do {
  50                 /* Fill the input buffer */
  51                 if ( (done_reading == 0) && (strlen(inbuf) < 128) ) {
  52
  53                         /* FIX ... genericize this */
  54                         ch = *inputmsg++;
  55                         if (ch > 0) {
  56                                 inbuf[strlen(inbuf)+1] = 0;
  57                                 inbuf[strlen(inbuf)] = ch;
  58                         }
  59                         else {
  60                                 done_reading = 1;
  61                         }
  62
  63                 }
  64
  65                 /* Do some parsing */
  66                 if (strlen(inbuf)>0) {
  67
  68                     /* Fold in all the spacing */
  69                     for (i=0; i<strlen(inbuf); ++i) {
  70                         if (inbuf[i]==10) inbuf[i]=32;
  71                         if (inbuf[i]==13) inbuf[i]=32;
  72                         if (inbuf[i]==9) inbuf[i]=32;
  73                         if ((inbuf[i]<32) || (inbuf[i]>126))
  74                                 strcpy(&inbuf[i], &inbuf[i+1]);
  75                         while ((inbuf[i]==32)&&(inbuf[i+1]==32))
  76                                 strcpy(&inbuf[i], &inbuf[i+1]);
  77                     }
  78
  79                     for (i=0; i<strlen(inbuf); ++i) {
  80
  81                         ch = inbuf[i];
  82
  83                         if (ch == '<') {
  84                                 ++nest;
  85                                 strcpy(tag, "");
  86                         }
  87
  88                         else if (ch == '>') {
  89                                 if (nest > 0) --nest;
  90
  91                                 if (!strcasecmp(tag, "P")) {
  92                                         strcat(outbuf, "\n\n");
  93                                 }
  94
  95                                 else if (!strcasecmp(tag, "H1")) {
  96                                         strcat(outbuf, "\n\n");
  97                                 }
  98
  99                                 else if (!strcasecmp(tag, "H2")) {
 100                                         strcat(outbuf, "\n\n");
 101                                 }
 102
 103                                 else if (!strcasecmp(tag, "H3")) {
 104                                         strcat(outbuf, "\n\n");
 105                                 }
 106
 107                                 else if (!strcasecmp(tag, "H4")) {
 108                                         strcat(outbuf, "\n\n");
 109                                 }
 110
 111                                 else if (!strcasecmp(tag, "/H1")) {
 112                                         strcat(outbuf, "\n");
 113                                 }
 114
 115                                 else if (!strcasecmp(tag, "/H2")) {
 116                                         strcat(outbuf, "\n");
 117                                 }
 118
 119                                 else if (!strcasecmp(tag, "/H3")) {
 120                                         strcat(outbuf, "\n");
 121                                 }
 122
 123                                 else if (!strcasecmp(tag, "/H4")) {
 124                                         strcat(outbuf, "\n");
 125                                 }
 126
 127                                 else if (!strcasecmp(tag, "HR")) {
 128                                         strcat(outbuf, "\n ");
 129                                         for (j=0; j<screenwidth-2; ++j)
 130                                                 strcat(outbuf, "-");
 131                                         strcat(outbuf, "\n");
 132                                 }
 133
 134                                 else if (!strcasecmp(tag, "BR")) {
 135                                         strcat(outbuf, "\n");
 136                                 }
 137
 138                                 else if (!strcasecmp(tag, "TR")) {
 139                                         strcat(outbuf, "\n");
 140                                 }
 141
 142                                 else if (!strcasecmp(tag, "/TABLE")) {
 143                                         strcat(outbuf, "\n");
 144                                 }
 145
 146                         }
 147
 148                         else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
 149                                 tag[strlen(tag)+1] = 0;
 150                                 tag[strlen(tag)] = ch;
 151                         }
 152
 153                         else if (!nest) {
 154                                 outbuf[strlen(outbuf)+1] = 0;
 155                                 outbuf[strlen(outbuf)] = ch;
 156                         }
 157                     }
 158                     strcpy(inbuf, &inbuf[i]);
 159                 }
 160
 161                 /* Convert &; tags to the forbidden characters */
 162                 if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
 163
 164                         if (!strncasecmp(&outbuf[i], "&nbsp;", 6)) {
 165                                 outbuf[i] = ' ';
 166                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 167                         }
 168
 169                         else if (!strncasecmp(&outbuf[i], "&lt;", 4)) {
 170                                 outbuf[i] = '<';
 171                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 172                         }
 173
 174                         else if (!strncasecmp(&outbuf[i], "&gt;", 4)) {
 175                                 outbuf[i] = '>';
 176                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 177                         }
 178
 179                         else if (!strncasecmp(&outbuf[i], "&amp;", 5)) {
 180                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 181                         }
 182
 183                         else if (!strncasecmp(&outbuf[i], "&quot;", 6)) {
 184                                 outbuf[i] = '\"';
 185                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 186                         }
 187
 188                         else if (!strncasecmp(&outbuf[i], "&copy;", 6)) {
 189                                 outbuf[i] = '(';
 190                                 outbuf[i+1] = 'c';
 191                                 outbuf[i+2] = ')';
 192                                 strcpy(&outbuf[i+3], &outbuf[i+6]);
 193                         }
 194
 195                 }
 196
 197                 /* Output any lines terminated with hard line breaks */
 198                 do {
 199                         did_out = 0;
 200                         if (strlen(outbuf)>0)
 201                             for (i = 0; i<strlen(outbuf); ++i) {
 202                                 if ( (i<(screenwidth-2)) && (outbuf[i]=='\n')) {
 203                                         fwrite(outbuf, i+1, 1, stdout);
 204                                         strcpy(outbuf, &outbuf[i+1]);
 205                                         i = 0;
 206                                         did_out = 1;
 207                                 }
 208                         }
 209                 } while (did_out);
 210
 211                 /* Add soft line breaks */
 212                 if (strlen(outbuf) > (screenwidth - 2)) {
 213                         rb = (-1);
 214                         for (i=0; i<(screenwidth-2); ++i) {
 215                                 if (outbuf[i]==32) rb = i;
 216                         }
 217                         if (rb>=0) {
 218                                 fwrite(outbuf, rb, 1, stdout);
 219                                 fwrite("\n", 1, 1, stdout);
 220                                 strcpy(outbuf, &outbuf[rb+1]);
 221                         } else {
 222                                 fwrite(outbuf, screenwidth-2, 1, stdout);
 223                                 fwrite("\n", 1, 1, stdout);
 224                                 strcpy(outbuf, &outbuf[screenwidth-2]);
 225                         }
 226                 }
 227
 228         } while (done_reading == 0);
 229         fwrite(outbuf, strlen(outbuf), 1, stdout);
 230         fwrite("\n", 1, 1, stdout);
 231
 232         inptr = mallok(100);
 233         strcpy(inptr, "This is eekish.\n");
 234         return inptr;
 235
 236 }
 237