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  * Convert HTML to plain text.
  34  */
  35 char *html_to_ascii(char *inputmsg, int screenwidth) {
  36         char inbuf[256];
  37         char outbuf[256];
  38         char tag[1024];
  39         int done_reading = 0;
  40         char *readptr, *outputbuf;
  41         size_t input_length, this_read, output_length;
  42         int i, ch, did_out, rb;
  43         int nest = 0;           /* Bracket nesting level */
  44
  45         input_length = strlen(inputmsg);
  46         readptr = inputmsg;
  47         output_length = strlen(inputmsg);
  48         outputbuf = mallok(output_length);
  49         if (outputbuf==NULL) return NULL;
  50         strcpy(inbuf, "");
  51         strcpy(outbuf, "");
  52
  53         lprintf(9, "Decoding %d bytes of HTML\n", input_length);
  54
  55         do {
  56                 /* Fill the input buffer */
  57                 if ( (done_reading == 0) && (strlen(inbuf) < 128) ) {
  58
  59                         /* copy from the input buffer */
  60                         lprintf(9, "input loop\n");
  61                         this_read = strlen(readptr);
  62                         if (this_read > 127) this_read = 127;
  63                         lprintf(9, "%d bytes\n", this_read);
  64                         for (i=0; i<this_read; ++i) {
  65                                 inbuf[strlen(inbuf)+1] = 0;
  66                                 inbuf[strlen(inbuf)] = readptr[0];
  67                                 ++readptr;
  68                         }
  69
  70                         if (strlen(readptr)==0) done_reading = 1;
  71                 }
  72                 else {
  73                         lprintf(9, "skipped input loop\n");
  74                 }
  75
  76                 /* Do some parsing */
  77                 lprintf(9, "parse loop\n");
  78                 if (strlen(inbuf)>0) {
  79
  80                     /* Fold in all the spacing */
  81                         lprintf(9, "spacing loop\n");
  82                     for (i=0; i<strlen(inbuf); ++i) {
  83                         if (inbuf[i]==10) inbuf[i]=32;
  84                         if (inbuf[i]==13) inbuf[i]=32;
  85                         if (inbuf[i]==9) inbuf[i]=32;
  86                         if ((inbuf[i]<32) || (inbuf[i]>126))
  87                                 strcpy(&inbuf[i], &inbuf[i+1]);
  88                         while ((inbuf[i]==32)&&(inbuf[i+1]==32))
  89                                 strcpy(&inbuf[i], &inbuf[i+1]);
  90                     }
  91
  92                         lprintf(9, "foo loop\n");
  93                     for (i=0; i<strlen(inbuf); ++i) {
  94
  95                         ch = inbuf[i];
  96
  97                         if (ch == '<') {
  98                         lprintf(9, "bar loop\n");
  99                                 ++nest;
 100                                 strcpy(tag, "");
 101                         }
 102
 103                         else if (ch == '>') {
 104                         lprintf(9, "baz loop\n");
 105                                 if (nest > 0) --nest;
 106
 107                                 if (!strcasecmp(tag, "P")) {
 108                                         strcat(outbuf, "\n\n");
 109                                 }
 110
 111                                 else if (!strcasecmp(tag, "H1")) {
 112                                         strcat(outbuf, "\n\n");
 113                                 }
 114
 115                                 else if (!strcasecmp(tag, "H2")) {
 116                                         strcat(outbuf, "\n\n");
 117                                 }
 118
 119                                 else if (!strcasecmp(tag, "H3")) {
 120                                         strcat(outbuf, "\n\n");
 121                                 }
 122
 123                                 else if (!strcasecmp(tag, "H4")) {
 124                                         strcat(outbuf, "\n\n");
 125                                 }
 126
 127                                 else if (!strcasecmp(tag, "/H1")) {
 128                                         strcat(outbuf, "\n");
 129                                 }
 130
 131                                 else if (!strcasecmp(tag, "/H2")) {
 132                                         strcat(outbuf, "\n");
 133                                 }
 134
 135                                 else if (!strcasecmp(tag, "/H3")) {
 136                                         strcat(outbuf, "\n");
 137                                 }
 138
 139                                 else if (!strcasecmp(tag, "/H4")) {
 140                                         strcat(outbuf, "\n");
 141                                 }
 142
 143                                 else if (!strcasecmp(tag, "HR")) {
 144                                         strcat(outbuf, "\n ");
 145                                         for (i=0; i<screenwidth-2; ++i)
 146                                                 strcat(outbuf, "-");
 147                                         strcat(outbuf, "\n");
 148                                 }
 149
 150                                 else if (!strcasecmp(tag, "BR")) {
 151                                         strcat(outbuf, "\n");
 152                                 }
 153
 154                                 else if (!strcasecmp(tag, "TR")) {
 155                                         strcat(outbuf, "\n");
 156                                 }
 157
 158                                 else if (!strcasecmp(tag, "/TABLE")) {
 159                                         strcat(outbuf, "\n");
 160                                 }
 161
 162                         }
 163
 164                         else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
 165                                 tag[strlen(tag)+1] = 0;
 166                                 tag[strlen(tag)] = ch;
 167                         }
 168
 169                         else if (!nest) {
 170                                 outbuf[strlen(outbuf)+1] = 0;
 171                                 outbuf[strlen(outbuf)] = ch;
 172                         }
 173                     }
 174                     strcpy(inbuf, &inbuf[i]);
 175                 }
 176
 177                 lprintf(9, "checkquepoynte\n");
 178
 179                 /* Convert &; tags to the forbidden characters */
 180                 if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
 181                         lprintf(9, "eek loop\n");
 182
 183                         if (!strncasecmp(&outbuf[i], "&nbsp;", 6)) {
 184                                 outbuf[i] = ' ';
 185                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 186                         }
 187
 188                         else if (!strncasecmp(&outbuf[i], "&lb;", 4)) {
 189                                 outbuf[i] = '<';
 190                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 191                         }
 192
 193                         else if (!strncasecmp(&outbuf[i], "&rb;", 4)) {
 194                                 outbuf[i] = '>';
 195                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 196                         }
 197
 198                         else if (!strncasecmp(&outbuf[i], "&amp;", 5)) {
 199                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 200                         }
 201
 202                         else if (!strncasecmp(&outbuf[i], "&quot;", 6)) {
 203                                 outbuf[i] = '\"';
 204                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 205                         }
 206
 207                 }
 208
 209                 /* Make sure the output buffer is big enough */
 210                 if ((strlen(outputbuf) + strlen(outbuf) + 2) > output_length) {
 211                         lprintf(9, "realloc loop\n");
 212                         output_length = output_length + strlen(outbuf) + 2;
 213                         outputbuf = reallok(outputbuf, output_length);
 214                 }
 215
 216                 /* Output any lines terminated with hard line breaks */
 217                 lprintf(9, "output loop 1\n");
 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                                         strncat(outputbuf, outbuf, i+1);
 224                                         strcpy(outbuf, &outbuf[i+1]);
 225                                         i = 0;
 226                                         did_out = 1;
 227                                 }
 228                         }
 229                 } while (did_out);
 230
 231                 /* Add soft line breaks */
 232                 lprintf(9, "output loop 2\n");
 233                 if (strlen(outbuf) > (screenwidth - 2)) {
 234                         rb = (-1);
 235                         for (i=0; i<(screenwidth-2); ++i) {
 236                                 if (outbuf[i]==32) rb = i;
 237                         }
 238                         if (rb>=0) {
 239                                 strncat(outputbuf, outbuf, rb);
 240                                 strcat(outputbuf, "\n");
 241                                 strcpy(outbuf, &outbuf[rb+1]);
 242                         } else {
 243                                 strncat(outputbuf, outbuf, screenwidth-2);
 244                                 strcat(outputbuf, "\n");
 245                                 strcpy(outbuf, &outbuf[screenwidth-2]);
 246                         }
 247                 }
 248
 249         } while (done_reading == 0);
 250         lprintf(9, "output loop 3\n");
 251         strncat(outputbuf, outbuf, strlen(outbuf));
 252         strcat(outputbuf, "\n");
 253
 254         return outputbuf;
 255 }