citadel/html.c

   1 /*
   2  * html.c -- Functions which handle translation between HTML and plain text
   3  * $Id$
   4  */
   5
   6 #include <stdlib.h>
   7 #include <unistd.h>
   8 #include <stdio.h>
   9 #include <ctype.h>
  10 #include <string.h>
  11
  12
  13 /*
  14  * Convert HTML to plain text.
  15  */
  16 void html_to_ascii(int screenwidth) {
  17         char inbuf[256];
  18         char outbuf[256];
  19         char tag[1024];
  20         int done_reading = 0;
  21         char *ptr;
  22         int i, ch, did_out, rb;
  23         int nest = 0;           /* Bracket nesting level */
  24
  25         strcpy(inbuf, "");
  26         strcpy(outbuf, "");
  27
  28         do {
  29                 /* Fill the input buffer */
  30                 if ( (done_reading == 0) && (strlen(inbuf) < 128) ) {
  31                         /* FIX ... genericize this */
  32                         ptr = fgets(&inbuf[strlen(inbuf)], 127, stdin);
  33                         if (ptr == NULL) done_reading = 1;
  34                 }
  35
  36                 /* Do some parsing */
  37                 if (strlen(inbuf)>0) {
  38
  39                     /* Fold in all the spacing */
  40                     for (i=0; i<strlen(inbuf); ++i) {
  41                         if (inbuf[i]==10) inbuf[i]=32;
  42                         if (inbuf[i]==13) inbuf[i]=32;
  43                         if (inbuf[i]==9) inbuf[i]=32;
  44                         if ((inbuf[i]<32) || (inbuf[i]>126))
  45                                 strcpy(&inbuf[i], &inbuf[i+1]);
  46                         while ((inbuf[i]==32)&&(inbuf[i+1]==32))
  47                                 strcpy(&inbuf[i], &inbuf[i+1]);
  48                     }
  49
  50                     for (i=0; i<strlen(inbuf); ++i) {
  51
  52                         ch = inbuf[i];
  53
  54                         if (ch == '<') {
  55                                 ++nest;
  56                                 strcpy(tag, "");
  57                         }
  58
  59                         else if (ch == '>') {
  60                                 if (nest > 0) --nest;
  61
  62                                 if (!strcasecmp(tag, "P")) {
  63                                         strcat(outbuf, "\n\n");
  64                                 }
  65
  66                                 else if (!strcasecmp(tag, "H1")) {
  67                                         strcat(outbuf, "\n\n");
  68                                 }
  69
  70                                 else if (!strcasecmp(tag, "H2")) {
  71                                         strcat(outbuf, "\n\n");
  72                                 }
  73
  74                                 else if (!strcasecmp(tag, "H3")) {
  75                                         strcat(outbuf, "\n\n");
  76                                 }
  77
  78                                 else if (!strcasecmp(tag, "H4")) {
  79                                         strcat(outbuf, "\n\n");
  80                                 }
  81
  82                                 else if (!strcasecmp(tag, "/H1")) {
  83                                         strcat(outbuf, "\n");
  84                                 }
  85
  86                                 else if (!strcasecmp(tag, "/H2")) {
  87                                         strcat(outbuf, "\n");
  88                                 }
  89
  90                                 else if (!strcasecmp(tag, "/H3")) {
  91                                         strcat(outbuf, "\n");
  92                                 }
  93
  94                                 else if (!strcasecmp(tag, "/H4")) {
  95                                         strcat(outbuf, "\n");
  96                                 }
  97
  98                                 else if (!strcasecmp(tag, "HR")) {
  99                                         strcat(outbuf, "\n ");
 100                                         for (i=0; i<screenwidth-2; ++i)
 101                                                 strcat(outbuf, "-");
 102                                         strcat(outbuf, "\n");
 103                                 }
 104
 105                                 else if (!strcasecmp(tag, "BR")) {
 106                                         strcat(outbuf, "\n");
 107                                 }
 108
 109                                 else if (!strcasecmp(tag, "TR")) {
 110                                         strcat(outbuf, "\n");
 111                                 }
 112
 113                                 else if (!strcasecmp(tag, "/TABLE")) {
 114                                         strcat(outbuf, "\n");
 115                                 }
 116
 117                         }
 118
 119                         else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
 120                                 tag[strlen(tag)+1] = 0;
 121                                 tag[strlen(tag)] = ch;
 122                         }
 123
 124                         else if (!nest) {
 125                                 outbuf[strlen(outbuf)+1] = 0;
 126                                 outbuf[strlen(outbuf)] = ch;
 127                         }
 128                     }
 129                     strcpy(inbuf, &inbuf[i]);
 130                 }
 131
 132                 /* Convert &; tags to the forbidden characters */
 133                 if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
 134
 135                         if (!strncasecmp(&outbuf[i], "&nbsp;", 6)) {
 136                                 outbuf[i] = ' ';
 137                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 138                         }
 139
 140                         else if (!strncasecmp(&outbuf[i], "&lb;", 4)) {
 141                                 outbuf[i] = '<';
 142                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 143                         }
 144
 145                         else if (!strncasecmp(&outbuf[i], "&rb;", 4)) {
 146                                 outbuf[i] = '>';
 147                                 strcpy(&outbuf[i+1], &outbuf[i+4]);
 148                         }
 149
 150                         else if (!strncasecmp(&outbuf[i], "&amp;", 5)) {
 151                                 strcpy(&outbuf[i+1], &outbuf[i+5]);
 152                         }
 153
 154                         else if (!strncasecmp(&outbuf[i], "&quot;", 6)) {
 155                                 outbuf[i] = '\"';
 156                                 strcpy(&outbuf[i+1], &outbuf[i+6]);
 157                         }
 158
 159                 }
 160
 161                 /* Output any lines terminated with hard line breaks */
 162                 do {
 163                         did_out = 0;
 164                         if (strlen(outbuf)>0)
 165                             for (i = 0; i<strlen(outbuf); ++i) {
 166                                 if ( (i<(screenwidth-2)) && (outbuf[i]=='\n')) {
 167                                         fwrite(outbuf, i+1, 1, stdout);
 168                                         strcpy(outbuf, &outbuf[i+1]);
 169                                         i = 0;
 170                                         did_out = 1;
 171                                 }
 172                         }
 173                 } while (did_out);
 174
 175                 /* Add soft line breaks */
 176                 if (strlen(outbuf) > (screenwidth - 2)) {
 177                         rb = (-1);
 178                         for (i=0; i<(screenwidth-2); ++i) {
 179                                 if (outbuf[i]==32) rb = i;
 180                         }
 181                         if (rb>=0) {
 182                                 fwrite(outbuf, rb, 1, stdout);
 183                                 fwrite("\n", 1, 1, stdout);
 184                                 strcpy(outbuf, &outbuf[rb+1]);
 185                         } else {
 186                                 fwrite(outbuf, screenwidth-2, 1, stdout);
 187                                 fwrite("\n", 1, 1, stdout);
 188                                 strcpy(outbuf, &outbuf[screenwidth-2]);
 189                         }
 190                 }
 191
 192         } while (done_reading == 0);
 193         fwrite(outbuf, strlen(outbuf), 1, stdout);
 194         fwrite("\n", 1, 1, stdout);
 195
 196 }
 197
 198
 199 /*
 200  * Temporary main loop for testing
 201  */
 202 int main() {
 203         html_to_ascii(80);
 204         return 0;
 205 }