/*
 * Crd2txt -- convert Windows card files to text files
 *
 * 1997,2000,2002 Hannu.Aronsson@iki.fi
 *
 * File format looks basically like this (see code for details). This
 * Interpretation has been tested on card files of up to about 1100
 * cards and 97k of data total.
 *
 * Header: Contains magic number (3 bytes) and number of cards.
 *
 * Title records: number of card times from offset 11 are 52-byte
 * records with rest-of-text offset in 2 shorts (within-page
 * address and page number) and the title text (terminated by NULL).
 *
 * Rest of text records: At indicated offset, there is a text length
 * short and then the text, with CRLF as line separator.
 *
 * Character set in all strings is Windows, i.e. about Latin1 8859-1
 *
 * Updates:
 * 02/2002: Some cardfile files are packed tighter, relaxed check.
 * 02/2002: Some cardfile files do not hame magic number "MGC"?
 */

#include <fcntl.h>
#include <stdlib.h>
#include <stdio.h>
#ifdef TURBO_C
#include <io.h>
#else
#include <unistd.h>
#endif
#include <assert.h>

void seeknread(int f, unsigned long where,
               unsigned char *buf, unsigned long len)
{
  unsigned long r;

  r = lseek(f, where, SEEK_SET);
  if (!(r==where)) printf("seek r=%lu where=%lu\n", r, where);
  assert(r == where);

  r = read(f, buf, len);
  if (!(r==len)) printf("read r=%lu len=%lu\n", r, len);
  assert(r == len);
}

unsigned short readshort(int f, unsigned long where)
{
  unsigned char buf[2];

  seeknread(f, where, buf, 2);
  return buf[0] + buf[1]*256;   /* Windoze is big endian */
}

#define MAXTEXT 2048

main(int ac, char **av)
{
  unsigned char buf[MAXTEXT];
  int f;
  int ncards, i, textlen, last;
  unsigned long rec, textloc, filelen, loc;

  if (ac < 2) {
    printf("Usage: crd2html file.crd > output.htm\n");
    printf("(c) 1997,2000,2002 http://www.ihaa.com/english/crd2html.html\n");
    exit(0);
  }

  f = open(av[1], O_RDONLY
#ifdef TURBO_C
   | O_BINARY
#endif
   );
  assert(f >= 0);

  filelen = lseek(f, 0, SEEK_END);
  assert(filelen > 8);

  seeknread(f, 0, buf, 3);
  /* Too strict? assert(!strncmp(buf, "MGC", 3)); */
  if (strncmp(buf, "MGC", 3)) {
    printf("<!-- Warning: File magic number is not \"MGC\" -->\n");
  }

  ncards = readshort(f, 3);
  assert(ncards >= 0 && ncards <= 4096);

  printf("<!-- Converting .CRD file %s with %d cards -->\n",
         av[1], ncards);
  printf("<UL>\n");

  for (rec = 0; rec < ncards; rec++) {
    loc = 11 + rec*52;
    textloc = readshort(f, loc); /* This is address within 64K */
    textloc += readshort(f, loc+2) * 65536; /* This is "page" number */
    assert(textloc >= 57 && textloc < filelen);

    seeknread(f, loc+5, buf, 52-5);
    /*    printf("<!-- Converted card %d of %d -->\n", rec+1, ncards); */
    printf("<LI><B>%s</B><BR>\n", buf);

    textlen = readshort(f, textloc + 2);
    assert(textlen >= 0 && textlen < filelen - textloc &&
           textlen < MAXTEXT);

    seeknread(f, textloc + 4, buf, textlen);
    buf[textlen] = '\0';

    last = '\n';
    /* Text har CRLF line endings */
    for (i=0; i< textlen; i++)
      switch(buf[i]) {
      case '\n':
        if (last != '\n')       /* Print only one of consecutive LFs */
          printf("<BR>\n");
        last = '\n';
        break;
      case '\r':
        /* Ignore, no change to last */
        break;
      default:
        putchar(buf[i]);
        last = buf[i];
      }
    printf("\n", buf);
  }

  printf("</UL>\n");
  printf("<!-- .CRD to HTML converter http://www.ihaa.com/english/crd2html.html -->\n\n");

  close(f);

  return 0;
}

/* EOF */