/*
 *****************************************************************************
 *
 *	$RCSfile: xml2utf.c,v $
 *	$Date: 1999/05/18 18:36:20 $
 *	$Source: /home/richard/Xml/RCS/xml2utf.c,v $
 *	$Revision: 1.6 $
 *	$Author: richard $
 *
 *****************************************************************************
 *
 *  Copyright 1998, Brown University and Richard Goerwitz
 *
 *****************************************************************************
 *
 *  Simple utility for converting XML to UTF-8 or UTF-16.
 *
 *****************************************************************************
 */

#include "general.h"
#include "errabort.h"
#include "readcfg.h"
#include "utfutil.h"

xmlparse_environment xmlparse_env;

int
main (int argc, char **argv)
{

  u_int16_t uc;
  int output_utf_8 = 1;
  int i, j, c, were_errors;
  my_wchar_t *wp, charbuf[BUFSIZ];

  /* zero out the xmlparse_env structure */
  memset (&xmlparse_env, 0, sizeof (xmlparse_env));

  /* Set up xmlparse_env; parse command-line opts; read/parse config file */
  readcfg (argc, argv);

  xwrap (errdebug (3, "re-parsing command-line arguments (third pass)\n"));
  while ((c = getopt (argc, argv, "1:8c:C:d:E:hl:m:p:su:")) != EOF)
    {
      switch (c)
	{
	case 'h':
	  /* emit an extensive help message */
	  fprintf (stderr, "Usage:  %s [same arguments as for xmlparse] [-8] [-16] <filename>\n", argv[0]);
	  exit (1);
	  break;
	case '8':
	  output_utf_8 = 1;
	  break;
	case '1':
	  if (*optarg == '6')
	    {
	      output_utf_8 = 0;
	      break;
	    }
	case '?':
	case ':':
	  errabort (10, "syntax error; invoke w/ -h option\n");
	  break;
	default:
	  /* other command-line arguments are handled in readcfg() */
	  break;
	}
    }

  if ((xmlparse_env.filelen = argc - optind) == 0)
    errabort (11, "no files to process\n");
  else
    {
      /* Stuff all input files supplied on the cmd line into
       * xml_file structures within xmlparse_env.
       */
      xwrap (errdebug (5, "initializing %d xml_file structs\n", xmlparse_env.filelen));
      xmlparse_env.xml_files = malloc (sizeof (xml_file) * xmlparse_env.filelen);
      if (xmlparse_env.xml_files == NULL)
	errabort (40, "malloc error in %s\n", "main()");
      for (i = 0; xmlparse_env.filelen > i; i++)
	xmlparse_env.xml_files[i] = create_xml_file (argv[optind + i]);
    }

  /* process xml files */
  were_errors = 0;
  xwrap (errdebug (5, "converting files (qua xml_file structs)...\n"));
  for (i = 0; i < xmlparse_env.filelen; i++)
    {
      /* check for NULL here (indicates un-openable file) */
      if (xmlparse_env.xml_files[i])
	{
	  while ((c = read_xml_file (xmlparse_env.xml_files[i], charbuf, BUFSIZ)))
	    {
	      if (! output_utf_8)
		{
		  /* c (count) includes the trailing nil, 0; remove it */
		  c--;
		  /* my_wchar_t may be "int", so we have to downsize */
		  for (j = 0; j < c; j++)
		    {
		      uc = charbuf[j];
		      /* UTF-16 chars are sixteen bits wide by definition */
		      fwrite (&uc, sizeof (u_int16_t), 1, stdout);
		    }
		}
	      else
		{
		  wp = charbuf;
		  if (*wp == 0xFFFE || *wp == 0xFEFF)
		    /* strip out byte-order marker */
		    { wp++; c--; }
		  fputs (utf_16_to_utf_8 (wp), stdout);
		}
	    }
	}
    }

  /* We're done! */
  return 0;

}