#include <stdio.h>

/*
  read and HTML file as stdin and replace all HTML tags by a space
  unless for some special tags
  */

#define PRE -1
#define endPRE -2

int checkTag(char *s)
{
  /* check which HTML tag found */
  if (!strcasecmp(s,"<TR>")) return('\n');
  if (!strcasecmp(s,"</TH>")) return('\t');
  if (!strcasecmp(s,"</TD>")) return('\t');
  if (!strcasecmp(s,"<BR>")) return('\n');
  if (!strcasecmp(s,"<PRE>")) return(PRE);
  if (!strcasecmp(s,"</PRE>")) return(endPRE);
  return(' ');
}

void error(char *msg)
{
  fprintf(stderr,"%s\n", msg);
  exit(1);
}

main(int argc, char *argv[])
{
  char *buffer;
  int buff_max=1024;
  int buff_used=0;
  char *ptr;
  int c;
  int putit = 1;
  int blank = 0;
  int pre = 0;

  if ((buffer=(char*)malloc(buff_max))==NULL)
    error("can't allocate memory for line buffer");

  /* read the file */
  while ((c = getc(stdin)) != EOF)
    {
      switch (c)
	{
	case '<': /* begin of tag */
	  ptr = buffer;
	  buff_used=1;
	  *ptr = c;
	  ptr++;
	  putit = 0;
	  break;
	  
	case '>': /* end of tag */
	  ptr[0] = c;
	  ptr[1] = 0;
	  c = checkTag(buffer);
	  switch (c)
	    {
	    case PRE:
	      pre = 1; break;
	    case endPRE:
	      pre = 0; break;
	    case ' ':
	      if (!blank) putchar(c);
	      blank = 1;
	      break;
	    default:
	      blank = 0;
	      putchar(c);
	    }
	  putit = 1;
	  break;
	  
	case '\n': /* end of line is not considered unless we are not in PRE section */
	  if (pre) putchar(c);
	  break;
	  
	default:
	  if (putit) /* the character has to be printed */
	    {
	      if (c == ' ' && !pre)
		{
		  if (!blank) putchar(c);
		  blank = 1;
		}
	      else
		{
		  putchar(c);
		  blank = 0;
		}
	      }
	  else /* just store the character (in a tag) */
	    {
	      if (buff_used > (buff_max - 3))
		{
		  buff_max += 1024;
		  if ((buffer=(char*)realloc(buffer,buff_max))==NULL)
		    error("can't realloc line buffer");
		  ptr = buffer + buff_used;
		}
	      buff_used++;
	      *ptr = c;
	      ptr++;
	    }
	}
    }
  free(buffer);
}