/*
 * andsharp -d
 *
 * -d specifies decoding.  Someday, other options may be specifiable.
 */

#include <stdio.h>
#include <stdlib.h>
#include <strings.h>

extern const char *__progname;

#define MAXCHNAME 64

static int iline;

static void printch(const unsigned char *n, int l)
{
 switch (l)
  { case 2:
       switch (n[0] | (n[1] * 256))
	{ case 'l' | ('t' * 256): putchar('<'); return;
	  case 'g' | ('t' * 256): putchar('>'); return;
	}
       break;
    case 3:
       if (!bcmp(n,"amp",3))
	{ putchar('&');
	  return;
	}
       break;
    case 4:
       if (!bcmp(n,"quot",4))
	{ putchar('"');
	  return;
	}
       if (!bcmp(n,"nbsp",4))
	{ /* This theoretically should be ' ', but the way &nbsp; is
	     actually used, ' ' is a better approximation. */
	  putchar(' ');
	  return;
	}
       break;
  }
 fprintf(stderr,"%s: line %d: unknown character name %.*s\n",__progname,iline,l,n);
}

static void decode(void)
{
 enum { S_IDLE = 1, S_AND, S_SHARP, S_HEX, S_CHNAM } state;
 unsigned char chname[MAXCHNAME];
 int chnl;
 int ccode;
 int c;

 iline = 1;
 while (1)
  { c = getchar();
    if (c == EOF) break;
    switch (state)
     { case S_IDLE:
	  if (c == '&')
	   { state = S_AND;
	   }
	  else
	   { putchar(c);
	   }
	  break;
       case S_AND:
	  if (c == '#')
	   { state = S_SHARP;
	     ccode = 0;
	   }
	  else
	   { state = S_CHNAM;
	     chname[0] = c;
	     chnl = 1;
	   }
	  break;
       case S_SHARP:
	  switch (c)
	   { case '0' ... '9':
		ccode = (ccode * 10) + (c - '0');
		if (ccode > 255)
		 { if (ccode <= 2559) /* ccode/10 <= 255 */
		    { fprintf(stderr,"%s: line %d: out-of-range code\n",__progname,iline);
		    }
		   ccode = 10000;
		 }
		break;
	     case 'x':
		/* Slightly wrong; this permits &#0x.. to be taken as hex.
		   Not troublesome enough to worry about. */
		if (ccode == 0)
		 { state = S_HEX;
		   break;
		 }
		/* fall through */
	     default:
		fprintf(stderr,"%s: line %d: invalid character ",__progname,iline);
		if (((c >= 32) && (c <= 126)) || (c >= 160))
		 { fprintf(stderr,"`%c'",c);
		 }
		else
		 { fprintf(stderr,"0x%02x",c);
		 }
		fprintf(stderr," in code, inserting terminator\n");
		ungetc(c,stdin);
		c = 0;
		/* fall through */
	     case ';':
		if (ccode < 256) putchar(ccode);
		state = S_IDLE;
		break;
	   }
	  break;
       case S_HEX:
	  switch (c)
	   { case '0' ... '9':
		ccode = (ccode * 16) + (c - '0');
		if (0)
		 {
	     case 'a' ... 'f':
		   ccode = (ccode * 16) + (c - 'a') + 10;
		 }
		if (0)
		 {
	     case 'A' ... 'F':
		   ccode = (ccode * 16) + (c - 'A') + 10;
		 }
		if (ccode > 255)
		 { if (ccode <= 4095) /* ccode/16 <= 255 */
		    { fprintf(stderr,"%s: line %d: out-of-range code\n",__progname,iline);
		    }
		   ccode = 10000;
		 }
		break;
	     default:
		fprintf(stderr,"%s: line %d: invalid character ",__progname,iline);
		if (((c >= 32) && (c <= 126)) || (c >= 160))
		 { fprintf(stderr,"`%c'",c);
		 }
		else
		 { fprintf(stderr,"0x%02x",c);
		 }
		fprintf(stderr," in code, inserting terminator\n");
		ungetc(c,stdin);
		c = 0;
		/* fall through */
	     case ';':
		if (ccode < 256) putchar(ccode);
		state = S_IDLE;
		break;
	   }
	  break;
       case S_CHNAM:
	  if (c == ';')
	   { printch(&chname[0],chnl);
	     state = S_IDLE;
	   }
	  else if (chnl < MAXCHNAME)
	   { chname[chnl++] = c;
	   }
	  else
	   { fprintf(stderr,"%s: line %d: character name too long: %.*s\n",__progname,iline,MAXCHNAME,&chname[0]);
	     state = S_IDLE;
	   }
	  break;
       default:
	  abort();
	  break;
     }
    if (c == '\n') iline ++;
  }
 switch (state)
  { case S_IDLE:
       break;
    case S_AND:
       putchar('&');
       break;
    case S_SHARP:
    case S_HEX:
       printf("&#%d",ccode);
       break;
    case S_CHNAM:
       printch(&chname[0],chnl);
       break;
  }
}

int main(int, char **);
int main(int ac, char **av)
{
 if (0)
  {
usage:;
    fprintf(stderr,"Usage: %s -d\n",__progname);
    exit(1);
  }
 if (ac != 2) goto usage;
      if (!strcmp(av[1],"-d")) decode();
 else goto usage;
 exit(0);
}