Q&D html stripper for mutt

dehtml.c

#include <stdio.h>
#include <string.h>

actualspace(register c)
{
    return (c=='\r') || (c=='\n') || isspace(c);
}


int
thesame(char *a, char *b, int len)
{
    return (strlen(a) == len) && (strncasecmp(a,b,len) == 0);
}


eatuntil(char *block)
{
    char end[80];
    int i = 0;
    register c;

    do {
	while ( (c=getchar()) != EOF && c != '<' )
	    ;

	if ( c == EOF )
	    return;

	i = 0;
	while ( (c=getchar()) != EOF && c != '>' ) {
	    if ( actualspace(c) || i >= sizeof end-1 )
		break;
	    else 
		end[i++] = c;
	}
	
	while ( c != '>' && (c=getchar()) != EOF )
	    ;

	if ( thesame(block, end, i) )
	    return;
    } while (1);
}


comment()
{
    register c;
    int cd = 0;

    while ( (c=getchar()) != EOF ) {
	if ( c == '>' ) {
	    if ( cd == 2 )
		return;
	    cd = 0;
	}
	else  if ( c == '-')
	    cd++;
	else 
	    cd = 0;
    }
}


mimeblock()
{
    char tag[80];
    int i = 0;
    register c;

    while ( (c=getchar()) != EOF && c != '>' ) {
	if ( actualspace(c) || i >= sizeof tag-1 ) 
	    break;
	else
	    tag[i++] = c;
    }
    if ( thesame("!--", tag, i ) ) {
	comment();
	return;
    }
	
    while ( c != '>' && (c=getchar()) != EOF )
	;

    if ( thesame("head", tag, i) )
	eatuntil("/head");
    else if ( thesame("style", tag, i) )
	eatuntil("/style");
}

entity()
{
    register c;
    while ( (c=getchar()) != EOF && c != ';' && !isspace(c) )
	;
    if ( c != EOF && c != ';' )
	ungetc(c, stdin);
}


main()
{
    register c;
    int squashing = 0;


    while ( (c=getchar()) != EOF ) {
	if ( c == '<' )
	    mimeblock();
	else if ( c == '&' )
	    entity();
	else if ( actualspace(c) ) {
	    if ( !squashing )
		putchar(c);
	    squashing=1;
	}
	else {
	    squashing=0;
	    putchar(c);
	}
    }
}