#!/usr/bin/perl
# $Header: Penyemakan Ejaan (html2text) $

use integer;
use HTML::Parser;
use CGI;

my $parser = HTML::Parser->new('api_version' => 3,
			       'start_h' => [ \&start, "tagname, attr" ],
			       'end_h' => [ \&end, 'tagname' ],
			       'text_h' => [ \&text, 'dtext' ]);

##</code><code>##

# stack is a stack of $doc values
my @stack;

# $count is used to count section numbers within the document
# (each section being defined by an anchor)
my $count;

# $doc accumulates the text of the document.  It might be a string, or
# an array ref containing subsiduary $doc values or further array refs
# which contain a tag and a $doc value.
my $doc;
my $debug = 0;

my $width = 76;

my %names;

# FILE OUTPUT
open (Out, '/home/ayob/temp/test.txt') || die "Fail tidak ada\n";
   @file = <Out>;
   $file = @file;
close Out;

open (In, '>/home/ayob/projekmajor/cc') || die "File does not exists\n";
   for ($n = 0; $n <= $file; $n++) {
      $_ = $file[$n];
      print In $_;
   }
   $n--;
   print In "\n$n.\t$operator-$nombor\t$nama";
close In;

# restore the previous context, and return the old one
sub popit {
  my $d = $doc;
  $doc = pop(@stack);
  print In "POP[$#stack]: doc-> {$doc}\n" if $debug;
  return $d;
}

# handle a start tag

sub start {
  my ($tag, $attrs) = @_;

  return if $tag eq "ul";
  if($tag ne "a") {
    if(!defined $doc) {
      $doc = [];
    }
  }
  print In "<$tag>\n" if $debug;
  print In map("  $_=$$attrs{$_}\n", keys %$attrs) if $debug;
  if($tag eq "h1"
     || $tag eq "h2"
     || $tag eq "h3"
     || $tag eq "p"
     || $tag eq "cite"
     || $tag eq "pre"
     || $tag eq "li") {
    push(@stack, $doc);
    $doc = "";
    print In "NEW[$#stack]: doc-> {$doc}\n" if $debug;
  } elsif($tag eq "table"
	  || $tag eq "tr"
	  || $tag eq "td"
	  || $tag eq "th") {
    push(@stack, $doc);
    if($tag eq "td" || $tag eq "th") {
      undef $doc;
      print In "NEW[$#stack]: doc-> {undef}\n" if $debug;
    } else {
      $doc = [];
      print In "NEW[$#stack]: doc-> {$doc}\n" if $debug;
    }
  } elsif($tag eq "a") {
    if(exists $$attrs{'href'}
       && $$attrs{'href'} =~ /^\#/) {
      my $name = $$attrs{'href'};
      $name =~ s/^\#//;
      if(!exists $names{$name}) {
	$names{$name} = $count++;
      }
      text("[" . $names{$name} . "] ");
    } elsif(exists $$attrs{'name'}) {
      my $name = $$attrs{'name'};
      if(exists $names{$name}) {
	text("[" . $names{$name} . "] ");
      }
    }
  } elsif($tag eq "em") {
    text("_");
  }
}

# handle an end tag

sub end {
  my ($tag) = @_;

  print In "</$tag>\n" if $debug;
  if($tag eq "h1"
     || $tag eq "h2"
     || $tag eq "h3"
     || $tag eq "p"
     || $tag eq "cite"
     || $tag eq "pre"
     || $tag eq "li") {
    my $d = popit;
    push (@$doc, [$tag, $d]);
  } elsif($tag eq "em") {
    text("_");
  } elsif($tag eq "table") {
    my $table = popit;
    my $width = 0;
    for my $row (@$table) {
      my $thiswidth = $#$row + 1;
      $width = $thiswidth if($thiswidth > $width);
    }
    push(@$doc, ["table", $table, $width]);
  } elsif($tag eq "tr") {
    # XXX rowspan
    my $row = popit;
    push(@$doc, $row);
  } elsif($tag eq "td" || $tag eq "th") {
    # XXX colspan
    if(!defined $doc) {
      $doc = "";
    }
    my $cell = popit;
    push(@$doc, $cell);
  }
}

# handle text

sub text {
  my ($text) = @_;

  if(!defined $doc) {
    $doc = "";
  }
  if(!ref $doc) {
    $doc .= $text;
    print In "TXT[$#stack]: doc-> {$doc}\n" if $debug;
  }
}

sub format ($$) {
  my ($doc, $width) = @_;
  my @ret = ();

  if(!ref $doc) {
    return &format([["p", $doc]], $width);
  }
  for my $e (@$doc) {
    my ($type, $value, $cols) = @$e;

    if($type eq "h1") {
      $value =~ s/\s+/ /;
      $value =~ s/^ //;
      $value =~ s/ $//;
      my $l = length $value;
      push(@ret, "") if($#ret >= 0 && $ret[$#ret] ne "");
      my $padding = ($l < $width
		     ? (" " x (($width - $l) / 2))
		     : "");
      push(@ret, $padding . $value);
      push(@ret, $padding . ("-" x $l));
    } elsif($type eq "p"
	    || $type eq "cite"
	    || $type eq "h2"
	    || $type eq "h3"
	    || $type eq "li") {
      my @s = split(/\s+/, $value);

      my @break = ($type eq "p" || $type eq "li") ? ("") : ("", "");
      push(@ret, @break) if($#ret >= 0 && $ret[$#ret] ne "");
      my $x;
      my $line;
      my $prefix = "";
      if($type eq "li") {
	$prefix = "  ";
	$line = " *";
	$x = 2;
      }
      for my $word (@s) {
	if(defined $line && $x + 1 + length $word > $width) {
	  if(defined $line) {
	    push(@ret, $line);
	    push(@ret, '-' x length $line) if($type eq 'h2');
	  }
	  undef $line;
	}
	if(!defined $line) {
	  $line = $prefix;
	  $x = length $prefix;
	}
	if($x != 0) {
	  $line .= " ";
	  ++$x;
	}
	$line .= "$word";
	$x += length $word;
      }
      if(defined $line) {
	push(@ret, $line);
	push(@ret, '-' x length $line) if($type eq 'h2');
      }
    } elsif($type eq "pre") {
      push(@ret, "") if($#ret >= 0 && $ret[$#ret] ne "");
      push(@ret, split(/\n/, $value));
    } elsif($type eq "table") {
      push(@ret, "") if($#ret >= 0 && $ret[$#ret] ne "");
      # for each row, fit into 1/N of the space available
      # XXX do something more sophisticated
      my $colwidth = ($width - 1) / $cols - 1;

      # this will be the maximum column width
      my @max;

      for my $try (1, 2) {
	my $divider;
	my $total;
	if($try == 2) {
	  $total = -1;
	  for my $w (@max) {
	    $total += $w + 1;
	  }
	  push(@ret, "," . ("-" x $total) . ".");
	  $divider = "+";
	  for my $w (@max) {
	    $divider .= "-" x $w;
	    $divider .= "+";
	  }
	}
	my $first = 1;
	for my $row (@$value) {
	  if ($first) {
	    $first = 0;
	  } elsif($try == 2) {
	    push(@ret, $divider);
	  }
	  # format each cell
	  my @formatted = ();
	  for my $cell (@$row) {
	    my @f = &format($cell, $colwidth);
	    push(@formatted, \@f);
	  }
	  # pick apart the formatted cells row by row, and reassemble
	  my $r = 0;
	  for (;;) {
	    my $line = "|";
	    my $keepgoing = 0;
	    my $c = 0;
	    for my $cell (@formatted) {
	      my $part;
	      if ($r <= $#$cell) {
		$part = $$cell[$r];
		$keepgoing = 1;
	      } else {
		$part = "";
	      }
	      my $pl = length $part;
	      if($try == 1) {
		++$pl if($pl < $colwidth);
		if(!defined $max[$c] || $pl > $max[$c]) {
		  $max[$c] = $pl;
		}
	      } else {
		$part .= " " x ($max[$c] - $pl) if($pl < $max[$c]);
		$line .= "$part|";
	      }
	      ++$c;
	    }
	    last if !$keepgoing;
	    if($try == 2) {
	      push(@ret, $line);
	    }
	    ++$r;
	  }
	}
	if($try == 2) {
	  push(@ret, "`" . ("-" x $total) . "'");
	}
      }
    }
  }
  return @ret;
}

my $files = 0;

while($#ARGV >= 0) {
  local $_ = shift;

  if(/^-/) {
    last if($_ eq "--");
    if($_ eq "-debug") {
      ++$debug;
    } else {
      die "$0: unknown option '$_'\n";
    }
  } else {
    $doc = [];
    $count = 1;
    %names = ();
    $parser->parse_file($_);
    (print STDOUT map("$_\n", &format($doc, $width)))
      or die "$0: writing to stdout: $!\n";
    ++$files;
  }
}

if(!$files) {
  $doc = [];
  $count = 1;
  %names = ();
  $parser->parse_file(*STDIN);
  (print STDOUT map("$_\n", &format($doc, $width)))
    or die "$0: writing to stdout: $!\n";
}

(close STDOUT) or die "$0: closing stdout: $!\n";