#!/usr/bin/perl -w
# This Script downloads the Directory Information from Dhiraagu Edirectory Website
# 
# LICENSE: Feel Free to modify to suit your needs...but pls keep this header !
# FourThirty Inc. 2008
#
# by Typos v 0.2 -- typos@fourthirty.org
# 22-05-2008
#
# CHANGELOG -----
# * Has the ability to continue when invoked with -C or -c
# * Few bug fixes to work with the new version of Dhiraagu site
# ---------------
#
print "Edir v 0.2 by Typos:\n" ;

use strict;
use lib "/common";
use LWP;

my $page = 0 ;
my $cont = 0 ;
my $cont_post = 0 ;

# Checks whether edir is invoked with -C option
foreach(@ARGV) {
	if ($_ eq '-c' || $_ eq '-C') {

		$cont = 1 ;
		my $file = '.edir.tmp' ;

		# reads .edir.tmp
		open EDIR, $file
			or die  "Could not open '$file': $!\n" ;
		my @content = <EDIR> ;
		close EDIR ;

		# sets $page to start from last stopped
        	foreach (@content) {
                	if (/nextPage/) {
                        	m/index\.php\?action=edir&page=(\d*)"/ ;
                        	$page = $1 ;
                	}
			
		}
		# sets POST value to number that was procession last
		foreach (@content) {
			if (/<td align="center">(.).*<\/td>/) {
				$cont_post = $1 ;
			}
		}
		last ;
	}
}

# All dhiraagu numbers start with one of these
# seems useless to check all 0-9 
# -- updated in 0.2 to include cont. functionality
#
my @numbers ;
if ($cont == 1) {
	if ($cont_post == 3) {

		@numbers = qw { 3 6 7 } ;

		} elsif ($cont_post == 6) {
			@numbers = qw { 6 7 } ;

		} elsif ($cont_post == 7) {
			@numbers = qw { 7 } ;
		}
} else {
	@numbers = qw { 3 6 7 } ;
}

foreach (@numbers) {

	my $post = $_ ;
	my $lp_control = 0 ;	# Variable used to drop out of the while loop...

	# tells to continue from prev. page and prints status
	if ($cont == 1) {
		$post = $cont_post ;
		print "Continuing from Number: $post and Page: $page\n" ;
		$cont = 0 ;
	}
	select STDOUT ;
	print "Fetching entries starting with $post: \n";


# Fetch each page 1 by 1 starting with number
# 
while ($lp_control != 1) {

	my $browser = LWP::UserAgent->new;  
	$browser->default_headers->push_header('Accept' => "text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5");
	my $URL = "http://www.dhivehinet.net.mv/index.php?action=edir&page=$page";
	my $response = $browser->post( $URL,
    	[ 'query' => "$post",
    	]
	);
	my $content = $response->decoded_content ;
	my $file = '.edir.tmp' ;

	open EDIR, "> $file"
        	or die "Could not open '$file': $!\n" ;

	select EDIR ; print $content ; select STDOUT ;
	close EDIR ;

	open EDIR, $file
		or die  "Could not open '$file': $!\n" ;
	my @content = <EDIR> ;
	close EDIR ;

	foreach (@content) {
		if (/nextPage/) {
			m/index\.php\?action=edir&page=(\d*)"/ ;
			$page = $1 ;
		} elsif (/You are currently viewing page (\d*) of (\d*)/) {
			if ($1 == $2) {
				#print $1 ; print $2 ;
				$lp_control = 1 ;
				$page = 0 ;
				#die ;
			}
		}
			  
	}

	# Trim Out useless html and format DATA 
	#
	my $mark = 0 ;
	foreach (@content) {
		if ((!/Directory Entry/) && ($mark == 0)) {
			s/.*\n// ;
		} else {
			if (!$mark == 1) { 
				$mark = 1 ; 
			}
			if (/Directory Entry/) {
				s/.*\n// ;
			}
			if (!/<\/td>/) {
				 s/.*\n// ;
	                }
		}
	}

	# Clears useless blank entries from @content array
	#
	foreach (@content) { 
		if ($_ eq " ") {
			shift @content ;
		}
	}

	foreach (reverse @content) {
		if (!/<br \/><\/td>/) {
			#s/.*// ;
			pop @content ;
		} else {
			pop @content ;
			last ;
		}
	}
	
	# Dump the Entries to a file (tab delimited)
	# 
	chomp (my $out_file = `date +%D`) ;
	$out_file =~ s/\//_/g ;
	$out_file = "Edir-" . $out_file . ".txt" ;

        open OUT, ">> $out_file"
                or die "Could not open '$out_file': $!\n" ;
	select OUT ;

	foreach (@content) {
			if ((/.*<td>(.+)<\/td>/) && (!/http/)) {
				# print "$page\t" ;	# used to debug
				print "$1\t" ; # name
			}
	                if ((/.*<td align="center">(.+)<\/td>/) && (!/http/)) {
				print "$1\n" ; # number
	                }
	}
	select STDOUT ;
	# print ".$page" ;
	print "." ;
	select OUT ;
}
}
select STDOUT ;
close OUT ;
unlink '.edir.tmp' ;
print "\n" ;
print "Edir: Successfully Completed\n" ;