url_check.pl

#!/usr/bin/perl

use strict;
use warnings;
use Time::HiRes;
use WWW::Mechanize;
use LWP::Simple;

my @urls = ("http://www.you24h.com/");
my @errorMessages = ("ERROR", "error");
	
sub countParams {
	my $numParameters = @_;
	print("The number of urls is $numParameters\n");
}

sub constainsString{
	my $content = $_[0];
	my @stringsToBeFound = @{_[1]};
	my @stringsFound =();
	my $isFound = 0;
	foreach my $currentString (@stringsToBeFound){
		if (index($content, $currentString) != -1) {
			$isFound = 1;
			push(@stringsFound, $currentString);
		}
	}
	return ($isFound, @stringsFound);

}

sub processSubLinks{
	my @links = @_;
	my @filetypesToProcess = ("css", ".jpg", ".jpeg", ".gif", ".png", ".js");
	for my $link ( @links ) {
		my ($isFound) = constainsString($link->url, @filetypesToProcess);
		if($isFound){
			my $linkStart = Time::HiRes::gettimeofday();
			print("Getting link:", $link->url, "\n");
			get($link->url);
			my $linkEnd = Time::HiRes::gettimeofday();
			print("Time: ", ($linkEnd - $linkStart)*1000," ms\n");
		}else{
			print ">>>>>>>Skipping :", $link->url, "\n";
		}
	}
}

sub checkUrls{
	my @urls = @{$_[0]};
	my @errorMessages = @{$_[1]};
	my $retries = $_[2];

	print("Number of retries: " . $retries . " \n");
	countParams(@urls);
	for my $url (@urls){
		my ($start, $end, $totals);
		
		my $count = 0;
		$totals = 0;
		for($count; $count <= $retries; $count++)
		{
			my $mech = WWW::Mechanize->new();

			$start = Time::HiRes::gettimeofday();
			
			$mech->get($url);

			my @links = $mech->links();
			processSubLinks(@links);
			
			$end = Time::HiRes::gettimeofday();
			print($url,  " testing:\n");
			print(($end - $start)*1000," ms\n");
			$totals += ($end - $start)*1000;
			my ($isError, @errorMessagesFound) = constainsString($mech->content, @errorMessages);
			
			if($isError){
				print("URL ". $url. " contains error messages: ", @errorMessagesFound);
			}

			print("****************************** RETRY","\n");

		}

		print ("Average :", ($totals/($retries+1)), "\n");
		
	}
}

my $retries = 0;
checkUrls(\@urls, \@errorMessages, $retries);
#testingPassingParameter(\@urls, $retries);
#my ($test1, @errors1) = containsErrorMessage("This does not contain flaws.");
#my ($test2, @errors2) = containsErrorMessage("This does contain ERROR.");

#print(" The Results are :". $test1.", ". $test2. "\n");
#print("The errors1 are: ", @errors1, "\n");
#print("The errors2 are: ", @errors2, "\n");