#!/usr/bin/env perl
#
#usage:  get.pl anyfile
#or: get.pl http://anything.anywhere.com/anyfile.html
#
use LWP::Simple; #provides  ability to access files via http,

$filename=@ARGV[0]; #get filename from command line
if ($filename=~m+http://+) {
	print "connecting to: $filename\n";
	$content=get($filename) or die("no $filename, die");
}
else{
	print "reading local file: $filename\n";
	local(*IN, $/); #see http://perldoc.com/perl5.8.0/pod/perlvar.html for $/ 
	#entire file wll go into a string because of $/ above, otherwise <IN> makes array
	open IN,"<$file";
	$content=<IN> or die("no $filename, die");
	close IN;
}

#see http://perldoc.com/perl5.8.0/pod/perlrequick.html
#The pattern strings are placed within qw# # instead of " or ', 
#because " and ' are within the patterns
$pats{"http"}=qw#http://(.*?)[\"\s\']#; 
$pats{"relative url"}=qw#href=\s*[\"\'](.*?)[\"\']#; 
$notpats{"relative url"}=qw#http://|mailto:#;
$pats{"email address"}=qw#([\w\.]+\@[\w\.]+)#;

#your task: comment out the above and make the following functional:

#find url of frames, (check at wwww.ou.edu):
#$pats{'frame'}= 

#find "Bobby noncompliant images, i.e. images without alt tag
#check at:
#   http://soonersports.ocsn.com
#   http://it.metr.ou.edu/rgraphics/radar
#   http://weather.ou.edu, etc.
#$pats{'bobby violation'}=
#$notpats{'bobby violation'}=

#find downloadable pdf documents.  check at:
# http://www.ou.edu/provost/pronew/ 
# http://weather.ou.edu/grad.htm
#$pats{'pdf'}=

for $key (sort keys %pats){
	$pat=$pats{$key};
	$notpat=$notpats{$key};
	print "\n$key: looking for $pat ";
	if ($notpat) {print "without $notpat :\n"} else {print":\n"}
	$count=0;
	while ($content=~m#$pat#sg) { #cannot use ~m/$pat/ because / may be within $pat
		$thematch=$1;
		if ($notpat){
			next if $thematch=~m#$notpat#s;
		}
		$count++;
		print " $count: $thematch\n";
	}
}