#!C:\perl\perl.exe

# geturls.pl rifles through a directory, and
# extracts all the URL's and email addresses it can, and 
# appends them to the appropriate files.

# the name of the directory containing files to be parsed
$directory = "html";

# the name of the file to append URLs to
$u = "newUrls.txt";

# the name of the file to append email addresses to
$e = 'webmail.txt';

$mailSearch = '([^":\s<>()/;]*@[^":\s<>()/;\.]*)';
$URLsearch = '(http://[\/\.\w\-\_\=\?\%\d\&]*)';
$URLsearch2 = '(www[\/\.\w\-\_\=\?\%\d\&]*)';
# @email = ();  # (re)initialize array

chdir($directory) or die $!;
open(EMAIL, ">>../$e") or die $!;
open(NEW_URLs, ">>../$u") or die $!;

foreach (<*>) {
	open (INFILE, $_) or die $!;
	while(<INFILE>) {
		if (m,$mailSearch,ig)	{
			print EMAIL "$&\n";
			print STDOUT "Appending $& to file";
		}
		elsif ((m,$URLsearch,ig) or (m,$URLsearch2,ig))	{
			print NEW_URLs "$&\n";
			print STDOUT "Appending $& to file";
		}


	}
	close INFILE;
}