linkgetter

#!/usr/bin/env python
#Call this script with the -u flag followed by a url and the -d flag followed by a domain (without the http://, the www. is optional)
#This script returns all unique links from the given domain inside the URL's source.
import urllib2, sys, re;

#FLAGS
URL = "";
DOMAIN = "";
valid = True;
args = sys.argv;
#TODO: URL validation, but whatever! It's easy if you really want it.
#Could've used getopt, but with just 2 parameters and a simple script, no point.
try:
    for i in range(1, len(args)):
        if args[i] == "-u": URL = args[i+1];
        if args[i] == "-d": DOMAIN = args[i+1];
except IndexError:
    valid = False;
#EOFLAGS

#TODO: Exception handling, but whatever! It's easy if you really want it. :P
if URL != "" and DOMAIN != "" and valid:
    pattern = re.compile("(https?://(www.)?%s[-A-Za-z0-9\?=\._!/]*)" % (DOMAIN));
    res = set(pattern.findall((urllib2.urlopen(URL)).read()));
    print "Found %s links." % (len(res));
    ls = "";
    for i in res:
        ls += "%s%s" % (i[0], "\n");
    print ls.rstrip();
else: print "Invalid Parameters.";