#!/usr/bin/python

# Minas Gjoka

import urllib,re, sys, time, random, datetime, time, signal

import urllib2, cookielib, httplib, gzip

from mechanize import Browser

def sleeprandom(minsecs,maxsecs):
    num = minsecs + random.random()*maxsecs
    print 'Sleeping for %f' % (num)
    time.sleep(num)
    

try:
    filename = sys.argv[1]
    netid = int(sys.argv[2])
    email = sys.argv[3]
    password = sys.argv[4]
except:
    print 'Usage:', sys.argv[0], 'uidfilename networkid email password'
    sys.exit(1)
    


USER_AGENT = "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.1) Gecko/20071106 Iceweasel/2.0.0.5 (Debian-2.0.0.1+dfsg-1)"

br = Browser()
br.addheaders = [("User-agent", USER_AGENT)]

br.open("https://www.facebook.com")
br.select_form("loginform")
br['email'] = email              
br['pass'] = password          
br['persistent'] = ["1"]


response = br.submit()


hash_uids = {}
fin = open(filename, "r")
for line in fin:
    hash_uids[int(line)] = 0
fin.close()

print "%d root uids found in file %s\n" % (len(hash_uids), filename)

#####read and set to append mode the list of FOFs
friend_filename = "%s-friends" % (filename)
friends_hash_uids = {}
f = open(friend_filename, "r")
for line in f:
    friends_hash_uids[int(line)] = 0
f.close()

print "%d friend uids found in file %s" % (len(friends_hash_uids), friend_filename)
f = open(friend_filename, "a")  
#####

fof_RegExp = re.compile(r"<a href=\\\"http:\\/\\/\w+.facebook.com\\/profile.php\?id=(\d+)\\\" class=\\\"fname\\\">")
counter = 0
k = hash_uids.keys()
random.shuffle(k)
for uid in k:
    samenet_friends = "http://www.facebook.com/friends/?id=%d&nk=%d" % (uid,netid)
    br.clear_history()
    while 1:
        ok = 0
        try :
            sleeprandom(0,1.5)
            
            network = br.open(samenet_friends)   # random selection of up to 10 people each time
            htmlSource = network.read()
    
            ok = 1 
        except:
            print "Error encountered: retrying downloading friends list"
            ok = 0
        
        if ok==1:
            break
    
    counter = counter + 1
    n_openprofiles =  len(fof_RegExp.findall(htmlSource))
    
    print "Random retrieval successful: %d unique friend UIDs in iteration %d \t\t %d open profiles found to process next" % \
                  (len(friends_hash_uids), counter, n_openprofiles)
    
    iterator = fof_RegExp.finditer(htmlSource)

    for match in iterator:
        f_uid = int(match.group(1))
        f.write('%d\n' % f_uid)
        f.flush()
        if (f_uid not in friends_hash_uids) and (f_uid not in hash_uids):
            url_profile = "http://www.facebook.com/profile.php?id=%d" % f_uid 
            max_waittime = 0.25
            while 1:
                 ok = 0
                 try :
                     sleeprandom(0,max_waittime)
                     
                     network = br.open(url_profile)   # random selection of up to 10 people each time
                     htmlSource = network.read()
             
                     ok = 1                  
                 except:
                     print "Error encountered: retrying"
                     ok = 0
                     max_waittime = max_waittime*2
                 
                 if ok==1:
                     break
            
            print "Friend Profile %d retrieved" % f_uid
            f_profile = gzip.open("profile-%d.gz" % f_uid,"w")
            f_profile.write("%s\n" % htmlSource)
            f_profile.close()
            friends_hash_uids[f_uid] = 1