#!/usr/bin/python

# Minas Gjoka

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

import urllib2, cookielib, httplib, gzip

from mechanize import Browser

try:
    filename = sys.argv[1]
    email = sys.argv[2]
    password = sys.argv[3]
except:
    print 'Usage:', sys.argv[0], 'outfilename_wuids email password'
    sys.exit(1)

def handler(signum, frame_unused):
    if signum ==signal.SIGINT:
        sys.exit()

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


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.addheaders = [("Accept-Encoding", "gzip, identity")]
#br.set_handle_gzip(True)

# remove comment if you get debug output
#br.set_debug_redirects(True)
#br.set_debug_responses(True)
#br.set_debug_http(True)


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


response = br.submit()
#print response.read()

appfinderRegExp = re.compile("<div class=\"container\" id=\"(\d+)\">")
openprofileRegExp = re.compile("<a href=\"http://\w+.facebook.com/profile.php\?id=(\d+)\" class=\"url fn\">")
closeprofileRegExp = re.compile("<span class=\"url fn\">(.*?)</span>")
my_networks = "http://www.facebook.com/b.php/"


hash_uids = {}
f = open(filename, "r")
for line in f:
    hash_uids[int(line)] = 1

f.close()
init_size = len(hash_uids)
print "%d uids found in file %s" % (init_size, filename)

f = open(filename, "a")  

counter = 0
while 1:
    br.clear_history()
    while 1:
        ok = 0
        try :
            sleeprandom(0,3)
            
            network = br.open(my_networks)   # random selection of up to 10 people each time
            htmlSource = network.read()
    
            ok = 1 
            

        except:
            print "Error encountered: retrying"
            ok = 0
        
        if ok==1:
            break

    counter = counter + 1

    n_closedprofiles =  len(closeprofileRegExp.findall(htmlSource))
    print "Random retrieval successful: %d/%d unique UIDs in iteration %d \t\t %d closed profiles" % \
                  (len(hash_uids), counter*10, counter, n_closedprofiles)
    

    iterator = openprofileRegExp.finditer(htmlSource)
    for match in iterator:
        uid = int(match.group(1))
        f.write('%d\n' % uid)
        f.flush()
        if uid not in hash_uids:
            url_profile = "http://www.facebook.com/profile.php?id=%d" % 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 "Profile %d retrieved" % uid
            f_profile = gzip.open("profile-%d.gz" % uid,"w")
            f_profile.write("%s\n" % htmlSource)
            f_profile.close()
            hash_uids[uid] = 1

        
    #iterator = appfinderRegExp.finditer(htmlSource)
    #for match in iterator:
    #print match.group(1)