IP Casanova Script (GUI v3.3)

#!/usr/bin/env python

# Version 3.3:
# added deep and shallow filter
# moved login out of search function
# fixed offset bug in original script
# removed unnecessary wait condition

################################################################################
# IMPORTANT THINGS YOU SHOULD TOTALLY READ:
#
# Country codes: http://en.wikipedia.org/wiki/ISO_3166-1#Officially_assigned_code_elements
# Language codes: http://en.wikipedia.org/wiki/List_of_ISO_639-1_codes
# If there are multiple codes available, you want the 2-letter one.
#
# "Deep filter" will do a specific search for your filtered keywords and ensures
# they will definitely not be viewed. It cannot be restricted to only those who
# are online though, so select this if:
# - you're also searching offline users
# - you're filtering out a pretty obscure term (not "korea")
# - you don't mind waiting a while for the script to build its blacklist
#
# Otherwise, the shallow filter just checks the "about me" snippet provided,
# so some can easily slip through the cracks.
#
# The save button will save everything you have entered in the config window at
# that time, including your password, in plaintext. So if you're the kind of
# person who cares about that, make sure the password box is empty when you save.
#
# If you put in incorrect values, the script won't warn you, you just won't
# get the results you want. At the moment anyway.
################################################################################

import time
import sys
import os
import requests
import re
import random
import getpass
from Tkinter import *
import tkFileDialog
import ast

# We keep already visited profile in a file to avoid visiting
# a profile multiple times on consecutive execution of this script
# YOU CAN DELETE THIS FILE IF YOU WANT TO VISIT EVERYBODY ONCE AGAIN
visitedUsersFilename = "users_visited.txt"

# Set this variable to True to output debug log,
# or False for more concise output
DEBUG = True
##
## END OF CONFIG VARIABLES
##

def login(s, user, pw):
    payload = {'action': 'login',
               'login': user,
               'auto_login' : '0',
               'password': pw}

    print "Logging in..."
    s.get("http://www.interpals.net/")
    r = s.post("http://www.interpals.net/login.php", data=payload)
    if "<li>Authentication failed. Please try again." in r.text:
        print "Login failed, check your details and try again"
        exit()
    print "Logged in: Starting the dance \o/"
    time.sleep(2)

def search(s, url, mindel, maxdel, rundel, filtered_users, fterms):
    offset1 = -20
    offset2 = 0
    sessioncount = 0
    run_number = 1
    min_delay = float(mindel)
    max_delay = float(maxdel)
    run_delay = float(rundel)
    f = open(visitedUsersFilename, 'a+')
    processedUsers = [line.strip() for line in f]
    print "Already processed " + str(len(processedUsers)) + " users."

    while True:
        runcount = 0
        if DEBUG:
            print "Fetching the search page..."
        r = s.get(url + "&offset=%d&offset=%d" % (offset1, offset2))
        data = r.text
        for m in re.finditer('<b><a href=', data):
            username = data[m.start()+13:m.start()+28]
            if '?' in username:
                mesto = username.index('?')
                username = username[:mesto]
            culoc = data.find("iso/16/", m.start())
            cucode = data[culoc+7:culoc+9]
            if len(fterms) > 0:
                # Run the shallow filter
                donotwant = False
                bioloc = data.find("sResTxtField", m.start())
                bioend = data.find("</div>", bioloc)
                for fil in fterms:
                    filloc = data.find(fil, bioloc+14, bioend)
                    if filloc != -1:
                        print("Avoiding " + username + " due to their bio snippet")
                        donotwant = True
                if donotwant:
                    continue
            if username not in processedUsers and username not in filtered_users:
                if DEBUG:
                    print("Visiting profile of " + username + " from " + cucode + " (" + str(sessioncount) + ")")
                runcount += 1
                sessioncount += 1
                r = s.get("http://www.interpals.net/" + username)
                waitTime = random.randrange(min_delay*10, max_delay*10) / float(10)
                if DEBUG:
                    print("Waiting " + str(waitTime) + "s")
                else:
                    os.system('cls' if os.name=='nt' else 'clear')
                    print('\rRun %d - Fetched %d users (%d total)' % (run_number, runcount, sessioncount))
                time.sleep(waitTime)
                processedUsers.append(username)
                f.write(username + "\n")
            elif DEBUG and username in processedUsers:
                print("Already visited " + username)
            elif DEBUG:
                print("Avoiding " + username + ", they're blacklisted")
        run_number += 1
        offset1=offset1+20
        offset2=offset2+20
        print ('Waiting (%d)s before next run (fetched %d users this run, %d total)\n' % (run_delay, runcount, sessioncount))
        time.sleep(run_delay)


def get_filtered_users(s, url, rundel):
    offset1 = -20
    offset2 = 0
    runcount = 0
    run_delay = float(rundel)
    f = open(visitedUsersFilename, 'a+')
    filtered_users = []
    while True:
        runcount = 0
        if DEBUG:
            print("Finding out who to avoid...")
        r = s.get(url + "&offset=%d&offset=%d" % (offset1, offset2))
        data = r.text

        numfound = data.find("Showing matches ")
        endfound = data.find("</div>", numfound)
        textfound = data[numfound+16:endfound]
        print("Processing users " + textfound)
        for m in re.finditer('<b><a href=', data):
            username = data[m.start()+13:m.start()+28]
            if '?' in username:
                mesto = username.index('?')
                username = username[:mesto]
            if DEBUG:
                print("Blacklisting " + username + "...")
            filtered_users.append(username)
            runcount += 1
        if runcount == 0:
            print("Finished building blacklist!")
            return filtered_users
        else:
            offset1=offset1+20
            offset2=offset2+20
            # Wait 3x as long as a courtesy, search goes down enough as is
            print ('Waiting (%d)s before getting the next page of undesirables' % (run_delay * 3))
            time.sleep(run_delay * 3)

"""
GUI SHIT INCOMING, LOOK AWAY IT'S HIDEOUS
"""

class Controller():
    """Controls all GUI elements"""
    def __init__(self, frame):
        """
        Sadly this giant mess of declarations can't be done on the same line
        since that binds all the checkboxes to the same value
        """
        male = IntVar()
        female = IntVar()
        photo = IntVar()
        online = IntVar()
        af = IntVar()
        asia = IntVar()
        eu = IntVar()
        na = IntVar()
        oc = IntVar()
        sa = IntVar()
        email = IntVar()
        snail = IntVar()
        langex = IntVar()
        friend = IntVar()
        sluzza = IntVar()
        relo = IntVar()
        deep = IntVar()

        main_frame = Frame(frame)


        top_frame = Frame(main_frame)
        age_frame = Frame(top_frame)
        Label(age_frame, text="Age").pack(side=LEFT, padx=10, pady=5)
        tlowage = Entry(age_frame, width=5)
        tlowage.pack(side=LEFT, anchor=E)
        Label(age_frame, text="to").pack(side=LEFT, padx=10, pady=5)
        thighage = Entry(age_frame, width=5)
        thighage.pack(side=LEFT, anchor=E)
        age_frame.pack()

        sex_frame = Frame(top_frame)
        malebox = Checkbutton(sex_frame, text="Male", var=male)
        malebox.pack(side=LEFT)
        femalebox = Checkbutton(sex_frame, text="Female", var=female)
        femalebox.pack(side=LEFT)
        sex_frame.pack()

        bool_frame = Frame(top_frame)
        photobox = Checkbutton(bool_frame, text="With photo only", var=photo)
        photobox.pack(side=LEFT)
        onlinebox = Checkbutton(bool_frame, text="Online only", var=online)
        onlinebox.pack(side=RIGHT)
        bool_frame.pack()

        top_frame.pack()

        left_frame = Frame(main_frame)


        look_frame = Frame(left_frame)
        Label(look_frame, text="Looking for").pack(pady=(0, 5))
        emailbox = Checkbutton(look_frame, text="Email penpals", var=email)
        emailbox.pack()
        snailbox = Checkbutton(look_frame, text="Snail mail penpals", var=snail)
        snailbox.pack()
        langbox = Checkbutton(look_frame, text="Language exchange", var=langex)
        langbox.pack()
        friendbox = Checkbutton(look_frame, text="Friendship", var=friend)
        friendbox.pack()
        sluzzabox = Checkbutton(look_frame, text="Romance/flirting", var=sluzza)
        sluzzabox.pack()
        relobox = Checkbutton(look_frame, text="Relationship", var=relo)
        relobox.pack()
        look_frame.pack()

        left_frame.pack(side=LEFT)

        right_frame = Frame(main_frame)
        country_frame = Frame(right_frame)
        Label(country_frame, text="Countries").pack()
        countryent = Entry(country_frame, width=15)
        countryent.pack()
        cbutt_frame = Frame(country_frame)
        countrybox = Listbox(country_frame, width=20)
        baddcount = Button(cbutt_frame, text="Add", command=lambda countrybox=countrybox: countrybox.insert(END, countryent.get()))
        baddcount.pack(side=LEFT)
        bremcount = Button(cbutt_frame, text="Remove", command=lambda countrybox=countrybox: countrybox.delete(ANCHOR))
        bremcount.pack(side=RIGHT)
        cbutt_frame.pack()

        countrybox.pack()
        country_frame.pack(padx=5, pady=20, side=LEFT)

        filter_frame = Frame(right_frame)
        Label(filter_frame, text="Filters").pack()
        filterent = Entry(filter_frame, width=15)
        filterent.pack()
        fbutt_frame = Frame(filter_frame)
        filterbox = Listbox(filter_frame, width=20, height=8)
        baddfilter = Button(fbutt_frame, text="Add", command=lambda filterbox=filterbox: filterbox.insert(END, filterent.get()))
        baddfilter.pack(side=LEFT)
        bremfilter = Button(fbutt_frame, text="Remove", command=lambda filterbox=filterbox: filterbox.delete(ANCHOR))
        bremfilter.pack(side=RIGHT)
        deepbox = Checkbutton(filter_frame, text="Deep filter", var=deep)
        deepbox.pack(side=BOTTOM)
        fbutt_frame.pack()
        filterbox.pack()
        filter_frame.pack(padx=5, pady=20, side=RIGHT)

        lang_frame = Frame(right_frame)
        Label(lang_frame, text="Languages").pack()
        langent = Entry(lang_frame, width=15)
        langent.pack()
        lbutt_frame = Frame(lang_frame)
        langbox = Listbox(lang_frame, width=20)
        baddlang = Button(lbutt_frame, text="Add", command=lambda langbox=langbox: langbox.insert(END, langent.get()))
        baddlang.pack(side=LEFT)
        bremlang = Button(lbutt_frame, text="Remove", command=lambda langbox=langbox: langbox.delete(ANCHOR))
        bremlang.pack(side=RIGHT)
        lbutt_frame.pack()
        langbox.pack()
        lang_frame.pack(padx=5, pady=20, side=RIGHT)


        right_frame.pack(side=RIGHT)

        mid_frame = Frame(main_frame)

        cont_frame = Frame(mid_frame)
        Label(cont_frame, text="Only select continents or countries, not both").pack()
        afbox = Checkbutton(cont_frame, text="Africa", var=af)
        afbox.pack()
        asbox = Checkbutton(cont_frame, text="Asia", var=asia)
        asbox.pack()
        eubox = Checkbutton(cont_frame, text="Europe", var=eu)
        eubox.pack()
        nabox = Checkbutton(cont_frame, text="North America", var=na)
        nabox.pack()
        ocbox = Checkbutton(cont_frame, text="Australia/Oceania", var=oc)
        ocbox.pack()
        sabox = Checkbutton(cont_frame, text="South America", var=sa)
        sabox.pack()
        cont_frame.pack()

        mid_frame.pack(side=RIGHT)

        main_frame.pack()
        buttons = Frame(frame)
        botent_frame = Frame(buttons)
        user_frame = Frame(botent_frame)
        Label(user_frame, text="Username").pack()
        euser = Entry(user_frame, width=15)
        euser.pack()
        Label(user_frame, text="Password").pack()
        epass = Entry(user_frame, width=15, show="*")
        epass.pack()
        user_frame.pack(side=LEFT, padx=30)
        delay_frame = Frame(botent_frame)
        Label(delay_frame, text="Min delay (views)").pack()
        emindel = Entry(delay_frame, width=3)
        emindel.insert(0, "1")
        emindel.pack()
        Label(delay_frame, text="Max delay (views)").pack()
        emaxdel = Entry(delay_frame, width=3)
        emaxdel.insert(0, "3")
        emaxdel.pack()
        Label(delay_frame, text="Delay between runs").pack()
        erundel = Entry(delay_frame, width=3)
        erundel.insert(0, "5")
        erundel.pack()
        delay_frame.pack(side=RIGHT)
        botent_frame.pack()
        # Proper OOP is for losers
        bstart = Button(buttons, text="Start", command=lambda:self.prepare(
            male.get(), female.get(), photo.get(), online.get(), af.get(),
            asia.get(), eu.get(), na.get(), oc.get(), sa.get(), email.get(),
            snail.get(), langex.get(), friend.get(), sluzza.get(), relo.get(),
            tlowage.get(), thighage.get(), countrybox.get(0, END),
            langbox.get(0, END), euser.get(), epass.get(), emindel.get(),
            emaxdel.get(), erundel.get(), frame, filterbox.get(0, END),
            deep.get()))
        bstart.pack(side=LEFT, padx=10, pady=10, expand=True)
        bload = Button(buttons, text="Load Preset", command=lambda:self.load(
            male, female, photo, online, af, asia, eu, na, oc, sa, email,
            snail, langex, friend, sluzza, relo, tlowage, thighage, countrybox,
            langbox, euser, epass, emindel, emaxdel, erundel, filterbox, deep))
        bload.pack(side=LEFT, padx=10, pady=10, expand=True)
        bsave = Button(buttons, text="Save Preset", command=lambda:self.save(
            male.get(), female.get(), photo.get(), online.get(), af.get(),
            asia.get(), eu.get(), na.get(), oc.get(), sa.get(), email.get(),
            snail.get(), langex.get(), friend.get(), sluzza.get(), relo.get(),
            tlowage.get(), thighage.get(), countrybox.get(0, END),
            langbox.get(0, END), euser.get(), epass.get(), emindel.get(),
            emaxdel.get(), erundel.get(), filterbox.get(0, END), deep.get()))
        bsave.pack(side=LEFT, padx=10, pady=10, expand=True)

        buttons.pack()

    def save(self, *args):
        name = tkFileDialog.asksaveasfilename()
        if name != "":
            f = open(name, 'w+')
            for a in args:
                f.write(str(a).rstrip()+"\n")

    def load(self, *args):
        name = tkFileDialog.askopenfilename()
        if name != "":
            f = open(name, 'r')
            counter = 0
            entries = [16, 17, 20, 21, 22, 23, 24]
            for line in f:
                if counter == 18 or counter == 19 or counter == 25:
                    args[counter].delete(0, END)
                    tup = ast.literal_eval(line)
                    for e in tup:
                        args[counter].insert(END, e)
                elif counter in entries:
                    args[counter].delete(0, END)
                    args[counter].insert(0, line.strip())
                else:
                    args[counter].set(int(line))
                counter += 1

    def prepare(self, *args):
        url = self.build_url(args, False)
        frame = args[25]
        frame.destroy()
        s = requests.Session()
        login(s, args[20], args[21])
        if args[27] == 1:
            filter_url = ""
            filtered_users = []
            if len(args[26]) > 0:
                filter_url = self.build_url(args, True)
                filtered_users = get_filtered_users(s, filter_url, args[24])
            search(s, url, args[22], args[23], args[24], filtered_users, [])
        else:
            fterms = args[26]
            search(s, url, args[22], args[23], args[24], [], fterms)

    def build_url(self, args, needs_filter):
        # Try and get this clusterfuck into something workable
        url = "http://www.interpals.net/search.php?todo=search&sort=last_login"
        # What ages do we want?
        url += "&age1=%s&age2=%s" % (args[16], args[17])
        # Do we want dudes or ladies?
        if args[0] == 1 and args[1] == 1:
            url += "&sex[0]=FEMALE&sex[1]=MALE"
        elif args[0] == 1:
            url += "&sex[0]=MALE"
        else:
            url += "&sex[0]=FEMALE"
        # Are we looking for people with just photos or who are online?
        if args[2] == 1:
            url += "&photo=1"
        if args[3] == 1:
            url += "&online=true"
        else:
            url += "&online=false"
        # Are we looking for continents or countries? And which?
        if len(args[18]) > 0:
            url += "&countries%5B%5D=---"
            countries = args[18]
            for c in countries:
                url += "&countries[%d]=%s" % (countries.index(c), c.upper())
        else:
            # Warning: masterful code reuse ahead
            counter = 0
            if args[4] == 1:
                url += "&continents[%d]=AF" % counter
                counter += 1
            if args[5] == 1:
                url += "&continents[%d]=AS" % counter
                counter += 1
            if args[6] == 1:
                url += "&continents[%d]=EU" % counter
                counter += 1
            if args[7] == 1:
                url += "&continents[%d]=NA" % counter
                counter += 1
            if args[8] == 1:
                url += "&continents[%d]=OC" % counter
                counter += 1
            if args[9] == 1:
                url += "&continents[%d]=SA" % counter
                counter += 1
        # Any language preference?
        if len(args[19]) > 0:
            url += "&languages%5B%5D=---"
            languages = args[19]
            for l in languages:
                url += "&languages[%d]=%s" % (languages.index(l), l.upper())
        # Do we care what they're looking for?
        counter = 0
        if args[10] == 1:
            url += "&lfor[%d]=lfor_email" % counter
            counter += 1
        if args[11] == 1:
            url += "&lfor[%d]=lfor_snail" % counter
            counter += 1
        if args[12] == 1:
            url += "&lfor[%d]=lfor_langex" % counter
            counter += 1
        if args[13] == 1:
            url += "&lfor[%d]=lfor_friend" % counter
            counter += 1
        if args[14] == 1:
            url += "&lfor[%d]=lfor_flirt" % counter
            counter += 1
        if args[15] == 1:
            url += "&lfor[%d]=lfor_relation" % counter
            counter += 1
        if needs_filter:
            filters = args[26]
            for f in filters:
                url += "&keywords=%s" % f
        if DEBUG:
            print("Built a search url successfully: " + url)
        return url


class CasaApp():
    def __init__(self, master=None):
        master.title("Casanova config v3.3")
        self.controller = Controller(master)


def main():
    root = Tk()
    app = CasaApp(root)
    root.mainloop()

if  __name__ == '__main__':
    main()