Merge pull request #31 from y-mehta/dev

Feature_Save_To_File

Author: PSNAppz

modules/__init__.py

@@ -3,5 +3,7 @@
 from .getweblinks import *
 from .pagereader import *
 from .updater import *
+from .savefile import *
 
-__all__ = (bcolors.__all__ + getemails.__all__ + getweblinks.__all__ + pagereader.__all__ + updater.__all__)
+
+__all__ = (bcolors.__all__ + getemails.__all__ + getweblinks.__all__ + pagereader.__all__ + updater.__all__ + savefile.__all__ )

modules/getemails.py

@@ -2,13 +2,14 @@
 import os
 sys.path.append(os.path.abspath('../'))
 from modules.bcolors import Bcolors
+from modules.savefile import saveJson
 import bs4
 
 __all__ = ['getMails']
 
 """Get all emails from the website"""
 
-def getMails(soup):
+def getMails(soup,save=0):
     _soup_instance = bs4.BeautifulSoup
     if isinstance(type(soup), type(_soup_instance)):
         emails = []
@@ -27,6 +28,8 @@ def getMails(soup):
         print ('-------------------------------')
         for mail in emails:
             print (mail)
+        if save:
+            saveJson("Extracted-Mail-IDs",emails)
         return ''
     else:
         raise(Bcolors.FAIL+'Method parameter is not of instance bs4.BeautifulSoup'+Bcolors.ENDC)

modules/getweblinks.py

@@ -1,6 +1,7 @@
 import sys
 import os
 sys.path.append(os.path.abspath('../'))
+from modules.savefile import saveJson 
 import urllib.request 
 from modules.bcolors import Bcolors
 import bs4
@@ -10,23 +11,26 @@
 
 __all__ = ['getLinks']
 
-def link_status(web):
+
+def link_status(web,out_queue,index):
     link_live = False
+    out_queue[index] = web + " is_live = False "
     try:
         urllib.request.urlopen(web)    
         link_live = True
+        out_queue[index] = web + " is_live = True "
         print(web)
     except urllib.error.HTTPError as e:
         print(Bcolors.On_Red+web+Bcolors.ENDC)
     except urllib.error.URLError as e:
         print(Bcolors.On_Red+web+Bcolors.ENDC)
     except http.client.RemoteDisconnected as e:
         print(Bcolors.On_Red+web+Bcolors.ENDC)
-    return 
+    return
 
 
 """Get all onion links from the website"""
-def getLinks(soup,ext,live=0):
+def getLinks(soup,ext,live=0,save=0):
     _soup_instance = bs4.BeautifulSoup
     extensions = []
     if ext:
@@ -53,13 +57,23 @@ def getLinks(soup,ext,live=0):
         print ('-------------------------------')
         if live:
             threads = []
+            result = [{} for x in websites]
             for web in websites:
-                t = threading.Thread(target=link_status, args=(web,))
-                threads.append(t)
+                t = threading.Thread(target=link_status, args=(web,result,websites.index(web)))
                 t.start()
+                threads.append(t)
+            try:
+                for t in threads:
+                    t.join()
+                if save:
+                    saveJson("Live-Onion-Links",result)
+            except:
+                pass
         else:
             for web in websites:
                 print(web)
+            if save:
+                saveJson("Onion-Links",websites)
         return websites           
     else:
         raise('Method parameter is not of instance bs4.BeautifulSoup')

modules/savefile.py

@@ -0,0 +1,21 @@
+import json
+import time
+
+__all__ = ['saveJson']
+
+# open the file "TorBoT-Export" in write ("a") mode
+def saveJson(datatype,data):
+   "function_docstring"
+   timestr = time.strftime("%Y%m%d-%H%M%S")
+   #Json File Creation
+   file = open("TorBoT-Export-"+datatype+timestr+".json", "a")
+   #Store data in Json format
+   output = {datatype : data}
+   #Dump output to file 
+   json.dump(output, file, indent=2)
+   file.close()
+   print("\nData will be saved with a File Name :"+ "TorBoT-Export-"+datatype+timestr+".json") 
+   return
+
+
+

torBot.py

@@ -14,12 +14,12 @@
 from stem.control import Controller
 
 with Controller.from_port(port = 9051) as controller:
-    controller.authenticate("16:872860B76453A77D60CA2BB8C1A7042072093276A3D701AD684053EC4C")
+    controller.authenticate("16:3BEA46EB6C489B90608A65120BD7CF0C7BA709513AB8ACF212B9537183")
     controller.signal(Signal.NEWNYM)
 #TorBot VERSION    
 _VERSION_ = "1.0.1"
 #TOR SETUP GLOBAL Vars
-SOCKS_PORT = 9050  # TOR proxy port that is default from torrc, change to whatever torrc is configured to
+SOCKS_PORT = 9050	  # TOR proxy port that is default from torrc, change to whatever torrc is configured to
 socks.set_default_proxy(socks.SOCKS5, "127.0.0.1",SOCKS_PORT)
 socket.socket = socks.socksocket
 # Perform DNS resolution through the socket
@@ -83,6 +83,7 @@ def main():
  parser.add_argument("--update",action="store_true",help="Update TorBot to the latest stable version")
  parser.add_argument("-q","--quiet",action="store_true")
  parser.add_argument("-u","--url",help="Specifiy a website link to crawl")
+ parser.add_argument("-s","--save",action="store_true", help="Save results in a file")
  parser.add_argument("-m","--mail",action="store_true", help="Get e-mail addresses from the crawled sites")
  parser.add_argument("-e","--extension",action='append',dest='extension',default=[],help="Specifiy additional website extensions to the list(.com or .org etc)")
  parser.add_argument("-l","--live",action="store_true",help="Check if websites are live or not (slow)")
@@ -100,16 +101,18 @@ def main():
  link = args.url
  ext = 0
  live = 0
+ save=0
  live = args.live
  ext = args.extension
+ save = args.save
  a = readPage("https://check.torproject.org/",1)
  if link:
  	b = readPage(link)
  else:
     b = readPage("http://torlinkbgs6aabns.onion/")
  if args.mail:
- 	getMails(b)
- getLinks(b,ext,live)
+ 	getMails(b,save)
+ getLinks(b,ext,live,save)
  print ("\n\n")
  return 0