add python 2 to 3 bash + fix Duplicate + clean

bin/Base64.py

@@ -3,7 +3,7 @@
 """
     Base64 module
 
-    Dectect Base64, decode it and send to XXX for reprocess
+    Dectect Base64 and decode it
 """
 import time
 import os

bin/Categ.py

@@ -77,7 +77,6 @@ if __name__ == "__main__":
             tmp_dict[bname] = re.compile('|'.join(patterns), re.IGNORECASE)
 
     prec_filename = None
-    print(tmp_dict)
 
     while True:
         filename = p.get_from_set()

bin/Duplicates.py

@@ -54,7 +54,6 @@ if __name__ == "__main__":
             dico_redis[str(year)+str(month).zfill(2)] = redis.StrictRedis(
                 host=p.config.get("Redis_Level_DB", "host"), port=year,
                 db=month)
-            #print("dup: "+str(year)+str(month).zfill(2)+"\n")
 
     # FUNCTIONS #
     publisher.info("Script duplicate started")
@@ -142,11 +141,12 @@ if __name__ == "__main__":
                                 paste_date = paste_date.decode('utf8')
                                 paste_date = paste_date if paste_date != None else "No date available"
                                 if paste_path != None:
+                                    if paste_path != PST.p_path:
                                         hash_dico[dico_hash] = (hash_type, paste_path, percent, paste_date)
 
                                         print('['+hash_type+'] '+'comparing: ' + str(PST.p_path[44:]) + '  and  ' + str(paste_path[44:]) + ' percentage: ' + str(percent))
+
                         except Exception:
-                            #print(str(e))
                             print('hash not comparable, bad hash: '+dico_hash+' , current_hash: '+paste_hash)
 
             # Add paste in DB after checking to prevent its analysis twice
@@ -181,7 +181,6 @@ if __name__ == "__main__":
                 y = time.time()
 
                 publisher.debug('{}Processed in {} sec'.format(to_print, y-x))
-                #print '{}Processed in {} sec'.format(to_print, y-x)
 
         except IOError:
             to_print = 'Duplicate;{};{};{};'.format(

bin/ModuleStats.py

@@ -29,7 +29,6 @@ def get_date_range(num_day):
 
 
 def compute_most_posted(server, message):
-    print(message)
     module, num, keyword, paste_date = message.split(';')
 
     redis_progression_name_set = 'top_'+ module +'_set_' + paste_date

bin/packages/Paste.py

@@ -112,9 +112,6 @@ class Paste(object):
         paste = self.cache.get(self.p_path)
         if paste is None:
             try:
-            #print('----------------------------------------------------------------')
-            #print(self.p_name)
-            #print('----------------------------------------------------------------')
                 with gzip.open(self.p_path, 'rb') as f:
                     paste = f.read()
                     self.cache.set(self.p_path, paste)
@@ -334,7 +331,6 @@ class Paste(object):
         Save a new duplicate on others pastes
         """
         for hash_type, path, percent, date in list_value:
-            print(hash_type, path, percent, date)
             #get json
             json_duplicate = self.store.hget(path, attr_name)
             #json save on redis

bin/packages/config.cfg.sample

@@ -54,6 +54,9 @@ criticalNumberToAlert=8
 #Will be considered as false positive if less that X matches from the top password list
 minTopPassList=5
 
+[Curve]
+max_execution_time = 90
+
 [Base64]
 path = Base64/
 max_execution_time = 60

bin/packages/modules.cfg

@@ -74,10 +74,6 @@ subscribe = Redis_Url
 subscribe = Redis_Url
 publish = Redis_alertHandler,Redis_Duplicate
 
-[Dox]
-subscribe = Redis_Dox
-publish = Redis_Duplicate,Redis_alertHandler
-
 [ModuleStats]
 subscribe = Redis_ModuleStats
 

pip_packages_requirement.txt

@@ -1,51 +0,0 @@
-#Essential
-redis
-pyzmq
-dnspython
-logbook
-pubsublogger
-textblob
-
-#Graph
-numpy
-matplotlib
-networkx
-terminaltables
-colorama
-asciimatics
-
-#Tokeniser
-nltk
-
-# Hashlib
-crcmod
-mmh3
-ssdeep
-python-Levenshtein
-
-#Others
-python-magic
-pybloomfiltermmap
-psutil
-phonenumbers
-
-ipython
-flask
-texttable
-
-#DomainClassifier
-DomainClassifier
-#Indexer requirements
-whoosh
-
-ipaddress
-pycountry
-
-# To fetch Onion urls
-PySocks
-
-#ASN lookup requirements
-https://storage.googleapis.com/google-code-archive-downloads/v2/code.google.com/adns-python/adns-python-1.2.1.tar.gz
-https://github.com/trolldbois/python-cymru-services/archive/master.zip
-
-https://github.com/saffsd/langid.py/archive/master.zip

var/www/modules/browsepastes/Flask_browsepastes.py

@@ -48,6 +48,7 @@ def getPastebyType(server, module_name):
     all_path = []
     for path in server.smembers('WARNING_'+module_name):
         all_path.append(path)
+
     return all_path
 
 
@@ -95,8 +96,6 @@ def importantPasteByModule():
     for path in allPastes[0:10]:
         path = path.decode('utf8')
         all_path.append(path)
-        #print(path)
-        #print(type(path))
         paste = Paste.Paste(path)
         content = paste.get_p_content()
         content_range = max_preview_char if len(content)>max_preview_char else len(content)-1