chg: [Tracker term] track terms by sources

bin/lib/Tracker.py

@@ -74,7 +74,7 @@ def get_tracker_level(tracker_uuid):
 def get_tracker_user_id(tracker_uuid):
     return r_serv_tracker.hget('tracker:{}'.format(tracker_uuid), 'user_id')
 
-def get_tracker_uuid_list(tracker, tracker_type):
+def get_tracker_uuid_list(tracker, tracker_type):  ######################################################### USE ME
     return list(r_serv_tracker.smembers('all:tracker_uuid:{}:{}'.format(tracker_type, tracker)))
 
 def get_tracker_tags(tracker_uuid):
@@ -83,6 +83,9 @@ def get_tracker_tags(tracker_uuid):
 def get_tracker_mails(tracker_uuid):
     return list(r_serv_tracker.smembers('tracker:mail:{}'.format(tracker_uuid)))
 
+def get_tracker_uuid_sources(tracker_uuid):
+    return list(r_serv_tracker.smembers(f'tracker:sources:{tracker_uuid}'))
+
 def get_tracker_description(tracker_uuid):
     return r_serv_tracker.hget('tracker:{}'.format(tracker_uuid), 'description')
 
@@ -133,10 +136,11 @@ def get_tracker_sparkline(tracker_uuid, num_day=6):
 
 def add_tracked_item(tracker_uuid, item_id, item_date):
     # track item
-    r_serv_tracker.sadd('tracker:item:{}:{}'.format(tracker_uuid, item_date), item_id)
+    res = r_serv_tracker.sadd(f'tracker:item:{tracker_uuid}:{item_date}', item_id)
     # track nb item by date
+    if res == 1:
         r_serv_tracker.zadd('tracker:stat:{}'.format(tracker_uuid), item_date, int(item_date))
-
+bin/lib/Tracker.py
 def get_email_subject(tracker_uuid):
     tracker_description = get_tracker_description(tracker_uuid)
     if not tracker_description:
@@ -150,6 +154,10 @@ def get_tracker_last_updated_by_type(tracker_type):
         epoch_update = 0
     return float(epoch_update)
 
+# # TODO: check type API
+def trigger_trackers_refresh(tracker_type):
+    r_serv_tracker.set(f'tracker:refresh:{tracker_type}', time.time())
+
 ######################
 #### TRACKERS ACL ####
 
@@ -235,7 +243,7 @@ def api_validate_tracker_to_add(tracker , tracker_type, nb_words=1):
         return ({"status": "error", "reason": "Incorrect type"}, 400)
     return ({"status": "success", "tracker": tracker, "type": tracker_type}, 200)
 
-def create_tracker(tracker, tracker_type, user_id, level, tags, mails, description, dashboard=0, tracker_uuid=None):
+def create_tracker(tracker, tracker_type, user_id, level, tags, mails, description, dashboard=0, tracker_uuid=None, sources=[]):
     # edit tracker
     if tracker_uuid:
         edit_tracker = True
@@ -255,7 +263,7 @@ def create_tracker(tracker, tracker_type, user_id, level, tags, mails, descripti
 
     # YARA
     if tracker_type == 'yara_custom' or tracker_type == 'yara_default':
-        # delete yara rule
+        # create yara rule
         if tracker_type == 'yara_default' and old_type == 'yara':
             if not is_default_yara_rule(old_tracker):
                 filepath = get_yara_rule_file_by_tracker_name(old_tracker)
@@ -318,6 +326,11 @@ def create_tracker(tracker, tracker_type, user_id, level, tags, mails, descripti
     for mail in mails:
         r_serv_tracker.sadd('tracker:mail:{}'.format(tracker_uuid), escape(mail) )
 
+    # create tracker sources filter
+    for source in sources:
+        # escape source ?
+        r_serv_tracker.sadd(f'tracker:sources:{tracker_uuid}', escape(source) )
+
     # toggle refresh module tracker list/set
     r_serv_tracker.set('tracker:refresh:{}'.format(tracker_type), time.time())
     if tracker_type != old_type: # toggle old type refresh
@@ -346,6 +359,7 @@ def api_add_tracker(dict_input, user_id):
     res = verify_mail_list(mails)
     if res:
         return res
+    sources = dict_input.get('sources', [])
 
     ## TODO: add dashboard key
     level = dict_input.get('level', 1)
@@ -371,7 +385,7 @@ def api_add_tracker(dict_input, user_id):
             if is_tracker_in_user_level(tracker, tracker_type, user_id) and not tracker_uuid:
                 return ({"status": "error", "reason": "Tracker already exist"}, 409)
 
-    tracker_uuid = create_tracker(tracker , tracker_type, user_id, level, tags, mails, description, tracker_uuid=tracker_uuid)
+    tracker_uuid = create_tracker(tracker , tracker_type, user_id, level, tags, mails, description, tracker_uuid=tracker_uuid, sources=sources)
 
     return ({'tracker': tracker, 'type': tracker_type, 'uuid': tracker_uuid}, 200)
 
@@ -407,10 +421,12 @@ def get_all_default_yara_rules_by_type(yara_types):
     else:
         return []
 
-def get_all_tracked_yara_files():
+def get_all_tracked_yara_files(filter_disabled=False):
     yara_files = r_serv_tracker.smembers('all:tracker:yara')
     if not yara_files:
         yara_files = []
+    if filter_disabled:
+        pass
     return yara_files
 
 def reload_yara_rules():
@@ -424,6 +440,22 @@ def reload_yara_rules():
     rules = yara.compile(filepaths=rule_dict)
     return rules
 
+# # TODO:
+# Avoid useless CHECK
+# Empty list == ALL SOURCES
+# FIXME MOOVE ME
+def get_tracker_sources(tracker, tracker_type):
+    l_sources = set()
+    for tracker_uuid in get_tracker_uuid_list(tracker, tracker_type):
+        sources = get_tracker_uuid_sources(tracker_uuid)
+        if sources:
+            for source in get_tracker_uuid_sources(tracker_uuid):
+                l_sources.add(source)
+        else:
+            l_sources = []
+            break
+    return l_sources
+
 def is_valid_yara_rule(yara_rule):
     try:
         yara.compile(source=yara_rule)
@@ -518,5 +550,14 @@ def api_get_default_rule_content(default_yara_rule):
 ##-- YARA --##
 
 if __name__ == '__main__':
-    res = is_valid_yara_rule('rule dummy {  }')
+    #res = is_valid_yara_rule('rule dummy {  }')
+
+    # res = create_tracker('test', 'word', 'admin@admin.test', 1, [], [], None, sources=['crawled', 'pastebin.com', 'rt/pastebin.com'])
+    res = create_tracker('test', 'word', 'admin@admin.test', 1, [], [], None)
+    # print(res)
+
+    t_uuid = '1c2d35b0-9330-4feb-b454-da13007aa9f7'
+    res = get_tracker_sources('test', 'word')
+
+
     print(res)

bin/lib/item_basic.py

@@ -64,7 +64,8 @@ def get_item_content(item_id):
                 item_content = f.read().decode()
                 r_cache.set(item_full_path, item_content)
                 r_cache.expire(item_full_path, 300)
-        except:
+        except Exception as e:
+            print(e)
             item_content = ''
     return str(item_content)
 
@@ -176,7 +177,7 @@ def add_map_obj_id_item_id(obj_id, item_id, obj_type):
 ##--  --##
 
 ## COMMON ##
-def _get_dir_source_name(directory, source_name=None, l_sources_name=set()):
+def _get_dir_source_name(directory, source_name=None, l_sources_name=set(), filter_dir=False):
     if source_name:
         l_dir = os.listdir(os.path.join(directory, source_name))
     else:
@@ -188,12 +189,16 @@ def _get_dir_source_name(directory, source_name=None, l_sources_name=set()):
     else:
         for src_name in l_dir:
             if len(src_name) == 4:
-                try:
+                #try:
                 int(src_name)
-                    l_sources_name.add(os.path.join(source_name))
+                to_add = os.path.join(source_name)
+                # filter sources, remove first directory
+                if filter_dir:
+                    to_add = to_add.replace('archive/', '').replace('alerts/', '')
+                l_sources_name.add(to_add)
                 return l_sources_name
-                except:
+                #except:
-                    pass
+                #    pass
             if source_name:
                 src_name = os.path.join(source_name, src_name)
             l_sources_name = _get_dir_source_name(directory, source_name=src_name, l_sources_name=l_sources_name)

bin/packages/Item.py

@@ -570,7 +570,9 @@ class Item(AbstractObject):
         """
         Returns Item source/feeder name
         """
-        return item_basic.get_source(self.id)
+        #return self.id.split('/')[-5]
+        l_source = self.id.split('/')[:-4]
+        return os.path.join(*l_source)
 
     def get_basename(self):
         return os.path.basename(self.id)
@@ -605,11 +607,8 @@ class Item(AbstractObject):
         except FileNotFoundError:
             return False
 
-# if __name__ == '__main__':
+#if __name__ == '__main__':
-#
+
-#     item = Item('')
-#     res = item.get_date(separator=True)
-#     print(res)
 
 #     import Domain
 #     domain = Domain.Domain('domain.onion')

bin/packages/Term.py

@@ -319,6 +319,9 @@ def delete_term(term_uuid):
     # remove mails
     r_serv_term.delete('tracker:mail:{}'.format(term_uuid))
 
+    # remove sources
+    r_serv_term.delete('tracker:sources:{}'.format(term_uuid))
+
     # remove item set
     all_item_date = r_serv_term.zrange('tracker:stat:{}'.format(term_uuid), 0, -1)
     for date in all_item_date:

bin/trackers/Tracker_Term.py

@@ -90,11 +90,13 @@ class Tracker_Term(AbstractModule):
             # create token statistics
             #for word in dict_words_freq:
             #    Term.create_token_statistics(item_date, word, dict_words_freq[word])
+            item_source = item.get_source()
 
             # check solo words
+            ####### # TODO: check if source needed #######
             for word in self.list_tracked_words:
                 if word in dict_words_freq:
-                    self.new_term_found(word, 'word', item.get_id(), item_date)
+                    self.new_term_found(word, 'word', item.get_id(), item_date, item_source)
 
             # check words set
             for elem in self.set_tracked_words_list:
@@ -107,15 +109,18 @@ class Tracker_Term(AbstractModule):
                     if word in dict_words_freq:
                         nb_uniq_word += 1
                 if nb_uniq_word >= nb_words_threshold:
-                        self.new_term_found(word_set, 'set', item.get_id(), item_date)
+                    self.new_term_found(word_set, 'set', item.get_id(), item_date, item_source)
 
-    def new_term_found(self, term, term_type, item_id, item_date):
+    def new_term_found(self, term, term_type, item_id, item_date, item_source):
         uuid_list = Term.get_term_uuid_list(term, term_type)
         self.redis_logger.info(f'new tracked term found: {term} in {item_id}')
         print(f'new tracked term found: {term} in {item_id}')
 
         for term_uuid in uuid_list:
-            Term.add_tracked_item(term_uuid, item_id, item_date)
+            tracker_sources = Tracker.get_tracker_uuid_sources(term_uuid)
+            if not tracker_sources or item_source in tracker_sources:
+                print(not tracker_sources or item_source in tracker_sources)
+                Tracker.add_tracked_item(term_uuid, item_id, item_date)
 
                 tags_to_add = Term.get_term_tags(term_uuid)
                 for tag in tags_to_add:
@@ -128,7 +133,7 @@ class Tracker_Term(AbstractModule):
                     mail_body = Tracker_Term.mail_body_template.format(term, item_id, self.full_item_url, item_id)
                 for mail in mail_to_notify:
                     self.redis_logger.debug(f'Send Mail {mail_subject}')
-                print(f'Send Mail {mail_subject}')
+                    print(f'S        print(item_content)end Mail {mail_subject}')
                     NotificationHelper.sendEmailNotification(mail, mail_subject, mail_body)