Merge remote-tracking branch 'upstream/master'

bin/CreditCards.py

@@ -66,7 +66,7 @@ if __name__ == "__main__":
                     publisher.warning('{}Checked {} valid number(s)'.format(
                         to_print, len(creditcard_set)))
                     #Send to duplicate
-                    p.populate_set_out(filepath, 'Duplicate')
+                    p.populate_set_out(filename, 'Duplicate')
                     #send to Browse_warning_paste
                     p.populate_set_out('creditcard;{}'.format(filename), 'BrowseWarningPaste')
                 else:

bin/CurveManageTopSets.py

@@ -22,8 +22,8 @@ from pubsublogger import publisher
 from packages import lib_words
 import datetime
 import calendar
-
+import os
-from Helper import Process
+import ConfigParser
 
 # Config Variables
 Refresh_rate = 60*5 #sec
@@ -96,13 +96,19 @@ if __name__ == '__main__':
     # Script is the default channel used for the modules.
     publisher.channel = 'Script'
 
-    config_section = 'CurveManageTopSets'
+    configfile = os.path.join(os.environ['AIL_BIN'], 'packages/config.cfg')
-    p = Process(config_section)
+    if not os.path.exists(configfile):
+        raise Exception('Unable to find the configuration file. \
+                        Did you set environment variables? \
+                        Or activate the virtualenv.')
+    
+    cfg = ConfigParser.ConfigParser()
+    cfg.read(configfile)
 
     server_term = redis.StrictRedis(
-        host=p.config.get("Redis_Level_DB_TermFreq", "host"),
+        host=cfg.get("Redis_Level_DB_TermFreq", "host"),
-        port=p.config.get("Redis_Level_DB_TermFreq", "port"),
+        port=cfg.getint("Redis_Level_DB_TermFreq", "port"),
-        db=p.config.get("Redis_Level_DB_TermFreq", "db"))
+        db=cfg.getint("Redis_Level_DB_TermFreq", "db"))
 
     publisher.info("Script Curve_manage_top_set started")
 
@@ -113,11 +119,6 @@ if __name__ == '__main__':
 
     while True:
         # Get one message from the input queue (module only work if linked with a queue)
-        message = p.get_from_set()
-        if message is None:
-            publisher.debug("{} queue is empty, waiting".format(config_section))
-            print 'sleeping'
         time.sleep(Refresh_rate) # sleep a long time then manage the set
         manage_top_set()
-            continue
 

bin/Duplicate_ssdeep.py

@@ -1,182 +0,0 @@
-#!/usr/bin/env python2
-# -*-coding:UTF-8 -*
-
-"""
-The Duplicate module
-====================
-
-This huge module is, in short term, checking duplicates.
-
-Requirements:
--------------
-
-
-"""
-import redis
-import os
-import time
-import datetime
-import json
-import ssdeep
-from packages import Paste
-from pubsublogger import publisher
-
-from Helper import Process
-
-if __name__ == "__main__":
-    publisher.port = 6380
-    publisher.channel = "Script"
-
-    config_section = 'Duplicates'
-    save_dico_and_reload = 1 #min
-    time_1 = time.time()
-    flag_reload_from_disk = True
-    flag_write_to_disk = False
-
-    p = Process(config_section)
-
-    # REDIS #
-    # DB OBJECT & HASHS ( DISK )
-    # FIXME increase flexibility
-    dico_redis = {}
-    for year in xrange(2013, datetime.date.today().year+1):
-        for month in xrange(0, 16):
-            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")
-
-    dicopath = os.path.join(os.environ['AIL_HOME'],
-                             p.config.get("Directories", "dicofilters"))
-
-    dico_path_set = set()
-    while True:
-        try:
-            hash_dico = {}
-            dupl = []
-
-            x = time.time()
-
-            message = p.get_from_set()
-            if message is not None:
-                path = message
-                PST = Paste.Paste(path)
-            else:
-                publisher.debug("Script Attribute is idling 10s")
-                time.sleep(10)
-                continue
-
-            PST._set_p_hash_kind("ssdeep")
-
-            # Assignate the correct redis connexion
-            r_serv1 = dico_redis[PST.p_date.year + PST.p_date.month]
-
-            # Creating the dicor name: dicoyyyymm
-            filedicopath = os.path.join(dicopath, 'dico' + PST.p_date.year +
-                                         PST.p_date.month)
-            filedicopath_today = filedicopath
-
-            # Save I/O
-            if time.time() - time_1 > save_dico_and_reload*60:
-                flag_write_to_disk = True
-
-            if os.path.exists(filedicopath):
-                if flag_reload_from_disk == True:
-                    flag_reload_from_disk = False
-                    print 'Reloading'
-                    with open(filedicopath, 'r') as fp:
-                        today_dico = json.load(fp)
-            else:
-                today_dico = {}
-                with open(filedicopath, 'w') as fp:
-                    json.dump(today_dico, fp)
-
-            # For now, just use monthly dico
-            dico_path_set.add(filedicopath)
-
-            # UNIQUE INDEX HASHS TABLE
-            yearly_index = str(datetime.date.today().year)+'00'
-            r_serv0 = dico_redis[yearly_index]
-            r_serv0.incr("current_index")
-            index = r_serv0.get("current_index")+str(PST.p_date)
-            
-            # For each dico
-            opened_dico = []
-            for dico in dico_path_set:
-                # Opening dico
-                if dico == filedicopath_today:
-                    opened_dico.append([dico, today_dico])
-                else:
-                    with open(dico, 'r') as fp:
-                        opened_dico.append([dico, json.load(fp)])
-
-              
-            #retrieve hash from paste
-            paste_hash = PST._get_p_hash()
-            
-            # Go throught the Database of the dico (of the month)
-            threshold_dup = 99 
-            for dico_name, dico in opened_dico:
-                for dico_key, dico_hash in dico.items():
-                    percent = ssdeep.compare(dico_hash, paste_hash)
-                    if percent > threshold_dup:
-                        db = dico_name[-6:]
-                        # Go throught the Database of the dico filter (month)
-                        r_serv_dico = dico_redis[db]
-                        
-                        # index of paste
-                        index_current = r_serv_dico.get(dico_hash)
-                        paste_path = r_serv_dico.get(index_current)
-                        if paste_path != None:
-                            hash_dico[dico_hash] = (paste_path, percent)
-
-                        #print 'comparing: ' + str(dico_hash[:20]) + '  and  ' + str(paste_hash[:20]) + ' percentage: ' + str(percent)
-                        print '   '+ PST.p_path[44:]  +', '+ paste_path[44:] + ', ' + str(percent)
-
-            # Add paste in DB to prevent its analyse twice
-            # HASHTABLES PER MONTH (because of r_serv1 changing db)
-            r_serv1.set(index, PST.p_path)
-            r_serv1.sadd("INDEX", index)
-            # Adding the hash in Redis
-            r_serv1.set(paste_hash, index)
-            r_serv1.sadd("HASHS", paste_hash)
-    ##################### Similarity found  #######################
-
-            # if there is data in this dictionnary
-            if len(hash_dico) != 0:
-                for dico_hash, paste_tuple in hash_dico.items():
-                    paste_path, percent = paste_tuple
-                    dupl.append((paste_path, percent))
-
-                # Creating the object attribute and save it.
-                to_print = 'Duplicate;{};{};{};'.format(
-                    PST.p_source, PST.p_date, PST.p_name)
-                if dupl != []:
-                    PST.__setattr__("p_duplicate", dupl)
-                    PST.save_attribute_redis("p_duplicate", dupl)
-                    publisher.info('{}Detected {}'.format(to_print, len(dupl)))
-                    print '{}Detected {}'.format(to_print, len(dupl))
-
-                y = time.time()
-
-                publisher.debug('{}Processed in {} sec'.format(to_print, y-x))
-           
-
-            # Adding the hash in the dico of the month
-            today_dico[index] = paste_hash
-
-            if flag_write_to_disk:
-                time_1 = time.time()
-                flag_write_to_disk = False
-                flag_reload_from_disk = True
-                print 'writing'
-                with open(filedicopath, 'w') as fp:
-                    json.dump(today_dico, fp)
-        except IOError:
-            to_print = 'Duplicate;{};{};{};'.format(
-                PST.p_source, PST.p_date, PST.p_name)
-            print "CRC Checksum Failed on :", PST.p_path
-            publisher.error('{}CRC Checksum Failed'.format(to_print))

bin/Helper.py

@@ -16,6 +16,7 @@ import ConfigParser
 import os
 import zmq
 import time
+import datetime
 import json
 
 
@@ -107,6 +108,7 @@ class Process(object):
         self.modules = ConfigParser.ConfigParser()
         self.modules.read(modulesfile)
         self.subscriber_name = conf_section
+
         self.pubsub = None
         if self.modules.has_section(conf_section):
             self.pubsub = PubSub()
@@ -117,6 +119,15 @@ class Process(object):
             port=self.config.get('RedisPubSub', 'port'),
             db=self.config.get('RedisPubSub', 'db'))
 
+        self.moduleNum = 1
+        for i in range(1, 50):
+            curr_num = self.r_temp.get("MODULE_"+self.subscriber_name + "_" + str(i))
+            if curr_num is None:
+                self.moduleNum = i
+                break
+
+
+
     def populate_set_in(self):
         # monoproc
         src = self.modules.get(self.subscriber_name, 'subscribe')
@@ -132,7 +143,28 @@ class Process(object):
         in_set = self.subscriber_name + 'in'
         self.r_temp.hset('queues', self.subscriber_name,
                          int(self.r_temp.scard(in_set)))
-        return self.r_temp.spop(in_set)
+        message = self.r_temp.spop(in_set)
+        timestamp = int(time.mktime(datetime.datetime.now().timetuple()))
+        dir_name = os.environ['AIL_HOME']+self.config.get('Directories', 'pastes')
+
+        if message is None:
+            return None
+
+        else:
+            try:
+                if ".gz" in message:
+                    path = message.split(".")[-2].split("/")[-1]
+                else:
+                    path = "?"
+                value = str(timestamp) + ", " + path
+                self.r_temp.set("MODULE_"+self.subscriber_name + "_" + str(self.moduleNum), value)
+                return message
+
+            except:
+                path = "?"
+                value = str(timestamp) + ", " + path
+                self.r_temp.set("MODULE_"+self.subscriber_name + "_" + str(self.moduleNum), value)
+                return message
 
     def populate_set_out(self, msg, channel=None):
         # multiproc

bin/LAUNCH.sh

@@ -112,33 +112,35 @@ function launching_scripts {
 
     echo -e $GREEN"\t* Launching ZMQ scripts"$DEFAULT
 
+    screen -S "Script" -X screen -t "ModuleInformation" bash -c './ModuleInformation.py -k 0 -c 1; read x'
+    sleep 0.1
     screen -S "Script" -X screen -t "Global" bash -c './Global.py; read x'
     sleep 0.1
-    screen -S "Script" -X screen -t "Duplicate" bash -c './Duplicate_ssdeep_v2.py; read x'
+    screen -S "Script" -X screen -t "Duplicates" bash -c './Duplicates.py; read x'
     sleep 0.1
-    screen -S "Script" -X screen -t "Attribute" bash -c './Attribute.py; read x'
+    screen -S "Script" -X screen -t "Attributes" bash -c './Attributes.py; read x'
     sleep 0.1
-    screen -S "Script" -X screen -t "Line" bash -c './Line.py; read x'
+    screen -S "Script" -X screen -t "Lines" bash -c './Lines.py; read x'
     sleep 0.1
-    screen -S "Script" -X screen -t "DomainClassifier" bash -c './DomClassifier.py; read x'
+    screen -S "Script" -X screen -t "DomClassifier" bash -c './DomClassifier.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Categ" bash -c './Categ.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Tokenize" bash -c './Tokenize.py; read x'
     sleep 0.1
-    screen -S "Script" -X screen -t "CreditCard" bash -c './CreditCard.py; read x'
+    screen -S "Script" -X screen -t "CreditCards" bash -c './CreditCards.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Onion" bash -c './Onion.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Mail" bash -c './Mail.py; read x'
     sleep 0.1
-    screen -S "Script" -X screen -t "Url" bash -c './Url.py; read x'
+    screen -S "Script" -X screen -t "Web" bash -c './Web.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Credential" bash -c './Credential.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Curve" bash -c './Curve.py; read x'
     sleep 0.1
-    screen -S "Script" -X screen -t "Curve_topsets_manager" bash -c './Curve_manage_top_sets.py; read x'
+    screen -S "Script" -X screen -t "CurveManageTopSets" bash -c './CurveManageTopSets.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Indexer" bash -c './Indexer.py; read x'
     sleep 0.1
@@ -158,7 +160,7 @@ function launching_scripts {
     sleep 0.1
     screen -S "Script" -X screen -t "Browse_warning_paste" bash -c './Browse_warning_paste.py; read x'
     sleep 0.1
-    screen -S "Script" -X screen -t "SentimentAnalyser" bash -c './SentimentAnalyser.py; read x'
+    screen -S "Script" -X screen -t "SentimentAnalysis" bash -c './SentimentAnalysis.py; read x'
 
 }
 

bin/ModuleInformation.py

@@ -0,0 +1,198 @@
+#!/usr/bin/env python2
+# -*-coding:UTF-8 -*
+
+'''
+
+This module can be use to see information of running modules.
+These information are logged in "logs/moduleInfo.log"
+
+It can also try to manage them by killing inactive one.
+However, it does not support mutliple occurence of the same module
+(It will kill the first one obtained by get)
+
+
+'''
+
+import time
+import datetime
+import redis
+import os
+import signal
+import argparse
+from subprocess import PIPE, Popen
+import ConfigParser
+import json
+from terminaltables import AsciiTable
+import textwrap
+
+# CONFIG VARIABLES
+threshold_stucked_module = 60*60*1 #1 hour
+log_filename = "../logs/moduleInfo.log"
+command_search_pid = "ps a -o pid,cmd | grep {}"
+command_restart_module = "screen -S \"Script\" -X screen -t \"{}\" bash -c \"./{}.py; read x\""
+
+
+def getPid(module):
+    p = Popen([command_search_pid.format(module+".py")], stdin=PIPE, stdout=PIPE, bufsize=1, shell=True)
+    for line in p.stdout:
+        splittedLine = line.split()
+        if 'python2' in splittedLine:
+            return int(splittedLine[0])
+        else:
+            return None
+
+def clearRedisModuleInfo():
+    for k in server.keys("MODULE_*"):
+        server.delete(k)
+
+def kill_module(module):
+    print ''
+    print '-> trying to kill module:', module
+
+    pid = getPid(module)
+    if pid is not None:
+        os.kill(pid, signal.SIGUSR1)
+        time.sleep(1)
+        if getPid(module) is None:
+            print module, 'has been killed'
+            print 'restarting', module, '...'
+            p2 = Popen([command_restart_module.format(module, module)], stdin=PIPE, stdout=PIPE, bufsize=1, shell=True)
+
+        else:
+            print 'killing failed, retrying...'
+            time.sleep(3)
+            os.kill(pid, signal.SIGUSR1)
+            time.sleep(1)
+            if getPid(module) is None:
+                print module, 'has been killed'
+                print 'restarting', module, '...'
+                p2 = Popen([command_restart_module.format(module, module)], stdin=PIPE, stdout=PIPE, bufsize=1, shell=True)
+            else:
+                print 'killing failed!'
+    time.sleep(7)
+
+
+if __name__ == "__main__":
+
+    parser = argparse.ArgumentParser(description='Show info concerning running modules and log suspected stucked modules. May be use to automatically kill and restart stucked one.')
+    parser.add_argument('-r', '--refresh', type=int, required=False, default=1, help='Refresh rate')
+    parser.add_argument('-k', '--autokill', type=int, required=True, default=1, help='Enable auto kill option (1 for TRUE, anything else for FALSE)')
+    parser.add_argument('-c', '--clear', type=int, required=False, default=1, help='Clear the current module information (Used to clear data from old launched modules)')
+
+    args = parser.parse_args()
+
+    configfile = os.path.join(os.environ['AIL_BIN'], 'packages/config.cfg')
+    if not os.path.exists(configfile):
+        raise Exception('Unable to find the configuration file. \
+                        Did you set environment variables? \
+                        Or activate the virtualenv.')
+
+    cfg = ConfigParser.ConfigParser()
+    cfg.read(configfile)
+
+    # REDIS #
+    server = redis.StrictRedis(
+        host=cfg.get("Redis_Queues", "host"),
+        port=cfg.getint("Redis_Queues", "port"),
+        db=cfg.getint("Redis_Queues", "db"))
+
+    if args.clear == 1:
+        clearRedisModuleInfo()
+
+
+    module_file_array = set()
+    with open('../doc/all_modules.txt', 'r') as module_file:
+        for line in module_file:
+            module_file_array.add(line[:-1])
+
+        while True:
+
+            all_queue = set()
+            curr_range = 50
+            printarray1 = []
+            printarray2 = []
+            printarray3 = []
+            for queue, card in server.hgetall("queues").iteritems():
+                all_queue.add(queue)
+                key = "MODULE_" + queue + "_"
+                for i in range(1, 50):
+                    curr_num = server.get("MODULE_"+ queue + "_" + str(i))
+                    if curr_num is None:
+                        curr_range = i
+                        break
+
+                for moduleNum in range(1, curr_range):
+                    value = server.get(key + str(moduleNum))
+                    if value is not None:
+                        timestamp, path = value.split(", ")
+                        if timestamp is not None and path is not None:
+                            startTime_readable = datetime.datetime.fromtimestamp(int(timestamp))
+                            processed_time_readable = str((datetime.datetime.now() - startTime_readable)).split('.')[0]
+
+                            if int(card) > 0:
+                                if int((datetime.datetime.now() - startTime_readable).total_seconds()) > threshold_stucked_module:
+                                    log = open(log_filename, 'a')
+                                    log.write(json.dumps([queue, card, str(startTime_readable), str(processed_time_readable), path]) + "\n")
+                                    if args.autokill == 1:
+                                        kill_module(queue)
+
+                                printarray1.append([str(queue), str(moduleNum), str(card), str(startTime_readable), str(processed_time_readable), str(path)])
+
+                            else:
+                                printarray2.append([str(queue), str(moduleNum), str(card), str(startTime_readable), str(processed_time_readable), str(path)])
+
+            for curr_queue in module_file_array:
+                if curr_queue not in all_queue:
+                    printarray3.append([curr_queue, "Not running"])
+
+            printarray1.sort(lambda x,y: cmp(x[4], y[4]), reverse=True)
+            printarray2.sort(lambda x,y: cmp(x[4], y[4]), reverse=True)
+            printarray1.insert(0,["Queue", "#", "Amount", "Paste start time", "Processing time for current paste (H:M:S)", "Paste hash"])
+            printarray2.insert(0,["Queue", "#","Amount", "Paste start time", "Time since idle (H:M:S)", "Last paste hash"])
+            printarray3.insert(0,["Queue", "State"])
+
+            os.system('clear')
+            t1 = AsciiTable(printarray1, title="Working queues")
+            t1.column_max_width(1)
+            if not t1.ok:
+                    longest_col = t1.column_widths.index(max(t1.column_widths))
+                    max_length_col = t1.column_max_width(longest_col)
+                    if max_length_col > 0:
+                        for i, content in enumerate(t1.table_data):
+                            if len(content[longest_col]) > max_length_col:
+                                temp = ''
+                                for l in content[longest_col].splitlines():
+                                    if len(l) > max_length_col:
+                                        temp += '\n'.join(textwrap.wrap(l, max_length_col)) + '\n'
+                                    else:
+                                        temp += l + '\n'
+                                    content[longest_col] = temp.strip()
+                            t1.table_data[i] = content
+
+            t2 = AsciiTable(printarray2, title="Idling queues")
+            t2.column_max_width(1)
+            if not t2.ok:
+                    longest_col = t2.column_widths.index(max(t2.column_widths))
+                    max_length_col = t2.column_max_width(longest_col)
+                    if max_length_col > 0:
+                        for i, content in enumerate(t2.table_data):
+                            if len(content[longest_col]) > max_length_col:
+                                temp = ''
+                                for l in content[longest_col].splitlines():
+                                    if len(l) > max_length_col:
+                                        temp += '\n'.join(textwrap.wrap(l, max_length_col)) + '\n'
+                                    else:
+                                        temp += l + '\n'
+                                    content[longest_col] = temp.strip()
+                            t2.table_data[i] = content
+
+            t3 = AsciiTable(printarray3, title="Not running queues")
+            t3.column_max_width(1)
+
+            print t1.table
+            print '\n'
+            print t2.table
+            print '\n'
+            print t3.table
+
+            time.sleep(args.refresh)

bin/WebStats.py

@@ -66,23 +66,7 @@ def compute_progression(server, field_name, num_day, url_parsed):
         # filter
         if (keyword_total_sum > threshold_total_sum) and (keyword_increase > threshold_increase):
             
-            if server.sismember(redis_progression_name_set, keyword): #if keyword is in the set
+            server.zadd("z_top_progression_"+field_name, float(keyword_increase), keyword)
-                server.hset(redis_progression_name, keyword, keyword_increase) #update its value
-
-            elif (server.scard(redis_progression_name_set) < max_set_cardinality):
-                server.sadd(redis_progression_name_set, keyword)
-
-            else: #not in the set
-                #Check value for all members
-                member_set = []
-                for keyw in server.smembers(redis_progression_name_set):
-                    member_set.append((keyw, int(server.hget(redis_progression_name, keyw))))
-                print member_set
-                member_set.sort(key=lambda tup: tup[1])
-                if member_set[0][1] < keyword_increase:
-                    #remove min from set and add the new one
-                    server.srem(redis_progression_name_set, member_set[0])
-                    server.sadd(redis_progression_name_set, keyword)
 
 
 if __name__ == '__main__':

bin/launch_scripts.sh

@@ -8,33 +8,35 @@ sleep 0.1
 
 echo -e $GREEN"\t* Launching ZMQ scripts"$DEFAULT
 
+    screen -S "Script" -X screen -t "ModuleInformation" bash -c './ModuleInformation.py -k 0 -c 1; read x'
+    sleep 0.1
     screen -S "Script" -X screen -t "Global" bash -c './Global.py; read x'
     sleep 0.1
-screen -S "Script" -X screen -t "Duplicate" bash -c './Duplicate_ssdeep_v2.py; read x'
+    screen -S "Script" -X screen -t "Duplicates" bash -c './Duplicates.py; read x'
     sleep 0.1
-screen -S "Script" -X screen -t "Attribute" bash -c './Attribute.py; read x'
+    screen -S "Script" -X screen -t "Attributes" bash -c './Attributes.py; read x'
     sleep 0.1
-screen -S "Script" -X screen -t "Line" bash -c './Line.py; read x'
+    screen -S "Script" -X screen -t "Lines" bash -c './Lines.py; read x'
     sleep 0.1
-screen -S "Script" -X screen -t "DomainClassifier" bash -c './DomClassifier.py; read x'
+    screen -S "Script" -X screen -t "DomClassifier" bash -c './DomClassifier.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Categ" bash -c './Categ.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Tokenize" bash -c './Tokenize.py; read x'
     sleep 0.1
-screen -S "Script" -X screen -t "CreditCard" bash -c './CreditCard.py; read x'
+    screen -S "Script" -X screen -t "CreditCards" bash -c './CreditCards.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Onion" bash -c './Onion.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Mail" bash -c './Mail.py; read x'
     sleep 0.1
-screen -S "Script" -X screen -t "Url" bash -c './Url.py; read x'
+    screen -S "Script" -X screen -t "Web" bash -c './Web.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Credential" bash -c './Credential.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Curve" bash -c './Curve.py; read x'
     sleep 0.1
-screen -S "Script" -X screen -t "Curve_topsets_manager" bash -c './Curve_manage_top_sets.py; read x'
+    screen -S "Script" -X screen -t "CurveManageTopSets" bash -c './CurveManageTopSets.py; read x'
     sleep 0.1
     screen -S "Script" -X screen -t "Indexer" bash -c './Indexer.py; read x'
     sleep 0.1
@@ -54,4 +56,4 @@ screen -S "Script" -X screen -t "SQLInjectionDetection" bash -c './SQLInjectionD
     sleep 0.1
     screen -S "Script" -X screen -t "Browse_warning_paste" bash -c './Browse_warning_paste.py; read x'
     sleep 0.1
-screen -S "Script" -X screen -t "SentimentAnalyser" bash -c './SentimentAnalyser.py; read x'
+    screen -S "Script" -X screen -t "SentimentAnalysis" bash -c './SentimentAnalysis.py; read x'

bin/packages/modules.cfg

@@ -34,7 +34,7 @@ subscribe = Redis_Global
 publish = Redis_CreditCards,Redis_Mail,Redis_Onion,Redis_Web,Redis_Credential,Redis_SourceCode,Redis_Cve
 
 [CreditCards]
-subscribe = Redis_CreditCard
+subscribe = Redis_CreditCards
 publish = Redis_Duplicate,Redis_ModuleStats,Redis_BrowseWarningPaste
 
 [Mail]

doc/generate_graph_data.py

@@ -0,0 +1,65 @@
+#!/usr/bin/env python2
+# -*-coding:UTF-8 -*
+
+content = ""
+modules = {}
+all_modules = []
+curr_module = ""
+streamingPub = {}
+streamingSub = {}
+
+with open('../bin/packages/modules.cfg', 'r') as f:
+    for line in f:
+        if line[0] != '#':
+            if line[0] == '[':
+                curr_name = line.replace('[','').replace(']','').replace('\n', '').replace(' ', '')
+                all_modules.append(curr_name)
+                modules[curr_name] = {'sub': [], 'pub': []}
+                curr_module = curr_name
+            elif curr_module != "": # searching for sub or pub
+                if line.startswith("subscribe"):
+                    curr_subscribers = [w for w in line.replace('\n', '').replace(' ', '').split('=')[1].split(',')]
+                    modules[curr_module]['sub'] = curr_subscribers
+                    for sub in curr_subscribers:
+                        streamingSub[sub] = curr_module
+
+                elif line.startswith("publish"):
+                    curr_publishers = [w for w in line.replace('\n', '').replace(' ', '').split('=')[1].split(',')]
+                    modules[curr_module]['pub'] = curr_publishers
+                    for pub in curr_publishers:
+                        streamingPub[pub] = curr_module
+                else:
+                    continue
+
+    output_set_graph = set()
+    with open('all_modules.txt', 'w') as f2:
+        for e in all_modules:
+            f2.write(e+"\n")
+
+    for module in modules.keys():
+        for stream_in in modules[module]['sub']:
+            if stream_in not in streamingPub.keys():
+                output_set_graph.add("\"" + stream_in + "\" [color=darkorange1] ;\n")
+                output_set_graph.add("\"" + stream_in + "\"" + "->" + module + ";\n")
+            else:
+                output_set_graph.add("\"" + streamingPub[stream_in] + "\"" + "->" + module + ";\n")
+
+        for stream_out in modules[module]['pub']:
+            if stream_out not in streamingSub.keys():
+                output_set_graph.add("\"" + stream_out + "\" [color=darkorange1] ;\n")
+                output_set_graph.add("\"" + stream_out + "\"" + "->" + module + ";\n")
+            else:
+                output_set_graph.add("\"" + module + "\"" + "->" + streamingSub[stream_out] + ";\n")
+
+
+    output_text_graph = ""
+    output_text_graph += "digraph unix {\n"\
+                              "graph [pad=\"0.5\"];\n"\
+                              "size=\"25,25\";\n"\
+                              "node [color=lightblue2, style=filled];\n"
+
+    for elem in output_set_graph:
+        output_text_graph += elem
+
+    output_text_graph += "}"
+    print output_text_graph

doc/generate_modules_data_flow_graph.sh

@@ -0,0 +1,3 @@
+#!/bin/bash
+
+python generate_graph_data.py | dot -T png -o module-data-flow.png

installing_deps.sh

@@ -17,6 +17,9 @@ sudo apt-get install libadns1 libadns1-dev
 #Needed for redis-lvlDB
 sudo apt-get install libev-dev libgmp-dev
 
+#Need for generate-data-flow graph
+sudo apt-get install graphviz
+
 #needed for mathplotlib
 test ! -L /usr/include/ft2build.h && sudo ln -s freetype2/ft2build.h /usr/include/
 sudo easy_install -U distribute

pip_packages_requirement.txt

@@ -10,6 +10,7 @@ textblob
 numpy
 matplotlib
 networkx
+terminaltables
 
 #Tokeniser
 nltk

var/www/Flask_server.py

@@ -81,48 +81,37 @@ def event_stream():
 
 def get_queues(r):
     # We may want to put the llen in a pipeline to do only one query.
-    return [(queue, int(card)) for queue, card in
+    data = [(queue, int(card)) for queue, card in r.hgetall("queues").iteritems()]
-            r.hgetall("queues").iteritems()]
+    newData = []
+
+    curr_range = 50
+    for queue, card in data:
+        key = "MODULE_" + queue + "_"
+        for i in range(1, 50):
+            curr_num = r.get("MODULE_"+ queue + "_" + str(i))
+            if curr_num is None:
+                curr_range = i
+                break
+
+        for moduleNum in range(1, curr_range):
+            value = r.get(key + str(moduleNum))
+            if value is not None:
+                timestamp, path = value.split(", ")
+                if timestamp is not None:
+                    startTime_readable = datetime.datetime.fromtimestamp(int(timestamp))
+                    processed_time_readable = str((datetime.datetime.now() - startTime_readable)).split('.')[0]
+                    seconds = int((datetime.datetime.now() - startTime_readable).total_seconds())
+                    newData.append( (queue, card, seconds, moduleNum) )
+                else:
+                    newData.append( (queue, cards, 0, moduleNum) )
+
+    return newData
 
 
 def list_len(s):
     return len(s)
 app.jinja_env.filters['list_len'] = list_len
 
-def parseStringToList(the_string):
-    strList = ""
-    elemList = []
-    for c in the_string:
-        if c != ']':
-            if c != '[' and c !=' ' and c != '"':
-                strList += c
-        else:
-            the_list = strList.split(',')
-            if len(the_list) == 3:
-               elemList = elemList + the_list
-            elif len(the_list) == 2:
-               elemList.append(the_list)
-            elif len(the_list) > 1:
-               elemList.append(the_list[1:])
-            strList = ""
-    return elemList
-
-def parseStringToList2(the_string):
-    if the_string == []:
-        return []
-    else:
-        res = []
-        tab_str = the_string.split('], [')
-        tab_str[0] = tab_str[0][1:]+']'
-        tab_str[len(tab_str)-1] = '['+tab_str[len(tab_str)-1][:-1]
-        res.append(parseStringToList(tab_str[0]))
-        for i in range(1, len(tab_str)-2):
-            tab_str[i] = '['+tab_str[i]+']'
-            res.append(parseStringToList(tab_str[i]))
-        if len(tab_str) > 1:
-            res.append(parseStringToList(tab_str[len(tab_str)-1]))
-        return res
-
 
 def showpaste(content_range):
     requested_path = request.args.get('paste', '')
@@ -136,7 +125,7 @@ def showpaste(content_range):
     p_mime = paste.p_mime
     p_lineinfo = paste.get_lines_info()
     p_content = paste.get_p_content().decode('utf-8', 'ignore')
-    p_duplicate_full_list = parseStringToList2(paste._get_p_duplicate())
+    p_duplicate_full_list = json.loads(paste._get_p_duplicate())
     p_duplicate_list = []
     p_simil_list = []
     p_hashtype_list = []
@@ -160,7 +149,7 @@ def showpaste(content_range):
         hash_types = []
         comp_vals = []
         for i in indices:
-            hash_types.append(p_duplicate_full_list[i][0])
+            hash_types.append(p_duplicate_full_list[i][0].encode('utf8'))
             comp_vals.append(p_duplicate_full_list[i][2])
             dup_list_removed.append(i)
 
@@ -267,19 +256,9 @@ def progressionCharts():
         return jsonify(bar_values)
 
     else:
-        redis_progression_name = 'top_progression_'+trending_name
+        redis_progression_name = "z_top_progression_" + trending_name
-        redis_progression_name_set = 'top_progression_'+trending_name+'_set'
+        keyw_value = r_serv_charts.zrevrangebyscore(redis_progression_name, '+inf', '-inf', withscores=True, start=0, num=10)
-
+        return jsonify(keyw_value)
-        # Iterate over element in top_x_set and retreive their value
-        member_set = []
-        for keyw in r_serv_charts.smembers(redis_progression_name_set):
-            keyw_value = r_serv_charts.hget(redis_progression_name, keyw)
-            keyw_value = keyw_value if keyw_value is not None else 0
-            member_set.append((keyw, int(keyw_value)))
-        member_set.sort(key=lambda tup: tup[1], reverse=True)
-        if len(member_set) == 0:
-            member_set.append(("No relevant data", int(100)))
-        return jsonify(member_set)
 
 @app.route("/_moduleCharts", methods=['GET'])
 def modulesCharts():
@@ -458,7 +437,7 @@ def sentiment_analysis_trending():
     return render_template("sentiment_analysis_trending.html")
 
 
-@app.route("/sentiment_analysis_getplotdata/")
+@app.route("/sentiment_analysis_getplotdata/", methods=['GET'])
 def sentiment_analysis_getplotdata():
     # Get the top providers based on number of pastes
     oneHour = 60*60
@@ -467,19 +446,27 @@ def sentiment_analysis_getplotdata():
     dateStart = dateStart.replace(minute=0, second=0, microsecond=0)
     dateStart_timestamp = calendar.timegm(dateStart.timetuple())
 
-    to_return = {}
+    getAllProviders = request.args.get('getProviders')
+    provider = request.args.get('provider')
+    allProvider = request.args.get('all')
+    if getAllProviders == 'True':
+        if allProvider == "True":
+            range_providers = r_serv_charts.smembers('all_provider_set')
+            return jsonify(list(range_providers))
+        else:
             range_providers = r_serv_charts.zrevrangebyscore('providers_set_'+ get_date_range(0)[0], '+inf', '-inf', start=0, num=8)
             # if empty, get yesterday top providers
-    print 'providers_set_'+ get_date_range(1)[1]
             range_providers = r_serv_charts.zrevrangebyscore('providers_set_'+ get_date_range(1)[1], '+inf', '-inf', start=0, num=8) if range_providers == [] else range_providers
             # if still empty, takes from all providers
             if range_providers == []:
                 print 'today provider empty'
                 range_providers = r_serv_charts.smembers('all_provider_set')
+            return jsonify(range_providers)
 
-    for cur_provider in range_providers:
+    elif provider is not None:
-        print cur_provider
+        to_return = {}
-        cur_provider_name = cur_provider + '_'
+
+        cur_provider_name = provider + '_'
         list_date = {}
         for cur_timestamp in range(int(dateStart_timestamp), int(dateStart_timestamp)-sevenDays-oneHour, -oneHour):
             cur_set_name = cur_provider_name + str(cur_timestamp)
@@ -489,9 +476,10 @@ def sentiment_analysis_getplotdata():
                 cur_value = r_serv_sentiment.get(cur_id)
                 list_value.append(cur_value)
             list_date[cur_timestamp] = list_value
-        to_return[cur_provider] = list_date
+        to_return[provider] = list_date
 
         return jsonify(to_return)
+    return "Bad request"
 
 
 

var/www/static/js/indexjavascript.js

@@ -221,11 +221,24 @@ function create_queue_table() {
 
     for(i = 0; i < (glob_tabvar.row1).length;i++){
         var tr = document.createElement('TR')
-        for(j = 0; j < (glob_tabvar.row1[i]).length; j++){
+        for(j = 0; j < 2; j++){
             var td = document.createElement('TD')
-            td.appendChild(document.createTextNode(glob_tabvar.row1[i][j]));
+            var moduleNum = j == 0 ? "." + glob_tabvar.row1[i][3] : "";
+            td.appendChild(document.createTextNode(glob_tabvar.row1[i][j] + moduleNum));
             tr.appendChild(td)
         }
+        // Used to decide the color of the row
+        // We have glob_tabvar.row1[][j] with:
+        // - j=0: ModuleName
+        // - j=1: queueLength
+        // - j=2: LastProcessedPasteTime
+        // - j=3: Number of the module belonging in the same category
+        if (parseInt(glob_tabvar.row1[i][2]) > 60*2 && parseInt(glob_tabvar.row1[i][1]) > 2)
+            tr.className += " danger";
+        else if (parseInt(glob_tabvar.row1[i][2]) > 60*1)
+            tr.className += " warning";
+        else
+            tr.className += " success";
         tableBody.appendChild(tr);
     }
     Tablediv.appendChild(table);

var/www/static/js/sentiment_trending.js

@@ -1,4 +1,42 @@
 
+/* Functions and config */
+function add_new_graph_today(id) {
+    return "<div id=\"panel-today\" class=\"panel panel-default pannelToday"+id+"\">" +
+                "<div class=\"panel-heading\">" +
+                    "<strong class=\"sparkLineStatsToday"+id+"t\">Graph "+id+"</strong>" +
+                    "<strong class=\"sparkLineStatsToday"+id+"s pull-right\">Avg</strong>" +
+                "</div>" +
+                "<div class=\"panel-body panelInside\">" +
+                    "<table class=\"table\">" +
+                        "<tbody>" +
+                            "<tr>" +
+                                "<td style=\"border-top: 0px solid #ddd;\"><div class=\"sparkLineStatsToday"+id+"\"></div></td> " +
+                                "<td style=\"border-top: 0px solid #ddd;\"><div class=\"sparkLineStatsToday"+id+"b\"></div></td> " +
+                            "</tr>" +
+                         "</tbody>" +
+                    "</table>" +
+                "</div>" +
+            "</div>";
+};
+function add_new_graph_week(id) {
+    return "<div id=\"panel-week\" class=\"panel panel-default pannelWeek"+id+"\">" +
+                "<div class=\"panel-heading\">" +
+                    "<strong class=\"sparkLineStatsWeek"+id+"t\">Graph "+id+"</strong>" +
+                    "<strong class=\"sparkLineStatsWeek"+id+"s pull-right\">Avg</strong>" +
+                "</div>" +
+                "<div class=\"panel-body panelInside\">" +
+                    "<table class=\"table\">" +
+                        "<tbody>" +
+                            "<tr>" +
+                                "<td style=\"border-top: 0px solid #ddd;\"><div class=\"sparkLineStatsWeek"+id+"\"></div></td> " +
+                                "<td style=\"border-top: 0px solid #ddd;\"><div class=\"sparkLineStatsWeek"+id+"b\"></div></td> " +
+                            "</tr>" +
+                         "</tbody>" +
+                    "</table>" +
+                "</div>" +
+            "</div>";
+}
+
  function generate_offset_to_time(num){
      var to_ret = {};
      for(i=0; i<=num; i++) {
@@ -45,29 +83,74 @@
 };
 
 
-$.getJSON("/sentiment_analysis_getplotdata/",
+/* Plot and queries */
-    function(data) {
+
-        var all_data = [];
+var all_graph_day_sum = 0.0;
-        var plot_data = [];
+var all_graph_hour_sum = 0.0;
+var all_graph_hour_sum_minus = 0.0;
+var all_graph_hour_maxVal = 0.0;
+var all_day_avg = 0.0;
+var all_day_avg_maxVal = 0.0;
 var graph_avg = [];
+var all_data = [];
+var provider_already_loaded = [];
+var totNumGraph = 0;
+
+// Query all providers name then launch the query and plot process for each of them.
+// When everything is terminated, plot the widgets (Gauge, canvasJS, table)
+// input: all - set to 'True' if you take all providers
+function draw_page(all) {
+    $.getJSON("/sentiment_analysis_getplotdata/?getProviders=True&all="+all,
+        function(data) {
+            var promises = [];
+
+            var the_length = provider_already_loaded.length == 0 ? 0 : provider_already_loaded.length;
+            for(i=0; i<data.length; i++) {
+                if(provider_already_loaded.indexOf(data[i]) != -1) {
+                    continue;
+                } else {
+                    totNumGraph++;
+                    if(i % 2 == 0) {
+                        $("#today_divl").append(add_new_graph_today(i+the_length+1));
+                        $("#week_divl").append(add_new_graph_week(i+the_length+1));
+                    }
+                    else {
+                        $("#today_divr").append(add_new_graph_today(i+the_length+1));
+                        $("#week_divr").append(add_new_graph_week(i+the_length+1));
+                    }
+                    provider_already_loaded.push(data[i])
+                    promises.push(query_and_plot(data[i], i+the_length));
+
+                }
+            }
+
+            $.when.apply($, promises).done( function (arg) {
+                draw_widgets();
+                $("#LoadAll").show('fast');
+            });
+        }
+    );
+}
+
+
+// Query data and plot it for a given provider
+// input - provider: The povider name to be plotted
+// input - graphNum: The number of the graph (Use to plot on correct div)
+function query_and_plot(provider, graphNum) {
+    var query_plot = $.getJSON("/sentiment_analysis_getplotdata/?provider="+provider,
+        function(data) {
+            var plot_data = [];
             var array_provider = Object.keys(data);
             var dates_providers = Object.keys(data[array_provider[0]]);
             var dateStart = parseInt(dates_providers[0]);
             var oneHour = 60*60;
             var oneWeek = oneHour*24*7;
 
-        var all_graph_day_sum = 0.0;
-        var all_graph_hour_sum = 0.0;
-        var all_graph_hour_maxVal = 0.0;
-        var all_day_avg = 0.0;
-        var all_day_avg_maxVal = 0.0;
-
-        for (graphNum=0; graphNum<8; graphNum++) {
             var max_value = 0.0;
             var max_value_day = 0.0;
             var graph_data = [];
             var spark_data = [];
-            var curr_provider = array_provider[graphNum];
+            var curr_provider = array_provider[0];
             var curr_sum = 0.0;
             var curr_sum_elem = 0.0;
             var day_sum = 0.0;
@@ -88,7 +171,6 @@ $.getJSON("/sentiment_analysis_getplotdata/",
                     var neu = 0;
 
                     for(i=0; i<data_array.length; i++){
-                        //console.log(data_array[i].replace(/\'/g, '\"'));
                         var curr_data = jQuery.parseJSON(data_array[i].replace(/\'/g, '\"'));
                         compPosAvg += curr_data['compoundPos'];
                         compNegAvg += curr_data['compoundNeg'];
@@ -109,8 +191,7 @@ $.getJSON("/sentiment_analysis_getplotdata/",
                     max_value = Math.abs(pos-neg) > max_value ? Math.abs(pos-neg) : max_value;
 
                     if(curr_date >= dateStart+oneWeek-23*oneHour){
-                        max_value_day = Math.abs(pos-neg) > max_value_day ? Math.abs(pos-neg) : max_value_day;
+                        max_value_day = Math.abs(pos-neg) > max_value_day ? Math.abs(pos-neg) : max_value_day; day_sum += (pos-neg);
-                        day_sum += (pos-neg);
                         day_sum_elem++;
                     }
                     if(curr_date > dateStart+oneWeek-2*oneHour && curr_date <=dateStart+oneWeek-oneHour){
@@ -121,13 +202,12 @@ $.getJSON("/sentiment_analysis_getplotdata/",
             }
             all_graph_day_sum += day_sum;
             all_graph_hour_sum += hour_sum;
+            all_graph_hour_sum_minus += hour_sum > 0 ? 0 : 1;
             all_graph_hour_maxVal = Math.abs(hour_sum) > all_graph_hour_maxVal ? Math.abs(hour_sum) : all_graph_hour_maxVal;
 
             var curr_avg = curr_sum / (curr_sum_elem);
             if(isNaN(curr_avg))
                 curr_avg = 0.0
-            //var curr_avg = curr_sum / (oneWeek/oneHour); 
-            //var curr_avg = curr_sum / (spark_data.length); 
             graph_avg.push([curr_provider, curr_avg]);
             plot_data.push(spark_data);
             all_data.push(graph_data);
@@ -141,7 +221,7 @@ $.getJSON("/sentiment_analysis_getplotdata/",
             var num = graphNum + 1;
             var placeholder = '.sparkLineStatsWeek' + num;
             sparklineOptions.barWidth = 2;
-            $(placeholder).sparkline(plot_data[graphNum], sparklineOptions);
+            $(placeholder).sparkline(plot_data[0], sparklineOptions);
             $(placeholder+'t').text(curr_provider);
             var curr_avg_text = isNaN(curr_avg) ? "No data" : curr_avg.toFixed(5);
             $(placeholder+'s').text(curr_avg_text);
@@ -169,8 +249,8 @@ $.getJSON("/sentiment_analysis_getplotdata/",
 
 
             // print today
-            var data_length = plot_data[graphNum].length;
+            var data_length = plot_data[0].length;
-            var data_today = plot_data[graphNum].slice(data_length-24, data_length);
+            var data_today = plot_data[0].slice(data_length-24, data_length);
 
             placeholder = '.sparkLineStatsToday' + num;
             sparklineOptions.barWidth = 14;
@@ -199,10 +279,16 @@ $.getJSON("/sentiment_analysis_getplotdata/",
                 $(avgName).addClass("panel-warning")
             }
 
-        }//for loop
+
+        }
+    );
+    return query_plot
+}
 
 
 
+function draw_widgets() {
+
     /* ---------------- Gauge ---------------- */
     var gaugeOptions = {
         animateEasing: true,
@@ -233,15 +319,14 @@ $.getJSON("/sentiment_analysis_getplotdata/",
     gaugeOptions.appendTo = '#gauge_today_last_hour';
     gaugeOptions.dialLabel = 'Last hour';
     gaugeOptions.elementId = 'gauge1';
-        var piePercent = (all_graph_hour_sum / 8) / all_graph_hour_maxVal;
+    var piePercent = (all_graph_hour_sum / (totNumGraph - all_graph_hour_sum_minus)) / all_graph_hour_maxVal;
     gaugeOptions.inc = piePercent;
     var gauge_today_last_hour = new FlexGauge(gaugeOptions);
 
     gaugeOptions2.appendTo = '#gauge_today_last_days';
     gaugeOptions2.dialLabel = 'Today';
     gaugeOptions2.elementId = 'gauge2';
-        //piePercent = (all_graph_day_sum / (8*24)) / max_value;
+    piePercent = (all_day_avg / totNumGraph) / all_day_avg_maxVal;
-        piePercent = (all_day_avg / 8) / all_day_avg_maxVal;
     gaugeOptions2.inc = piePercent;
     var gauge_today_last_days = new FlexGauge(gaugeOptions2);
 
@@ -276,7 +361,7 @@ $.getJSON("/sentiment_analysis_getplotdata/",
     var comp_sum_day_neg = 0.0;
     var comp_sum_hour_pos = 0.0;
     var comp_sum_hour_neg = 0.0;
-        for(graphNum=0; graphNum<8; graphNum++){
+    for(graphNum=0; graphNum<totNumGraph; graphNum++){
         curr_graphData = all_data[graphNum];
         var gauge_data = curr_graphData.slice(curr_graphData.length-24, curr_graphData.length);
         for (i=1; i< gauge_data.length; i++){
@@ -311,7 +396,7 @@ $.getJSON("/sentiment_analysis_getplotdata/",
                 type: "bar",
                 color: "green",
                 dataPoints: [
-                        {y: comp_sum_hour_pos/8}
+                    {y: comp_sum_hour_pos/totNumGraph}
                 ]
             },
             {
@@ -319,7 +404,7 @@ $.getJSON("/sentiment_analysis_getplotdata/",
                 type: "bar",
                 color: "red",
                 dataPoints: [
-                        {y: comp_sum_hour_neg/8}
+                    {y: comp_sum_hour_neg/totNumGraph}
                 ]
             }
         ]
@@ -347,7 +432,7 @@ $.getJSON("/sentiment_analysis_getplotdata/",
             type: "bar",
             color: "green",
             dataPoints: [
-                    {y: comp_sum_day_pos/8}
+                {y: comp_sum_day_pos/totNumGraph}
             ]
         },
         {
@@ -355,7 +440,7 @@ $.getJSON("/sentiment_analysis_getplotdata/",
             type: "bar",
             color: "red",
             dataPoints: [
-                    {y: comp_sum_day_neg/8}
+                {y: comp_sum_day_neg/totNumGraph}
             ]
         }
         ]
@@ -366,21 +451,4 @@ $.getJSON("/sentiment_analysis_getplotdata/",
     chart_canvas1.render();
     chart_canvas2.render();
 
-
-
 }
-);
-
-
-
-
-
-
-
-
-
-
-
-
-
-

var/www/templates/index.html

@@ -154,7 +154,6 @@
     </script>
 
     <script src="{{ url_for('static', filename='js/bootstrap.min.js') }}"></script>
-    <script src="{{ url_for('static', filename='js/plugins/metisMenu/metisMenu.js') }}"></script>
 
 </body>
 

var/www/templates/sentiment_analysis_trending.html

@@ -113,6 +113,7 @@
   </div>
     <!-- /.row -->
     <div class="row">
+        <button id="LoadAll" class="btn btn-info" style="margin: 5px;"><span class="glyphicon glyphicon-download"> </span> Load data from all providers </button>
 
     <!-- Pannel TODAY -->
     <div class="row">
@@ -128,139 +129,13 @@
                             <div class="col-lg-9" style="padding-left: 0px;">
                                 <!-- providers charts -->
                                 <div class="col-lg-6">
-                                   <div class="sparkLineStats">
+                                   <div id="today_divl" class="sparkLineStats">
-                                       <div id="panel-today" class="panel panel-default pannelToday1">
+
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsToday1t">Graph 1</strong>
-                                               <strong class="sparkLineStatsToday1s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday1"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday1b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                       </div>
-                                       <div id="panel-today" class="panel panel-default pannelToday2">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsToday2t">Graph 2</strong>
-                                               <strong class="sparkLineStatsToday2s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday2"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday2b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                       </div>
-                                       <div id="panel-today" class="panel panel-default pannelToday3">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsToday3t">Graph 3</strong>
-                                               <strong class="sparkLineStatsToday3s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday3"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday3b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                        </div>
-                                       <div id="panel-today" class="panel panel-default pannelToday4">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsToday4t">Graph 4</strong>
-                                               <strong class="sparkLineStatsToday4s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday4"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday4b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                        </div>
                                    </div>
                                 </div>
                                 <div class="col-lg-6">
-                                   <div class="sparkLineStats">
+                                   <div id="today_divr" class="sparkLineStats">
-                                       <div id="panel-today" class="panel panel-default pannelToday5">
+
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsToday5t">Graph 5</strong>
-                                               <strong class="sparkLineStatsToday5s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday5"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday5b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                        </div>
-                                       <div id="panel-today" class="panel panel-default pannelToday6">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsToday6t">Graph 6</strong>
-                                               <strong class="sparkLineStatsToday6s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday6"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday6b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                        </div>
-                                       <div id="panel-today" class="panel panel-default pannelToday7">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsToday7t">Graph 7</strong>
-                                               <strong class="sparkLineStatsToday7s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday7"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday7b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                        </div>
-                                       <div id="panel-today" class="panel panel-default pannelToday8">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsToday8t">Graph 8</strong>
-                                               <strong class="sparkLineStatsToday8s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday8"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsToday8b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                        </div>
                                    </div>
                                 </div> 
                             </div>
@@ -304,139 +179,13 @@
                             <div class="col-lg-9" style="padding-left: 0px;">
                                 <!-- providers charts -->
                                 <div class="col-lg-6">
-                                   <div class="sparkLineStats">
+                                   <div id="week_divl" class="sparkLineStats">
-                                       <div id="panel-week" class="panel panel-default pannelWeek1">
+
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsWeek1t">Graph 1</strong>
-                                               <strong class="sparkLineStatsWeek1s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek1"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek1b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                       </div>
-                                       <div id="panel-week" class="panel panel-default pannelWeek2">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsWeek2t">Graph 2</strong>
-                                               <strong class="sparkLineStatsWeek2s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek2"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek2b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                       </div>
-                                       <div id="panel-week" class="panel panel-default pannelWeek3">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsWeek3t">Graph 3</strong>
-                                               <strong class="sparkLineStatsWeek3s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek3"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek3b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                       </div>
-                                       <div id="panel-week" class="panel panel-default pannelWeek4">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsWeek4t">Graph 4</strong>
-                                               <strong class="sparkLineStatsWeek4s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek4"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek4b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                       </div>
                                    </div>
                                 </div>
                                 <div class="col-lg-6">
-                                   <div class="sparkLineStats">
+                                   <div id="week_divr" class="sparkLineStats">
-                                       <div id="panel-week" class="panel panel-default pannelWeek5">
+                                       
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsWeek5t">Graph 5</strong>
-                                               <strong class="sparkLineStatsWeek5s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek5"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek5b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                       </div>
-                                       <div id="panel-week" class="panel panel-default pannelWeek6">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsWeek6t">Graph 6</strong>
-                                               <strong class="sparkLineStatsWeek6s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek6"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek6b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                       </div>
-                                       <div id="panel-week" class="panel panel-default pannelWeek7">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsWeek7t">Graph 7</strong>
-                                               <strong class="sparkLineStatsWeek7s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek7"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek7b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                       </div>
-                                       <div id="panel-week" class="panel panel-default pannelWeek8">
-                                           <div class="panel-heading">
-                                               <strong class="sparkLineStatsWeek8t">Graph 8</strong>
-                                               <strong class="sparkLineStatsWeek8s pull-right">Avg</strong>
-                                           </div>
-                                           <div class="panel-body panelInside">
-                                               <table class="table">
-                                                   <tbody>
-                                                       <tr>
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek8"></div></td> 
-                                                           <td style="border-top: 0px solid #ddd;"><div class="sparkLineStatsWeek8b"></div></td> 
-                                                       </tr>
-                                                    </tbody>
-                                               </table>
-                                           </div>
-                                       </div>
                                    </div>
                                 </div> 
                             </div>
@@ -491,8 +240,6 @@
     </div>
 
     
-
-    
     <!-- /.row -->
     </div>
 <!-- /#page-wrapper -->
@@ -502,13 +249,15 @@
 <!-- import graph function -->
 <script src="{{ url_for('static', filename='js/sentiment_trending.js') }}"></script>
 <script>
+    $("#LoadAll").hide();
 
     $(document).ready(function(){
         activePage = $('h1.page-header').attr('data-page');
         $("#"+activePage).addClass("active");
         $('[data-toggle="tooltip"]').tooltip(); 
 
-
+        $("#LoadAll").click(function(){ draw_page("True"); });
+        draw_page("False");
         // Reload every 30min
         setTimeout(function(){ location.reload(); }, 30*60*1000);
    });

var/www/templates/terms_management.html

@@ -233,15 +233,15 @@
             $.getJSON(url, function (data) {
                 if (data.length != 0) {
                     var html_to_add = "";
-                    html_to_add += "<table class=\"table table-striped\">";
+                    html_to_add += "<table id=\"modal-table\" class=\"table table-striped\">";
                     html_to_add += "<thead>";
                       html_to_add += "<tr>";
                         html_to_add += "<th>Source</th>";
                         html_to_add += "<th>Date</th>";
                         html_to_add += "<th>Encoding</th>";
                         html_to_add += "<th>Size (Kb)</th>";
-                        html_to_add += "<th>Mime</th>";
+                        html_to_add += "<th># lines</th>";
-                        html_to_add += "<th>(# lines, Max line length)</th>";
+                        html_to_add += "<th>Max length</th>";
                         html_to_add += "<th>Preview</th>";
                       html_to_add += "</tr>";
                     html_to_add += "</thead>";
@@ -253,8 +253,8 @@
                                 html_to_add += "<td>"+curr_data.date+"</td>";
                                 html_to_add += "<td>"+curr_data.encoding+"</td>";
                                 html_to_add += "<td>"+curr_data.size+"</td>";
-                                html_to_add += "<td>"+curr_data.mime+"</td>";
+                                html_to_add += "<td>"+curr_data.lineinfo[0]+"</td>";
-                                html_to_add += "<td>("+curr_data.lineinfo[0]+", "+curr_data.lineinfo[1]+")</td>";
+                                html_to_add += "<td>"+curr_data.lineinfo[1]+"</td>";
                                 html_to_add += "<td><div class=\"row\"><button class=\"btn btn-xs btn-default\" data-toggle=\"popover\" data-placement=\"left\" data-content=\""+curr_data.content+"\">Preview content</button><a target=\"_blank\" href=\"{{ url_for('showsavedpaste') }}?paste="+curr_data.path+"&num=0\">  <button type=\"button\" class=\"btn btn-xs btn-info\">Show Paste</button></a></div></td>";
 
                             html_to_add += "</tr>";
@@ -264,6 +264,7 @@
                     $("#mymodalbody").html(html_to_add);
                     $("[data-toggle=popover]").popover();
                     $("#button_show_plot").attr("href", "{{ url_for('terms_plot_tool')}}"+"?term="+the_modal.attr('data-term') );
+	            $('#modal-table').DataTable();
                 } else {
                     $("#mymodalbody").html("No paste containing this term has been received yet.");
                     $("#button_show_plot").attr("href", "{{ url_for('terms_plot_tool')}}"+"?term="+the_modal.attr('data-term') );