Merge branch 'terms-improvements'

bin/Curve.py

@@ -117,12 +117,18 @@ if __name__ == "__main__":
             r_serv1.hincrby(low_word, date, int(score))
 
             # Update redis
+            #consider the num of occurence of this term
             curr_word_value = int(server_term.hincrby(timestamp, low_word, int(score)))
+            #1 term per paste
+            curr_word_value_perPaste = int(server_term.hincrby("per_paste_" + str(timestamp), low_word, int(1)))
 
             # Add in set only if term is not in the blacklist
             if low_word not in server_term.smembers(BlackListTermsSet_Name):
+                #consider the num of occurence of this term
                 server_term.zincrby(curr_set, low_word, float(score))
-            
+                #1 term per paste
+                server_term.zincrby("per_paste_" + curr_set, low_word, float(1))
+             
             #Add more info for tracked terms
             check_if_tracked_term(low_word, filename)
 

bin/CurveManageTopSets.py

@@ -48,11 +48,13 @@ def manage_top_set():
     blacklist_size = int(server_term.scard(BlackListTermsSet_Name))
 
     dico = {}
+    dico_per_paste = {}
 
     # Retreive top data (max_card + blacklist_size) from days sets
     for timestamp in range(startDate, startDate - top_termFreq_setName_month[1]*oneDay, -oneDay):
         curr_set = top_termFreq_setName_day[0] + str(timestamp)
         array_top_day = server_term.zrevrangebyscore(curr_set, '+inf', '-inf', withscores=True, start=0, num=top_term_freq_max_set_cardinality+blacklist_size)
+        array_top_day_per_paste = server_term.zrevrangebyscore("per_paste_" + curr_set, '+inf', '-inf', withscores=True, start=0, num=top_term_freq_max_set_cardinality+blacklist_size)
 
         for word, value in array_top_day:
             if word not in server_term.smembers(BlackListTermsSet_Name):
@@ -61,8 +63,16 @@ def manage_top_set():
                 else:
                     dico[word] = value
 
+        for word, value in array_top_day_per_paste:
+            if word not in server_term.smembers(BlackListTermsSet_Name):
+                if word in dico_per_paste.keys():
+                    dico_per_paste[word] += value
+                else:
+                    dico_per_paste[word] = value
+
         if timestamp == startDate - num_day_week*oneDay:
             dico_week = copy.deepcopy(dico)
+            dico_week_per_paste = copy.deepcopy(dico_per_paste)
 
     # convert dico into sorted array
     array_month = []
@@ -77,17 +87,37 @@ def manage_top_set():
     array_week.sort(key=lambda tup: -tup[1])
     array_week = array_week[0:20]
 
+    # convert dico_per_paste into sorted array
+    array_month_per_paste = []
+    for w, v in dico_per_paste.iteritems():
+        array_month_per_paste.append((w, v))
+    array_month_per_paste.sort(key=lambda tup: -tup[1])
+    array_month_per_paste = array_month_per_paste[0:20]
+
+    array_week_per_paste = []
+    for w, v in dico_week_per_paste.iteritems():
+        array_week_per_paste.append((w, v))
+    array_week_per_paste.sort(key=lambda tup: -tup[1])
+    array_week_per_paste = array_week_per_paste[0:20]
+
+
     # suppress every terms in top sets
     for curr_set, curr_num_day in top_termFreq_set_array[1:3]:
         for w in server_term.zrange(curr_set, 0, -1):
             server_term.zrem(curr_set, w)
+        for w in server_term.zrange("per_paste_" + curr_set, 0, -1):
+            server_term.zrem("per_paste_" + curr_set, w)
 
     # Add top term from sorted array in their respective sorted sets
     for elem in array_week:
         server_term.zadd(top_termFreq_setName_week[0], float(elem[1]), elem[0])
+    for elem in array_week_per_paste:
+        server_term.zadd("per_paste_" + top_termFreq_setName_week[0], float(elem[1]), elem[0])
   
     for elem in array_month:
         server_term.zadd(top_termFreq_setName_month[0], float(elem[1]), elem[0])
+    for elem in array_month_per_paste:
+        server_term.zadd("per_paste_" + top_termFreq_setName_month[0], float(elem[1]), elem[0])
 
     timestamp = int(time.mktime(datetime.datetime.now().timetuple()))
     value = str(timestamp) + ", " + "-"

doc/all_modules.txt

@@ -1,31 +0,0 @@
-Attributes
-BrowseWarningPaste
-Categ
-Credential
-CreditCards
-Curve
-CurveManageTopSets
-Cve
-DomClassifier
-Duplicates
-Global
-Indexer
-Keys
-Lines
-Mail
-Mixer
-ModuleInformation
-Keys
-Lines
-Mail
-Mixer
-ModuleInformation
-ModuleStats
-Onion
-Phone
-Release
-SentimentAnalysis
-SQLInjectionDetection
-Tokenize
-Web
-WebStats

var/www/Flasks/Flask_terms.py

@@ -20,13 +20,13 @@ cfg = Flask_config.cfg
 r_serv_term = Flask_config.r_serv_term
 # ============ FUNCTIONS ============
 
-def Term_getValueOverRange(word, startDate, num_day):
+def Term_getValueOverRange(word, startDate, num_day, per_paste=""):
     passed_days = 0
     oneDay = 60*60*24
     to_return = []
     curr_to_return = 0
     for timestamp in range(startDate, startDate - max(num_day)*oneDay, -oneDay):
-        value = r_serv_term.hget(timestamp, word)
+        value = r_serv_term.hget(per_paste+str(timestamp), word)
         curr_to_return += int(value) if value is not None else 0
         for i in num_day:
             if passed_days == i-1:
@@ -39,6 +39,14 @@ def Term_getValueOverRange(word, startDate, num_day):
 
 @app.route("/terms_management/")
 def terms_management():
+    per_paste = request.args.get('per_paste')
+    if per_paste == "1" or per_paste is None:
+        per_paste_text = "per_paste_"
+        per_paste = 1
+    else:
+        per_paste_text = ""
+        per_paste = 0
+
     TrackedTermsSet_Name = "TrackedSetTermSet"
     BlackListTermsSet_Name = "BlackListSetTermSet"
     TrackedTermsDate_Name = "TrackedTermDate"
@@ -53,7 +61,7 @@ def terms_management():
     track_list_num_of_paste = []
     for tracked_term in r_serv_term.smembers(TrackedTermsSet_Name):
         track_list.append(tracked_term)
-        value_range = Term_getValueOverRange(tracked_term, today_timestamp, [1, 7, 31])
+        value_range = Term_getValueOverRange(tracked_term, today_timestamp, [1, 7, 31], per_paste=per_paste_text)
 
         term_date = r_serv_term.hget(TrackedTermsDate_Name, tracked_term)
 
@@ -70,7 +78,7 @@ def terms_management():
         term_date = datetime.datetime.utcfromtimestamp(int(term_date)) if term_date is not None else "No date recorded"
         black_list.append([blacked_term, term_date])
 
-    return render_template("terms_management.html", black_list=black_list, track_list=track_list, track_list_values=track_list_values,  track_list_num_of_paste=track_list_num_of_paste)
+    return render_template("terms_management.html", black_list=black_list, track_list=track_list, track_list_values=track_list_values,  track_list_num_of_paste=track_list_num_of_paste, per_paste=per_paste)
 
 
 @app.route("/terms_management_query_paste/")
@@ -182,12 +190,19 @@ def terms_plot_tool_data():
     range_end = calendar.timegm(range_end.timetuple())
     term =  request.args.get('term')
 
+    per_paste = request.args.get('per_paste')
+    if per_paste == "1" or per_paste is None:
+        per_paste = "per_paste_"
+    else:
+        per_paste = ""
+
+
     if term is None:
         return "None"
     else:
         value_range = []
         for timestamp in range(range_start, range_end+oneDay, oneDay):
-            value = r_serv_term.hget(timestamp, term)
+            value = r_serv_term.hget(per_paste+str(timestamp), term)
             curr_value_range = int(value) if value is not None else 0
             value_range.append([timestamp, curr_value_range])
         value_range.insert(0,term)
@@ -196,7 +211,9 @@ def terms_plot_tool_data():
 
 @app.route("/terms_plot_top/")
 def terms_plot_top():
-    return render_template("terms_plot_top.html")
+    per_paste = request.args.get('per_paste')
+    per_paste = per_paste if per_paste is not None else 1
+    return render_template("terms_plot_top.html", per_paste=per_paste)
 
 
 @app.route("/terms_plot_top_data/")
@@ -206,17 +223,24 @@ def terms_plot_top_data():
     today = today.replace(hour=0, minute=0, second=0, microsecond=0)
     today_timestamp = calendar.timegm(today.timetuple())
 
-    set_day = "TopTermFreq_set_day_" + str(today_timestamp)
-    set_week = "TopTermFreq_set_week";
-    set_month = "TopTermFreq_set_month";
+    per_paste = request.args.get('per_paste')
+    if per_paste == "1" or per_paste is None:
+        per_paste = "per_paste_"
+    else:
+        per_paste = ""
 
-    the_set = request.args.get('set')
+    set_day = per_paste + "TopTermFreq_set_day_" + str(today_timestamp)
+    set_week = per_paste + "TopTermFreq_set_week";
+    set_month = per_paste + "TopTermFreq_set_month";
+
+    the_set = per_paste + request.args.get('set')
     num_day = int(request.args.get('num_day'))
+
     if the_set is None:
         return "None"
     else:
         to_return = []
-        if the_set == "TopTermFreq_set_day":
+        if "TopTermFreq_set_day" in the_set:
             the_set += "_" + str(today_timestamp)
 
         for term, tot_value in r_serv_term.zrevrangebyscore(the_set, '+inf', '-inf', withscores=True, start=0, num=20):
@@ -229,7 +253,7 @@ def terms_plot_top_data():
             position['month'] = position['month']+1 if position['month'] is not None else "<20"
             value_range = []
             for timestamp in range(today_timestamp, today_timestamp - num_day*oneDay, -oneDay):
-                value = r_serv_term.hget(timestamp, term)
+                value = r_serv_term.hget(per_paste+str(timestamp), term)
                 curr_value_range = int(value) if value is not None else 0
                 value_range.append([timestamp, curr_value_range])
                 

var/www/templates/terms_management.html

@@ -12,6 +12,7 @@
   <link href="{{ url_for('static', filename='font-awesome/css/font-awesome.css') }}" rel="stylesheet">
   <link href="{{ url_for('static', filename='css/sb-admin-2.css') }}" rel="stylesheet">
   <link href="{{ url_for('static', filename='css/dataTables.bootstrap.css') }}" rel="stylesheet" type="text/css" />
+  <link href="{{ url_for('static', filename='css/switch_checkbox.css') }}" rel="stylesheet" type="text/css" />
   <script language="javascript" src="{{ url_for('static', filename='js/jquery.js')}}"></script>
   <script src="{{ url_for('static', filename='js/bootstrap.min.js') }}"></script>
   <script src="{{ url_for('static', filename='js/jquery.dataTables.min.js') }}"></script>
@@ -91,6 +92,13 @@
         <div class="col-lg-12">
             <div class="row">
                 <div class="col-lg-12">
+                   <label class="switch">
+                       <input id="per_paste" class="switch-input" value="per_paste" type="checkbox" onclick="reload_per_paste()">
+                       <span class="switch-label" data-on="On" data-off="Off"></span>
+                       <span class="switch-handle"></span>
+                   </label>
+                   <strong style="top: 3px; position: relative;">1 term per paste</strong>
+
                     <div id="panel-today" class="panel panel-success">
                         <div class="panel-heading">
                             <strong>Manage tracked terms</strong>
@@ -116,21 +124,19 @@
                                      </tr>
                                 </thead>
                                 <tbody>
-                                {% set i = 0 %}
                                 {% for term in track_list %}
                                   <tr>
                                       <td>{{ term }}</td>
-                                      <td>{{ track_list_values[i][3] }}</td>
-                                      <td>{{ track_list_values[i][0] }}</td>
-                                      <td>{{ track_list_values[i][1] }}</td>
-                                      <td>{{ track_list_values[i][2] }}</td>
-                                      <td>{{ track_list_num_of_paste[i] }}</td>
+                                      <td>{{ track_list_values[loop.index0][3] }}</td>
+                                      <td>{{ track_list_values[loop.index0][0] }}</td>
+                                      <td>{{ track_list_values[loop.index0][1] }}</td>
+                                      <td>{{ track_list_values[loop.index0][2] }}</td>
+                                      <td>{{ track_list_num_of_paste[loop.index0] }}</td>
                                       <td><p style="margin: 0px;">
                                           <span data-toggle="modal" data-target="#mymodal" data-term="{{ term }}" ><button class="btn-link" data-toggle="tooltip" data-placement="right" title="Show concerned paste(s)"><span class="glyphicon glyphicon-info-sign"></span></button></span>
                                           <button class="btn-link btn-interaction" data-toggle="tooltip" data-placement="left" title="Remove this term" data-content="{{ term }}" data-section="followTerm" data-action="delete"><span class="glyphicon glyphicon-trash"></span></button>
                                       </p></td>
                                   </tr>
-                                {% set i = i + 1 %}
                                 {% endfor %}
                                 </tbody>
                             </table>
@@ -162,22 +168,20 @@
                             <table class="table table-striped table-bordered table-hover" id="myTable2">
                                 <thead>
                                      <tr>
-                                        <th style="max-width: 800px;">Termx</th>
+                                        <th style="max-width: 800px;">Term</th>
                                         <th>Added date</th>
                                         <th>Action</th>
                                      </tr>
                                 </thead>
                                 <tbody>
-                                {% set i = 0 %}
-                                {% for term in black_list %}
+                                {% for term, date in black_list %}
                                   <tr>
-                                      <td>{{ black_list[i][0] }}</td>
-                                      <td>{{ black_list[i][1] }}</td>
+                                      <td>{{ term }}</td>
+                                      <td>{{ date }}</td>
                                       <td><p style="margin: 0px;">
-                                          <button class="btn-link btn-interaction" data-toggle="tooltip" data-placement="right" title="Remove this term" data-content="{{ black_list[i][0] }}" data-section="blacklistTerm" data-action="delete"><span class="glyphicon glyphicon-trash"></span></button>
+                                          <button class="btn-link btn-interaction" data-toggle="tooltip" data-placement="right" title="Remove this term" data-content="{{ term }}" data-section="blacklistTerm" data-action="delete"><span class="glyphicon glyphicon-trash"></span></button>
                                       </p></td>
                                   </tr>
-                                {% set i = i + 1 %}
                                 {% endfor %}
                                 </tbody>
                             </table>
@@ -198,13 +202,21 @@
 
 <!-- import graph function -->
 <script>
+    function reload_per_paste() {
+        var checked = $("#per_paste").prop( "checked" ) ? 1 : 0;
+        window.location.href = {{ url_for('terms_management') }}+"?per_paste="+checked;
+    }
+
+
     var table_track;
     var table_black;
 
     $(document).ready(function(){
         activePage = $('h1.page-header').attr('data-page');
         $("#"+activePage).addClass("active");
-
+        if({{ per_paste }} ==  1) {
+            $("#per_paste").attr('checked', true)
+        }
 
         $('[data-toggle="tooltip"]').tooltip();
 	table_track = $('#myTable').DataTable();

var/www/templates/terms_plot_tool.html

@@ -13,6 +13,7 @@
   <link href="{{ url_for('static', filename='css/sb-admin-2.css') }}" rel="stylesheet">
   <link href="{{ url_for('static', filename='css/dataTables.bootstrap.css') }}" rel="stylesheet" type="text/css" />
   <link href="{{ url_for('static', filename='css/jquery-ui.min.css') }}" rel="stylesheet" type="text/css" />
+  <link href="{{ url_for('static', filename='css/switch_checkbox.css') }}" rel="stylesheet" type="text/css" />
   <script language="javascript" src="{{ url_for('static', filename='js/jquery.js')}}"></script>
   <script src="{{ url_for('static', filename='js/bootstrap.min.js') }}"></script>
   <script src="{{ url_for('static', filename='js/jquery.dataTables.min.js') }}"></script>
@@ -81,6 +82,15 @@
                                 <button id="plot-btn-add" class="btn btn-success" style="margin-left: 6px;"><span class="fa fa-plus"></span> Add the term to the chart</button>
                             </div>
 
+                            <div class='pull_right' style="float: right;">
+                                <label class="switch">
+                                    <input id="per_paste" class="switch-input" value="per_paste" type="checkbox" checked>
+                                    <span class="switch-label" data-on="On" data-off="Off"></span>
+                                    <span class="switch-handle"></span>
+                                </label>
+                                <strong style="top: 3px; position: relative;">1 term per paste</strong>
+                            </div>
+
                         </div>
                         <!-- /.panel-body -->
                     </div>
@@ -202,8 +212,9 @@ function plotData() {
     plotted_terms = [term]
     var range_start = new Date($( ".sliderRange" ).slider( "values", 0 )).getTime() / 1000;
     var range_end = new Date($( ".sliderRange" ).slider( "values", 1 )).getTime() / 1000;
+    var checked = $("#per_paste").prop( "checked" ) ? 1 : 0;
 
-    $.getJSON("{{ url_for('terms_plot_tool_data') }}", { range_start: range_start, range_end: range_end, term: term }, function(data, status){
+    $.getJSON("{{ url_for('terms_plot_tool_data') }}", { range_start: range_start, range_end: range_end, term: term, per_paste: checked }, function(data, status){
         graph_data = [];
         var to_plot = [];
         var curr_data = [];
@@ -239,8 +250,9 @@ function addData() {
     plotted_terms.push(term)
     var range_start = new Date($( ".sliderRange" ).slider( "values", 0 )).getTime() / 1000;
     var range_end = new Date($( ".sliderRange" ).slider( "values", 1 )).getTime() / 1000;
+    var checked = $("#per_paste").prop( "checked" ) ? 1 : 0;
 
-    $.getJSON("{{ url_for('terms_plot_tool_data') }}", { range_start: range_start, range_end: range_end, term: term }, function(data, status){
+    $.getJSON("{{ url_for('terms_plot_tool_data') }}", { range_start: range_start, range_end: range_end, term: term, per_paste: checked }, function(data, status){
         var curr_data = [];
         for(i=1; i<data.length; i++) {
                 curr_data.push([data[i][0]*1000, data[i][1]]);
@@ -259,8 +271,9 @@ function replot() {
         var term = plotted_terms[i];
         var range_start = new Date($( ".sliderRange" ).slider( "values", 0 )).getTime() / 1000;
         var range_end = new Date($( ".sliderRange" ).slider( "values", 1 )).getTime() / 1000;
+        var checked = $("#per_paste").prop( "checked" ) ? 1 : 0;
 
-        promises.push($.getJSON("{{ url_for('terms_plot_tool_data') }}", { range_start: range_start, range_end: range_end, term: term }, function(data, status){
+        promises.push($.getJSON("{{ url_for('terms_plot_tool_data') }}", { range_start: range_start, range_end: range_end, term: term, per_paste: checked }, function(data, status){
             var curr_data = [];
             for(i=1; i<data.length; i++) {
                     curr_data.push([data[i][0]*1000, data[i][1]]);
@@ -279,5 +292,3 @@ function replot() {
 </script>
 
 
-</body>
-</html>

var/www/templates/terms_plot_top.html

@@ -12,6 +12,7 @@
   <link href="{{ url_for('static', filename='font-awesome/css/font-awesome.css') }}" rel="stylesheet">
   <link href="{{ url_for('static', filename='css/sb-admin-2.css') }}" rel="stylesheet">
   <link href="{{ url_for('static', filename='css/dataTables.bootstrap.css') }}" rel="stylesheet" type="text/css" />
+  <link href="{{ url_for('static', filename='css/switch_checkbox.css') }}" rel="stylesheet" type="text/css" />
   <script language="javascript" src="{{ url_for('static', filename='js/jquery.js')}}"></script>
   <script src="{{ url_for('static', filename='js/bootstrap.min.js') }}"></script>
   <script src="{{ url_for('static', filename='js/jquery.dataTables.min.js') }}"></script>
@@ -67,6 +68,14 @@
         <div class="col-lg-12">
             <div class="row">
                 <div class="col-lg-12">
+                    <div class='pull_right'>
+                        <label class="switch">
+                            <input id="per_paste" class="switch-input" value="per_paste" type="checkbox" onclick="reload_per_paste()">
+                            <span class="switch-label" data-on="On" data-off="Off"></span>
+                            <span class="switch-handle"></span>
+                        </label>
+                        <strong style="top: 3px; position: relative;">1 term per paste</strong>
+                    <div>
                     <div id="panel-today" class="panel panel-info">
                         <div class="panel-heading">
                             <strong>Today</strong>
@@ -281,7 +290,15 @@
     $(document).ready(function(){
         activePage = $('h1.page-header').attr('data-page');
         $("#"+activePage).addClass("active");
+        if({{ per_paste }} ==  1) {
+            $("#per_paste").attr('checked', true)
+        }
    });
+
+    function reload_per_paste() {
+        var checked = $("#per_paste").prop( "checked" ) ? 1 : 0;
+        window.location.href = {{ url_for('terms_plot_top') }}+"?per_paste="+checked;
+    }
 </script>
 
 
@@ -332,6 +349,7 @@ set_today = "TopTermFreq_set_day";
 set_week = "TopTermFreq_set_week";
 set_month = "TopTermFreq_set_month";
 default_num_curves = 8;
+per_paste = {{ per_paste }}
 
 var plot_today;
 var plot_week;
@@ -339,7 +357,7 @@ var plot_month;
 
 var promises = []; // Used to know when everything has been received
 
-promises.push($.getJSON("{{ url_for('terms_plot_top_data') }}", { set: set_today, num_day: 5 }, function(data, status){
+promises.push($.getJSON("{{ url_for('terms_plot_top_data') }}", { set: set_today, num_day: 5, per_paste: per_paste }, function(data, status){
     data.sort(function(a, b){return b[2]-a[2];});
     // Sort data
     var table_today = $("#table-today")
@@ -380,7 +398,7 @@ promises.push($.getJSON("{{ url_for('terms_plot_top_data') }}", { set: set_today
 
 }));
 
-promises.push($.getJSON("{{ url_for('terms_plot_top_data') }}", { set: set_week, num_day: 7 }, function(data, status){
+promises.push($.getJSON("{{ url_for('terms_plot_top_data') }}", { set: set_week, num_day: 7, per_paste: per_paste }, function(data, status){
     data.sort(function(a, b){return b[2]-a[2];});
     // Sort data
     var table = $("#table-week")
@@ -420,7 +438,7 @@ promises.push($.getJSON("{{ url_for('terms_plot_top_data') }}", { set: set_week,
     });
 }));
 
-promises.push($.getJSON("{{ url_for('terms_plot_top_data') }}", { set: set_month, num_day: 31 }, function(data, status){
+promises.push($.getJSON("{{ url_for('terms_plot_top_data') }}", { set: set_month, num_day: 31, per_paste: per_paste }, function(data, status){
     data.sort(function(a, b){return b[2]-a[2];});
     // Sort data
     var table = $("#table-month")