Merge branch 'website-improvement' into production

bin/CurveManageTopSets.py

@@ -45,13 +45,14 @@ def manage_top_set():
     startDate = datetime.datetime.now()
     startDate = startDate.replace(hour=0, minute=0, second=0, microsecond=0)
     startDate = calendar.timegm(startDate.timetuple())
+    blacklist_size = int(server_term.scard(BlackListTermsSet_Name))
 
     dico = {}
 
-    # Retreive top data (2*max_card) from days sets
+    # 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*2)
+        array_top_day = server_term.zrevrangebyscore(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):

bin/WebStats.py

@@ -38,35 +38,55 @@ def get_date_range(num_day):
         date_list.append(date.substract_day(i))
     return date_list
 
+# Compute the progression for one keyword
+def compute_progression_word(keyword):
+    date_range = get_date_range(num_day)
+    # check if this keyword is eligible for progression
+    keyword_total_sum = 0
+    value_list = []
+    for date in date_range: # get value up to date_range
+        curr_value = server.hget(keyword, date)
+        value_list.append(int(curr_value if curr_value is not None else 0))
+        keyword_total_sum += int(curr_value) if curr_value is not None else 0
+    oldest_value = value_list[-1] if value_list[-1] != 0 else 1 #Avoid zero division
+
+    # The progression is based on the ratio: value[i] / value[i-1]
+    keyword_increase = 0
+    value_list_reversed = value_list[:]
+    value_list_reversed.reverse()
+    for i in range(1, len(value_list_reversed)):
+        divisor = value_list_reversed[i-1] if value_list_reversed[i-1] != 0 else 1
+        keyword_increase += value_list_reversed[i] / divisor
+
+    return (keyword_increase, keyword_total_sum)
+
+
+'''
+    recompute the set top_progression zset
+        - Compute the current field progression
+        - re-compute the current progression for each first 2*max_set_cardinality fields in the top_progression_zset
+'''
 def compute_progression(server, field_name, num_day, url_parsed):
-    redis_progression_name = 'top_progression_'+field_name
+    redis_progression_name_set = "z_top_progression_"+field_name
-    redis_progression_name_set = 'top_progression_'+field_name+'_set'
 
     keyword = url_parsed[field_name]
     if keyword is not None:
-        date_range = get_date_range(num_day) 
 
-        # check if this keyword is eligible for progression
+        #compute the progression of the current word
-        keyword_total_sum = 0 
+        keyword_increase, keyword_total_sum = compute_progression_word(keyword)
-        value_list = []
-        for date in date_range: # get value up to date_range
-            curr_value = server.hget(keyword, date)
-            value_list.append(int(curr_value if curr_value is not None else 0))
-            keyword_total_sum += int(curr_value) if curr_value is not None else 0
-        oldest_value = value_list[-1] if value_list[-1] != 0 else 1 #Avoid zero division
 
-        # The progression is based on the ratio: value[i] / value[i-1]
+        #re-compute the progression of 2*max_set_cardinality
-        keyword_increase = 0
+        current_top = server.zrevrangebyscore(redis_progression_name_set, '+inf', '-inf', withscores=True, start=0, num=2*max_set_cardinality)
-        value_list_reversed = value_list[:]
+        for word, value in array_top_day:
-        value_list_reversed.reverse()
+            word_inc, word_tot_sum = compute_progression_word(word)
-        for i in range(1, len(value_list_reversed)):
+            server.zrem(redis_progression_name_set, word)
-            divisor = value_list_reversed[i-1] if value_list_reversed[i-1] != 0 else 1
+            if (word_tot_sum > threshold_total_sum) and (word_inc > threshold_increase):
-            keyword_increase += value_list_reversed[i] / divisor
+                server.zadd(redis_progression_name_set, float(word_inc), word)
 
-        # filter
+        # filter before adding
         if (keyword_total_sum > threshold_total_sum) and (keyword_increase > threshold_increase):
-            
+            server.zadd(redis_progression_name_set, float(keyword_increase), keyword)
-            server.zadd("z_top_progression_"+field_name, float(keyword_increase), keyword)
+
 
 
 if __name__ == '__main__':

var/www/Flask_server.py

@@ -772,6 +772,7 @@ def terms_plot_tool_data():
             value = r_serv_term.hget(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)
         return jsonify(value_range)
 
 

var/www/templates/terms_plot_tool.html

@@ -72,7 +72,7 @@
                         </div>
                         <div class="panel-body">
                             <div aria-disabled="false" class="slider sliderRange sliderBlue ui-slider ui-slider-horizontal ui-widget ui-widget-content ui-corner-all" style="margin-bottom: 5px;"></div>
-                            <strong>Date:</strong> <input type="text" id="amount" readonly style="border:0; color:#f6931f; font-weight:bold;">
+                            <strong>Date:</strong> <input type="text" id="amount" readonly style="border:0; color:#f6931f; font-weight:bold;"> <button id="plot-month" onclick="replot();" class="btn btn-info" style="float: right;">Replot</button>
 
                             <div class="form-group input-group" style="margin-top: 30px;">
                                 <span class="input-group-addon"><span class="glyphicon glyphicon-stats"></span></span>
@@ -168,6 +168,7 @@
 <script>
 var plot;
 var graph_data = [];
+var plotted_terms = [];
 var graph_options = {
                         series: {
                             lines: { 
@@ -198,6 +199,7 @@ function plotData() {
     $('#plot-btn-add').show("fast");
     var curthis = $(this);
     var term = $('#TermInput').val();
+    plotted_terms = [term]
     var range_start = new Date($( ".sliderRange" ).slider( "values", 0 )).getTime() / 1000;
     var range_end = new Date($( ".sliderRange" ).slider( "values", 1 )).getTime() / 1000;
 
@@ -205,8 +207,8 @@ function plotData() {
         graph_data = [];
         var to_plot = [];
         var curr_data = [];
-        for(i=0; i<data.length; i++) {
+        for(i=1; i<data.length; i++) {
-                curr_data.push([data[i][0]*1000, data[i][1]]);
+            curr_data.push([data[i][0]*1000, data[i][1]]);
         }
         to_plot.push({ data: curr_data, label: term}); 
         graph_data.push({ data: curr_data, label: term});
@@ -234,22 +236,46 @@ function plotData() {
 function addData() {
     var curthis = $(this);
     var term = $('#TermInput').val();
+    plotted_terms.push(term)
     var range_start = new Date($( ".sliderRange" ).slider( "values", 0 )).getTime() / 1000;
     var range_end = new Date($( ".sliderRange" ).slider( "values", 1 )).getTime() / 1000;
 
     $.getJSON("{{ url_for('terms_plot_tool_data') }}", { range_start: range_start, range_end: range_end, term: term }, function(data, status){
-        var to_plot = [];
         var curr_data = [];
-        for(i=0; i<data.length; i++) {
+        for(i=1; i<data.length; i++) {
                 curr_data.push([data[i][0]*1000, data[i][1]]);
         }
-        to_plot.push({ data: curr_data, label: term}); 
         graph_data.push({ data: curr_data, label: term});
         plot = $.plot($("#graph"), graph_data, graph_options);
         $("#TermInput").val("");
     })
 }
 
+
+function replot() {
+    graph_data = [];
+    promises = [];
+    for(i=0; i<plotted_terms.length; i++) {
+        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;
+
+        promises.push($.getJSON("{{ url_for('terms_plot_tool_data') }}", { range_start: range_start, range_end: range_end, term: term }, function(data, status){
+            var curr_data = [];
+            for(i=1; i<data.length; i++) {
+                    curr_data.push([data[i][0]*1000, data[i][1]]);
+            }
+            graph_data.push({ data: curr_data, label: data[0]});
+            $("#TermInput").val("");
+        }))
+    }
+    
+    $.when.apply($, promises).done( function () {
+        plot = $.plot($("#graph"), graph_data, graph_options);
+    });
+    
+}
+
 </script>