Handling duplicate urls in ga_url processing
[ckanext-ga-report.git] / ckanext / ga_report / controller.py
blob:a/ckanext/ga_report/controller.py -> blob:b/ckanext/ga_report/controller.py
--- a/ckanext/ga_report/controller.py
+++ b/ckanext/ga_report/controller.py
@@ -1,6 +1,7 @@
 import re
 import csv
 import sys
+import json
 import logging
 import operator
 import collections
@@ -13,6 +14,7 @@
 
 log = logging.getLogger('ckanext.ga-report')
 
+DOWNLOADS_AVAILABLE_FROM = '2012-12'
 
 def _get_month_name(strdate):
     import calendar
@@ -20,14 +22,39 @@
     d = strptime(strdate, '%Y-%m')
     return '%s %s' % (calendar.month_name[d.tm_mon], d.tm_year)
 
-
-def _month_details(cls):
-    '''Returns a list of all the month names'''
+def _get_unix_epoch(strdate):
+    from time import strptime,mktime
+    d = strptime(strdate, '%Y-%m')
+    return int(mktime(d))
+
+def _month_details(cls, stat_key=None):
+    '''
+    Returns a list of all the periods for which we have data, unfortunately
+    knows too much about the type of the cls being passed as GA_Url has a
+    more complex query
+
+    This may need extending if we add a period_name to the stats
+    '''
     months = []
-    vals = model.Session.query(cls.period_name).filter(cls.period_name!='All').distinct().all()
+    day = None
+
+    q = model.Session.query(cls.period_name,cls.period_complete_day)\
+        .filter(cls.period_name!='All').distinct(cls.period_name)
+    if stat_key:
+        q=  q.filter(cls.stat_name==stat_key)
+
+    vals = q.order_by("period_name desc").all()
+
+    if vals and vals[0][1]:
+        day = int(vals[0][1])
+        ordinal = 'th' if 11 <= day <= 13 \
+            else {1:'st',2:'nd',3:'rd'}.get(day % 10, 'th')
+        day = "{day}{ordinal}".format(day=day, ordinal=ordinal)
+
     for m in vals:
         months.append( (m[0], _get_month_name(m[0])))
-    return sorted(months, key=operator.itemgetter(0), reverse=True)
+
+    return months, day
 
 
 class GaReport(BaseController):
@@ -35,7 +62,7 @@
     def csv(self, month):
         import csv
 
-        q = model.Session.query(GA_Stat)
+        q = model.Session.query(GA_Stat).filter(GA_Stat.stat_name!='Downloads')
         if month != 'all':
             q = q.filter(GA_Stat.period_name==month)
         entries = q.order_by('GA_Stat.period_name, GA_Stat.stat_name, GA_Stat.key').all()
@@ -52,11 +79,12 @@
                              entry.key.encode('utf-8'),
                              entry.value.encode('utf-8')])
 
+
     def index(self):
 
         # Get the month details by fetching distinct values and determining the
         # month names from the values.
-        c.months = _month_details(GA_Stat)
+        c.months, c.day = _month_details(GA_Stat)
 
         # Work out which month to show, based on query params of the first item
         c.month_desc = 'all months'
@@ -71,24 +99,39 @@
         entries = q.order_by('ga_stat.key').all()
 
         def clean_key(key, val):
-            if key in ['Average time on site', 'Pages per visit', 'New visits', 'Bounces']:
+            if key in ['Average time on site', 'Pages per visit', 'New visits', 'Bounce rate (home page)']:
                 val =  "%.2f" % round(float(val), 2)
                 if key == 'Average time on site':
                     mins, secs = divmod(float(val), 60)
                     hours, mins = divmod(mins, 60)
                     val = '%02d:%02d:%02d (%s seconds) ' % (hours, mins, secs, val)
-                if key in ['New visits','Bounces']:
+                if key in ['New visits','Bounce rate (home page)']:
                     val = "%s%%" % val
             if key in ['Total page views', 'Total visits']:
                 val = int(val)
 
             return key, val
+
+        # Query historic values for sparkline rendering
+        sparkline_query = model.Session.query(GA_Stat)\
+                .filter(GA_Stat.stat_name=='Totals')\
+                .order_by(GA_Stat.period_name)
+        sparkline_data = {}
+        for x in sparkline_query:
+            sparkline_data[x.key] = sparkline_data.get(x.key,[])
+            key, val = clean_key(x.key,float(x.value))
+            tooltip = '%s: %s' % (_get_month_name(x.period_name), val)
+            sparkline_data[x.key].append( (tooltip,x.value) )
+        # Trim the latest month, as it looks like a huge dropoff
+        for key in sparkline_data:
+            sparkline_data[key] = sparkline_data[key][:-1]
 
         c.global_totals = []
         if c.month:
             for e in entries:
                 key, val = clean_key(e.key, e.value)
-                c.global_totals.append((key, val))
+                sparkline = sparkline_data[e.key]
+                c.global_totals.append((key, val, sparkline))
         else:
             d = collections.defaultdict(list)
             for e in entries:
@@ -97,11 +140,19 @@
                 if k in ['Total page views', 'Total visits']:
                     v = sum(v)
                 else:
-                    v = float(sum(v))/len(v)
+                    v = float(sum(v))/float(len(v))
+                sparkline = sparkline_data[k]
                 key, val = clean_key(k,v)
 
-                c.global_totals.append((key, val))
-                c.global_totals = sorted(c.global_totals, key=operator.itemgetter(0))
+                c.global_totals.append((key, val, sparkline))
+        # Sort the global totals into a more pleasant order
+        def sort_func(x):
+            key = x[0]
+            total_order = ['Total page views','Total visits','Pages per visit']
+            if key in total_order:
+                return total_order.index(key)
+            return 999
+        c.global_totals = sorted(c.global_totals, key=sort_func)
 
         keys = {
             'Browser versions': 'browser_versions',
@@ -138,7 +189,29 @@
 
         for k, v in keys.iteritems():
             q = model.Session.query(GA_Stat).\
-                filter(GA_Stat.stat_name==k)
+                filter(GA_Stat.stat_name==k).\
+                order_by(GA_Stat.period_name)
+            # Run the query on all months to gather graph data
+            series = {}
+            x_axis = set()
+            for stat in q:
+                x_val = _get_unix_epoch(stat.period_name)
+                series[ stat.key ] = series.get(stat.key,{})
+                series[ stat.key ][x_val] = float(stat.value)
+                x_axis.add(x_val)
+            # Common x-axis for all series. Exclude this month (incomplete data)
+            x_axis = sorted(list(x_axis))[:-1]
+            # Buffer a rickshaw dataset from the series
+            def create_graph(series_name, series_data):
+                return { 
+                    'name':series_name, 
+                    'data':[ {'x':x,'y':series_data.get(x,0)} for x in x_axis ]
+                    }
+            rickshaw = [ create_graph(name,data) for name, data in series.items() ]
+            rickshaw = sorted(rickshaw,key=lambda x:x['data'][-1]['y'])
+            setattr(c, v+'_graph', json.dumps(rickshaw))
+
+            # Buffer the tabular data
             if c.month:
                 entries = []
                 q = q.filter(GA_Stat.period_name==c.month).\
@@ -155,7 +228,7 @@
             # Get the total for each set of values and then set the value as
             # a percentage of the total
             if k == 'Social sources':
-                total = sum([x for n,x in c.global_totals if n == 'Total visits'])
+                total = sum([x for n,x,graph in c.global_totals if n == 'Total visits'])
             else:
                 total = sum([num for _,num in entries])
             setattr(c, v, [(k,_percent(v,total)) for k,v in entries ])
@@ -206,13 +279,14 @@
             str('attachment; filename=datasets_%s_%s.csv' % (c.publisher_name, month,))
 
         writer = csv.writer(response)
-        writer.writerow(["Dataset Title", "Dataset Name", "Views", "Visits", "Period Name"])
-
-        for package,view,visit in packages:
+        writer.writerow(["Dataset Title", "Dataset Name", "Views", "Visits", "Resource downloads", "Period Name"])
+
+        for package,view,visit,downloads in packages:
             writer.writerow([package.title.encode('utf-8'),
                              package.name.encode('utf-8'),
                              view,
                              visit,
+                             downloads,
                              month])
 
     def publishers(self):
@@ -220,7 +294,7 @@
 
         # Get the month details by fetching distinct values and determining the
         # month names from the values.
-        c.months = _month_details(GA_Url)
+        c.months, c.day = _month_details(GA_Url)
 
         # Work out which month to show, based on query params of the first item
         c.month = request.params.get('month', '')
@@ -232,11 +306,11 @@
         return render('ga_report/publisher/index.html')
 
     def _get_packages(self, publisher=None, count=-1):
-        '''Returns the datasets in order of visits'''
-        if count == -1:
-            count = sys.maxint
-
+        '''Returns the datasets in order of views'''
+        have_download_data = True
         month = c.month or 'All'
+        if month != 'All':
+            have_download_data = month >= DOWNLOADS_AVAILABLE_FROM
 
         q = model.Session.query(GA_Url,model.Package)\
             .filter(model.Package.name==GA_Url.package_id)\
@@ -244,11 +318,28 @@
         if publisher:
             q = q.filter(GA_Url.department_id==publisher.name)
         q = q.filter(GA_Url.period_name==month)
-        q = q.order_by('ga_url.visitors::int desc')
+        q = q.order_by('ga_url.pageviews::int desc')
         top_packages = []
-        for entry,package in q.limit(count):
+        if count == -1:
+            entries = q.all()
+        else:
+            entries = q.limit(count)
+
+        for entry,package in entries:
             if package:
-                top_packages.append((package, entry.pageviews, entry.visitors))
+                # Downloads ....
+                if have_download_data:
+                    dls = model.Session.query(GA_Stat).\
+                        filter(GA_Stat.stat_name=='Downloads').\
+                        filter(GA_Stat.key==package.name)
+                    if month != 'All':  # Fetch everything unless the month is specific
+                        dls = dls.filter(GA_Stat.period_name==month)
+                    downloads = 0
+                    for x in dls:
+                        downloads += int(x.value)
+                else:
+                    downloads = 'No data'
+                top_packages.append((package, entry.pageviews, entry.visits, downloads))
             else:
                 log.warning('Could not find package associated package')
 
@@ -278,7 +369,7 @@
 
         # Get the month details by fetching distinct values and determining the
         # month names from the values.
-        c.months = _month_details(GA_Url)
+        c.months, c.day = _month_details(GA_Url)
 
         # Work out which month to show, based on query params of the first item
         c.month = request.params.get('month', '')
@@ -296,6 +387,26 @@
 
         c.top_packages = self._get_packages(c.publisher, 20)
 
+        # Graph query
+        top_package_names = [ x[0].name for x in c.top_packages ]
+        graph_query = model.Session.query(GA_Url,model.Package)\
+            .filter(model.Package.name==GA_Url.package_id)\
+            .filter(GA_Url.url.like('/dataset/%'))\
+            .filter(GA_Url.package_id.in_(top_package_names))
+        graph_data = {}
+        for entry,package in graph_query:
+            if not package: continue
+            if entry.period_name=='All': continue
+            graph_data[package.id] = graph_data.get(package.id,{
+                'name':package.title,
+                'data':[]
+                })
+            graph_data[package.id]['data'].append({
+                'x':_get_unix_epoch(entry.period_name),
+                'y':int(entry.pageviews),
+                })
+        c.graph_data = json.dumps(graph_data.values())
+
         return render('ga_report/publisher/read.html')
 
 def _get_top_publishers(limit=20):
@@ -306,13 +417,13 @@
     month = c.month or 'All'
     connection = model.Session.connection()
     q = """
-        select department_id, sum(pageviews::int) views, sum(visitors::int) visits
+        select department_id, sum(pageviews::int) views, sum(visits::int) visits
         from ga_url
         where department_id <> ''
           and package_id <> ''
           and url like '/dataset/%%'
           and period_name=%s
-        group by department_id order by visits desc
+        group by department_id order by views desc
         """
     if limit:
         q = q + " limit %s;" % (limit)
@@ -329,7 +440,7 @@
 def _get_publishers():
     '''
     Returns a list of all publishers. Each item is a tuple:
-      (names, title)
+      (name, title)
     '''
     publishers = []
     for pub in model.Session.query(model.Group).\