Merge branch 'master' of github.com:datagovuk/ckanext-ga-report
Merge branch 'master' of github.com:datagovuk/ckanext-ga-report

file:a/.gitignore -> file:b/.gitignore
--- a/.gitignore
+++ b/.gitignore
@@ -1,6 +1,7 @@
 *.py[co]
 *.py~
 .gitignore
+ckan.log
 
 # Packages
 *.egg

--- a/ckanext/ga_report/command.py
+++ b/ckanext/ga_report/command.py
@@ -23,7 +23,7 @@
         import ckan.model as model
         model.Session.remove()
         model.Session.configure(bind=model.meta.engine)
-        log = logging.getLogger('ckanext.ga-report')
+        log = logging.getLogger('ckanext.ga_report')
 
         import ga_model
         ga_model.init_tables()
@@ -55,6 +55,36 @@
         init_service('token.dat',
                       self.args[0] if self.args
                                    else 'credentials.json')
+
+class FixTimePeriods(CkanCommand):
+    """
+    Fixes the 'All' records for GA_Urls
+
+    It is possible that older urls that haven't recently been visited
+    do not have All records.  This command will traverse through those
+    records and generate valid All records for them.
+    """
+    summary = __doc__.split('\n')[0]
+    usage = __doc__
+    max_args = 0
+    min_args = 0
+
+    def __init__(self, name):
+        super(FixTimePeriods, self).__init__(name)
+
+    def command(self):
+        import ckan.model as model
+        from ga_model import post_update_url_stats
+        self._load_config()
+        model.Session.remove()
+        model.Session.configure(bind=model.meta.engine)
+
+        log = logging.getLogger('ckanext.ga_report')
+
+        log.info("Updating 'All' records for old URLs")
+        post_update_url_stats()
+        log.info("Processing complete")
+
 
 
 class LoadAnalytics(CkanCommand):

--- 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
@@ -20,8 +22,12 @@
     d = strptime(strdate, '%Y-%m')
     return '%s %s' % (calendar.month_name[d.tm_mon], d.tm_year)
 
-
-def _month_details(cls):
+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
@@ -32,9 +38,13 @@
     months = []
     day = None
 
-    vals = model.Session.query(cls.period_name,cls.period_complete_day)\
-        .filter(cls.period_name!='All').distinct(cls.period_name)\
-        .order_by("period_name desc").all()
+    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 \
@@ -52,7 +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()
@@ -68,6 +78,7 @@
                              entry.stat_name.encode('utf-8'),
                              entry.key.encode('utf-8'),
                              entry.value.encode('utf-8')])
+
 
     def index(self):
 
@@ -101,11 +112,26 @@
 
             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:
@@ -114,11 +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',
@@ -155,12 +189,13 @@
 
         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)
+            # Buffer the tabular data
             if c.month:
                 entries = []
                 q = q.filter(GA_Stat.period_name==c.month).\
                           order_by('ga_stat.value::int desc')
-
             d = collections.defaultdict(int)
             for e in q.all():
                 d[e.key] += int(e.value)
@@ -169,10 +204,27 @@
                 entries.append((key,val,))
             entries = sorted(entries, key=operator.itemgetter(1), reverse=True)
 
+            # Run a query on all months to gather graph data
+            graph_query = model.Session.query(GA_Stat).\
+                filter(GA_Stat.stat_name==k).\
+                order_by(GA_Stat.period_name)
+            graph_dict = {}
+            for stat in graph_query:
+                graph_dict[ stat.key ] = graph_dict.get(stat.key,{
+                    'name':stat.key, 
+                    'raw': {}
+                    })
+                graph_dict[ stat.key ]['raw'][stat.period_name] = float(stat.value)
+            stats_in_table = [x[0] for x in entries]
+            stats_not_in_table = set(graph_dict.keys()) - set(stats_in_table)
+            stats = stats_in_table + sorted(list(stats_not_in_table))
+            graph = [graph_dict[x] for x in stats]
+            setattr(c, v+'_graph', json.dumps( _to_rickshaw(graph,percentageMode=True) ))
+
             # 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 ])
@@ -197,7 +249,9 @@
         writer = csv.writer(response)
         writer.writerow(["Publisher Title", "Publisher Name", "Views", "Visits", "Period Name"])
 
-        for publisher,view,visit in _get_top_publishers(None):
+        top_publishers = _get_top_publishers(limit=None)
+
+        for publisher,view,visit in top_publishers:
             writer.writerow([publisher.title.encode('utf-8'),
                              publisher.name.encode('utf-8'),
                              view,
@@ -217,19 +271,20 @@
             if not c.publisher:
                 abort(404, 'A publisher with that name could not be found')
 
-        packages = self._get_packages(c.publisher)
+        packages = self._get_packages(publisher=c.publisher, month=c.month)
         response.headers['Content-Type'] = "text/csv; charset=utf-8"
         response.headers['Content-Disposition'] = \
             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):
@@ -246,14 +301,17 @@
             c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month])
 
         c.top_publishers = _get_top_publishers()
+        graph_data = _get_top_publishers_graph()
+        c.top_publishers_graph = json.dumps( _to_rickshaw(graph_data) )
+
         return render('ga_report/publisher/index.html')
 
-    def _get_packages(self, publisher=None, count=-1):
+    def _get_packages(self, publisher=None, month='', count=-1):
         '''Returns the datasets in order of views'''
-        if count == -1:
-            count = sys.maxint
-
-        month = c.month or 'All'
+        have_download_data = True
+        month = 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)\
@@ -263,9 +321,26 @@
         q = q.filter(GA_Url.period_name==month)
         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.visits))
+                # 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')
 
@@ -311,9 +386,73 @@
         entry = q.filter(GA_Url.period_name==c.month).first()
         c.publisher_page_views = entry.pageviews if entry else 0
 
-        c.top_packages = self._get_packages(c.publisher, 20)
+        c.top_packages = self._get_packages(publisher=c.publisher, count=20, month=c.month)
+
+        # Graph query
+        top_packages_all_time = self._get_packages(publisher=c.publisher, count=20, month='All')
+        top_package_names = [ x[0].name for x in top_packages_all_time ]
+        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))
+        all_series = {}
+        for entry,package in graph_query:
+            if not package: continue
+            if entry.period_name=='All': continue
+            all_series[package.name] = all_series.get(package.name,{
+                'name':package.title,
+                'raw': {}
+                })
+            all_series[package.name]['raw'][entry.period_name] = int(entry.pageviews)
+        graph = [ all_series[series_name] for series_name in top_package_names ]
+        c.graph_data = json.dumps( _to_rickshaw(graph) )
 
         return render('ga_report/publisher/read.html')
+
+def _to_rickshaw(data, percentageMode=False):
+    if data==[]:
+        return data
+    # x-axis is every month in c.months. Note that data might not exist 
+    # for entire history, eg. for recently-added datasets
+    x_axis = [x[0] for x in c.months]
+    x_axis.reverse() # Ascending order
+    x_axis = x_axis[:-1] # Remove latest month
+    totals = {}
+    for series in data:
+        series['data'] = []
+        for x_string in x_axis:
+            x = _get_unix_epoch( x_string )
+            y = series['raw'].get(x_string,0)
+            series['data'].append({'x':x,'y':y})
+            totals[x] = totals.get(x,0)+y
+    if not percentageMode:
+        return data
+    # Turn all data into percentages
+    # Roll insignificant series into a catch-all
+    THRESHOLD = 1
+    raw_data = data
+    data = []
+    for series in raw_data:
+        for point in series['data']:
+            percentage = (100*float(point['y'])) / totals[point['x']]
+            if not (series in data) and percentage>THRESHOLD:
+                data.append(series)
+            point['y'] = percentage
+    others = [ x for x in raw_data if not (x in data) ]
+    if len(others):
+        data_other = []
+        for i in range(len(x_axis)):
+            x = _get_unix_epoch(x_axis[i])
+            y = 0
+            for series in others: 
+                y += series['data'][i]['y']
+            data_other.append({'x':x,'y':y})
+        data.append({ 
+            'name':'Other',
+            'data': data_other
+            })
+    return data
+
 
 def _get_top_publishers(limit=20):
     '''
@@ -343,6 +482,46 @@
     return top_publishers
 
 
+def _get_top_publishers_graph(limit=20):
+    '''
+    Returns a list of the top 20 publishers by dataset visits.
+    (The number to show can be varied with 'limit')
+    '''
+    connection = model.Session.connection()
+    q = """
+        select department_id, sum(pageviews::int) views
+        from ga_url
+        where department_id <> ''
+          and package_id <> ''
+          and url like '/dataset/%%'
+          and period_name='All'
+        group by department_id order by views desc
+        """
+    if limit:
+        q = q + " limit %s;" % (limit)
+
+    res = connection.execute(q)
+    department_ids = [ row[0] for row in res ]
+
+    # Query for a history graph of these department ids
+    q = model.Session.query(
+            GA_Url.department_id, 
+            GA_Url.period_name, 
+            func.sum(cast(GA_Url.pageviews,sqlalchemy.types.INT)))\
+        .filter( GA_Url.department_id.in_(department_ids) )\
+        .filter( GA_Url.url.like('/dataset/%') )\
+        .filter( GA_Url.package_id!='' )\
+        .group_by( GA_Url.department_id, GA_Url.period_name )
+    graph_dict = {}
+    for dept_id,period_name,views in q:
+        graph_dict[dept_id] = graph_dict.get( dept_id, {
+            'name' : model.Group.get(dept_id).title,
+            'raw' : {}
+            })
+        graph_dict[dept_id]['raw'][period_name] = views
+    return [ graph_dict[id] for id in department_ids ]
+
+
 def _get_publishers():
     '''
     Returns a list of all publishers. Each item is a tuple:

--- a/ckanext/ga_report/download_analytics.py
+++ b/ckanext/ga_report/download_analytics.py
@@ -1,6 +1,7 @@
 import os
 import logging
 import datetime
+import httplib
 import collections
 from pylons import config
 from ga_model import _normalize_url
@@ -13,6 +14,7 @@
 FORMAT_MONTH = '%Y-%m'
 MIN_VIEWS = 50
 MIN_VISITS = 20
+MIN_DOWNLOADS = 10
 
 class DownloadAnalytics(object):
     '''Downloads and stores analytics info'''
@@ -31,6 +33,11 @@
         first_of_this_month = datetime.datetime(date.year, date.month, 1)
         _, last_day_of_month = calendar.monthrange(int(date.year), int(date.month))
         last_of_this_month =  datetime.datetime(date.year, date.month, last_day_of_month)
+        # if this is the latest month, note that it is only up until today
+        now = datetime.datetime.now()
+        if now.year == date.year and now.month == date.month:
+            last_day_of_month = now.day
+            last_of_this_month = now
         periods = ((date.strftime(FORMAT_MONTH),
                     last_day_of_month,
                     first_of_this_month, last_of_this_month),)
@@ -122,8 +129,12 @@
                 log.info('Storing publisher views (%i rows)', len(data.get('url')))
                 self.store(period_name, period_complete_day, data,)
 
-                log.info('Aggregating datasets by publisher')
+                # Make sure the All records are correct.
+                ga_model.post_update_url_stats()
+
+                log.info('Associating datasets with their publisher')
                 ga_model.update_publisher_stats(period_name) # about 30 seconds.
+
 
             log.info('Downloading and storing analytics for site-wide stats')
             self.sitewide_stats( period_name, period_complete_day )
@@ -168,17 +179,23 @@
 
         # Supported query params at
         # https://developers.google.com/analytics/devguides/reporting/core/v3/reference
-        results = self.service.data().ga().get(
-                                 ids='ga:' + self.profile_id,
-                                 filters=query,
-                                 start_date=start_date,
-                                 metrics=metrics,
-                                 sort=sort,
-                                 dimensions="ga:pagePath",
-                                 max_results=10000,
-                                 end_date=end_date).execute()
+        try:
+            results = self.service.data().ga().get(
+                                     ids='ga:' + self.profile_id,
+                                     filters=query,
+                                     start_date=start_date,
+                                     metrics=metrics,
+                                     sort=sort,
+                                     dimensions="ga:pagePath",
+                                     max_results=10000,
+                                     end_date=end_date).execute()
+        except httplib.BadStatusLine:
+            log.error(u"Failed to download data=> ids: ga:{0}, filters: {1}, start_date: {2}, end_date: {3}, metrics: {4}, sort: {5}, dimensions: ga:pagePath".format(
+                self.profile_id, query, start_date, end_date, metrics, sort ))
+            return dict(url=[])
 
         packages = []
+        log.info("There are %d results" % results['totalResults'])
         for entry in results.get('rows'):
             (loc,pageviews,visits) = entry
             url = _normalize_url('http:/' + loc) # strips off domain e.g. www.data.gov.uk or data.gov.uk
@@ -203,7 +220,7 @@
         start_date = '%s-01' % period_name
         end_date = '%s-%s' % (period_name, last_day_of_month)
         funcs = ['_totals_stats', '_social_stats', '_os_stats',
-                 '_locale_stats', '_browser_stats', '_mobile_stats']
+                 '_locale_stats', '_browser_stats', '_mobile_stats', '_download_stats']
         for f in funcs:
             log.info('Downloading analytics for %s' % f.split('_')[1])
             getattr(self, f)(start_date, end_date, period_name, period_complete_day)
@@ -250,7 +267,7 @@
                                  ids='ga:' + self.profile_id,
                                  filters='ga:pagePath==%s' % (path,),
                                  start_date=start_date,
-                                 metrics='ga:bounces,ga:pageviews',
+                                 metrics='ga:visitBounceRate',
                                  dimensions='ga:pagePath',
                                  max_results=10000,
                                  end_date=end_date).execute()
@@ -260,10 +277,10 @@
                       path, result_data)
             return
         results = result_data[0]
-        bounces, total = [float(x) for x in result_data[0][1:]]
-        pct = 100 * bounces/total
-        log.info('%d bounces from %d total == %s', bounces, total, pct)
-        ga_model.update_sitewide_stats(period_name, "Totals", {'Bounce rate (home page)': pct},
+        bounces = float(results[1])
+        # visitBounceRate is already a %
+        log.info('Google reports visitBounceRate as %s', bounces)
+        ga_model.update_sitewide_stats(period_name, "Totals", {'Bounce rate (home page)': float(bounces)},
             period_complete_day)
 
 
@@ -290,6 +307,74 @@
         self._filter_out_long_tail(data, MIN_VIEWS)
         ga_model.update_sitewide_stats(period_name, "Country", data, period_complete_day)
 
+
+    def _download_stats(self, start_date, end_date, period_name, period_complete_day):
+        """ Fetches stats about data downloads """
+        import ckan.model as model
+
+        data = {}
+
+        results = self.service.data().ga().get(
+                                 ids='ga:' + self.profile_id,
+                                 start_date=start_date,
+                                 filters='ga:eventAction==download',
+                                 metrics='ga:totalEvents',
+                                 sort='-ga:totalEvents',
+                                 dimensions="ga:eventLabel",
+                                 max_results=10000,
+                                 end_date=end_date).execute()
+        result_data = results.get('rows')
+        if not result_data:
+            # We may not have data for this time period, so we need to bail
+            # early.
+            log.info("There is no download data for this time period")
+            return
+
+        def process_result_data(result_data, cached=False):
+            progress_total = len(result_data)
+            progress_count = 0
+            resources_not_matched = []
+            for result in result_data:
+                progress_count += 1
+                if progress_count % 100 == 0:
+                    log.debug('.. %d/%d done so far', progress_count, progress_total)
+
+                url = result[0].strip()
+
+                # Get package id associated with the resource that has this URL.
+                q = model.Session.query(model.Resource)
+                if cached:
+                    r = q.filter(model.Resource.cache_url.like("%s%%" % url)).first()
+                else:
+                    r = q.filter(model.Resource.url.like("%s%%" % url)).first()
+
+                package_name = r.resource_group.package.name if r else ""
+                if package_name:
+                    data[package_name] = data.get(package_name, 0) + int(result[1])
+                else:
+                    resources_not_matched.append(url)
+                    continue
+            if resources_not_matched:
+                log.debug('Could not match %i or %i resource URLs to datasets. e.g. %r',
+                          len(resources_not_matched), progress_total, resources_not_matched[:3])
+
+        log.info('Associating downloads of resource URLs with their respective datasets')
+        process_result_data(results.get('rows'))
+
+        results = self.service.data().ga().get(
+                                 ids='ga:' + self.profile_id,
+                                 start_date=start_date,
+                                 filters='ga:eventAction==download-cache',
+                                 metrics='ga:totalEvents',
+                                 sort='-ga:totalEvents',
+                                 dimensions="ga:eventLabel",
+                                 max_results=10000,
+                                 end_date=end_date).execute()
+        log.info('Associating downloads of cache resource URLs with their respective datasets')
+        process_result_data(results.get('rows'), cached=False)
+
+        self._filter_out_long_tail(data, MIN_DOWNLOADS)
+        ga_model.update_sitewide_stats(period_name, "Downloads", data, period_complete_day)
 
     def _social_stats(self, start_date, end_date, period_name, period_complete_day):
         """ Finds out which social sites people are referred from """

--- a/ckanext/ga_report/ga_model.py
+++ b/ckanext/ga_report/ga_model.py
@@ -161,13 +161,64 @@
 
 
 def pre_update_url_stats(period_name):
-    log.debug("Deleting '%s' records" % period_name)
-    model.Session.query(GA_Url).\
-            filter(GA_Url.period_name==period_name).delete()
-    log.debug("Deleting 'All' records")
-    model.Session.query(GA_Url).\
-            filter(GA_Url.period_name == 'All').delete()
+    q = model.Session.query(GA_Url).\
+        filter(GA_Url.period_name==period_name)
+    log.debug("Deleting %d '%s' records" % (q.count(), period_name))
+    q.delete()
+
+    q = model.Session.query(GA_Url).\
+        filter(GA_Url.period_name == 'All')
+    log.debug("Deleting %d 'All' records..." % q.count())
+    q.delete()
+
+    model.Session.flush()
+    model.Session.commit()
     model.repo.commit_and_remove()
+    log.debug('...done')
+
+def post_update_url_stats():
+
+    """ Check the distinct url field in ga_url and make sure
+        it has an All record.  If not then create one.
+
+        After running this then every URL should have an All
+        record regardless of whether the URL has an entry for
+        the month being currently processed.
+    """
+    log.debug('Post-processing "All" records...')
+    query = """select url, pageviews::int, visits::int
+               from ga_url
+               where url not in (select url from ga_url where period_name ='All')"""
+    connection = model.Session.connection()
+    res = connection.execute(query)
+
+    views, visits = {}, {}
+    # url, views, visits
+    for row in res:
+        views[row[0]] = views.get(row[0], 0) + row[1]
+        visits[row[0]] = visits.get(row[0], 0) + row[2]
+
+    progress_total = len(views.keys())
+    progress_count = 0
+    for key in views.keys():
+        progress_count += 1
+        if progress_count % 100 == 0:
+            log.debug('.. %d/%d done so far', progress_count, progress_total)
+
+        package, publisher = _get_package_and_publisher(key)
+
+        values = {'id': make_uuid(),
+                  'period_name': "All",
+                  'period_complete_day': 0,
+                  'url': key,
+                  'pageviews': views[key],
+                  'visits': visits[key],
+                  'department_id': publisher,
+                  'package_id': package
+                  }
+        model.Session.add(GA_Url(**values))
+    model.Session.commit()
+    log.debug('..done')
 
 
 def update_url_stats(period_name, period_complete_day, url_data):
@@ -176,9 +227,14 @@
     stores them in GA_Url under the period and recalculates the totals for
     the 'All' period.
     '''
+    progress_total = len(url_data)
+    progress_count = 0
     for url, views, visits in url_data:
+        progress_count += 1
+        if progress_count % 100 == 0:
+            log.debug('.. %d/%d done so far', progress_count, progress_total)
+
         package, publisher = _get_package_and_publisher(url)
-
 
         item = model.Session.query(GA_Url).\
             filter(GA_Url.period_name==period_name).\
@@ -325,11 +381,8 @@
            order_by(model.Group.name).all()
 
 def get_children(publisher):
-    '''Finds child publishers for the given publisher (object). (Not recursive)'''
-    from ckan.model.group import HIERARCHY_CTE
-    return model.Session.query(model.Group).\
-           from_statement(HIERARCHY_CTE).params(id=publisher.id, type='publisher').\
-           all()
+    '''Finds child publishers for the given publisher (object). (Not recursive i.e. returns one level)'''
+    return publisher.get_children_groups(type='organization')
 
 def go_down_tree(publisher):
     '''Provided with a publisher object, it walks down the hierarchy and yields each publisher,
@@ -349,7 +402,7 @@
         if period_name != 'All':
             q = q.filter_by(period_name=period_name)
         q.delete()
-    model.Session.commit()
+    model.repo.commit_and_remove()
 
 def get_score_for_dataset(dataset_name):
     '''

--- a/ckanext/ga_report/helpers.py
+++ b/ckanext/ga_report/helpers.py
@@ -80,7 +80,7 @@
     return base.render_snippet('ga_report/ga_popular_single.html', **context)
 
 
-def most_popular_datasets(publisher, count=20):
+def most_popular_datasets(publisher, count=20, preview_image=None):
 
     if not publisher:
         _log.error("No valid publisher passed to 'most_popular_datasets'")
@@ -92,7 +92,8 @@
         'dataset_count': len(results),
         'datasets': results,
 
-        'publisher': publisher
+        'publisher': publisher,
+        'preview_image': preview_image
     }
 
     return base.render_snippet('ga_report/publisher/popular.html', **ctx)
@@ -106,8 +107,18 @@
     for entry in entries:
         if len(datasets) < count:
             p = model.Package.get(entry.url[len('/dataset/'):])
+
+            if not p:
+                _log.warning("Could not find Package for {url}".format(url=entry.url))
+                continue
+
+            if not p.state == 'active':
+                _log.warning("Package {0} is not active, it is {1}".format(p.name, p.state))
+                continue
+
             if not p in datasets:
                 datasets[p] = {'views':0, 'visits': 0}
+                
             datasets[p]['views'] = datasets[p]['views'] + int(entry.pageviews)
             datasets[p]['visits'] = datasets[p]['visits'] + int(entry.visits)
 
@@ -117,3 +128,17 @@
 
     return sorted(results, key=operator.itemgetter(1), reverse=True)
 
+def month_option_title(month_iso, months, day):
+    month_isos = [ iso_code for (iso_code,name) in months ]
+    try:
+        index = month_isos.index(month_iso)
+    except ValueError:
+        _log.error('Month "%s" not found in list of months.' % month_iso)
+        return month_iso
+    month_name = months[index][1]
+    if index==0:
+        return month_name + (' (up to %s)'%day)
+    return month_name
+
+
+

--- a/ckanext/ga_report/plugin.py
+++ b/ckanext/ga_report/plugin.py
@@ -5,7 +5,8 @@
 
 from ckanext.ga_report.helpers import (most_popular_datasets,
                                        popular_datasets,
-                                       single_popular_dataset)
+                                       single_popular_dataset,
+                                       month_option_title)
 
 log = logging.getLogger('ckanext.ga-report')
 
@@ -27,7 +28,8 @@
             'ga_report_installed': lambda: True,
             'popular_datasets': popular_datasets,
             'most_popular_datasets': most_popular_datasets,
-            'single_popular_dataset': single_popular_dataset
+            'single_popular_dataset': single_popular_dataset,
+            'month_option_title': month_option_title
         }
 
     def after_map(self, map):
@@ -41,6 +43,16 @@
             '/data/site-usage/data_{month}.csv',
             controller='ckanext.ga_report.controller:GaReport',
             action='csv'
+        )
+        map.connect(</