Make sure we total the vals for 'All' rather than just duplicate the entry
Make sure we total the vals for 'All' rather than just duplicate the entry

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

file:a/README.rst -> file:b/README.rst
--- a/README.rst
+++ b/README.rst
@@ -31,10 +31,12 @@
 2. Ensure you development.ini (or similar) contains the info about your Google Analytics account and configuration::
 
       googleanalytics.id = UA-1010101-1
-      googleanalytics.account = Account name (i.e. data.gov.uk, see top level item at https://www.google.com/analytics)
+      googleanalytics.account = Account name (e.g. data.gov.uk, see top level item at https://www.google.com/analytics)
+      googleanalytics.token.filepath = ~/pyenv/token.dat
       ga-report.period = monthly
+      ga-report.bounce_url = /
 
-   Note that your credentials will be readable by system administrators on your server. Rather than use sensitive account details, it is suggested you give access to the GA account to a new Google account that you create just for this purpose.
+   The ga-report.bounce_url specifies a particular path to record the bounce rate for. Typically it is / (the home page).
 
 3. Set up this extension's database tables using a paster command. (Ensure your CKAN pyenv is still activated, run the command from ``src/ckanext-ga-report``, alter the ``--config`` option to point to your site config file)::
 
@@ -43,6 +45,12 @@
 4. Enable the extension in your CKAN config file by adding it to ``ckan.plugins``::
 
     ckan.plugins = ga-report
+
+Problem shooting
+----------------
+
+* ``(ProgrammingError) relation "ga_url" does not exist``
+  This means that the ``paster initdb`` step has not been run successfully. Refer to the installation instructions for this extension.
 
 
 Authorization
@@ -75,13 +83,17 @@
 
     $ paster getauthtoken --config=../ckan/development.ini
 
+Now ensure you reference the correct path to your token.dat in your CKAN config file (e.g. development.ini)::
+
+    googleanalytics.token.filepath = ~/pyenv/token.dat
+
 
 Tutorial
 --------
 
-Download some GA data and store it in CKAN's db. (Ensure your CKAN pyenv is still activated, run the command from ``src/ckanext-ga-report``, alter the ``--config`` option to point to your site config file) and specifying the name of your auth file (token.dat by default) from the previous step::
+Download some GA data and store it in CKAN's database. (Ensure your CKAN pyenv is still activated, run the command from ``src/ckanext-ga-report``, alter the ``--config`` option to point to your site config file) and specifying the name of your auth file (token.dat by default) from the previous step::
 
-    $ paster loadanalytics token.dat latest --config=../ckan/development.ini
+    $ paster loadanalytics latest --config=../ckan/development.ini
 
 The value after the token file is how much data you want to retrieve, this can be
 

--- a/ckanext/ga_report/command.py
+++ b/ckanext/ga_report/command.py
@@ -1,5 +1,8 @@
 import logging
 import datetime
+import os
+
+from pylons import config
 
 from ckan.lib.cli import CkanCommand
 # No other CKAN imports allowed until _load_config is run,
@@ -58,20 +61,30 @@
     """Get data from Google Analytics API and save it
     in the ga_model
 
-    Usage: paster loadanalytics <tokenfile> <time-period>
+    Usage: paster loadanalytics <time-period>
 
-    Where <tokenfile> is the name of the auth token file from
-    the getauthtoken step.
-
-    And where <time-period> is:
+    Where <time-period> is:
         all         - data for all time
         latest      - (default) just the 'latest' data
         YYYY-MM     - just data for the specific month
     """
     summary = __doc__.split('\n')[0]
     usage = __doc__
-    max_args = 2
-    min_args = 1
+    max_args = 1
+    min_args = 0
+
+    def __init__(self, name):
+        super(LoadAnalytics, self).__init__(name)
+        self.parser.add_option('-d', '--delete-first',
+                               action='store_true',
+                               default=False,
+                               dest='delete_first',
+                               help='Delete data for the period first')
+        self.parser.add_option('-s', '--skip_url_stats',
+                               action='store_true',
+                               default=False,
+                               dest='skip_url_stats',
+                               help='Skip the download of URL data - just do site-wide stats')
 
     def command(self):
         self._load_config()
@@ -79,17 +92,25 @@
         from download_analytics import DownloadAnalytics
         from ga_auth import (init_service, get_profile_id)
 
+        ga_token_filepath = os.path.expanduser(config.get('googleanalytics.token.filepath', ''))
+        if not ga_token_filepath:
+            print 'ERROR: In the CKAN config you need to specify the filepath of the ' \
+                  'Google Analytics token file under key: googleanalytics.token.filepath'
+            return
+
         try:
-            svc = init_service(self.args[0], None)
+            svc = init_service(ga_token_filepath, None)
         except TypeError:
             print ('Have you correctly run the getauthtoken task and '
-                   'specified the correct file here')
+                   'specified the correct token file in the CKAN config under '
+                   '"googleanalytics.token.filepath"?')
             return
 
-        downloader = DownloadAnalytics(svc, profile_id=get_profile_id(svc))
+        downloader = DownloadAnalytics(svc, profile_id=get_profile_id(svc),
+                                       delete_first=self.options.delete_first,
+                                       skip_url_stats=self.options.skip_url_stats)
 
-        time_period = self.args[1] if self.args and len(self.args) > 1 \
-            else 'latest'
+        time_period = self.args[0] if self.args else 'latest'
         if time_period == 'all':
             downloader.all_()
         elif time_period == 'latest':

--- a/ckanext/ga_report/controller.py
+++ b/ckanext/ga_report/controller.py
@@ -4,12 +4,12 @@
 import logging
 import operator
 import collections
-from ckan.lib.base import BaseController, c, render, request, response, abort
+from ckan.lib.base import (BaseController, c, g, render, request, response, abort)
 
 import sqlalchemy
 from sqlalchemy import func, cast, Integer
 import ckan.model as model
-from ga_model import GA_Url, GA_Stat
+from ga_model import GA_Url, GA_Stat, GA_ReferralStat, GA_Publisher
 
 log = logging.getLogger('ckanext.ga-report')
 
@@ -22,11 +22,29 @@
 
 
 def _month_details(cls):
+    '''
+    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).distinct().all()
+    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()
+    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):
@@ -34,7 +52,7 @@
     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()
@@ -51,14 +69,34 @@
                              entry.key.encode('utf-8'),
                              entry.value.encode('utf-8')])
 
+    def csv_downloads(self, month):
+        import csv
+
+        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.key').all()
+
+        response.headers['Content-Type'] = "text/csv; charset=utf-8"
+        response.headers['Content-Disposition'] = str('attachment; filename=downloads_%s.csv' % (month,))
+
+        writer = csv.writer(response)
+        writer.writerow(["Period", "Resource URL", "Count"])
+
+        for entry in entries:
+            writer.writerow([entry.period_name.encode('utf-8'),
+                             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 time'
+        c.month_desc = 'all months'
         c.month = request.params.get('month', '')
         if c.month:
             c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month])
@@ -70,14 +108,17 @@
         entries = q.order_by('ga_stat.key').all()
 
         def clean_key(key, val):
-            if key in ['Average time on site', 'Pages per visit', 'Percent new visits']:
+            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 ['Bounces', 'Total pageviews']:
+                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
 
         c.global_totals = []
@@ -90,44 +131,49 @@
             for e in entries:
                 d[e.key].append(float(e.value))
             for k, v in d.iteritems():
-                if k in ['Bounces', 'Total pageviews']:
+                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))
                 key, val = clean_key(k,v)
+
                 c.global_totals.append((key, val))
                 c.global_totals = sorted(c.global_totals, key=operator.itemgetter(0))
 
         keys = {
-            'Browser versions': 'browsers',
-            'Operating Systems versions': 'os',
+            'Browser versions': 'browser_versions',
+            'Browsers': 'browsers',
+            'Operating Systems versions': 'os_versions',
+            'Operating Systems': 'os',
             'Social sources': 'social_networks',
             'Languages': 'languages',
             'Country': 'country'
         }
 
-        browser_version_re = re.compile("(.*)\((.*)\)")
+        def shorten_name(name, length=60):
+            return (name[:length] + '..') if len(name) > 60 else name
+
+        def fill_out_url(url):
+            import urlparse
+            return urlparse.urljoin(g.site_url, url)
+
+        c.social_referrer_totals, c.social_referrers = [], []
+        q = model.Session.query(GA_ReferralStat)
+        q = q.filter(GA_ReferralStat.period_name==c.month) if c.month else q
+        q = q.order_by('ga_referrer.count::int desc')
+        for entry in q.all():
+            c.social_referrers.append((shorten_name(entry.url), fill_out_url(entry.url),
+                                       entry.source,entry.count))
+
+        q = model.Session.query(GA_ReferralStat.url,
+                                func.sum(GA_ReferralStat.count).label('count'))
+        q = q.filter(GA_ReferralStat.period_name==c.month) if c.month else q
+        q = q.order_by('count desc').group_by(GA_ReferralStat.url)
+        for entry in q.all():
+            c.social_referrer_totals.append((shorten_name(entry[0]), fill_out_url(entry[0]),'',
+                                            entry[1]))
+
         for k, v in keys.iteritems():
-
-            def clean_field(key):
-                if k != 'Browser versions':
-                    return key
-                m = browser_version_re.match(key)
-                browser = m.groups()[0].strip()
-                ver = m.groups()[1]
-                parts = ver.split('.')
-                if len(parts) > 1:
-                    if parts[1][0] == '0':
-                        ver = parts[0]
-                    else:
-                        ver = "%s.%s" % (parts[0],parts[1])
-                if browser in ['Safari','Android Browser']:  # Special case complex version nums
-                    ver = parts[0]
-                    if len(ver) > 2:
-                        ver = "%s%sX" % (ver[0], ver[1])
-
-                return "%s (%s)" % (browser, ver,)
-
             q = model.Session.query(GA_Stat).\
                 filter(GA_Stat.stat_name==k)
             if c.month:
@@ -137,174 +183,231 @@
 
             d = collections.defaultdict(int)
             for e in q.all():
-                d[clean_field(e.key)] += int(e.value)
+                d[e.key] += int(e.value)
             entries = []
             for key, val in d.iteritems():
                 entries.append((key,val,))
             entries = sorted(entries, key=operator.itemgetter(1), reverse=True)
 
-            setattr(c, v, [(k,v) for k,v in entries ])
-
-
+            # 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'])
+            else:
+                total = sum([num for _,num in entries])
+            setattr(c, v, [(k,_percent(v,total)) for k,v in entries ])
 
         return render('ga_report/site/index.html')
 
-
-class GaPublisherReport(BaseController):
+    def downloads(self):
+
+        # Get the month details by fetching distinct values and determining the
+        # month names from the values.
+        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'
+        c.month = request.params.get('month', '')
+        if c.month:
+            c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month])
+
+        c.downloads = []
+        q = model.Session.query(GA_Stat).filter(GA_Stat.stat_name=='Downloads')
+        q = q.filter(GA_Stat.period_name==c.month) if c.month else q
+        q = q.order_by("ga_stat.value::int desc")
+
+        data = collections.defaultdict(int)
+        for entry in q.all():
+            r = model.Session.query(model.Resource).filter(model.Resource.url==entry.key).first()
+            if not r:
+                continue
+            data[r] += int(entry.value)
+
+        for k,v in data.iteritems():
+            c.downloads.append((k,v))
+
+        c.downloads = sorted(c.downloads, key=operator.itemgetter(1), reverse=True)
+
+        return render('ga_report/site/downloads.html')
+
+
+class GaDatasetReport(BaseController):
     """
-    Displays the pageview and visit count for specific publishers based on
-    the datasets associated with the publisher.
+    Displays the pageview and visit count for datasets
+    with options to filter by publisher and time period.
     """
-    def csv(self, month):
-
-        c.month = month if not month =='all' else ''
+    def publisher_csv(self, month):
+        '''
+        Returns a CSV of each publisher with the total number of dataset
+        views & visits.
+        '''
+        c.month = month if not month == 'all' else ''
         response.headers['Content-Type'] = "text/csv; charset=utf-8"
         response.headers['Content-Disposition'] = str('attachment; filename=publishers_%s.csv' % (month,))
 
         writer = csv.writer(response)
-        writer.writerow(["Publisher", "Views", "Visits", "Period Name"])
-
-        for publisher,view,visit in self._get_publishers(None):
+        writer.writerow(["Publisher Title", "Publisher Name", "Views", "Visits", "Period Name"])
+
+        for publisher,view,visit in _get_top_publishers(None):
             writer.writerow([publisher.title.encode('utf-8'),
+                             publisher.name.encode('utf-8'),
                              view,
                              visit,
                              month])
 
-
-
-    def publisher_csv(self, id, month):
-
-        c.month = month if not month =='all' else ''
-        c.publisher = model.Group.get(id)
-        if not c.publisher:
-            abort(404, 'A publisher with that name could not be found')
+    def dataset_csv(self, id='all', month='all'):
+        '''
+        Returns a CSV with the number of views & visits for each dataset.
+
+        :param id: A Publisher ID or None if you want for all
+        :param month: The time period, or 'all'
+        '''
+        c.month = month if not month == 'all' else ''
+        if id != 'all':
+            c.publisher = model.Group.get(id)
+            if not c.publisher:
+                abort(404, 'A publisher with that name could not be found')
 
         packages = self._get_packages(c.publisher)
         response.headers['Content-Type'] = "text/csv; charset=utf-8"
         response.headers['Content-Disposition'] = \
-            str('attachment; filename=%s_%s.csv' % (c.publisher.name, month,))
+            str('attachment; filename=datasets_%s_%s.csv' % (c.publisher_name, month,))
 
         writer = csv.writer(response)
-        writer.writerow(["Publisher", "Views", "Visits", "Period Name"])
+        writer.writerow(["Dataset Title", "Dataset Name", "Views", "Visits", "Period Name"])
 
         for package,view,visit in packages:
             writer.writerow([package.title.encode('utf-8'),
+                             package.name.encode('utf-8'),
                              view,
                              visit,
                              month])
 
-
-
-    def index(self):
+    def publishers(self):
+        '''A list of publishers and the number of views/visits for each'''
 
         # 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', '')
-        c.month_desc = 'all time'
+        c.month_desc = 'all months'
         if c.month:
             c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month])
 
-        c.top_publishers = self._get_publishers()
-
+        c.top_publishers = _get_top_publishers()
         return render('ga_report/publisher/index.html')
 
-    def _get_publishers(self, limit=20):
-        connection = model.Session.connection()
-        q = """
-            select department_id, sum(pageviews::int) views, sum(visitors::int) visits
-            from ga_url
-            where department_id <> ''"""
-        if c.month:
-            q = q + """
-                    and period_name=%s
-            """
-        q = q + """
-                group by department_id order by views desc
-            """
-        if limit:
-            q = q + " limit %s;" % (limit)
-
-        # Add this back (before and period_name =%s) if you want to ignore publisher
-        # homepage views
-        # and not url like '/publisher/%%'
-
-        top_publishers = []
-        res = connection.execute(q, c.month)
-
-        for row in res:
-            g = model.Group.get(row[0])
-            if g:
-                top_publishers.append((g, row[1], row[2]))
-        return top_publishers
-
-    def _get_packages(self, publisher, count=-1):
+    def _get_packages(self, publisher=None, count=-1):
+        '''Returns the datasets in order of views'''
         if count == -1:
             count = sys.maxint
 
+        month = c.month or 'All'
+
+        q = model.Session.query(GA_Url,model.Package)\
+            .filter(model.Package.name==GA_Url.package_id)\
+            .filter(GA_Url.url.like('/dataset/%'))
+        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.pageviews::int desc')
         top_packages = []
-        q =  model.Session.query(GA_Url).\
-            filter(GA_Url.department_id==publisher.name).\
-            filter(GA_Url.url.like('/dataset/%'))
-        if c.month:
-            q = q.filter(GA_Url.period_name==c.month)
-        q = q.order_by('ga_url.pageviews::int desc')
-
-        if c.month:
-            for entry in q[:count]:
-                p = model.Package.get(entry.url[len('/dataset/'):])
-                top_packages.append((p,entry.pageviews,entry.visitors))
-        else:
-            ds = {}
-            for entry in q.all():
-                if len(ds) >= count:
-                    break
-                p = model.Package.get(entry.url[len('/dataset/'):])
-                if not p in ds:
-                    ds[p] = {'views':0, 'visits': 0}
-                ds[p]['views'] = ds[p]['views'] + int(entry.pageviews)
-                ds[p]['visits'] = ds[p]['visits'] + int(entry.visitors)
-
-            results = []
-            for k, v in ds.iteritems():
-                results.append((k,v['views'],v['visits']))
-
-            top_packages = sorted(results, key=operator.itemgetter(1), reverse=True)
+        for entry,package in q.limit(count):
+            if package:
+                top_packages.append((package, entry.pageviews, entry.visits))
+            else:
+                log.warning('Could not find package associated package')
+
         return top_packages
 
-
-    def read(self, id):
+    def read(self):
+        '''
+        Lists the most popular datasets across all publishers
+        '''
+        return self.read_publisher(None)
+
+    def read_publisher(self, id):
+        '''
+        Lists the most popular datasets for a publisher (or across all publishers)
+        '''
         count = 20
 
-        c.publisher = model.Group.get(id)
-        if not c.publisher:
-            abort(404, 'A publisher with that name could not be found')
+        c.publishers = _get_publishers()
+
+        id = request.params.get('publisher', id)
+        if id and id != 'all':
+            c.publisher = model.Group.get(id)
+            if not c.publisher:
+                abort(404, 'A publisher with that name could not be found')
+            c.publisher_name = c.publisher.name
         c.top_packages = [] # package, dataset_views in c.top_packages
 
         # 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', '')
         if not c.month:
-            c.month_desc = 'all time'
+            c.month_desc = 'all months'
         else:
             c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month])
 
+        month = c.month or 'All'
         c.publisher_page_views = 0
         q = model.Session.query(GA_Url).\
-            filter(GA_Url.url=='/publisher/%s' % c.publisher.name)
-        if c.month:
-            entry = q.filter(GA_Url.period_name==c.month).first()
-            c.publisher_page_views = entry.pageviews if entry else 0
-        else:
-            for e in q.all():
-                c.publisher_page_views = c.publisher_page_views  + int(e.pageviews)
+            filter(GA_Url.url=='/publisher/%s' % c.publisher_name)
+        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)
 
         return render('ga_report/publisher/read.html')
 
+def _get_top_publishers(limit=20):
+    '''
+    Returns a list of the top 20 publishers by dataset visits.
+    (The number to show can be varied with 'limit')
+    '''
+    month = c.month or 'All'
+    connection = model.Session.connection()
+    q = """
+        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 views desc
+        """
+    if limit:
+        q = q + " limit %s;" % (limit)
+
+    top_publishers = []
+    res = connection.execute(q, month)
+    for row in res:
+        g = model.Group.get(row[0])
+        if g:
+            top_publishers.append((g, row[1], row[2]))
+    return top_publishers
+
+
+def _get_publishers():
+    '''
+    Returns a list of all publishers. Each item is a tuple:
+      (name, title)
+    '''
+    publishers = []
+    for pub in model.Session.query(model.Group).\
+               filter(model.Group.type=='publisher').\
+               filter(model.Group.state=='active').\
+               order_by(model.Group.name):
+        publishers.append((pub.name, pub.title))
+    return publishers
+
+def _percent(num, total):
+    p = 100 * float(num)/float(total)
+    return "%.2f%%" % round(p, 2)
+

--- a/ckanext/ga_report/download_analytics.py
+++ b/ckanext/ga_report/download_analytics.py
@@ -1,9 +1,9 @@
 import os
 import logging
 import datetime
-
+import collections
 from pylons import config
-
+from ga_model import _normalize_url
 import ga_model
 
 #from ga_client import GA
@@ -11,15 +11,20 @@
 log = logging.getLogger('ckanext.ga-report')
 
 FORMAT_MONTH = '%Y-%m'
+MIN_VIEWS = 50
+MIN_VISITS = 20
+MIN_DOWNLOADS = 10
 
 class DownloadAnalytics(object):
     '''Downloads and stores analytics info'''
 
-    def __init__(self, service=None, profile_id=None):
+    def __init__(self, service=None, profile_id=None, delete_first=False,
+                 skip_url_stats=False):
         self.period = config['ga-report.period']
         self.service = service
         self.profile_id = profile_id
-
+        self.delete_first = delete_first
+        self.skip_url_stats = skip_url_stats
 
     def specific_month(self, date):
         import calendar
@@ -90,32 +95,77 @@
 
     def download_and_store(self, periods):
         for period_name, period_complete_day, start_date, end_date in periods:
-            log.info('Downloading Analytics for period "%s" (%s - %s)',
+            log.info('Period "%s" (%s - %s)',
                      self.get_full_period_name(period_name, period_complete_day),
-                     start_date.strftime('%Y %m %d'),
-                     end_date.strftime('%Y %m %d'))
-
-            data = self.download(start_date, end_date, '~/dataset/[a-z0-9-_]+')
-            log.info('Storing Dataset Analytics for period "%s"',
-                     self.get_full_period_name(period_name, period_complete_day))
-            self.store(period_name, period_complete_day, data, )
-
-            data = self.download(start_date, end_date, '~/publisher/[a-z0-9-_]+')
-            log.info('Storing Publisher Analytics for period "%s"',
-                     self.get_full_period_name(period_name, period_complete_day))
-            self.store(period_name, period_complete_day, data,)
-
-            ga_model.update_publisher_stats(period_name) # about 30 seconds.
-            self.sitewide_stats( period_name )
-
-
-    def download(self, start_date, end_date, path='~/dataset/[a-z0-9-_]+'):
+                     start_date.strftime('%Y-%m-%d'),
+                     end_date.strftime('%Y-%m-%d'))
+
+            if self.delete_first:
+                log.info('Deleting existing Analytics for this period "%s"',
+                         period_name)
+                ga_model.delete(period_name)
+
+            if not self.skip_url_stats:
+                # Clean out old url data before storing the new
+                ga_model.pre_update_url_stats(period_name)
+
+                accountName = config.get('googleanalytics.account')
+
+                log.info('Downloading analytics for dataset views')
+                data = self.download(start_date, end_date, '~/%s/dataset/[a-z0-9-_]+' % accountName)
+
+                log.info('Storing dataset views (%i rows)', len(data.get('url')))
+                self.store(period_name, period_complete_day, data, )
+
+                log.info('Downloading analytics for publisher views')
+                data = self.download(start_date, end_date, '~/%s/publisher/[a-z0-9-_]+' % accountName)
+
+                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')
+                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 )
+
+            log.info('Downloading and storing analytics for social networks')
+            self.update_social_info(period_name, start_date, end_date)
+
+
+    def update_social_info(self, period_name, start_date, end_date):
+        start_date = start_date.strftime('%Y-%m-%d')
+        end_date = end_date.strftime('%Y-%m-%d')
+        query = 'ga:hasSocialSourceReferral=~Yes$'
+        metrics = 'ga:entrances'
+        sort = '-ga:entrances'
+
+        # Supported query params at
+        # https://developers.google.com/analytics/devguides/reporting/core/v3/reference
+        results = self.service.data().ga().get(
+                                 ids='ga:' + self.profile_id,
+                                 filters=query,
+                                 start_date=start_date,
+                                 metrics=metrics,
+                                 sort=sort,
+                                 dimensions="ga:landingPagePath,ga:socialNetwork",
+                                 max_results=10000,
+                                 end_date=end_date).execute()
+        data = collections.defaultdict(list)
+        rows = results.get('rows',[])
+        for row in rows:
+            url = _normalize_url('http:/' + row[0])
+            data[url].append( (row[1], int(row[2]),) )
+        ga_model.update_social(period_name, data)
+
+
+    def download(self, start_date, end_date, path=None):
         '''Get data from GA for a given time period'''
         start_date = start_date.strftime('%Y-%m-%d')
         end_date = end_date.strftime('%Y-%m-%d')
         query = 'ga:pagePath=%s$' % path
-        metrics = 'ga:uniquePageviews, ga:visitors'
-        sort = '-ga:uniquePageviews'
+        metrics = 'ga:pageviews, ga:visits'
+        sort = '-ga:pageviews'
 
         # Supported query params at
         # https://developers.google.com/analytics/devguides/reporting/core/v3/reference
@@ -129,35 +179,35 @@
                                  max_results=10000,
                                  end_date=end_date).execute()
 
-        if os.getenv('DEBUG'):
-            import pprint
-            pprint.pprint(results)
-            print 'Total results: %s' % results.get('totalResults')
-
         packages = []
         for entry in results.get('rows'):
             (loc,pageviews,visits) = entry
-            packages.append( ('http:/' + loc, pageviews, visits,) ) # Temporary hack
+            url = _normalize_url('http:/' + loc) # strips off domain e.g. www.data.gov.uk or data.gov.uk
+
+            if not url.startswith('/dataset/') and not url.startswith('/publisher/'):
+                # filter out strays like: