Changed to pageviews instead of unique pageviews
[ckanext-ga-report.git] / ckanext / ga_report / ga_model.py
blob:a/ckanext/ga_report/ga_model.py -> blob:b/ckanext/ga_report/ga_model.py
--- a/ckanext/ga_report/ga_model.py
+++ b/ckanext/ga_report/ga_model.py
@@ -9,6 +9,8 @@
 
 import ckan.model as model
 from ckan.lib.base import *
+
+log = __import__('logging').getLogger(__name__)
 
 def make_uuid():
     return unicode(uuid.uuid4())
@@ -45,6 +47,7 @@
                   Column('id', types.UnicodeText, primary_key=True,
                          default=make_uuid),
                   Column('period_name', types.UnicodeText),
+                  Column('period_complete_day', types.UnicodeText),
                   Column('stat_name', types.UnicodeText),
                   Column('key', types.UnicodeText),
                   Column('value', types.UnicodeText), )
@@ -111,9 +114,7 @@
     >>> normalize_url('http://data.gov.uk/dataset/weekly_fuel_prices')
     '/dataset/weekly_fuel_prices'
     '''
-    # Deliberately leaving a /
-    url = url.replace('http:/','')
-    return '/' + '/'.join(url.split('/')[2:])
+    return '/' + '/'.join(url.split('/')[3:])
 
 
 def _get_package_and_publisher(url):
@@ -134,7 +135,7 @@
             return None, publisher_match.groups()[0]
     return None, None
 
-def update_sitewide_stats(period_name, stat_name, data):
+def update_sitewide_stats(period_name, stat_name, data, period_complete_day):
     for k,v in data.iteritems():
         item = model.Session.query(GA_Stat).\
             filter(GA_Stat.period_name==period_name).\
@@ -144,11 +145,13 @@
             item.period_name = period_name
             item.key = k
             item.value = v
+            item.period_complete_day = period_complete_day
             model.Session.add(item)
         else:
             # create the row
             values = {'id': make_uuid(),
                      'period_name': period_name,
+                     'period_complete_day': period_complete_day,
                      'key': k,
                      'value': v,
                      'stat_name': stat_name
@@ -165,7 +168,11 @@
 
 
 def update_url_stats(period_name, period_complete_day, url_data):
-
+    '''
+    Given a list of urls and number of hits for each during a given period,
+    stores them in GA_Url under the period and recalculates the totals for
+    the 'All' period.
+    '''
     for url, views, visits in url_data:
         package, publisher = _get_package_and_publisher(url)
 
@@ -210,7 +217,7 @@
                       'period_complete_day': 0,
                       'url': url,
                       'pageviews': sum([int(e.pageviews) for e in entries]) + old_pageviews,
-                      'visits': sum([int(e.visits) for e in entries]) + old_visits,
+                      'visits': sum([int(e.visits or 0) for e in entries]) + old_visits,
                       'department_id': publisher,
                       'package_id': package
                      }
@@ -341,3 +348,34 @@
         q.delete()
     model.Session.commit()
 
+def get_score_for_dataset(dataset_name):
+    '''
+    Returns a "current popularity" score for a dataset,
+    based on how many views it has had recently.
+    '''
+    import datetime
+    now = datetime.datetime.now()
+    last_month = now - datetime.timedelta(days=30)
+    period_names = ['%s-%02d' % (last_month.year, last_month.month),
+                    '%s-%02d' % (now.year, now.month),
+                    ]
+
+    score = 0
+    for period_name in period_names:
+        score /= 2 # previous periods are discounted by 50%
+        entry = model.Session.query(GA_Url)\
+                .filter(GA_Url.period_name==period_name)\
+                .filter(GA_Url.package_id==dataset_name).first()
+        # score
+        if entry:
+            views = float(entry.pageviews)
+            if entry.period_complete_day:
+                views_per_day = views / entry.period_complete_day
+            else:
+                views_per_day = views / 15 # guess
+            score += views_per_day
+
+    score = int(score * 100)
+    log.debug('Popularity %s: %s', score, dataset_name)
+    return score
+