Changed to pageviews instead of unique pageviews
[ckanext-ga-report.git] / ckanext / ga_report / helpers.py
blob:a/ckanext/ga_report/helpers.py -> blob:b/ckanext/ga_report/helpers.py
--- a/ckanext/ga_report/helpers.py
+++ b/ckanext/ga_report/helpers.py
@@ -50,9 +50,12 @@
             dataset = model.Package.get(ga_url.url[len('/dataset/'):])
             if dataset and not dataset.state == 'active':
                 dataset = None
-                count += 1
-                if count > 10:
-                    break
+            # When testing, it is possible that top datasets are not available
+            # so only go round this loop a few times before falling back on
+            # a random dataset.
+            count += 1
+            if count > 10:
+                break
     if not dataset:
         # fallback
         dataset = model.Session.query(model.Package)\
@@ -60,14 +63,15 @@
         if not dataset:
             return None
     dataset_dict = get_action('package_show')({'model': model,
-                                               'session': model.Session},
+                                               'session': model.Session,
+                                               'validate': False},
                                               {'id':dataset.id})
     return dataset_dict
 
 def single_popular_dataset_html(top=20):
     dataset_dict = single_popular_dataset(top)
     groups = package.get('groups', [])
-    publishers = [ g for g in groups if g.get('type') == 'publisher' ]
+    publishers = [ g for g in groups if g.get('type') == 'organization' ]
     publisher = publishers[0] if publishers else {'name':'', 'title': ''}
     context = {
         'dataset': dataset_dict,
@@ -76,7 +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'")
@@ -88,7 +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)
@@ -102,8 +107,22 @@
     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.private == False:
+                _log.warning("Package {0} is private {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)
 
@@ -113,3 +132,17 @@
 
     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
+
+
+