add siteanalytics, exclude empty datasets from some stats
add siteanalytics, exclude empty datasets from some stats

--- a/ckanext/dga_stats/controller.py
+++ b/ckanext/dga_stats/controller.py
@@ -12,7 +12,6 @@
         c.top_rated_packages = stats.top_rated_packages()
         c.most_edited_packages = stats.most_edited_packages()
         c.largest_groups = stats.largest_groups()
-        c.top_tags = stats.top_tags()
         c.top_package_owners = stats.top_package_owners()
         c.summary_stats = stats.summary_stats()
         c.activity_counts = stats.activity_counts()

--- a/ckanext/dga_stats/stats.py
+++ b/ckanext/dga_stats/stats.py
@@ -72,6 +72,7 @@
     def by_org(cls, limit=10):
         connection = model.Session.connection()
         res = connection.execute("select package.owner_org, package.private, count(*) from package \
+		inner join (select distinct package_id from resource_group inner join resource on resource.resource_group_id = resource_group.id) as r on package.id = r.package_id \
 		inner join \"group\" on package.owner_org = \"group\".id \
 		where package.state='active'\
 		group by package.owner_org,\"group\".name, package.private \
@@ -107,6 +108,7 @@
     def top_active_orgs(cls, limit=10):
         connection = model.Session.connection()
         res = connection.execute("select package.owner_org, count(*) from package \
+		inner join (select distinct package_id from resource_group inner join resource on resource.resource_group_id = resource_group.id) as r on package.id = r.package_id \
 		inner join \"group\" on package.owner_org = \"group\".id \
                 inner join (select distinct object_id from activity where activity.timestamp > (now() - interval '60 day')) \
                 latestactivities on latestactivities.object_id = package.id \
@@ -116,32 +118,6 @@
                 order by count(*) desc;").fetchall();
         res_groups = [(model.Session.query(model.Group).get(unicode(group_id)), val) for group_id, val in res]
         return res_groups
-
-    @classmethod
-    def top_tags(cls, limit=10, returned_tag_info='object'): # by package
-        assert returned_tag_info in ('name', 'id', 'object')
-        tag = table('tag')
-        package_tag = table('package_tag')
-        package = table('package')
-        #TODO filter out tags with state=deleted
-        if returned_tag_info == 'name':
-            from_obj = [package_tag.join(tag)]
-            tag_column = tag.c.name
-        else:
-            from_obj = None
-            tag_column = package_tag.c.tag_id
-        s = select([tag_column, func.count(package_tag.c.package_id)],
-                    from_obj=from_obj)
-        s = s.group_by(tag_column).\
-	      where(package.c.private == 'f').\
-            order_by(func.count(package_tag.c.package_id).desc()).\
-            limit(limit)
-        res_col = model.Session.execute(s).fetchall()
-        if returned_tag_info in ('id', 'name'):
-            return res_col
-        elif returned_tag_info == 'object':
-            res_tags = [(model.Session.query(model.Tag).get(unicode(tag_id)), val) for tag_id, val in res_col]
-            return res_tags
 
     @classmethod
     def top_package_owners(cls, limit=10):
@@ -164,10 +140,10 @@
        connection = model.Session.connection()
 
        res = connection.execute("SELECT 'Total Organisations', count(*) from \"group\" where type = 'organization' and state = 'active' union \
-				select 'Total Datasets', count(*) from package where (state='active' or state='draft' or state='draft-complete') and private = 'f' union \
+				select 'Total Datasets', count(*) from package inner join (select distinct package_id from resource_group inner join resource on resource.resource_group_id = resource_group.id) as r on package.id = r.package_id where (package.state='active' or package.state='draft' or package.state='draft-complete') and private = 'f' union \
 				select 'Total Archived Datasets', count(*) from package where (state='active' or state='draft' or state='draft-complete') and private = 't' union \
 				select 'Total Data Files/Resources', count(*) from resource where state='active' union \
-				select 'Total Machine Readable/Data API Resources', count(*) from resource where state='active' and webstore_url = 'active'\
+				select 'Total Machine Readable/Data API Resources', count(*) from resource where state='active' and (webstore_url = 'active' or format="wms")\
 				").fetchall();
        return res
 

--- a/ckanext/dga_stats/templates/ckanext/stats/index.html
+++ b/ckanext/dga_stats/templates/ckanext/stats/index.html
@@ -304,6 +304,8 @@
         <li class="nav-item"><a href="#stats-most-edited" data-toggle="tab">{{ _('Most Edited Datasets') }}</a></li>
         <li class="nav-item"><a href="#stats-by-org" data-toggle="tab">{{ _('Datasets by Organization') }}</a></li> 
         <li class="nav-item"><a href="#stats-res-by-org" data-toggle="tab">{{ _('Resources by Organization') }}</a></li> 
+        <li class="nav-item"><a href="#stats-activity-org" data-toggle="tab">{{ _('Most Active Organisations') }}</a></li>
+        <li class="nav-item active"><a href="//data.gov.au/site-usage" data-toggle="tab">Site Analytics</a></li>
         <li class="nav-item active"><a href="#stats-summary" data-toggle="tab">{{ _('Summary') }}</a></li> 
       </ul>
     </nav>