--- a/ckanext/dga_stats/stats.py +++ b/ckanext/dga_stats/stats.py @@ -2,6 +2,7 @@ from pylons import config from sqlalchemy import Table, select, func, and_ +from sqlalchemy.sql.expression import text import ckan.plugins as p import ckan.model as model @@ -55,9 +56,11 @@ member = table('member') s = select([member.c.group_id, func.count(member.c.table_id)]).\ group_by(member.c.group_id).\ - where(and_(member.c.group_id!=None, member.c.table_name=='package')).\ - order_by(func.count(member.c.table_id).desc()).\ - limit(limit) + where(member.c.group_id!=None).\ + where(member.c.table_name=='package').\ + where(member.c.capacity=='public').\ + order_by(func.count(member.c.table_id).desc()) + #limit(limit) res_ids = model.Session.execute(s).fetchall() res_groups = [(model.Session.query(model.Group).get(unicode(group_id)), val) for group_id, val in res_ids] @@ -65,15 +68,27 @@ @classmethod def by_org(cls, limit=10): - group = table('group') - package = table('package') - s = select([group.c.id, package.c.private, func.count(package.c.private)]).\ - group_by(group.c.id, package.c.private).\ - order_by(group.c.id).\ - limit(limit) - - res_ids = model.Session.execute(s).fetchall() - res_groups = [(model.Session.query(model.Group).get(unicode(group_id)), private, val) for group_id, private, val in res_ids] + connection = model.Session.connection() + res = connection.execute("select package.owner_org, package.private, count(*) from package \ + inner join \"group\" on package.owner_org = \"group\".id \ + where package.state='active'\ + group by package.owner_org,\"group\".name, package.private \ + order by \"group\".name, package.private;").fetchall(); + res_groups = [(model.Session.query(model.Group).get(unicode(group_id)), private, val) for group_id, private, val in res] + return res_groups + + @classmethod + def top_active_orgs(cls, limit=10): + connection = model.Session.connection() + res = connection.execute("select package.owner_org, count(*) from package \ + 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 \ + where package.state='active' \ + and package.private = 'f' \ + group by package.owner_org \ + 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 @@ -107,10 +122,9 @@ package_role = table('package_role') user_object_role = table('user_object_role') package = table('package') -# join(package, package_role.c.package_id == package.c.id).\ -# where(package.c.private == 'f').\ - s = select([user_object_role.c.user_id, func.count(user_object_role.c.role)], from_obj=[user_object_role.join(package_role)]).\ + s = select([user_object_role.c.user_id, func.count(user_object_role.c.role)], from_obj=[user_object_role.join(package_role).join(package, package_role.c.package_id == package.c.id)]).\ where(user_object_role.c.role==model.authz.Role.ADMIN).\ + where(package.c.private == 'f').\ where(user_object_role.c.user_id!=None).\ group_by(user_object_role.c.user_id).\ order_by(func.count(user_object_role.c.role).desc()).\ @@ -124,8 +138,11 @@ 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' union \ - select 'Total Data Files/Resources', count(*) from resource where state='active'").fetchall(); + select 'Total Datasets', count(*) from package where (state='active' or state='draft' or 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'\ + ").fetchall(); return res @@ -140,6 +157,16 @@ connection = model.Session.connection() res = connection.execute("select name,sysadmin,role from user_object_role right outer join \"user\" on user_object_role.user_id = \"user\".id where name not in ('logged_in','visitor') group by name,sysadmin,role order by sysadmin desc, role asc;").fetchall(); return res + + @classmethod + def recent_datasets(cls): + activity = table('activity') + package = table('package') + s = select([func.max(activity.c.timestamp),package.c.id, activity.c.activity_type], from_obj=[activity.join(package,activity.c.object_id == package.c.id)]).where(package.c.private == 'f').\ + where(activity.c.timestamp > func.now() - text("interval '60 day'")).group_by(package.c.id,activity.c.activity_type).order_by(func.max(activity.c.timestamp)) + result = model.Session.execute(s).fetchall() + return [(datetime2date(timestamp), model.Session.query(model.Package).get(unicode(package_id)), activity_type) for timestamp,package_id,activity_type in result] + class RevisionStats(object): @@ -211,7 +238,7 @@ package_revision = table('package_revision') revision = table('revision') package = table('package') - s = select([package_revision.c.id, func.min(revision.c.timestamp)], from_obj=[package_revision.join(revision)]).\ + s = select([package_revision.c.id, func.min(revision.c.timestamp)], from_obj=[package_revision.join(revision).join(package)]).\ where(package.c.private == 'f').\ group_by(package_revision.c.id).order_by(func.min(revision.c.timestamp)) res = model.Session.execute(s).fetchall() # [(id, datetime), ...] @@ -239,8 +266,10 @@ # be 'for all time' else you get first revision in the time period. package_revision = table('package_revision') revision = table('revision') - s = select([package_revision.c.id, func.min(revision.c.timestamp)], from_obj=[package_revision.join(revision)]).\ + package = table('package') + s = select([package_revision.c.id, func.min(revision.c.timestamp)], from_obj=[package_revision.join(revision).join(package)]).\ where(package_revision.c.state==model.State.DELETED).\ + where(package.c.private == 'f').\ group_by(package_revision.c.id).\ order_by(func.min(revision.c.timestamp)) res = model.Session.execute(s).fetchall() # [(id, datetime), ...]