From: root Date: Thu, 23 Jan 2014 11:15:48 +0000 Subject: remove top_tags stats for DB load X-Git-Url: https://maxious.lambdacomplex.org/git/?p=ckanext-dga-stats.git&a=commitdiff&h=5d9af40697a2166378beb90751247e1a4e8b9c51 --- remove top_tags stats for DB load --- --- a/ckanext/dga_stats/controller.py +++ b/ckanext/dga_stats/controller.py @@ -12,11 +12,12 @@ 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() c.by_org = stats.by_org() + c.res_by_org = stats.res_by_org() + c.top_active_orgs = stats.top_active_orgs() c.user_access_list = stats.user_access_list() c.recent_datasets = stats.recent_datasets() c.new_packages_by_week = rev_stats.get_by_week('new_packages') --- a/ckanext/dga_stats/stats.py +++ b/ckanext/dga_stats/stats.py @@ -6,6 +6,8 @@ import ckan.plugins as p import ckan.model as model + +import re cache_enabled = p.toolkit.asbool(config.get('ckanext.stats.cache_enabled', 'True')) @@ -78,30 +80,42 @@ 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 + def res_by_org(cls, limit=10): + connection = model.Session.connection() + reses = connection.execute("select owner_org,format,count(*) from \ + resource inner join resource_group on resource.resource_group_id = resource_group.id \ + inner join package on resource_group.package_id = package.id group by owner_org,format order by count desc;").fetchall(); + group_ids = [] + group_tab = {} + group_spatial = {} + group_other = {} + for group_id,format,count in reses: + if group_id not in group_ids: + group_ids.append(group_id) + group_tab[group_id] = 0 + group_spatial[group_id] = 0 + group_other[group_id] = 0 + if re.search('xls|csv|ms-excel|spreadsheetml.sheet|zip|netcdf',format, re.IGNORECASE): + group_tab[group_id] = group_tab[group_id] + count + elif re.search('wms|wfs|wcs|shp|kml|kmz',format, re.IGNORECASE): + group_spatial[group_id] = group_spatial[group_id] + count + else: + group_other[group_id] = group_other[group_id] + count + return [(model.Session.query(model.Group).get(unicode(group_id)), group_tab[group_id],group_spatial[group_id],group_other[group_id], group_tab[group_id]+group_spatial[group_id]+group_other[group_id]) for group_id in group_ids] + + @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 def top_package_owners(cls, limit=10): --- a/ckanext/dga_stats/templates/ckanext/stats/index.html +++ b/ckanext/dga_stats/templates/ckanext/stats/index.html @@ -185,7 +185,7 @@ - + @@ -203,6 +203,58 @@ {% endif %} + {% endfor %} + +
{{ _('Group') }}{{ _('Organisation') }} {{ _('Public/Archived') }} {{ _('Number of datasets') }}
{{ num_packages }}
+ {% else %} +

{{ _('No groups') }}

+ {% endif %} + +
+

{{ _('Resources by Organization') }}

+ {% if c.res_by_org %} + + + + + + + + + + + + {% for group,t,s,o,tot in c.res_by_org %} + + + + + + + + {% endfor %} + +
{{ _('Organisation') }}{{ _('Tabular') }}{{ _('Spatial') }}{{ _('Other') }}{{ _('Total') }}
{{ h.link_to(group.title or group.name, h.url_for(controller='organization', action='read', id=group.name)) }}{{ t }}{{ s }}{{ o }}{{ tot }}
+ {% else %} +

{{ _('No groups') }}

+ {% endif %} +
+
+

{{ _('Most Active Organisations') }}

+ {% if c.top_active_orgs %} + + + + + + + + + {% for group, num_packages in c.top_active_orgs %} + + + + {% endfor %}
{{ _('Organisation') }}{{ _('Number of datasets updated recently') }}
{{ h.link_to(group.title or group.name, h.url_for(controller='organization', action='read', id=group.name)) }}{{ num_packages }}
@@ -251,6 +303,8 @@
  • {{ _('Dataset Revisions per Week') }}
  • {{ _('Most Edited Datasets') }}
  • {{ _('Datasets by Organization') }}
    • +
  • {{ _('Resources by Organization') }}
    • +
  • {{ _('Most Active Organisations') }}
  • {{ _('Summary') }}