additional overview stats
--- a/ckanext/dga_stats/controller.py
+++ b/ckanext/dga_stats/controller.py
@@ -18,6 +18,7 @@
c.activity_counts = stats.activity_counts()
c.by_org = stats.by_org()
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')
c.deleted_packages_by_week = rev_stats.get_by_week('deleted_packages')
c.num_packages_by_week = rev_stats.get_num_packages_by_week()
--- 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 +68,13 @@
@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
@@ -107,10 +108,9 @@
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 +124,11 @@
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 +143,16 @@
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 +224,7 @@
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 +252,10 @@
# 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), ...]
--- a/ckanext/dga_stats/templates/ckanext/stats/index.html
+++ b/ckanext/dga_stats/templates/ckanext/stats/index.html
@@ -6,30 +6,8 @@
{% block primary_content %}
<article class="module">
- <section id="summary" class="module-content tab-content">
- <h2>{{ _('Summary') }}</h2>
- {% if c.summary_stats %}
- <table class="table table-chunky table-bordered table-striped">
- <thead>
- <tr>
- <th>{{ _('Measure') }}</th>
- <th class="metric">{{ _('Value') }}</th>
- </tr>
- </thead>
- <tbody>
- {% for measure,value in c.summary_stats %}
- <tr>
- <td>{{measure}}</td>
- <td class="metric">{{ value }}</td>
- </tr>
- {% endfor %}
- </tbody>
- </table>
- {% else %}
- <p class="empty">{{ _('No groups') }}</p>
- {% endif %}
- </section>
- <section id="activity-counts" class="module-content tab-content">
+{% if h.check_access('sysadmin') %}
+ <section id="stats-activity-counts" class="module-content tab-content">
<h2>{{ _('Site Activity Log') }}</h2>
{% if c.activity_counts %}
<table class="table table-chunky table-bordered table-striped">
@@ -54,8 +32,32 @@
<p class="empty">{{ _('No groups') }}</p>
{% endif %}
</section>
-{% if h.check_access('sysadmin') %}
- <section id="user-access-list" class="module-content tab-content">
+ <section id="stats-recent-datasets" class="module-content tab-content">
+ <h2>{{ _('Recent Datasets') }}</h2>
+ {% if c.recent_datasets %}
+ <table class="table table-chunky table-bordered table-striped">
+ <thead>
+ <tr>
+ <th>{{ _('Date') }}</th>
+ <th>{{ _('Dataset') }}</th>
+ <th>{{ _('New/Modified') }}</th>
+ </tr>
+ </thead>
+ <tbody>
+ {% for date,package,newmodified in c.recent_datasets %}
+ <tr>
+ <td>{{ date }}</td>
+ <td>{{ h.link_to(package.title or package.name, h.url_for(controller='package', action='read', id=package.name)) }}</td>
+ <td>{{ newmodified }}</td>
+ </tr>
+ {% endfor %}
+ </tbody>
+ </table>
+ {% else %}
+ <p class="empty">{{ _('No groups') }}</p>
+ {% endif %}
+ </section>
+ <section id="stats-user-access-list" class="module-content tab-content">
<h2>{{ _('User Access List') }}</h2>
{% if c.user_access_list %}
<table class="table table-chunky table-bordered table-striped">
@@ -129,31 +131,6 @@
</table>
</section>
- <section id="stats-top-rated" class="module-content tab-content">
- <h2>{{ _('Top Rated Datasets') }}</h2>
- {% if c.top_rated_packages %}
- <table class="table table-chunky table-bordered table-striped">
- <thead>
- <tr>
- <th>Dataset</th>
- <th class="metric">{{ _('Average rating') }}</th>
- <th class="metric">{{ _('Number of ratings') }}</th>
- </tr>
- </thead>
- <tbody>
- {% for package, rating, num_ratings in c.top_rated_packages %}
- <tr>
- <th>{{ h.link_to(package.title or package.name, h.url_for(controller='package', action='read', id=package.name)) }}</th>
- <td class="metric">{{ rating }}</td>
- <td class="metric">{{ num_ratings }}</td>
- </tr>
- {% endfor %}
- </tbody>
- </table>
- {% else %}
- <p class="empty">{{ _('No ratings') }}</p>
- {% endif %}
- </section>
<section id="stats-most-edited" class="module-content tab-content">
<h2>{{ _('Most Edited Datasets') }}</h2>
@@ -202,46 +179,6 @@
<p class="empty">{{ _('No groups') }}</p>
{% endif %}
</section>
-
- <section id="stats-top-tags" class="module-content tab-content">
- <h2>{{ _('Top Tags') }}</h2>
- <table class="table table-chunky table-bordered table-striped">
- <thead>
- <tr>
- <th>{{ _('Tag Name') }}</th>
- <th class="metric">{{ _('Number of Datasets') }}</th>
- </tr>
- </thead>
- <tbody>
- {% for tag, num_packages in c.top_tags %}
- <tr>
- <td>{{ h.link_to(tag.name, h.url_for(controller='package', action='search', tags=tag.name)) }}</td>
- <td class="metric">{{ num_packages }}</td>
- </tr>
- {% endfor %}
- </tbody>
- </table>
- </section>
-
- <section id="stats-most-owned" class="module-content tab-content">
- <h2>{{ _('Users Owning Most Datasets') }}</h2>
- <table class="table table-chunky table-bordered table-striped">
- <thead>
- <tr>
- <th>{{ _('User') }}</th>
- <th class="metric">{{ _('Number of Datasets') }}</th>
- </tr>
- </thead>
- <tbody>
- {% for user, num_packages in c.top_package_owners %}
- <tr>
- <td class="media">{{ h.linked_user(user) }}</td>
- <td class="metric">{{ num_packages }}</td>
- </tr>
- {% endfor %}
- </tbody>
- </table>
- </section>
<section id="stats-by-org" class="module-content tab-content">
<h2>{{ _('Datasets by Organization') }}</h2>
{% if c.by_org %}
@@ -255,6 +192,7 @@
</thead>
<tbody>
{% for group,private, num_packages in c.by_org %}
+ {% if private == False or h.check_access('sysadmin') %}
<tr>
<td>{{ h.link_to(group.title or group.name, h.url_for(controller='organization', action='read', id=group.name)) }}</td>
{% if private == True %}
@@ -264,14 +202,39 @@
{% endif %}
<td class="metric">{{ num_packages }}</td>
</tr>
- {% endfor %}
- </tbody>
- </table>
- {% else %}
- <p class="empty">{{ _('No groups') }}</p>
- {% endif %}
- </section>
-
+ {% endif %}
+ {% endfor %}
+ </tbody>
+ </table>
+ {% else %}
+ <p class="empty">{{ _('No groups') }}</p>
+ {% endif %}
+ </section>
+ <section id="stats-summary" class="module-content tab-content">
+ <h2>{{ _('Summary') }}</h2>
+ {% if c.summary_stats %}
+ <table class="table table-chunky table-bordered table-striped">
+ <thead>
+ <tr>
+ <th>{{ _('Measure') }}</th>
+ <th class="metric">{{ _('Value') }}</th>
+ </tr>
+ </thead>
+ <tbody>
+ {% for measure,value in c.summary_stats %}
+ {% if 'Archived' not in measure or h.check_access('sysadmin') %}
+ <tr>
+ <td>{{measure}}</td>
+ <td class="metric">{{ value }}</td>
+ </tr>
+ {% endif %}
+ {% endfor %}
+ </tbody>
+ </table>
+ {% else %}
+ <p class="empty">{{ _('No groups') }}</p>
+ {% endif %}
+ </section>
</article>
{% endblock %}
@@ -280,18 +243,15 @@
<h2 class="module-heading"><i class="icon-bar-chart icon-medium"></i> {{ _('Statistics Menu') }}</h2>
<nav data-module="stats-nav">
<ul class="unstyled nav nav-simple">
- <li class="nav-item active"><a href="#summary" data-toggle="tab">{{ _('Summary') }}</a></li>
- <li class="nav-item"><a href="#activity-counts" data-toggle="tab">{{ _('Site Activity Log') }}</a></li>
{% if h.check_access('sysadmin') %}
- <li class="nav-item"><a href="#user-access-list" data-toggle="tab">{{ _('User Access List') }}</a></li>
+ <li class="nav-item"><a href="#stats-recent-datasets" data-toggle="tab">{{ _('Recent Datasets') }}</a></li>
+ <li class="nav-item"><a href="#stats-user-access-list" data-toggle="tab">{{ _('User Access List') }}</a></li>
{% endif %}
<li class="nav-item"><a href="#stats-total-datasets" data-toggle="tab">{{ _('Total Number of Datasets') }}</a></li>
<li class="nav-item"><a href="#stats-dataset-revisions" data-toggle="tab">{{ _('Dataset Revisions per Week') }}</a></li>
-<!-- <li class="nav-item"><a href="#stats-top-rated" data-toggle="tab">{{ _('Top Rated Datasets') }}</a></li> -->
<li class="nav-item"><a href="#stats-most-edited" data-toggle="tab">{{ _('Most Edited Datasets') }}</a></li>
- <li class="nav-item"><a href="#stats-top-tags" data-toggle="tab">{{ _('Top Tags') }}</a></li>
- <li class="nav-item"><a href="#stats-most-owned" data-toggle="tab">{{ _('Users Owning Most 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-by-org" data-toggle="tab">{{ _('Datasets by Organization') }}</a></li>
+ <li class="nav-item active"><a href="#stats-summary" data-toggle="tab">{{ _('Summary') }}</a></li>
</ul>
</nav>
</section>