import logging | import logging |
import operator | import operator |
import ckan.lib.base as base | import ckan.lib.base as base |
import ckan.model as model | import ckan.model as model |
from ckan.logic import get_action | from ckan.logic import get_action |
from ckanext.ga_report.ga_model import GA_Url, GA_Publisher | from ckanext.ga_report.ga_model import GA_Url, GA_Publisher |
from ckanext.ga_report.controller import _get_publishers | from ckanext.ga_report.controller import _get_publishers |
_log = logging.getLogger(__name__) | _log = logging.getLogger(__name__) |
def popular_datasets(count=10): | def popular_datasets(count=10): |
import random | import random |
publisher = None | publisher = None |
publishers = _get_publishers(30) | publishers = _get_publishers(30) |
total = len(publishers) | total = len(publishers) |
while not publisher or not datasets: | while not publisher or not datasets: |
rand = random.randrange(0, total) | rand = random.randrange(0, total) |
publisher = publishers[rand][0] | publisher = publishers[rand][0] |
if not publisher.state == 'active': | if not publisher.state == 'active': |
publisher = None | publisher = None |
continue | continue |
datasets = _datasets_for_publisher(publisher, 10)[:count] | datasets = _datasets_for_publisher(publisher, 10)[:count] |
ctx = { | ctx = { |
'datasets': datasets, | 'datasets': datasets, |
'publisher': publisher | 'publisher': publisher |
} | } |
return base.render_snippet('ga_report/ga_popular_datasets.html', **ctx) | return base.render_snippet('ga_report/ga_popular_datasets.html', **ctx) |
def single_popular_dataset(top=20): | def single_popular_dataset(top=20): |
'''Returns a random dataset from the most popular ones. | '''Returns a random dataset from the most popular ones. |
:param top: the number of top datasets to select from | :param top: the number of top datasets to select from |
''' | ''' |
import random | import random |
top_datasets = model.Session.query(GA_Url).\ | top_datasets = model.Session.query(GA_Url).\ |
filter(GA_Url.url.like('/dataset/%')).\ | filter(GA_Url.url.like('/dataset/%')).\ |
order_by('ga_url.pageviews::int desc') | order_by('ga_url.pageviews::int desc') |
num_top_datasets = top_datasets.count() | num_top_datasets = top_datasets.count() |
dataset = None | dataset = None |
if num_top_datasets: | if num_top_datasets: |
count = 0 | count = 0 |
while not dataset: | while not dataset: |
rand = random.randrange(0, min(top, num_top_datasets)) | rand = random.randrange(0, min(top, num_top_datasets)) |
ga_url = top_datasets[rand] | ga_url = top_datasets[rand] |
dataset = model.Package.get(ga_url.url[len('/dataset/'):]) | dataset = model.Package.get(ga_url.url[len('/dataset/'):]) |
if dataset and not dataset.state == 'active': | if dataset and not dataset.state == 'active': |
dataset = None | dataset = None |
count += 1 | # When testing, it is possible that top datasets are not available |
if count > 10: | # so only go round this loop a few times before falling back on |
break | # a random dataset. |
count += 1 | |
if count > 10: | |
break | |
if not dataset: | if not dataset: |
# fallback | # fallback |
dataset = model.Session.query(model.Package)\ | dataset = model.Session.query(model.Package)\ |
.filter_by(state='active').first() | .filter_by(state='active').first() |
if not dataset: | if not dataset: |
return None | return None |
dataset_dict = get_action('package_show')({'model': model, | dataset_dict = get_action('package_show')({'model': model, |
'session': model.Session, | 'session': model.Session, |
'validate': False}, | 'validate': False}, |
{'id':dataset.id}) | {'id':dataset.id}) |
return dataset_dict | return dataset_dict |
def single_popular_dataset_html(top=20): | def single_popular_dataset_html(top=20): |
dataset_dict = single_popular_dataset(top) | dataset_dict = single_popular_dataset(top) |
groups = package.get('groups', []) | 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') == 'publisher' ] |
publisher = publishers[0] if publishers else {'name':'', 'title': ''} | publisher = publishers[0] if publishers else {'name':'', 'title': ''} |
context = { | context = { |
'dataset': dataset_dict, | 'dataset': dataset_dict, |
'publisher': publisher_dict | 'publisher': publisher_dict |
} | } |
return base.render_snippet('ga_report/ga_popular_single.html', **context) | 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): |
if not publisher: | if not publisher: |
_log.error("No valid publisher passed to 'most_popular_datasets'") | _log.error("No valid publisher passed to 'most_popular_datasets'") |
return "" | return "" |
results = _datasets_for_publisher(publisher, count) | results = _datasets_for_publisher(publisher, count) |
ctx = { | ctx = { |
'dataset_count': len(results), | 'dataset_count': len(results), |
'datasets': results, | 'datasets': results, |
'publisher': publisher | 'publisher': publisher |
} | } |
return base.render_snippet('ga_report/publisher/popular.html', **ctx) | return base.render_snippet('ga_report/publisher/popular.html', **ctx) |
def _datasets_for_publisher(publisher, count): | def _datasets_for_publisher(publisher, count): |
datasets = {} | datasets = {} |
entries = model.Session.query(GA_Url).\ | entries = model.Session.query(GA_Url).\ |
filter(GA_Url.department_id==publisher.name).\ | filter(GA_Url.department_id==publisher.name).\ |
filter(GA_Url.url.like('/dataset/%')).\ | filter(GA_Url.url.like('/dataset/%')).\ |
order_by('ga_url.pageviews::int desc').all() | order_by('ga_url.pageviews::int desc').all() |
for entry in entries: | for entry in entries: |
if len(datasets) < count: | if len(datasets) < count: |
p = model.Package.get(entry.url[len('/dataset/'):]) | p = model.Package.get(entry.url[len('/dataset/'):]) |
if not p in datasets: | if not p in datasets: |
datasets[p] = {'views':0, 'visits': 0} | datasets[p] = {'views':0, 'visits': 0} |
datasets[p]['views'] = datasets[p]['views'] + int(entry.pageviews) | datasets[p]['views'] = datasets[p]['views'] + int(entry.pageviews) |
datasets[p]['visits'] = datasets[p]['visits'] + int(entry.visits) | datasets[p]['visits'] = datasets[p]['visits'] + int(entry.visits) |
results = [] | results = [] |
for k, v in datasets.iteritems(): | for k, v in datasets.iteritems(): |
results.append((k,v['views'],v['visits'])) | results.append((k,v['views'],v['visits'])) |
return sorted(results, key=operator.itemgetter(1), reverse=True) | return sorted(results, key=operator.itemgetter(1), reverse=True) |
<html xmlns:py="http://genshi.edgewall.org/" | <html xmlns:py="http://genshi.edgewall.org/" |
xmlns:i18n="http://genshi.edgewall.org/i18n" | xmlns:i18n="http://genshi.edgewall.org/i18n" |
xmlns:xi="http://www.w3.org/2001/XInclude" | xmlns:xi="http://www.w3.org/2001/XInclude" |
py:strip=""> | py:strip=""> |
<li class="widget-container boxed widget_text"> | <li class="widget-container boxed widget_text"> |
<h4>Notes</h4> | <h4>Notes</h4> |
<hr/> | |
<ul> | <ul> |
<li>"Views" is the number of times a page was loaded in users' browsers.</li> | <li>"Views" is the number of times a page was loaded in users' browsers.</li> |
<li>"Visits" is the number of unique user visits to a page, counted once for each visitor for each of their browsing sessions.</li> | <li>"Visits" is the number of unique user visits to a page, counted once for each visitor for each of their browsing sessions.</li> |
<li>These usage statistics are confined to users with javascript enabled, which excludes web crawlers and API calls.</li> | <li>These usage statistics are confined to users with javascript enabled, which excludes web crawlers and API calls.</li> |
<li>The results are not shown when the number of views/visits is tiny. Where these relate to site pages, results are available in full in the CSV download. Where these relate to users' web browser information, results are not disclosed, for privacy reasons.</li> | <li>The results are not shown when the number of views/visits is tiny. Where these relate to site pages, results are available in full in the CSV download. Where these relate to users' web browser information, results are not disclosed, for privacy reasons.</li> |
</ul> | </ul> |
</li> | </li> |
</html> | </html> |
<html xmlns:py="http://genshi.edgewall.org/" | <html xmlns:py="http://genshi.edgewall.org/" |
xmlns:i18n="http://genshi.edgewall.org/i18n" | xmlns:i18n="http://genshi.edgewall.org/i18n" |
xmlns:xi="http://www.w3.org/2001/XInclude" | xmlns:xi="http://www.w3.org/2001/XInclude" |
py:strip=""> | py:strip=""> |
<xi:include href="../ga_util.html" /> | <xi:include href="../ga_util.html" /> |
<py:def function="page_title">Usage by Publisher</py:def> | <py:def function="page_title">Usage by Publisher</py:def> |
<py:match path="primarysidebar"> | <py:match path="primarysidebar"> |
<li class="widget-container boxed widget_text"> | <li class="widget-container boxed widget_text"> |
<h4>Download</h4> | <h4>Download</h4> |
<hr/> | |
<p><center> | <p><center> |
<a class="btn button btn-primary" href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='publisher_csv',month=c.month or 'all')}">Download as CSV</a></center> | <a class="btn button btn-primary" href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='publisher_csv',month=c.month or 'all')}">Download as CSV</a></center> |
</p> | </p> |
</li> | </li> |
<xi:include href="../notes.html" /> | <xi:include href="../notes.html" /> |
</py:match> | </py:match> |
<div py:match="content"> | <div py:match="content"> |
<h1>Site Usage</h1> | <h1>Site Usage</h1> |
${usage_nav('Publishers')} | ${usage_nav('Publishers')} |
<form class="form-inline" action="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='publishers')}" method="get"> | <form class="form-inline" action="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='publishers')}" method="get"> |
<div class="controls"> | <div class="controls"> |
${month_selector(c.month, c.months, c.day)} | ${month_selector(c.month, c.months, c.day)} |
<input class="btn button btn-primary" type='submit' value="Update"/> | <input class="btn button btn-primary" type='submit' value="Update"/> |
</div> | </div> |
</form> | </form> |
<table class="table table-condensed table-bordered table-striped"> | <table class="table table-condensed table-bordered table-striped"> |
<tr> | <tr> |
<th>Publisher</th> | <th>Publisher</th> |
<!-- <th>Dataset Visits</th>--> | <!-- <th>Dataset Visits</th>--> |
<th>Dataset Views</th> | <th>Dataset Views</th> |
</tr> | </tr> |
<py:for each="publisher, views, visits in c.top_publishers"> | <py:for each="publisher, views, visits in c.top_publishers"> |
<tr> | <tr> |
<td>${h.link_to(publisher.title, h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport', action='read_publisher', id=publisher.name) + ("?month=" + c.month) if c.month else '')} | <td>${h.link_to(publisher.title, h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport', action='read_publisher', id=publisher.name) + ("?month=" + c.month) if c.month else '')} |
</td> | </td> |
<!-- <td>${visits}</td> --> | <!-- <td>${visits}</td> --> |
<td>${views}</td> | <td>${views}</td> |
</tr> | </tr> |
</py:for> | </py:for> |
</table> | </table> |
</div> | </div> |
<xi:include href="../../layout.html" /> | <xi:include href="../../layout.html" /> |
<py:def function="optional_footer"> | <py:def function="optional_footer"> |
<script type='text/javascript'> | <script type='text/javascript'> |
$('.nav-tabs li a').click(function (e) { | $('.nav-tabs li a').click(function (e) { |
e.preventDefault(); | e.preventDefault(); |
$(this).tab('show'); | $(this).tab('show'); |
}) | }) |
</script> | </script> |
</py:def> | </py:def> |
</html> | </html> |
<html xmlns:py="http://genshi.edgewall.org/" | <html xmlns:py="http://genshi.edgewall.org/" |
xmlns:i18n="http://genshi.edgewall.org/i18n" | xmlns:i18n="http://genshi.edgewall.org/i18n" |
xmlns:xi="http://www.w3.org/2001/XInclude" | xmlns:xi="http://www.w3.org/2001/XInclude" |
py:strip=""> | py:strip=""> |
<xi:include href="../ga_util.html" /> | <xi:include href="../ga_util.html" /> |
<py:def function="page_title">Usage by Dataset</py:def> | <py:def function="page_title">Usage by Dataset</py:def> |
<py:match path="primarysidebar"> | <py:match path="primarysidebar"> |
<li class="widget-container boxed widget_text"> | <li class="widget-container boxed widget_text"> |
<h4>Download</h4> | <h4>Download</h4> |
<hr/> | |
<p><center> | <p><center> |
<a class="btn button btn-primary" href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='dataset_csv',id=c.publisher_name or 'all',month=c.month or 'all')}">Download as CSV</a></center> | <a class="btn button btn-primary" href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='dataset_csv',id=c.publisher_name or 'all',month=c.month or 'all')}">Download as CSV</a></center> |
</p> | </p> |
</li> | </li> |
<xi:include href="../notes.html" /> | <xi:include href="../notes.html" /> |
</py:match> | </py:match> |
<div py:match="content"> | <div py:match="content"> |
<h1>Site Usage</h1> | <h1>Site Usage</h1> |
${usage_nav('Datasets')} | ${usage_nav('Datasets')} |
<form class="form-inline" action="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='read')}" method="get"> | <form class="form-inline" action="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='read')}" method="get"> |
<div class="controls"> | <div class="controls"> |
${month_selector(c.month, c.months, c.day)} | ${month_selector(c.month, c.months, c.day)} |
<select name="publisher"> | <select name="publisher"> |
<option value='' py:attrs="{'selected': 'selected' if not c.publisher else None}">All publishers</option> | <option value='' py:attrs="{'selected': 'selected' if not c.publisher else None}">All publishers</option> |
<py:for each="val,desc in c.publishers"> | <py:for each="val,desc in c.publishers"> |
<option value='${val}' py:attrs="{'selected': 'selected' if c.publisher_name == val else None}">${desc}</option> | <option value='${val}' py:attrs="{'selected': 'selected' if c.publisher_name == val else None}">${desc}</option> |
</py:for> | </py:for> |
</select> | </select> |
<input class="btn button btn-primary" type='submit' value="Update"/> | <input class="btn button btn-primary" type='submit' value="Update"/> |
</div> | </div> |
</form> | </form> |
<h3 py:if="c.publisher"><a href="${h.url_for(controller='ckanext.dgu.controllers.publisher:PublisherController',action='read',id=c.publisher.name)}">${c.publisher.title}</a></h3> | <h3 py:if="c.publisher"><a href="${h.url_for(controller='ckanext.dgu.controllers.publisher:PublisherController',action='read',id=c.publisher.name)}">${c.publisher.title}</a></h3> |
<p py:if="not c.top_packages">No page views in this period</p> | <p py:if="not c.top_packages">No page views in this period</p> |
<table py:if="c.top_packages" class="table table-condensed table-bordered table-striped"> | <table py:if="c.top_packages" class="table table-condensed table-bordered table-striped"> |
<tr> | <tr> |
<th>Dataset</th> | <th>Dataset</th> |
<!-- <th>Visits</th> --> | <!-- <th>Visits</th> --> |
<th>Views</th> | <th>Views</th> |
</tr> | </tr> |
<py:for each="package, views, visits in c.top_packages"> | <py:for each="package, views, visits in c.top_packages"> |
<tr> | <tr> |
<td>${h.link_to(package.title or package.name, h.url_for(controller='package', action='read', id=package.name))} | <td>${h.link_to(package.title or package.name, h.url_for(controller='package', action='read', id=package.name))} |
</td> | </td> |
<!-- <td>${visits}</td> --> | <!-- <td>${visits}</td> --> |
<td>${views}</td> | <td>${views}</td> |
</tr> | </tr> |
</py:for> | </py:for> |
</table> | </table> |
</div> | </div> |
<xi:include href="../../layout.html" /> | <xi:include href="../../layout.html" /> |
</html> | </html> |
<html xmlns:py="http://genshi.edgewall.org/" | <html xmlns:py="http://genshi.edgewall.org/" |
xmlns:i18n="http://genshi.edgewall.org/i18n" | xmlns:i18n="http://genshi.edgewall.org/i18n" |
xmlns:xi="http://www.w3.org/2001/XInclude" | xmlns:xi="http://www.w3.org/2001/XInclude" |
py:strip=""> | py:strip=""> |
<xi:include href="../ga_util.html" /> | <xi:include href="../ga_util.html" /> |
<py:def function="page_title">Site usage</py:def> | <py:def function="page_title">Site usage</py:def> |
<py:match path="primarysidebar"> | <py:match path="primarysidebar"> |
<li class="widget-container boxed widget_text"> | <li class="widget-container boxed widget_text"> |
<h4>Download</h4> | <h4>Download</h4> |
<hr/> | |
<p><center> | <p><center> |
<a class="btn button btn-primary" href="${h.url_for(controller='ckanext.ga_report.controller:GaReport',action='csv',month=c.month or 'all')}">Download as CSV</a></center> | <a class="btn button btn-primary" href="${h.url_for(controller='ckanext.ga_report.controller:GaReport',action='csv',month=c.month or 'all')}">Download as CSV</a></center> |
</p> | </p> |
</li> | </li> |
<xi:include href="../notes.html" /> | <xi:include href="../notes.html" /> |
</py:match> | </py:match> |
<div py:match="content"> | <div py:match="content"> |
<h1>Site Usage</h1> | <h1>Site Usage</h1> |
${usage_nav('Site-wide')} | ${usage_nav('Site-wide')} |
<form class="form-inline" action="${h.url_for(controller='ckanext.ga_report.controller:GaReport',action='index')}" method="get"> | <form class="form-inline" action="${h.url_for(controller='ckanext.ga_report.controller:GaReport',action='index')}" method="get"> |
<div class="controls"> | <div class="controls"> |
${month_selector(c.month, c.months, c.day)} | ${month_selector(c.month, c.months, c.day)} |
<input class="btn button btn-primary" type='submit' value="Update"/> | <input |