| 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 |
| # When testing, it is possible that top datasets are not available | # When testing, it is possible that top datasets are not available |
| # so only go round this loop a few times before falling back on | # so only go round this loop a few times before falling back on |
| # a random dataset. | # a random dataset. |
| count += 1 | count += 1 |
| if count > 10: | if count > 10: |
| break | 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, preview_image=None): | 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 |
| 'preview_image': preview_image | |
| } | } |
| 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: | if not p: |
| _log.warning("Could not find Package for {url}".format(url=entry.url)) | _log.warning("Could not find Package for {url}".format(url=entry.url)) |
| continue | continue |
| 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 | <html |
| xmlns="http://www.w3.org/1999/xhtml" | xmlns="http://www.w3.org/1999/xhtml" |
| xmlns:i18n="http://genshi.edgewall.org/i18n" | xmlns:i18n="http://genshi.edgewall.org/i18n" |
| xmlns:py="http://genshi.edgewall.org/" | xmlns:py="http://genshi.edgewall.org/" |
| xmlns:xi="http://www.w3.org/2001/XInclude" | xmlns:xi="http://www.w3.org/2001/XInclude" |
| py:strip="" | py:strip="" |
| > | > |
| <select name="month" py:def="month_selector(current_month, months, day)"> | <select name="month" py:def="month_selector(current_month, months, day)"> |
| <option value='' py:attrs="{'selected': 'selected' if not current_month else None}">All months</option> | <option value='' py:attrs="{'selected': 'selected' if not current_month else None}">All months</option> |
| <py:for each="i, (val,desc) in enumerate(months)"> | <py:for each="i, (val,desc) in enumerate(months)"> |
| <option value='${val}' py:attrs="{'selected': 'selected' if current_month == val else None}">${desc}<py:if test="i == 0 and day"> (up to ${day})</py:if></option> | <option value='${val}' py:attrs="{'selected': 'selected' if current_month == val else None}">${desc}<py:if test="i == 0 and day"> (up to ${day})</py:if></option> |
| </py:for> | </py:for> |
| </select> | </select> |
| <table py:def="social_table(items, with_source=False)" class="ga-reports-table table table-condensed table-bordered table-striped"> | <table py:def="social_table(items, with_source=False)" class="ga-reports-table table table-condensed table-bordered table-striped"> |
| <tr> | <tr> |
| <th>Name</th> | <th>Name</th> |
| <th py:if="with_source">Source</th> | <th py:if="with_source">Source</th> |
| <th>Visits</th> | <th>Visits</th> |
| </tr> | </tr> |
| <py:for each="name, url, source, count in items"> | <py:for each="name, url, source, count in items"> |
| <tr> | <tr> |
| <td><a href="${url}">${name}</a></td> | <td><a href="${url}">${name}</a></td> |
| <td py:if="with_source">${source}</td> | <td py:if="with_source">${source}</td> |
| <td class="td-numeric">${count}</td> | <td class="td-numeric">${count}</td> |
| </tr> | </tr> |
| </py:for> | </py:for> |
| </table> | </table> |
| <div py:def="rickshaw_graph(items_json,id,mode='line',colorscheme='munin')"> | <div py:def="rickshaw_graph(items_json,id,mode='line',colorscheme='munin')"> |
| <div id="chart_container_$id" class="rickshaw_chart_container"> | <div id="chart_container_$id" class="rickshaw_chart_container"> |
| <div id="y_axis_$id" class="rickshaw_y_axis"></div> | <div id="y_axis_$id" class="rickshaw_y_axis"></div> |
| <div id="chart_$id" class="rickshaw_chart"></div> | <div id="chart_$id" class="rickshaw_chart"></div> |
| <script type="text/javascript"> | <script type="text/javascript"> |
| $(function() { | $(function() { |
| var items = $items_json; | var items = $items_json; |
| if (items.length==0) { return; } | if (items.length==0) { return; } |
| CKAN.GA_Reports.render_rickshaw('$id', items, '$mode', '$colorscheme'); | CKAN.GA_Reports.render_rickshaw('$id', items, '$mode', '$colorscheme'); |
| }); | }); |
| </script> | </script> |
| </div> | </div> |
| </div> | </div> |
| <table py:def="stat_table(items, title='Views')" class="ga-reports-table table table-condensed table-bordered table-striped"> | <table py:def="stat_table(items, title='Views')" class="ga-reports-table table table-condensed table-bordered table-striped"> |
| <tr> | <tr> |
| <th>Name</th> | <th>Name</th> |
| <th class="td-numeric">% ${title}</th> | <th class="td-numeric">% ${title}</th> |
| </tr> | </tr> |
| <py:for each="name, value in items"> | <py:for each="name, value in items"> |
| <tr> | <tr> |
| <td>${name}</td> | <td>${name}</td> |
| <td class="td-numeric">${value}</td> | <td class="td-numeric">${value}</td> |
| </tr> | </tr> |
| </py:for> | </py:for> |
| </table> | </table> |
| <div py:def="usage_nav(active_name)"> | <div py:def="usage_nav(active_name)" id="minornavigation"> |
| <span class="subheading"> | <div id="minornavigation-bg-left"> |
| <img src="/images/arrow-down-right.png" /> | <div id="minornavigation-bg-right"> |
| <a class="${if_(active_name=='Site-wide','active')}" href="${h.url_for(controller='ckanext.ga_report.controller:GaReport',action='index')}${if_(c.month,'?month='+c.month)}">Site-wide</a> | <ul class="nav nav-pills"> |
| <span class="divider">|</span> | <?python querystring='?month='+c.month if c.month else '' ?> |
| <a class="${if_(active_name=='Publishers','active')}" href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='publishers')}${if_(c.month,'?month='+c.month)}">Publishers</a> | <li py:attrs="{'class': 'active' if active_name=='Site-wide' else None}"><a py:attrs="{'class': 'active' if active_name=='Site-wide' else None}" href="${h.url_for(controller='ckanext.ga_report.controller:GaReport',action='index')}${querystring}"><img src="/images/icons/page_white.png" height="16px" width="16px" alt="None" class="inline-icon "/> Site-wide</a></li> |
| <span class="divider">|</span> | <li py:attrs="{'class': 'active' if active_name=='Publishers' else None}"> |
| <a class="${if_(active_name=='Datasets','active')}" href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='read')}${if_(c.month,'?month='+c.month)}">Datasets</a> | <a py:attrs="{'class': 'active' if active_name=='Publishers' else None}" href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='publishers')}${querystring}"><img src="/images/icons/page_white.png" height="16px" width="16px" alt="None" class="inline-icon "/> Publishers</a> |
| </span> | </li> |
| </div> | <li py:attrs="{'class': 'active' if active_name=='Datasets' else None}"> |
| <a py:attrs="{'class': 'active' if active_name=='Datasets' else None}" href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='read')}${querystring}"><img src="/images/icons/page_white.png" height="16px" width="16px" alt="None" class="inline-icon "/> Datasets</a> | |
| <div py:def="ga_sidebar(download_link)"> | </li> |
| <div class="boxed"> | |
| <div class="widget-container widget_text"> | |
| <h4>Download</h4> | |
| <p><center> | |
| <a class="btn button btn-primary" href="${download_link}">Download as CSV</a></center> | |
| </p> | |
| </div> | |
| <div class="widget-container widget_text"> | |
| <h4>Graph Legend</h4> | |
| <div id="graph-legend-container"> | |
| <div style="display: none;" id="legend_none">(No graph is loaded)</div> | |
| </div> | |
| </div> | |
| <div class="widget-container widget_text"> | |
| <h4>Notes</h4> | |
| <ul> | |
| <li>"Views" is the number of times a page was loaded in users' browsers.</li> | |
| <li>"Downloads" is the number of times a user has clicked to download either an original or cached resource for a particular dataset. Download information is only available from 2nd December 2012; 'No data' is shown for records before that date.</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> | |
| </ul> | </ul> |
| </div> | </div> |
| </div> | </div> |
| </div> | </div> |
| </html> | </html> |
| <html xmlns:py="http://genshi.edgewall.org/" | |
| xmlns:i18n="http://genshi.edgewall.org/i18n" | |
| xmlns:xi="http://www.w3.org/2001/XInclude" | |
| py:strip=""> | |
| <li class="widget-container boxed widget_text"> | |
| <h4>Notes</h4> | |
| <ul> | |
| <li>"Views" is the number of times a page was loaded in users' browsers.</li> | |
| <li>"Downloads" is the number of times a user has clicked to download either an original or cached resource for a particular dataset. Download information is only available from 2nd December 2012; 'No data' is shown for records before that date.</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> | |
| </ul> | |
| </li> | |
| </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"> |
| ${ga_sidebar(download_link=h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='publisher_csv',month=c.month or 'all'))} | <li class="widget-container boxed widget_text"> |
| <h4>Download</h4> | |
| <p><center> | |
| <a class="btn button btn-primary" href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='publisher_csv',month=c.month or 'all')}">Download as CSV</a></center> | |
| </p> | |
| </li> | |
| <li class="widget-container boxed widget_text"> | |
| <h4>Graph Legend</h4> | |
| <div id="graph-legend-container"> | |
| </div> | |
| </li> | |
| <xi:include href="../notes.html" /> | |
| </py:match> | </py:match> |
| <py:def function="optional_head"> | <py:def function="optional_head"> |
| <link rel="stylesheet" type="text/css" href="/scripts/vendor/rickshaw.min.css"/> | <link rel="stylesheet" type="text/css" href="/scripts/vendor/rickshaw.min.css"/> |
| <link rel="stylesheet" type="text/css" href="/css/ga_report.css?1"/> | <link rel="stylesheet" type="text/css" href="/css/ga_report.css?1"/> |
| <script type="text/javascript" src="/scripts/ckanext_ga_reports.js?1"></script> | <script type="text/javascript" src="/scripts/ckanext_ga_reports.js?1"></script> |
| <script type="text/javascript" src="/scripts/vendor/jquery.sparkline.modified.js"></script> | <script type="text/javascript" src="/scripts/vendor/jquery.sparkline.modified.js"></script> |
| <script type="text/javascript" src="/scripts/rickshaw_ie7_shim.js"></script> | <script type="text/javascript" src="/scripts/rickshaw_ie7_shim.js"></script> |
| <script type="text/javascript" src="/scripts/vendor/d3.v2.js"></script> | <script type="text/javascript" src="/scripts/vendor/d3.v2.js"></script> |
| <script type="text/javascript" src="/scripts/vendor/d3.layout.min.js"></script> | <script type="text/javascript" src="/scripts/vendor/d3.layout.min.js"></script> |
| <script type="text/javascript" src="/scripts/vendor/rickshaw.min.js"></script> | <script type="text/javascript" src="/scripts/vendor/rickshaw.min.js"></script> |
| </py:def> | </py:def> |
| <div py:match="content"> | <div py:match="content"> |
| <h1>Site Usage ${usage_nav('Publishers')}</h1> | <h1>Site Usage</h1> |
| <div class="boxed"> | ${usage_nav('Publishers')} |
| ${rickshaw_graph(c.top_publishers_graph,'publishers')} | ${rickshaw_graph(c.top_publishers_graph,'publishers')} |
| <form class="form-inline" action="${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"> |
| <label>Show stats table for:</label> | <label>Show stats table for:</label> |
| ${month_selector(c.month, c.months, c.day)} | ${month_selector(c.month, c.months, c.day)} |
| </div> | </div> |
| </form> | </form> |