Handling duplicate urls in ga_url processing
--- a/README.rst
+++ b/README.rst
@@ -33,6 +33,10 @@
googleanalytics.id = UA-1010101-1
googleanalytics.account = Account name (e.g. data.gov.uk, see top level item at https://www.google.com/analytics)
ga-report.period = monthly
+ ga-report.bounce_url = /data
+
+ The ga-report.bounce_url specifies the path to use when calculating bounces. For DGU this is /data
+ but you may want to set this to /.
3. Set up this extension's database tables using a paster command. (Ensure your CKAN pyenv is still activated, run the command from ``src/ckanext-ga-report``, alter the ``--config`` option to point to your site config file)::
--- a/ckanext/ga_report/controller.py
+++ b/ckanext/ga_report/controller.py
@@ -9,7 +9,7 @@
import sqlalchemy
from sqlalchemy import func, cast, Integer
import ckan.model as model
-from ga_model import GA_Url, GA_Stat, GA_ReferralStat
+from ga_model import GA_Url, GA_Stat, GA_ReferralStat, GA_Publisher
log = logging.getLogger('ckanext.ga-report')
@@ -22,8 +22,9 @@
def _month_details(cls):
+ '''Returns a list of all the month names'''
months = []
- vals = model.Session.query(cls.period_name).distinct().all()
+ vals = model.Session.query(cls.period_name).filter(cls.period_name!='All').distinct().all()
for m in vals:
months.append( (m[0], _get_month_name(m[0])))
return sorted(months, key=operator.itemgetter(0), reverse=True)
@@ -70,15 +71,15 @@
entries = q.order_by('ga_stat.key').all()
def clean_key(key, val):
- if key in ['Average time on site', 'Pages per visit', 'New visits']:
+ if key in ['Average time on site', 'Pages per visit', 'New visits', 'Bounces']:
val = "%.2f" % round(float(val), 2)
if key == 'Average time on site':
mins, secs = divmod(float(val), 60)
hours, mins = divmod(mins, 60)
val = '%02d:%02d:%02d (%s seconds) ' % (hours, mins, secs, val)
- if key == 'New visits':
+ if key in ['New visits','Bounces']:
val = "%s%%" % val
- if key in ['Bounces', 'Total page views', 'Total visits']:
+ if key in ['Total page views', 'Total visits']:
val = int(val)
return key, val
@@ -93,11 +94,12 @@
for e in entries:
d[e.key].append(float(e.value))
for k, v in d.iteritems():
- if k in ['Bounces', 'Total page views', 'Total visits']:
+ if k in ['Total page views', 'Total visits']:
v = sum(v)
else:
v = float(sum(v))/len(v)
key, val = clean_key(k,v)
+
c.global_totals.append((key, val))
c.global_totals = sorted(c.global_totals, key=operator.itemgetter(0))
@@ -134,29 +136,7 @@
c.social_referrer_totals.append((shorten_name(entry[0]), fill_out_url(entry[0]),'',
entry[1]))
-
- browser_version_re = re.compile("(.*)\((.*)\)")
for k, v in keys.iteritems():
-
- def clean_field(key):
- if k != 'Browser versions':
- return key
- m = browser_version_re.match(key)
- browser = m.groups()[0].strip()
- ver = m.groups()[1]
- parts = ver.split('.')
- if len(parts) > 1:
- if parts[1][0] == '0':
- ver = parts[0]
- else:
- ver = "%s.%s" % (parts[0],parts[1])
- if browser in ['Safari','Android Browser']: # Special case complex version nums
- ver = parts[0]
- if len(ver) > 2:
- ver = "%s%sX" % (ver[0], ver[1])
-
- return "%s (%s)" % (browser, ver,)
-
q = model.Session.query(GA_Stat).\
filter(GA_Stat.stat_name==k)
if c.month:
@@ -172,17 +152,13 @@
entries.append((key,val,))
entries = sorted(entries, key=operator.itemgetter(1), reverse=True)
- def percent(num, total):
- p = 100 * float(num)/float(total)
- return "%.2f%%" % round(p, 2)
-
# Get the total for each set of values and then set the value as
# a percentage of the total
if k == 'Social sources':
total = sum([x for n,x in c.global_totals if n == 'Total visits'])
else:
total = sum([num for _,num in entries])
- setattr(c, v, [(k,percent(v,total)) for k,v in entries ])
+ setattr(c, v, [(k,_percent(v,total)) for k,v in entries ])
return render('ga_report/site/index.html')
@@ -253,7 +229,6 @@
c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month])
c.top_publishers = _get_top_publishers()
-
return render('ga_report/publisher/index.html')
def _get_packages(self, publisher=None, count=-1):
@@ -261,43 +236,22 @@
if count == -1:
count = sys.maxint
- q = model.Session.query(GA_Url)\
+ month = c.month or 'All'
+
+ q = model.Session.query(GA_Url,model.Package)\
+ .filter(model.Package.name==GA_Url.package_id)\
.filter(GA_Url.url.like('/dataset/%'))
if publisher:
q = q.filter(GA_Url.department_id==publisher.name)
- if c.month:
- q = q.filter(GA_Url.period_name==c.month)
+ q = q.filter(GA_Url.period_name==month)
q = q.order_by('ga_url.visitors::int desc')
-
- if c.month:
- top_packages = []
- for entry in q.limit(count):
- package_name = entry.url[len('/dataset/'):]
- p = model.Package.get(package_name)
- if p:
- top_packages.append((p, entry.pageviews, entry.visitors))
- else:
- log.warning('Could not find package "%s"', package_name)
- else:
- ds = {}
- for entry in q:
- if len(ds) >= count:
- break
- package_name = entry.url[len('/dataset/'):]
- p = model.Package.get(package_name)
- if p:
- if not p in ds:
- ds[p] = {'views': 0, 'visits': 0}
- ds[p]['views'] = ds[p]['views'] + int(entry.pageviews)
- ds[p]['visits'] = ds[p]['visits'] + int(entry.visitors)
- else:
- log.warning('Could not find package "%s"', package_name)
-
- results = []
- for k, v in ds.iteritems():
- results.append((k,v['views'],v['visits']))
-
- top_packages = sorted(results, key=operator.itemgetter(1), reverse=True)
+ top_packages = []
+ for entry,package in q.limit(count):
+ if package:
+ top_packages.append((package, entry.pageviews, entry.visitors))
+ else:
+ log.warning('Could not find package associated package')
+
return top_packages
def read(self):
@@ -333,15 +287,12 @@
else:
c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month])
+ month = c.month or 'All'
c.publisher_page_views = 0
q = model.Session.query(GA_Url).\
filter(GA_Url.url=='/publisher/%s' % c.publisher_name)
- if c.month:
- entry = q.filter(GA_Url.period_name==c.month).first()
- c.publisher_page_views = entry.pageviews if entry else 0
- else:
- for e in q.all():
- c.publisher_page_views = c.publisher_page_views + int(e.pageviews)
+ entry = q.filter(GA_Url.period_name==c.month).first()
+ c.publisher_page_views = entry.pageviews if entry else 0
c.top_packages = self._get_packages(c.publisher, 20)
@@ -352,33 +303,28 @@
Returns a list of the top 20 publishers by dataset visits.
(The number to show can be varied with 'limit')
'''
+ month = c.month or 'All'
connection = model.Session.connection()
q = """
select department_id, sum(pageviews::int) views, sum(visitors::int) visits
from ga_url
- where department_id <> ''"""
- if c.month:
- q = q + """
- and period_name=%s
- """
- q = q + """
- group by department_id order by visits desc
+ where department_id <> ''
+ and package_id <> ''
+ and url like '/dataset/%%'
+ and period_name=%s
+ group by department_id order by visits desc
"""
if limit:
q = q + " limit %s;" % (limit)
- # Add this back (before and period_name =%s) if you want to ignore publisher
- # homepage views
- # and not url like '/publisher/%%'
-
top_publishers = []
- res = connection.execute(q, c.month)
-
+ res = connection.execute(q, month)
for row in res:
g = model.Group.get(row[0])
if g:
top_publishers.append((g, row[1], row[2]))
return top_publishers
+
def _get_publishers():
'''
@@ -393,3 +339,7 @@
publishers.append((pub.name, pub.title))
return publishers
+def _percent(num, total):
+ p = 100 * float(num)/float(total)
+ return "%.2f%%" % round(p, 2)
+
--- a/ckanext/ga_report/download_analytics.py
+++ b/ckanext/ga_report/download_analytics.py
@@ -3,7 +3,7 @@
import datetime
import collections
from pylons import config
-
+from ga_model import _normalize_url
import ga_model
#from ga_client import GA
@@ -100,12 +100,18 @@
self.get_full_period_name(period_name, period_complete_day),
start_date.strftime('%Y %m %d'),
end_date.strftime('%Y %m %d'))
- data = self.download(start_date, end_date, '~/dataset/[a-z0-9-_]+')
+
+ # Clean up the entries before we run this
+ ga_model.pre_update_url_stats(period_name)
+
+ accountName = config.get('googleanalytics.account')
+
+ data = self.download(start_date, end_date, '~/%s/dataset/[a-z0-9-_]+' % accountName)
log.info('Storing Dataset Analytics for period "%s"',
self.get_full_period_name(period_name, period_complete_day))
self.store(period_name, period_complete_day, data, )
- data = self.download(start_date, end_date, '~/publisher/[a-z0-9-_]+')
+ data = self.download(start_date, end_date, '~/%s/publisher/[a-z0-9-_]+' % accountName)
log.info('Storing Publisher Analytics for period "%s"',
self.get_full_period_name(period_name, period_complete_day))
self.store(period_name, period_complete_day, data,)
@@ -114,6 +120,7 @@
self.sitewide_stats( period_name )
self.update_social_info(period_name, start_date, end_date)
+
def update_social_info(self, period_name, start_date, end_date):
start_date = start_date.strftime('%Y-%m-%d')
@@ -136,17 +143,16 @@
data = collections.defaultdict(list)
rows = results.get('rows',[])
for row in rows:
- from ga_model import _normalize_url
data[_normalize_url(row[0])].append( (row[1], int(row[2]),) )
ga_model.update_social(period_name, data)
- def download(self, start_date, end_date, path='~/dataset/[a-z0-9-_]+'):
+ def download(self, start_date, end_date, path=None):
'''Get data from GA for a given time period'''
start_date = start_date.strftime('%Y-%m-%d')
end_date = end_date.strftime('%Y-%m-%d')
query = 'ga:pagePath=%s$' % path
- metrics = 'ga:uniquePageviews, ga:visitors'
+ metrics = 'ga:uniquePageviews, ga:visits'
sort = '-ga:uniquePageviews'
# Supported query params at
@@ -161,15 +167,13 @@
max_results=10000,
end_date=end_date).execute()
- if os.getenv('DEBUG'):
- import pprint
- pprint.pprint(results)
- print 'Total results: %s' % results.get('totalResults')
-
packages = []
for entry in results.get('rows'):
(loc,pageviews,visits) = entry
- packages.append( ('http:/' + loc, pageviews, visits,) ) # Temporary hack
+ url = _normalize_url('http:/' + loc)
+ if not url.startswith('/dataset/') and not url.startswith('/publisher/'):
+ continue
+ packages.append( (url, pageviews, visits,) ) # Temporary hack
return dict(url=packages)
def store(self, period_name, period_complete_day, data):
@@ -213,18 +217,37 @@
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
start_date=start_date,
- metrics='ga:pageviewsPerVisit,ga:bounces,ga:avgTimeOnSite,ga:percentNewVisits,ga:visitors',
+ metrics='ga:pageviewsPerVisit,ga:avgTimeOnSite,ga:percentNewVisits,ga:visits',
max_results=10000,
end_date=end_date).execute()
result_data = results.get('rows')
data = {
'Pages per visit': result_data[0][0],
- 'Bounces': result_data[0][1],
- 'Average time on site': result_data[0][2],
- 'New visits': result_data[0][3],
- 'Total visits': result_data[0][4],
+ 'Average time on site': result_data[0][1],
+ 'New visits': result_data[0][2],
+ 'Total visits': result_data[0][3],
}
ga_model.update_sitewide_stats(period_name, "Totals", data)
+
+ # Bounces from /data. This url is specified in configuration because
+ # for DGU we don't want /.
+ path = config.get('ga-report.bounce_url','/')
+ print path
+ results = self.service.data().ga().get(
+ ids='ga:' + self.profile_id,
+ filters='ga:pagePath=~%s$' % (path,),
+ start_date=start_date,
+ metrics='ga:bounces,ga:uniquePageviews',
+ dimensions='ga:pagePath',
+ max_results=10000,
+ end_date=end_date).execute()
+ result_data = results.get('rows')
+ for results in result_data:
+ if results[0] == path:
+ bounce, total = [float(x) for x in results[1:]]
+ pct = 100 * bounce/total
+ print "%d bounces from %d total == %s" % (bounce, total, pct)
+ ga_model.update_sitewide_stats(period_name, "Totals", {'Bounces': pct})
def _locale_stats(self, start_date, end_date, period_name):
--- a/ckanext/ga_report/ga_model.py
+++ b/ckanext/ga_report/ga_model.py
@@ -1,10 +1,10 @@
import re
import uuid
-from sqlalchemy import Table, Column, MetaData
+from sqlalchemy import Table, Column, MetaData, ForeignKey
from sqlalchemy import types
from sqlalchemy.sql import select
-from sqlalchemy.orm import mapper
+from sqlalchemy.orm import mapper, relation
from sqlalchemy import func
import ckan.model as model
@@ -14,8 +14,6 @@
return unicode(uuid.uuid4())
metadata = MetaData()
-
-
class GA_Url(object):
@@ -32,6 +30,7 @@
Column('visitors', types.UnicodeText),
Column('url', types.UnicodeText),
Column('department_id', types.UnicodeText),
+ Column('package_id', types.UnicodeText),
)
mapper(GA_Url, url_table)
@@ -117,7 +116,7 @@
return '/' + '/'.join(url.split('/')[2:])
-def _get_department_id_of_url(url):
+def _get_package_and_publisher(url):
# e.g. /dataset/fuel_prices
# e.g. /dataset/fuel_prices/resource/e63380d4
dataset_match = re.match('/dataset/([^/]+)(/.*)?', url)
@@ -127,12 +126,13 @@
if dataset:
publisher_groups = dataset.get_groups('publisher')
if publisher_groups:
- return publisher_groups[0].name
+ return dataset_ref,publisher_groups[0].name
+ return dataset_ref, None
else:
publisher_match = re.match('/publisher/([^/]+)(/.*)?', url)
if publisher_match:
- return publisher_match.groups()[0]
-
+ return None, publisher_match.groups()[0]
+ return None, None
def update_sitewide_stats(period_name, stat_name, data):
for k,v in data.iteritems():
@@ -157,34 +157,76 @@
model.Session.commit()
+def update_url_stat_totals(period_name):
+
+ """
+ items = model.Session.query(GA_Url).\
+ filter(GA_Url.period_name != "All").\
+ filter(GA_Url.url==url).all()
+ values = {'id': make_uuid(),
+ 'period_name': "All",
+ 'period_complete_day': "0",
+ 'url': url,
+ 'pageviews': sum([int(x.pageviews) for x in items]),
+ 'visitors': sum([int(x.visitors) for x in items]),
+ 'department_id': department_id,
+ 'package_id': package
+ }
+ model.Session.add(GA_Url(**values))
+ model.Session.commit()
+ """
+
+def pre_update_url_stats(period_name):
+ model.Session.query(GA_Url).\
+ filter(GA_Url.period_name==period_name).delete()
+ model.Session.query(GA_Url).\
+ filter(GA_Url.period_name=='All').delete()
+
def update_url_stats(period_name, period_complete_day, url_data):
+
for url, views, visitors in url_data:
- url = _normalize_url(url)
- department_id = _get_department_id_of_url(url)
-
- # see if the row for this url & month is in the table already
+ package, publisher = _get_package_and_publisher(url)
+
item = model.Session.query(GA_Url).\
filter(GA_Url.period_name==period_name).\
filter(GA_Url.url==url).first()
if item:
- item.period_name = period_name
- item.pageviews = views
- item.visitors = visitors
- item.department_id = department_id
+ item.pageviews = item.pageviews + views
+ item.visitors = item.visitors + visitors
model.Session.add(item)
else:
- # create the row
values = {'id': make_uuid(),
'period_name': period_name,
'period_complete_day': period_complete_day,
'url': url,
'pageviews': views,
'visitors': visitors,
- 'department_id': department_id
+ 'department_id': publisher,
+ 'package_id': package
}
model.Session.add(GA_Url(**values))
model.Session.commit()
+
+ if package:
+ entries = model.Session.query(GA_Url).\
+ filter(GA_Url.period_name!='All').\
+ filter(GA_Url.url==url).all()
+ values = {'id': make_uuid(),
+ 'period_name': 'All',
+ 'period_complete_day': 0,
+ 'url': url,
+ 'pageviews': sum([int(e.pageviews) for e in entries]),
+ 'visitors': sum([int(e.visitors) for e in entries]),
+ 'department_id': publisher,
+ 'package_id': package
+ }
+
+ model.Session.add(GA_Url(**values))
+ model.Session.commit()
+
+
+
def update_social(period_name, data):
--- a/ckanext/ga_report/helpers.py
+++ b/ckanext/ga_report/helpers.py
@@ -1,7 +1,9 @@
import logging
import operator
+
import ckan.lib.base as base
import ckan.model as model
+from ckan.logic import get_action
from ckanext.ga_report.ga_model import GA_Url, GA_Publisher
from ckanext.ga_report.controller import _get_publishers
@@ -39,25 +41,38 @@
order_by('ga_url.pageviews::int desc')
num_top_datasets = top_datasets.count()
+ dataset = None
if num_top_datasets:
- dataset = None
+ count = 0
while not dataset:
rand = random.randrange(0, min(top, num_top_datasets))
ga_url = top_datasets[rand]
dataset = model.Package.get(ga_url.url[len('/dataset/'):])
if dataset and not dataset.state == 'active':
dataset = None
- else:
+ count += 1
+ if count > 10:
+ break
+ if not dataset:
+ # fallback
dataset = model.Session.query(model.Package)\
.filter_by(state='active').first()
- publisher = dataset.get_groups('publisher')[0]
- return {
- 'dataset': dataset,
- 'publisher': publisher
- }
+ if not dataset:
+ return None
+ dataset_dict = get_action('package_show')({'model': model,
+ 'session': model.Session},
+ {'id':dataset.id})
+ return dataset_dict
def single_popular_dataset_html(top=20):
- context = single_popular_dataset(top)
+ dataset_dict = single_popular_dataset(top)
+ groups = package.get('groups', [])
+ publishers = [ g for g in groups if g.get('type') == 'publisher' ]
+ publisher = publishers[0] if publishers else {'name':'', 'title': ''}
+ context = {
+ 'dataset': dataset_dict,
+ 'publisher': publisher_dict
+ }
return base.render_snippet('ga_report/ga_popular_single.html', **context)
--- a/ckanext/ga_report/templates/ga_report/ga_popular_single.html
+++ b/ckanext/ga_report/templates/ga_report/ga_popular_single.html
@@ -8,14 +8,14 @@
<h2>Featured dataset</h2>
<div class="dataset-summary boxed">
- <a class="dataset-header" href="${h.url_for(controller='package', action='read', id=dataset.name)}">
- <h3>${dataset.title}</h3>
+ <a class="dataset-header" href="${h.url_for(controller='package', action='read', id=dataset['name'])}">
+ <h3>${dataset['title']}</h3>
</a>
<h4>
<strong>Publisher</strong> :
- <a href="/publisher/${publisher.name}">${publisher.title}</a>
+ <a href="/publisher/${publisher['name']}">${publisher['title']}</a>
</h4>
- <div>${h.truncate(dataset.notes, length=200, whole_word=True)}</div>
+ <div>${h.truncate(dataset['notes_rendered'], length=200, whole_word=True)}</div>
</div>
<div>
<a href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='read')}" class="btn">Other popular datasets</a>
--- a/ckanext/ga_report/templates/ga_report/publisher/index.html
+++ b/ckanext/ga_report/templates/ga_report/publisher/index.html
@@ -41,15 +41,15 @@
<table class="table table-condensed table-bordered table-striped">
<tr>
<th>Publisher</th>
+ <th>Dataset Visits</th>
<th>Dataset Views</th>
- <th>Dataset Visits</th>
</tr>
<py:for each="publisher, views, visits in c.top_publishers">
<tr>
<td>${h.link_to(publisher.title, h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport', action='read_publisher', id=publisher.name))}
</td>
+ <td>${visits}</td>
<td>${views}</td>
- <td>${visits}</td>
</tr>
</py:for>
</table>
--- a/ckanext/ga_report/templates/ga_report/publisher/read.html
+++ b/ckanext/ga_report/templates/ga_report/publisher/read.html
@@ -47,15 +47,15 @@
<table py:if="c.top_packages" class="table table-condensed table-bordered table-striped">
<tr>
<th>Dataset</th>
+ <th>Visits</th>
<th>Views</th>
- <th>Visits</th>
</tr>
<py:for each="package, views, visits in c.top_packages">
<tr>
<td>${h.link_to(package.title or package.name, h.url_for(controller='package', action='read', id=package.name))}
</td>
+ <td>${visits}</td>
<td>${views}</td>
- <td>${visits}</td>
</tr>
</py:for>
</table>