merge
merge

import re import re
import csv import csv
import sys import sys
import logging import logging
import operator import operator
import collections import collections
from ckan.lib.base import (BaseController, c, g, render, request, response, abort) from ckan.lib.base import (BaseController, c, g, render, request, response, abort)
   
import sqlalchemy import sqlalchemy
from sqlalchemy import func, cast, Integer from sqlalchemy import func, cast, Integer
import ckan.model as model 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') log = logging.getLogger('ckanext.ga-report')
   
   
def _get_month_name(strdate): def _get_month_name(strdate):
import calendar import calendar
from time import strptime from time import strptime
d = strptime(strdate, '%Y-%m') d = strptime(strdate, '%Y-%m')
return '%s %s' % (calendar.month_name[d.tm_mon], d.tm_year) return '%s %s' % (calendar.month_name[d.tm_mon], d.tm_year)
   
   
def _month_details(cls): def _month_details(cls):
'''Returns a list of all the month names''' '''Returns a list of all the month names'''
months = [] months = []
vals = model.Session.query(cls.period_name).filter(cls.period_name!='All').distinct().all() vals = model.Session.query(cls.period_name).filter(cls.period_name!='All').distinct().all()
for m in vals: for m in vals:
months.append( (m[0], _get_month_name(m[0]))) months.append( (m[0], _get_month_name(m[0])))
return sorted(months, key=operator.itemgetter(0), reverse=True) return sorted(months, key=operator.itemgetter(0), reverse=True)
   
   
class GaReport(BaseController): class GaReport(BaseController):
   
def csv(self, month): def csv(self, month):
import csv import csv
   
q = model.Session.query(GA_Stat) q = model.Session.query(GA_Stat)
if month != 'all': if month != 'all':
q = q.filter(GA_Stat.period_name==month) q = q.filter(GA_Stat.period_name==month)
entries = q.order_by('GA_Stat.period_name, GA_Stat.stat_name, GA_Stat.key').all() entries = q.order_by('GA_Stat.period_name, GA_Stat.stat_name, GA_Stat.key').all()
   
response.headers['Content-Type'] = "text/csv; charset=utf-8" response.headers['Content-Type'] = "text/csv; charset=utf-8"
response.headers['Content-Disposition'] = str('attachment; filename=stats_%s.csv' % (month,)) response.headers['Content-Disposition'] = str('attachment; filename=stats_%s.csv' % (month,))
   
writer = csv.writer(response) writer = csv.writer(response)
writer.writerow(["Period", "Statistic", "Key", "Value"]) writer.writerow(["Period", "Statistic", "Key", "Value"])
   
for entry in entries: for entry in entries:
writer.writerow([entry.period_name.encode('utf-8'), writer.writerow([entry.period_name.encode('utf-8'),
entry.stat_name.encode('utf-8'), entry.stat_name.encode('utf-8'),
entry.key.encode('utf-8'), entry.key.encode('utf-8'),
entry.value.encode('utf-8')]) entry.value.encode('utf-8')])
   
def index(self): def index(self):
   
# Get the month details by fetching distinct values and determining the # Get the month details by fetching distinct values and determining the
# month names from the values. # month names from the values.
c.months = _month_details(GA_Stat) c.months = _month_details(GA_Stat)
   
# Work out which month to show, based on query params of the first item # Work out which month to show, based on query params of the first item
c.month_desc = 'all months' c.month_desc = 'all months'
c.month = request.params.get('month', '') c.month = request.params.get('month', '')
if c.month: if c.month:
c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month])
   
q = model.Session.query(GA_Stat).\ q = model.Session.query(GA_Stat).\
filter(GA_Stat.stat_name=='Totals') filter(GA_Stat.stat_name=='Totals')
if c.month: if c.month:
q = q.filter(GA_Stat.period_name==c.month) q = q.filter(GA_Stat.period_name==c.month)
entries = q.order_by('ga_stat.key').all() entries = q.order_by('ga_stat.key').all()
   
def clean_key(key, val): def clean_key(key, val):
if key in ['Average time on site', 'Pages per visit', 'New visits', 'Bounces']: if key in ['Average time on site', 'Pages per visit', 'New visits', 'Bounces']:
val = "%.2f" % round(float(val), 2) val = "%.2f" % round(float(val), 2)
if key == 'Average time on site': if key == 'Average time on site':
mins, secs = divmod(float(val), 60) mins, secs = divmod(float(val), 60)
hours, mins = divmod(mins, 60) hours, mins = divmod(mins, 60)
val = '%02d:%02d:%02d (%s seconds) ' % (hours, mins, secs, val) val = '%02d:%02d:%02d (%s seconds) ' % (hours, mins, secs, val)
if key in ['New visits','Bounces']: if key in ['New visits','Bounces']:
val = "%s%%" % val val = "%s%%" % val
if key in ['Total page views', 'Total visits']: if key in ['Total page views', 'Total visits']:
val = int(val) val = int(val)
   
return key, val return key, val
   
c.global_totals = [] c.global_totals = []
if c.month: if c.month:
for e in entries: for e in entries:
key, val = clean_key(e.key, e.value) key, val = clean_key(e.key, e.value)
c.global_totals.append((key, val)) c.global_totals.append((key, val))
else: else:
d = collections.defaultdict(list) d = collections.defaultdict(list)
for e in entries: for e in entries:
d[e.key].append(float(e.value)) d[e.key].append(float(e.value))
for k, v in d.iteritems(): for k, v in d.iteritems():
if k in ['Total page views', 'Total visits']: if k in ['Total page views', 'Total visits']:
v = sum(v) v = sum(v)
else: else:
v = float(sum(v))/len(v) v = float(sum(v))/len(v)
key, val = clean_key(k,v) key, val = clean_key(k,v)
   
c.global_totals.append((key, val)) c.global_totals.append((key, val))
c.global_totals = sorted(c.global_totals, key=operator.itemgetter(0)) c.global_totals = sorted(c.global_totals, key=operator.itemgetter(0))
   
keys = { keys = {
'Browser versions': 'browser_versions', 'Browser versions': 'browser_versions',
'Browsers': 'browsers', 'Browsers': 'browsers',
'Operating Systems versions': 'os_versions', 'Operating Systems versions': 'os_versions',
'Operating Systems': 'os', 'Operating Systems': 'os',
'Social sources': 'social_networks', 'Social sources': 'social_networks',
'Languages': 'languages', 'Languages': 'languages',
'Country': 'country' 'Country': 'country'
} }
   
def shorten_name(name, length=60): def shorten_name(name, length=60):
return (name[:length] + '..') if len(name) > 60 else name return (name[:length] + '..') if len(name) > 60 else name
   
def fill_out_url(url): def fill_out_url(url):
import urlparse import urlparse
return urlparse.urljoin(g.site_url, url) return urlparse.urljoin(g.site_url, url)
   
c.social_referrer_totals, c.social_referrers = [], [] c.social_referrer_totals, c.social_referrers = [], []
q = model.Session.query(GA_ReferralStat) q = model.Session.query(GA_ReferralStat)
q = q.filter(GA_ReferralStat.period_name==c.month) if c.month else q q = q.filter(GA_ReferralStat.period_name==c.month) if c.month else q
q = q.order_by('ga_referrer.count::int desc') q = q.order_by('ga_referrer.count::int desc')
for entry in q.all(): for entry in q.all():
c.social_referrers.append((shorten_name(entry.url), fill_out_url(entry.url), c.social_referrers.append((shorten_name(entry.url), fill_out_url(entry.url),
entry.source,entry.count)) entry.source,entry.count))
   
q = model.Session.query(GA_ReferralStat.url, q = model.Session.query(GA_ReferralStat.url,
func.sum(GA_ReferralStat.count).label('count')) func.sum(GA_ReferralStat.count).label('count'))
q = q.filter(GA_ReferralStat.period_name==c.month) if c.month else q q = q.filter(GA_ReferralStat.period_name==c.month) if c.month else q
q = q.order_by('count desc').group_by(GA_ReferralStat.url) q = q.order_by('count desc').group_by(GA_ReferralStat.url)
for entry in q.all(): for entry in q.all():
c.social_referrer_totals.append((shorten_name(entry[0]), fill_out_url(entry[0]),'', c.social_referrer_totals.append((shorten_name(entry[0]), fill_out_url(entry[0]),'',
entry[1])) entry[1]))
   
for k, v in keys.iteritems(): for k, v in keys.iteritems():
q = model.Session.query(GA_Stat).\ q = model.Session.query(GA_Stat).\
filter(GA_Stat.stat_name==k) filter(GA_Stat.stat_name==k)
if c.month: if c.month:
entries = [] entries = []
q = q.filter(GA_Stat.period_name==c.month).\ q = q.filter(GA_Stat.period_name==c.month).\
order_by('ga_stat.value::int desc') order_by('ga_stat.value::int desc')
   
d = collections.defaultdict(int) d = collections.defaultdict(int)
for e in q.all(): for e in q.all():
d[e.key] += int(e.value) d[e.key] += int(e.value)
entries = [] entries = []
for key, val in d.iteritems(): for key, val in d.iteritems():
entries.append((key,val,)) entries.append((key,val,))
entries = sorted(entries, key=operator.itemgetter(1), reverse=True) entries = sorted(entries, key=operator.itemgetter(1), reverse=True)
   
# Get the total for each set of values and then set the value as # Get the total for each set of values and then set the value as
# a percentage of the total # a percentage of the total
if k == 'Social sources': if k == 'Social sources':
total = sum([x for n,x in c.global_totals if n == 'Total visits']) total = sum([x for n,x in c.global_totals if n == 'Total visits'])
else: else:
total = sum([num for _,num in entries]) 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') return render('ga_report/site/index.html')
   
   
class GaDatasetReport(BaseController): class GaDatasetReport(BaseController):
""" """
Displays the pageview and visit count for datasets Displays the pageview and visit count for datasets
with options to filter by publisher and time period. with options to filter by publisher and time period.
""" """
def publisher_csv(self, month): def publisher_csv(self, month):
''' '''
Returns a CSV of each publisher with the total number of dataset Returns a CSV of each publisher with the total number of dataset
views & visits. views & visits.
''' '''
c.month = month if not month == 'all' else '' c.month = month if not month == 'all' else ''
response.headers['Content-Type'] = "text/csv; charset=utf-8" response.headers['Content-Type'] = "text/csv; charset=utf-8"
response.headers['Content-Disposition'] = str('attachment; filename=publishers_%s.csv' % (month,)) response.headers['Content-Disposition'] = str('attachment; filename=publishers_%s.csv' % (month,))
   
writer = csv.writer(response) writer = csv.writer(response)
writer.writerow(["Publisher Title", "Publisher Name", "Views", "Visits", "Period Name"]) writer.writerow(["Publisher Title", "Publisher Name", "Views", "Visits", "Period Name"])
   
for publisher,view,visit in _get_top_publishers(None): for publisher,view,visit in _get_top_publishers(None):
writer.writerow([publisher.title.encode('utf-8'), writer.writerow([publisher.title.encode('utf-8'),
publisher.name.encode('utf-8'), publisher.name.encode('utf-8'),
view, view,
visit, visit,
month]) month])
   
def dataset_csv(self, id='all', month='all'): def dataset_csv(self, id='all', month='all'):
''' '''
Returns a CSV with the number of views & visits for each dataset. Returns a CSV with the number of views & visits for each dataset.
   
:param id: A Publisher ID or None if you want for all :param id: A Publisher ID or None if you want for all
:param month: The time period, or 'all' :param month: The time period, or 'all'
''' '''
c.month = month if not month == 'all' else '' c.month = month if not month == 'all' else ''
if id != 'all': if id != 'all':
c.publisher = model.Group.get(id) c.publisher = model.Group.get(id)
if not c.publisher: if not c.publisher:
abort(404, 'A publisher with that name could not be found') abort(404, 'A publisher with that name could not be found')
   
packages = self._get_packages(c.publisher) packages = self._get_packages(c.publisher)
response.headers['Content-Type'] = "text/csv; charset=utf-8" response.headers['Content-Type'] = "text/csv; charset=utf-8"
response.headers['Content-Disposition'] = \ response.headers['Content-Disposition'] = \
str('attachment; filename=datasets_%s_%s.csv' % (c.publisher_name, month,)) str('attachment; filename=datasets_%s_%s.csv' % (c.publisher_name, month,))
   
writer = csv.writer(response) writer = csv.writer(response)
writer.writerow(["Dataset Title", "Dataset Name", "Views", "Visits", "Period Name"]) writer.writerow(["Dataset Title", "Dataset Name", "Views", "Visits", "Period Name"])
   
for package,view,visit in packages: for package,view,visit in packages:
writer.writerow([package.title.encode('utf-8'), writer.writerow([package.title.encode('utf-8'),
package.name.encode('utf-8'), package.name.encode('utf-8'),
view, view,
visit, visit,
month]) month])
   
def publishers(self): def publishers(self):
'''A list of publishers and the number of views/visits for each''' '''A list of publishers and the number of views/visits for each'''
   
# Get the month details by fetching distinct values and determining the # Get the month details by fetching distinct values and determining the
# month names from the values. # month names from the values.
c.months = _month_details(GA_Url) c.months = _month_details(GA_Url)
   
# Work out which month to show, based on query params of the first item # Work out which month to show, based on query params of the first item
c.month = request.params.get('month', '') c.month = request.params.get('month', '')
c.month_desc = 'all months' c.month_desc = 'all months'
if c.month: if c.month:
c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month])
   
c.top_publishers = _get_top_publishers() c.top_publishers = _get_top_publishers()
   
return render('ga_report/publisher/index.html') return render('ga_report/publisher/index.html')
   
def _get_packages(self, publisher=None, count=-1): def _get_packages(self, publisher=None, count=-1):
'''Returns the datasets in order of visits''' '''Returns the datasets in order of visits'''
if count == -1: if count == -1:
count = sys.maxint count = sys.maxint
   
month = c.month or 'All' month = c.month or 'All'
   
q = model.Session.query(GA_Url,model.Package)\ q = model.Session.query(GA_Url,model.Package)\
.filter(model.Package.name==GA_Url.package_id)\ .filter(model.Package.name==GA_Url.package_id)\
.filter(GA_Url.url.like('/dataset/%')) .filter(GA_Url.url.like('/dataset/%'))
if publisher: if publisher:
q = q.filter(GA_Url.department_id==publisher.name) q = q.filter(GA_Url.department_id==publisher.name)
q = q.filter(GA_Url.period_name==month) q = q.filter(GA_Url.period_name==month)
q = q.order_by('ga_url.visitors::int desc') q = q.order_by('ga_url.visitors::int desc')
top_packages = [] top_packages = []
   
for entry,package in q.limit(count): for entry,package in q.limit(count):
if package: if package:
top_packages.append((package, entry.pageviews, entry.visitors)) top_packages.append((package, entry.pageviews, entry.visitors))
else: else:
log.warning('Could not find package associated package') log.warning('Could not find package associated package')
   
return top_packages return top_packages
   
def read(self): def read(self):
''' '''
Lists the most popular datasets across all publishers Lists the most popular datasets across all publishers
''' '''
return self.read_publisher(None) return self.read_publisher(None)
   
def read_publisher(self, id): def read_publisher(self, id):
''' '''
Lists the most popular datasets for a publisher (or across all publishers) Lists the most popular datasets for a publisher (or across all publishers)
''' '''
count = 20 count = 20
   
c.publishers = _get_publishers() c.publishers = _get_publishers()
   
id = request.params.get('publisher', id) id = request.params.get('publisher', id)
if id and id != 'all': if id and id != 'all':
c.publisher = model.Group.get(id) c.publisher = model.Group.get(id)
if not c.publisher: if not c.publisher:
abort(404, 'A publisher with that name could not be found') abort(404, 'A publisher with that name could not be found')
c.publisher_name = c.publisher.name c.publisher_name = c.publisher.name
c.top_packages = [] # package, dataset_views in c.top_packages c.top_packages = [] # package, dataset_views in c.top_packages
   
# Get the month details by fetching distinct values and determining the # Get the month details by fetching distinct values and determining the
# month names from the values. # month names from the values.
c.months = _month_details(GA_Url) c.months = _month_details(GA_Url)
   
# Work out which month to show, based on query params of the first item # Work out which month to show, based on query params of the first item
c.month = request.params.get('month', '') c.month = request.params.get('month', '')
if not c.month: if not c.month:
c.month_desc = 'all months' c.month_desc = 'all months'
else: else:
c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month])
   
month = c.mnth or 'All' month = c.month or 'All'
c.publisher_page_views = 0 c.publisher_page_views = 0
q = model.Session.query(GA_Url).\ q = model.Session.query(GA_Url).\
filter(GA_Url.url=='/publisher/%s' % c.publisher_name) filter(GA_Url.url=='/publisher/%s' % c.publisher_name)
entry = q.filter(GA_Url.period_name==c.month).first() entry = q.filter(GA_Url.period_name==c.month).first()
c.publisher_page_views = entry.pageviews if entry else 0 c.publisher_page_views = entry.pageviews if entry else 0
   
c.top_packages = self._get_packages(c.publisher, 20) c.top_packages = self._get_packages(c.publisher, 20)
   
return render('ga_report/publisher/read.html') return render('ga_report/publisher/read.html')
   
def _get_top_publishers(limit=20): def _get_top_publishers(limit=20):
''' '''
Returns a list of the top 20 publishers by dataset visits. Returns a list of the top 20 publishers by dataset visits.
(The number to show can be varied with 'limit') (The number to show can be varied with 'limit')
''' '''
  month = c.month or 'All'
connection = model.Session.connection() connection = model.Session.connection()
q = """ q = """
select department_id, sum(pageviews::int) views, sum(visitors::int) visits select department_id, sum(pageviews::int) views, sum(visitors::int) visits
from ga_url from ga_url
where department_id <> ''""" where department_id <> ''
if c.month: and period_name=%s
q = q + """ group by department_id order by visits desc
and period_name=%s  
"""  
q = q + """  
group by department_id order by visits desc  
""" """
if limit: if limit:
q = q + " limit %s;" % (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 = [] top_publishers = []
res = connection.execute(q, c.month) res = connection.execute(q, month)
   
for row in res: for row in res:
g = model.Group.get(row[0]) g = model.Group.get(row[0])
if g: if g:
top_publishers.append((g, row[1], row[2])) top_publishers.append((g, row[1], row[2]))
return top_publishers return top_publishers
   
   
def _get_publishers(): def _get_publishers():
''' '''
Returns a list of all publishers. Each item is a tuple: Returns a list of all publishers. Each item is a tuple:
(names, title) (names, title)
''' '''
publishers = [] publishers = []
for pub in model.Session.query(model.Group).\ for pub in model.Session.query(model.Group).\
filter(model.Group.type=='publisher').\ filter(model.Group.type=='publisher').\
filter(model.Group.state=='active').\ filter(model.Group.state=='active').\
order_by(model.Group.name): order_by(model.Group.name):
publishers.append((pub.name, pub.title)) publishers.append((pub.name, pub.title))
return publishers return publishers
   
def _percent(num, total): def _percent(num, total):
p = 100 * float(num)/float(total) p = 100 * float(num)/float(total)
return "%.2f%%" % round(p, 2) return "%.2f%%" % round(p, 2)
   
import os import os
import logging import logging
import datetime import datetime
import collections import collections
from pylons import config from pylons import config
  from ga_model import _normalize_url
import ga_model import ga_model
   
#from ga_client import GA #from ga_client import GA
   
log = logging.getLogger('ckanext.ga-report') log = logging.getLogger('ckanext.ga-report')
   
FORMAT_MONTH = '%Y-%m' FORMAT_MONTH = '%Y-%m'
MIN_VIEWS = 50 MIN_VIEWS = 50
MIN_VISITS = 20 MIN_VISITS = 20
   
class DownloadAnalytics(object): class DownloadAnalytics(object):
'''Downloads and stores analytics info''' '''Downloads and stores analytics info'''
   
def __init__(self, service=None, profile_id=None, delete_first=False): def __init__(self, service=None, profile_id=None, delete_first=False):
self.period = config['ga-report.period'] self.period = config['ga-report.period']
self.service = service self.service = service
self.profile_id = profile_id self.profile_id = profile_id
self.delete_first = delete_first self.delete_first = delete_first
   
def specific_month(self, date): def specific_month(self, date):
import calendar import calendar
   
first_of_this_month = datetime.datetime(date.year, date.month, 1) first_of_this_month = datetime.datetime(date.year, date.month, 1)
_, last_day_of_month = calendar.monthrange(int(date.year), int(date.month)) _, last_day_of_month = calendar.monthrange(int(date.year), int(date.month))
last_of_this_month = datetime.datetime(date.year, date.month, last_day_of_month) last_of_this_month = datetime.datetime(date.year, date.month, last_day_of_month)
periods = ((date.strftime(FORMAT_MONTH), periods = ((date.strftime(FORMAT_MONTH),
last_day_of_month, last_day_of_month,
first_of_this_month, last_of_this_month),) first_of_this_month, last_of_this_month),)
self.download_and_store(periods) self.download_and_store(periods)
   
   
def latest(self): def latest(self):
if self.period == 'monthly': if self.period == 'monthly':
# from first of this month to today # from first of this month to today
now = datetime.datetime.now() now = datetime.datetime.now()
first_of_this_month = datetime.datetime(now.year, now.month, 1) first_of_this_month = datetime.datetime(now.year, now.month, 1)
periods = ((now.strftime(FORMAT_MONTH), periods = ((now.strftime(FORMAT_MONTH),
now.day, now.day,
first_of_this_month, now),) first_of_this_month, now),)
else: else:
raise NotImplementedError raise NotImplementedError
self.download_and_store(periods) self.download_and_store(periods)
   
   
def for_date(self, for_date): def for_date(self, for_date):
assert isinstance(since_date, datetime.datetime) assert isinstance(since_date, datetime.datetime)
periods = [] # (period_name, period_complete_day, start_date, end_date) periods = [] # (period_name, period_complete_day, start_date, end_date)
if self.period == 'monthly': if self.period == 'monthly':
first_of_the_months_until_now = [] first_of_the_months_until_now = []
year = for_date.year year = for_date.year
month = for_date.month month = for_date.month
now = datetime.datetime.now() now = datetime.datetime.now()
first_of_this_month = datetime.datetime(now.year, now.month, 1) first_of_this_month = datetime.datetime(now.year, now.month, 1)
while True: while True:
first_of_the_month = datetime.datetime(year, month, 1) first_of_the_month = datetime.datetime(year, month, 1)
if first_of_the_month == first_of_this_month: if first_of_the_month == first_of_this_month:
periods.append((now.strftime(FORMAT_MONTH), periods.append((now.strftime(FORMAT_MONTH),
now.day, now.day,
first_of_this_month, now)) first_of_this_month, now))
break break
elif first_of_the_month < first_of_this_month: elif first_of_the_month < first_of_this_month:
in_the_next_month = first_of_the_month + datetime.timedelta(40) in_the_next_month = first_of_the_month + datetime.timedelta(40)
last_of_the_month = datetime.datetime(in_the_next_month.year, last_of_the_month = datetime.datetime(in_the_next_month.year,
in_the_next_month.month, 1)\ in_the_next_month.month, 1)\
- datetime.timedelta(1) - datetime.timedelta(1)
periods.append((now.strftime(FORMAT_MONTH), 0, periods.append((now.strftime(FORMAT_MONTH), 0,
first_of_the_month, last_of_the_month)) first_of_the_month, last_of_the_month))
else: else:
# first_of_the_month has got to the future somehow # first_of_the_month has got to the future somehow
break break
month += 1 month += 1
if month > 12: if month > 12:
year += 1 year += 1
month = 1 month = 1
else: else:
raise NotImplementedError raise NotImplementedError
self.download_and_store(periods) self.download_and_store(periods)
   
@staticmethod @staticmethod
def get_full_period_name(period_name, period_complete_day): def get_full_period_name(period_name, period_complete_day):
if period_complete_day: if period_complete_day:
return period_name + ' (up to %ith)' % period_complete_day return period_name + ' (up to %ith)' % period_complete_day
else: else:
return period_name return period_name
   
   
def download_and_store(self, periods): def download_and_store(self, periods):
for period_name, period_complete_day, start_date, end_date in periods: for period_name, period_complete_day, start_date, end_date in periods:
log.info('Period "%s" (%s - %s)', log.info('Period "%s" (%s - %s)',
self.get_full_period_name(period_name, period_complete_day), self.get_full_period_name(period_name, period_complete_day),
start_date.strftime('%Y-%m-%d'), start_date.strftime('%Y-%m-%d'),
end_date.strftime('%Y-%m-%d')) end_date.strftime('%Y-%m-%d'))
if self.delete_first: if self.delete_first:
log.info('Deleting existing Analytics for this period "%s"', log.info('Deleting existing Analytics for this period "%s"',
period_name) period_name)
ga_model.delete(period_name) ga_model.delete(period_name)
   
  # Clean up the entries before we run this
  ga_model.pre_update_url_stats(period_name)
   
  accountName = config.get('googleanalytics.account')
   
log.info('Downloading analytics for dataset views') log.info('Downloading analytics for dataset views')
data = self.download(start_date, end_date, '~/dataset/[a-z0-9-_]+') data = self.download(start_date, end_date, '~/%s/dataset/[a-z0-9-_]+' % accountName)
   
log.info('Storing dataset views (%i rows)', len(data.get('url'))) log.info('Storing dataset views (%i rows)', len(data.get('url')))
self.store(period_name, period_complete_day, data, ) self.store(period_name, period_complete_day, data, )
   
log.info('Downloading analytics for publisher views') log.info('Downloading analytics for publisher views')
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 views (%i rows)', len(data.get('url'))) log.info('Storing publisher views (%i rows)', len(data.get('url')))
self.store(period_name, period_complete_day, data,) self.store(period_name, period_complete_day, data,)
   
log.info('Aggregating datasets by publisher') log.info('Aggregating datasets by publisher')
ga_model.update_publisher_stats(period_name) # about 30 seconds. ga_model.update_publisher_stats(period_name) # about 30 seconds.
   
log.info('Downloading and storing analytics for site-wide stats') log.info('Downloading and storing analytics for site-wide stats')
self.sitewide_stats( period_name ) self.sitewide_stats( period_name )
   
log.info('Downloading and storing analytics for social networks') log.info('Downloading and storing analytics for social networks')
self.update_social_info(period_name, start_date, end_date) self.update_social_info(period_name, start_date, end_date)
   
   
def update_social_info(self, 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') start_date = start_date.strftime('%Y-%m-%d')
end_date = end_date.strftime('%Y-%m-%d') end_date = end_date.strftime('%Y-%m-%d')
query = 'ga:hasSocialSourceReferral=~Yes$' query = 'ga:hasSocialSourceReferral=~Yes$'
metrics = 'ga:entrances' metrics = 'ga:entrances'
sort = '-ga:entrances' sort = '-ga:entrances'
   
# Supported query params at # Supported query params at
# https://developers.google.com/analytics/devguides/reporting/core/v3/reference # https://developers.google.com/analytics/devguides/reporting/core/v3/reference
results = self.service.data().ga().get( results = self.service.data().ga().get(
ids='ga:' + self.profile_id, ids='ga:' + self.profile_id,
filters=query, filters=query,
start_date=start_date, start_date=start_date,
metrics=metrics, metrics=metrics,
sort=sort, sort=sort,
dimensions="ga:landingPagePath,ga:socialNetwork", dimensions="ga:landingPagePath,ga:socialNetwork",
max_results=10000, max_results=10000,
end_date=end_date).execute() end_date=end_date).execute()
data = collections.defaultdict(list) data = collections.defaultdict(list)
rows = results.get('rows',[]) rows = results.get('rows',[])
for row in rows: for row in rows:
from ga_model import _normalize_url  
data[_normalize_url(row[0])].append( (row[1], int(row[2]),) ) data[_normalize_url(row[0])].append( (row[1], int(row[2]),) )
ga_model.update_social(period_name, data) 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''' '''Get data from GA for a given time period'''
start_date = start_date.strftime('%Y-%m-%d') start_date = start_date.strftime('%Y-%m-%d')
end_date = end_date.strftime('%Y-%m-%d') end_date = end_date.strftime('%Y-%m-%d')
query = 'ga:pagePath=%s$' % path query = 'ga:pagePath=%s$' % path
metrics = 'ga:uniquePageviews, ga:visits' metrics = 'ga:uniquePageviews, ga:visits'
sort = '-ga:uniquePageviews' sort = '-ga:uniquePageviews'
   
# Supported query params at # Supported query params at
# https://developers.google.com/analytics/devguides/reporting/core/v3/reference # https://developers.google.com/analytics/devguides/reporting/core/v3/reference
results = self.service.data().ga().get( results = self.service.data().ga().get(
ids='ga:' + self.profile_id, ids='ga:' + self.profile_id,
filters=query, filters=query,
start_date=start_date, start_date=start_date,
metrics=metrics, metrics=metrics,
sort=sort, sort=sort,
dimensions="ga:pagePath", dimensions="ga:pagePath",
max_results=10000, max_results=10000,
end_date=end_date).execute() end_date=end_date).execute()
   
packages = [] packages = []
for entry in results.get('rows'): for entry in results.get('rows'):
(loc,pageviews,visits) = entry (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) return dict(url=packages)
   
def store(self, period_name, period_complete_day, data): def store(self, period_name, period_complete_day, data):
if 'url' in data: if 'url' in data:
ga_model.update_url_stats(period_name, period_complete_day, data['url']) ga_model.update_url_stats(period_name, period_complete_day, data['url'])
   
def sitewide_stats(self, period_name): def sitewide_stats(self, period_name):
import calendar import calendar
year, month = period_name.split('-') year, month = period_name.split('-')
_, last_day_of_month = calendar.monthrange(int(year), int(month)) _, last_day_of_month = calendar.monthrange(int(year), int(month))
   
start_date = '%s-01' % period_name start_date = '%s-01' % period_name
end_date = '%s-%s' % (period_name, last_day_of_month) end_date = '%s-%s' % (period_name, last_day_of_month)
funcs = ['_totals_stats', '_social_stats', '_os_stats', funcs = ['_totals_stats', '_social_stats', '_os_stats',
'_locale_stats', '_browser_stats', '_mobile_stats'] '_locale_stats', '_browser_stats', '_mobile_stats']
for f in funcs: for f in funcs:
log.info('Downloading analytics for %s' % f.split('_')[1]) log.info('Downloading analytics for %s' % f.split('_')[1])
getattr(self, f)(start_date, end_date, period_name) getattr(self, f)(start_date, end_date, period_name)
   
def _get_results(result_data, f): def _get_results(result_data, f):
data = {} data = {}
for result in result_data: for result in result_data:
key = f(result) key = f(result)
data[key] = data.get(key,0) + result[1] data[key] = data.get(key,0) + result[1]
return data return data
   
def _totals_stats(self, start_date, end_date, period_name): def _totals_stats(self, start_date, end_date, period_name):
""" Fetches distinct totals, total pageviews etc """ """ Fetches distinct totals, total pageviews etc """
results = self.service.data().ga().get( results = self.service.data().ga().get(
ids='ga:' + self.profile_id, ids='ga:' + self.profile_id,
start_date=start_date, start_date=start_date,
metrics='ga:uniquePageviews', metrics='ga:uniquePageviews',
sort='-ga:uniquePageviews', sort='-ga:uniquePageviews',
max_results=10000, max_results=10000,
end_date=end_date).execute() end_date=end_date).execute()
result_data = results.get('rows') result_data = results.get('rows')
ga_model.update_sitewide_stats(period_name, "Totals", {'Total page views': result_data[0][0]}) ga_model.update_sitewide_stats(period_name, "Totals", {'Total page views': result_data[0][0]})
   
results = self.service.data().ga().get( results = self.service.data().ga().get(
ids='ga:' + self.profile_id, ids='ga:' + self.profile_id,
start_date=start_date, start_date=start_date,
metrics='ga:pageviewsPerVisit,ga:avgTimeOnSite,ga:percentNewVisits,ga:visits', metrics='ga:pageviewsPerVisit,ga:avgTimeOnSite,ga:percentNewVisits,ga:visits',
max_results=10000, max_results=10000,
end_date=end_date).execute() end_date=end_date).execute()
result_data = results.get('rows') result_data = results.get('rows')
data = { data = {
'Pages per visit': result_data[0][0], 'Pages per visit': result_data[0][0],
'Average time on site': result_data[0][1], 'Average time on site': result_data[0][1],
'New visits': result_data[0][2], 'New visits': result_data[0][2],
'Total visits': result_data[0][3], 'Total visits': result_data[0][3],
} }
ga_model.update_sitewide_stats(period_name, "Totals", data) ga_model.update_sitewide_stats(period_name, "Totals", data)
   
# Bounces from /data. This url is specified in configuration because # Bounces from /data. This url is specified in configuration because
# for DGU we don't want /. # for DGU we don't want /.
path = config.get('ga-report.bounce_url','/') path = config.get('ga-report.bounce_url','/')
print path print path
results = self.service.data().ga().get( results = self.service.data().ga().get(
ids='ga:' + self.profile_id, ids='ga:' + self.profile_id,
filters='ga:pagePath=~%s$' % (path,), filters='ga:pagePath=~%s$' % (path,),
start_date=start_date, start_date=start_date,
metrics='ga:bounces,ga:uniquePageviews', metrics='ga:bounces,ga:uniquePageviews',
dimensions='ga:pagePath', dimensions='ga:pagePath',
max_results=10000, max_results=10000,
end_date=end_date).execute() end_date=end_date).execute()
result_data = results.get('rows') result_data = results.get('rows')
for results in result_data: for results in result_data:
if results[0] == path: if results[0] == path:
bounce, total = [float(x) for x in results[1:]] bounce, total = [float(x) for x in results[1:]]
pct = 100 * bounce/total pct = 100 * bounce/total
print "%d bounces from %d total == %s" % (bounce, total, pct) print "%d bounces from %d total == %s" % (bounce, total, pct)
ga_model.update_sitewide_stats(period_name, "Totals", {'Bounces': pct}) ga_model.update_sitewide_stats(period_name, "Totals", {'Bounces': pct})
   
   
def _locale_stats(self, start_date, end_date, period_name): def _locale_stats(self, start_date, end_date, period_name):
""" Fetches stats about language and country """ """ Fetches stats about language and country """
results = self.service.data().ga().get( results = self.service.data().ga().get(
ids='ga:' + self.profile_id, ids='ga:' + self.profile_id,
start_date=start_date, start_date=start_date,
metrics='ga:uniquePageviews', metrics='ga:uniquePageviews',
sort='-ga:uniquePageviews', sort='-ga:uniquePageviews',
dimensions="ga:language,ga:country", dimensions="ga:language,ga:country",
max_results=10000, max_results=10000,
end_date=end_date).execute() end_date=end_date).execute()
result_data = results.get('rows') result_data = results.get('rows')
data = {} data = {}
for result in result_data: for result in result_data:
data[result[0]] = data.get(result[0], 0) + int(result[2]) data[result[0]] = data.get(result[0], 0) + int(result[2])
self._filter_out_long_tail(data, MIN_VIEWS) self._filter_out_long_tail(data, MIN_VIEWS)
ga_model.update_sitewide_stats(period_name, "Languages", data) ga_model.update_sitewide_stats(period_name, "Languages", data)
   
data = {} data = {}
for result in result_data: for result in result_data:
data[result[1]] = data.get(result[1], 0) + int(result[2]) data[result[1]] = data.get(result[1], 0) + int(result[2])
self._filter_out_long_tail(data, MIN_VIEWS) self._filter_out_long_tail(data, MIN_VIEWS)
ga_model.update_sitewide_stats(period_name, "Country", data) ga_model.update_sitewide_stats(period_name, "Country", data)
   
   
def _social_stats(self, start_date, end_date, period_name): def _social_stats(self, start_date, end_date, period_name):
""" Finds out which social sites people are referred from """ """ Finds out which social sites people are referred from """
results = self.service.data().ga().get( results = self.service.data().ga().get(
ids='ga:' + self.profile_id, ids='ga:' + self.profile_id,
start_date=start_date, start_date=start_date,
metrics='ga:uniquePageviews', metrics='ga:uniquePageviews',
sort='-ga:uniquePageviews', sort='-ga:uniquePageviews',
dimensions="ga:socialNetwork,ga:referralPath", dimensions="ga:socialNetwork,ga:referralPath",
max_results=10000, max_results=10000,
end_date=end_date).execute() end_date=end_date).execute()
result_data = results.get('rows') result_data = results.get('rows')
data = {} data = {}
for result in result_data: for result in result_data:
if not result[0] == '(not set)': if not result[0] == '(not set)':
data[result[0]] = data.get(result[0], 0) + int(result[2]) data[result[0]] = data.get(result[0], 0) + int(result[2])
self._filter_out_long_tail(data, 3) self._filter_out_long_tail(data, 3)
ga_model.update_sitewide_stats(period_name, "Social sources", data) ga_model.update_sitewide_stats(period_name, "Social sources", data)
   
   
def _os_stats(self, start_date, end_date, period_name): def _os_stats(self, start_date, end_date, period_name):
""" Operating system stats """ """ Operating system stats """
results = self.service.data().ga().get( results = self.service.data().ga().get(
ids='ga:' + self.profile_id, ids='ga:' + self.profile_id,
start_date=start_date, start_date=start_date,
metrics='ga:uniquePageviews', metrics='ga:uniquePageviews',
sort='-ga:uniquePageviews', sort='-ga:uniquePageviews',
dimensions="ga:operatingSystem,ga:operatingSystemVersion", dimensions="ga:operatingSystem,ga:operatingSystemVersion",
max_results=10000, max_results=10000,
end_date=end_date).execute() end_date=end_date).execute()
result_data = results.get('rows') result_data = results.get('rows')
data = {} data = {}
for result in result_data: for result in result_data:
data[result[0]] = data.get(result[0], 0) + int(result[2]) data[result[0]] = data.get(result[0], 0) + int(result[2])
self._filter_out_long_tail(data, MIN_VIEWS) self._filter_out_long_tail(data, MIN_VIEWS)
ga_model.update_sitewide_stats(period_name, "Operating Systems", data) ga_model.update_sitewide_stats(period_name, "Operating Systems", data)
   
data = {} data = {}
for result in result_data: for result in result_data:
if int(result[2]) >= MIN_VIEWS: if int(result[2]) >= MIN_VIEWS:
key = "%s %s" % (result[0],result[1]) key = "%s %s" % (result[0],result[1])
data[key] = result[2] data[key] = result[2]
ga_model.update_sitewide_stats(period_name, "Operating Systems versions", data) ga_model.update_sitewide_stats(period_name, "Operating Systems versions", data)
   
   
def _browser_stats(self, start_date, end_date, period_name): def _browser_stats(self, start_date, end_date, period_name):
""" Information about browsers and browser versions """ """ Information about browsers and browser versions """
results = self.service.data().ga().get( results = self.service.data().ga().get(
ids='ga:' + self.profile_id, ids='ga:' + self.profile_id,
start_date=start_date, start_date=start_date,
metrics='ga:uniquePageviews', metrics='ga:uniquePageviews',
sort='-ga:uniquePageviews', sort='-ga:uniquePageviews',
dimensions="ga:browser,ga:browserVersion", dimensions="ga:browser,ga:browserVersion",
max_results=10000, max_results=10000,
end_date=end_date).execute() end_date=end_date).execute()
result_data = results.get('rows') result_data = results.get('rows')
# e.g. [u'Firefox', u'19.0', u'20'] # e.g. [u'Firefox', u'19.0', u'20']
   
data = {} data = {}
for result in result_data: for result in result_data:
data[result[0]] = data.get(result[0], 0) + int(result[2]) data[result[0]] = data.get(result[0], 0) + int(result[2])
self._filter_out_long_tail(data, MIN_VIEWS) self._filter_out_long_tail(data, MIN_VIEWS)
ga_model.update_sitewide_stats(period_name, "Browsers", data) ga_model.update_sitewide_stats(period_name, "Browsers", data)
   
data = {} data = {}
for result in result_data: for result in result_data:
key = "%s %s" % (result[0], self._filter_browser_version(result[0], result[1])) key = "%s %s" % (result[0], self._filter_browser_version(result[0], result[1]))
data[key] = data.get(key, 0) + int(result[2]) data[key] = data.get(key, 0) + int(result[2])
self._filter_out_long_tail(data, MIN_VIEWS) self._filter_out_long_tail(data, MIN_VIEWS)
ga_model.update_sitewide_stats(period_name, "Browser versions", data) ga_model.update_sitewide_stats(period_name, "Browser versions", data)
   
@classmethod @classmethod
def _filter_browser_version(cls, browser, version_str): def _filter_browser_version(cls, browser, version_str):
''' '''
Simplifies a browser version string if it is detailed. Simplifies a browser version string if it is detailed.
i.e. groups together Firefox 3.5.1 and 3.5.2 to be just 3. i.e. groups together Firefox 3.5.1 and 3.5.2 to be just 3.
This is helpful when viewing stats and good to protect privacy. This is helpful when viewing stats and good to protect privacy.
''' '''
ver = version_str ver = version_str
parts = ver.split('.') parts = ver.split('.')
if len(parts) > 1: if len(parts) > 1:
if parts[1][0] == '0': if parts[1][0] == '0':
ver = parts[0] ver = parts[0]
else: else:
ver = "%s" % (parts[0]) ver = "%s" % (parts[0])
# Special case complex version nums # Special case complex version nums
if browser in ['Safari', 'Android Browser']: if browser in ['Safari', 'Android Browser']:
ver = parts[0] ver = parts[0]
if len(ver) > 2: if len(ver) > 2:
num_hidden_digits = len(ver) - 2 num_hidden_digits = len(ver) - 2
ver = ver[0] + ver[1] + 'X' * num_hidden_digits ver = ver[0] + ver[1] + 'X' * num_hidden_digits
return ver return ver
   
def _mobile_stats(self, start_date, end_date, period_name): def _mobile_stats(self, start_date, end_date, period_name):
""" Info about mobile devices """ """ Info about mobile devices """
   
results = self.service.data().ga().get( results = self.service.data().ga().get(
ids='ga:' + self.profile_id, ids='ga:' + self.profile_id,
start_date=start_date, start_date=start_date,
metrics='ga:uniquePageviews', metrics='ga:uniquePageviews',
sort='-ga:uniquePageviews', sort='-ga:uniquePageviews',
dimensions="ga:mobileDeviceBranding, ga:mobileDeviceInfo", dimensions="ga:mobileDeviceBranding, ga:mobileDeviceInfo",
max_results=10000, max_results=10000,
end_date=end_date).execute() end_date=end_date).execute()
   
result_data = results.get('rows') result_data = results.get('rows')
data = {} data = {}
for result in result_data: for result in result_data:
data[result[0]] = data.get(result[0], 0) + int(result[2]) data[result[0]] = data.get(result[0], 0) + int(result[2])
self._filter_out_long_tail(data, MIN_VIEWS) self._filter_out_long_tail(data, MIN_VIEWS)
ga_model.update_sitewide_stats(period_name, "Mobile brands", data) ga_model.update_sitewide_stats(period_name, "Mobile brands", data)
   
data = {} data = {}
for result in result_data: for result in result_data:
data[result[1]] = data.get(result[1], 0) + int(result[2]) data[result[1]] = data.get(result[1], 0) + int(result[2])
self._filter_out_long_tail(data, MIN_VIEWS) self._filter_out_long_tail(data, MIN_VIEWS)
ga_model.update_sitewide_stats(period_name, "Mobile devices", data) ga_model.update_sitewide_stats(period_name, "Mobile devices", data)
   
@classmethod @classmethod
def _filter_out_long_tail(cls, data, threshold=10): def _filter_out_long_tail(cls, data, threshold=10):
''' '''
Given data which is a frequency distribution, filter out Given data which is a frequency distribution, filter out
results which are below a threshold count. This is good to protect results which are below a threshold count. This is good to protect
privacy. privacy.
''' '''
for key, value in data.items(): for key, value in data.items():
if value < threshold: if value < threshold:
del data[key] del data[key]
   
import re import re
import uuid import uuid
   
from sqlalchemy import Table, Column, MetaData, ForeignKey from sqlalchemy import Table, Column, MetaData, ForeignKey
from sqlalchemy import types from sqlalchemy import types
from sqlalchemy.sql import select from sqlalchemy.sql import select
from sqlalchemy.orm import mapper, relation from sqlalchemy.orm import mapper, relation
from sqlalchemy import func from sqlalchemy import func
   
import ckan.model as model import ckan.model as model
from ckan.lib.base import * from ckan.lib.base import *
   
def make_uuid(): def make_uuid():
return unicode(uuid.uuid4()) return unicode(uuid.uuid4())
   
metadata = MetaData() metadata = MetaData()
   
class GA_Url(object): class GA_Url(object):
   
def __init__(self, **kwargs): def __init__(self, **kwargs):
for k,v in kwargs.items(): for k,v in kwargs.items():
setattr(self, k, v) setattr(self, k, v)
   
url_table = Table('ga_url', metadata, url_table = Table('ga_url', metadata,
Column('id', types.UnicodeText, primary_key=True, Column('id', types.UnicodeText, primary_key=True,
default=make_uuid), default=make_uuid),
Column('period_name', types.UnicodeText), Column('period_name', types.UnicodeText),
Column('period_complete_day', types.Integer), Column('period_complete_day', types.Integer),
Column('pageviews', types.UnicodeText), Column('pageviews', types.UnicodeText),
Column('visitors', types.UnicodeText), Column('visitors', types.UnicodeText),
Column('url', types.UnicodeText), Column('url', types.UnicodeText),
Column('department_id', types.UnicodeText), Column('department_id', types.UnicodeText),
Column('package_id', types.UnicodeText), Column('package_id', types.UnicodeText),
) )
mapper(GA_Url, url_table) mapper(GA_Url, url_table)
   
   
class GA_Stat(object): class GA_Stat(object):
   
def __init__(self, **kwargs): def __init__(self, **kwargs):
for k,v in kwargs.items(): for k,v in kwargs.items():
setattr(self, k, v) setattr(self, k, v)
   
stat_table = Table('ga_stat', metadata, stat_table = Table('ga_stat', metadata,
Column('id', types.UnicodeText, primary_key=True, Column('id', types.UnicodeText, primary_key=True,
default=make_uuid), default=make_uuid),
Column('period_name', types.UnicodeText), Column('period_name', types.UnicodeText),
Column('stat_name', types.UnicodeText), Column('stat_name', types.UnicodeText),
Column('key', types.UnicodeText), Column('key', types.UnicodeText),
Column('value', types.UnicodeText), ) Column('value', types.UnicodeText), )
mapper(GA_Stat, stat_table) mapper(GA_Stat, stat_table)
   
   
class GA_Publisher(object): class GA_Publisher(object):
   
def __init__(self, **kwargs): def __init__(self, **kwargs):
for k,v in kwargs.items(): for k,v in kwargs.items():
setattr(self, k, v) setattr(self, k, v)
   
pub_table = Table('ga_publisher', metadata, pub_table = Table('ga_publisher', metadata,
Column('id', types.UnicodeText, primary_key=True, Column('id', types.UnicodeText, primary_key=True,
default=make_uuid), default=make_uuid),
Column('period_name', types.UnicodeText), Column('period_name', types.UnicodeText),
Column('publisher_name', types.UnicodeText), Column('publisher_name', types.UnicodeText),
Column('views', types.UnicodeText), Column('views', types.UnicodeText),
Column('visitors', types.UnicodeText), Column('visitors', types.UnicodeText),
Column('toplevel', types.Boolean, default=False), Column('toplevel', types.Boolean, default=False),
Column('subpublishercount', types.Integer, default=0), Column('subpublishercount', types.Integer, default=0),
Column('parent', types.UnicodeText), Column('parent', types.UnicodeText),
) )
mapper(GA_Publisher, pub_table) mapper(GA_Publisher, pub_table)
   
   
class GA_ReferralStat(object): class GA_ReferralStat(object):
   
def __init__(self, **kwargs): def __init__(self, **kwargs):
for k,v in kwargs.items(): for k,v in kwargs.items():
setattr(self, k, v) setattr(self, k, v)
   
referrer_table = Table('ga_referrer', metadata, referrer_table = Table('ga_referrer', metadata,
Column('id', types.UnicodeText, primary_key=True, Column('id', types.UnicodeText, primary_key=True,
default=make_uuid), default=make_uuid),
Column('period_name', types.UnicodeText), Column('period_name', types.UnicodeText),
Column('source', types.UnicodeText), Column('source', types.UnicodeText),
Column('url', types.UnicodeText), Column('url', types.UnicodeText),
Column('count', types.Integer), Column('count', types.Integer),
) )
mapper(GA_ReferralStat, referrer_table) mapper(GA_ReferralStat, referrer_table)
   
   
   
def init_tables(): def init_tables():
metadata.create_all(model.meta.engine) metadata.create_all(model.meta.engine)
   
   
cached_tables = {} cached_tables = {}
   
   
def get_table(name): def get_table(name):
if name not in cached_tables: if name not in cached_tables:
meta = MetaData() meta = MetaData()
meta.reflect(bind=model.meta.engine) meta.reflect(bind=model.meta.engine)
table = meta.tables[name] table = meta.tables[name]
cached_tables[name] = table cached_tables[name] = table
return cached_tables[name] return cached_tables[name]
   
   
def _normalize_url(url): def _normalize_url(url):
'''Strip off the hostname etc. Do this before storing it. '''Strip off the hostname etc. Do this before storing it.
   
>>> normalize_url('http://data.gov.uk/dataset/weekly_fuel_prices') >>> normalize_url('http://data.gov.uk/dataset/weekly_fuel_prices')
'/dataset/weekly_fuel_prices' '/dataset/weekly_fuel_prices'
''' '''
# Deliberately leaving a / # Deliberately leaving a /
url = url.replace('http:/','') url = url.replace('http:/','')
return '/' + '/'.join(url.split('/')[2:]) return '/' + '/'.join(url.split('/')[2:])
   
   
def _get_department_id_of_url(url): def _get_department_id_of_url(url):
# e.g. /dataset/fuel_prices # e.g. /dataset/fuel_prices
# e.g. /dataset/fuel_prices/resource/e63380d4 # e.g. /dataset/fuel_prices/resource/e63380d4
dataset_match = re.match('/dataset/([^/]+)(/.*)?', url) dataset_match = re.match('/dataset/([^/]+)(/.*)?', url)
if dataset_match: if dataset_match:
dataset_ref = dataset_match.groups()[0] dataset_ref = dataset_match.groups()[0]
dataset = model.Package.get(dataset_ref) dataset = model.Package.get(dataset_ref)
if dataset: if dataset:
publisher_groups = dataset.get_groups('publisher') publisher_groups = dataset.get_groups('publisher')
if publisher_groups: if publisher_groups:
return publisher_groups[0].name return publisher_groups[0].name
else: else:
publisher_match = re.match('/publisher/([^/]+)(/.*)?', url) publisher_match = re.match('/publisher/([^/]+)(/.*)?', url)
if publisher_match: if publisher_match:
return publisher_match.groups()[0] return publisher_match.groups()[0]
   
   
def update_sitewide_stats(period_name, stat_name, data): def update_sitewide_stats(period_name, stat_name, data):
for k,v in data.iteritems(): for k,v in data.iteritems():
item = model.Session.query(GA_Stat).\ item = model.Session.query(GA_Stat).\
filter(GA_Stat.period_name==period_name).\ filter(GA_Stat.period_name==period_name).\
filter(GA_Stat.key==k).\ filter(GA_Stat.key==k).\
filter(GA_Stat.stat_name==stat_name).first() filter(GA_Stat.stat_name==stat_name).first()
if item: if item:
item.period_name = period_name item.period_name = period_name
item.key = k item.key = k
item.value = v item.value = v
model.Session.add(item) model.Session.add(item)
else: else:
# create the row # create the row
values = {'id': make_uuid(), values = {'id': make_uuid(),
'period_name': period_name, 'period_name': period_name,
'key': k, 'key': k,
'value': v, 'value': v,
'stat_name': stat_name 'stat_name': stat_name
} }
model.Session.add(GA_Stat(**values)) model.Session.add(GA_Stat(**values))
model.Session.commit() model.Session.commit()
   
   
  def update_url_stat_totals(period_name):
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)  
   
package = None  
if url.startswith('/dataset/'):  
package = url[len('/dataset/'):]  
   
# see if the row for this url & month is in the table already  
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.package_id = package  
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,  
'package_id': package  
}  
model.Session.add(GA_Url(**values))  
   
# We now need to recaculate the ALL time_period from the data we have  
# Delete the old 'All'  
old = model.Session.query(GA_Url).\  
filter(GA_Url.period_name == "All").\  
filter(GA_Url.url==url).delete()  
   
items = model.Session.query(GA_Url).\ items = model.Session.query(GA_Url).\
filter(GA_Url.period_name != "All").\ filter(GA_Url.period_name != "All").\
filter(GA_Url.url==url).all() filter(GA_Url.url==url).all()
values = {'id': make_uuid(), values = {'id': make_uuid(),
'period_name': "All", 'period_name': "All",
'period_complete_day': "0", 'period_complete_day': "0",
'url': url, 'url': url,
'pageviews': sum([int(x.pageviews) for x in items]), 'pageviews': sum([int(x.pageviews) for x in items]),
'visitors': sum([int(x.visitors) for x in items]), 'visitors': sum([int(x.visitors) for x in items]),
'department_id': department_id, 'department_id': department_id,
'package_id': package 'package_id': package
} }
model.Session.add(GA_Url(**values)) model.Session.add(GA_Url(**values))
   
model.Session.commit() 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:
  department_id = _get_department_id_of_url(url)
   
  package = None
  if url.startswith('/dataset/'):
  package = url[len('/dataset/'):]
   
  values = {'id': make_uuid(),
  'period_name': period_name,
  'period_complete_day': period_complete_day,
  'url': url,
  'pageviews': views,
  'visitors': visitors,
  'department_id': department_id,
  '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': department_id,
  'package_id': package
  }
  model.Session.add(GA_Url(**values))
  model.Session.commit()
   
   
   
   
   
def update_social(period_name, data): def update_social(period_name, data):
# Clean up first. # Clean up first.
model.Session.query(GA_ReferralStat).\ model.Session.query(GA_ReferralStat).\
filter(GA_ReferralStat.period_name==period_name).delete() filter(GA_ReferralStat.period_name==period_name).delete()
   
for url,data in data.iteritems(): for url,data in data.iteritems():
for entry in data: for entry in data:
source = entry[0] source = entry[0]
count = entry[1] count = entry[1]
   
item = model.Session.query(GA_ReferralStat).\ item = model.Session.query(GA_ReferralStat).\
filter(GA_ReferralStat.period_name==period_name).\ filter(GA_ReferralStat.period_name==period_name).\
filter(GA_ReferralStat.source==source).\ filter(GA_ReferralStat.source==source).\
filter(GA_ReferralStat.url==url).first() filter(GA_ReferralStat.url==url).first()
if item: if item:
item.count = item.count + count item.count = item.count + count
model.Session.add(item) model.Session.add(item)
else: else:
# create the row # create the row
values = {'id': make_uuid(), values = {'id': make_uuid(),
'period_name': period_name, 'period_name': period_name,
'source': source, 'source': source,
'url': url, 'url': url,
'count': count, 'count': count,
} }
model.Session.add(GA_ReferralStat(**values)) model.Session.add(GA_ReferralStat(**values))
model.Session.commit() model.Session.commit()
   
def update_publisher_stats(period_name): def update_publisher_stats(period_name):
""" """
Updates the publisher stats from the data retrieved for /dataset/* Updates the publisher stats from the data retrieved for /dataset/*
and /publisher/*. Will run against each dataset and generates the and /publisher/*. Will run against each dataset and generates the
totals for the entire tree beneath each publisher. totals for the entire tree beneath each publisher.
""" """
toplevel = get_top_level() toplevel = get_top_level()
publishers = model.Session.query(model.Group).\ publishers = model.Session.query(model.Group).\
filter(model.Group.type=='publisher').\ filter(model.Group.type=='publisher').\
filter(model.Group.state=='active').all() filter(model.Group.state=='active').all()
for publisher in publishers: for publisher in publishers:
views, visitors, subpub = update_publisher(period_name, publisher, publisher.name) views, visitors, subpub = update_publisher(period_name, publisher, publisher.name)
parent, parents = '', publisher.get_groups('publisher') parent, parents = '', publisher.get_groups('publisher')
if parents: if parents:
parent = parents[0].name parent = parents[0].name
item = model.Session.query(GA_Publisher).\ item = model.Session.query(GA_Publisher).\
filter(GA_Publisher.period_name==period_name).\ filter(GA_Publisher.period_name==period_name).\
filter(GA_Publisher.publisher_name==publisher.name).first() filter(GA_Publisher.publisher_name==publisher.name).first()
if item: if item:
item.views = views item.views = views
item.visitors = visitors item.visitors = visitors
item.publisher_name = publisher.name item.publisher_name = publisher.name
item.toplevel = publisher in toplevel item.toplevel = publisher in toplevel
item.subpublishercount = subpub item.subpublishercount = subpub
item.parent = parent item.parent = parent
model.Session.add(item) model.Session.add(item)
else: else:
# create the row # create the row
values = {'id': make_uuid(), values = {'id': make_uuid(),
'period_name': period_name, 'period_name': period_name,
'publisher_name': publisher.name, 'publisher_name': publisher.name,
'views': views, 'views': views,
'visitors': visitors, 'visitors': visitors,
'toplevel': publisher in toplevel, 'toplevel': publisher in toplevel,
'subpublishercount': subpub, 'subpublishercount': subpub,
'parent': parent 'parent': parent
} }
model.Session.add(GA_Publisher(**values)) model.Session.add(GA_Publisher(**values))
model.Session.commit() model.Session.commit()
   
   
def update_publisher(period_name, pub, part=''): def update_publisher(period_name, pub, part=''):
views,visitors,subpub = 0, 0, 0 views,visitors,subpub = 0, 0, 0
for publisher in go_down_tree(pub): for publisher in go_down_tree(pub):
subpub = subpub + 1 subpub = subpub + 1
items = model.Session.query(GA_Url).\ items = model.Session.query(GA_Url).\
filter(GA_Url.period_name==period_name).\ filter(GA_Url.period_name==period_name).\
filter(GA_Url.department_id==publisher.name).all() filter(GA_Url.department_id==publisher.name).all()
for item in items: for item in items:
views = views + int(item.pageviews) views = views + int(item.pageviews)
visitors = visitors + int(item.visitors) visitors = visitors + int(item.visitors)
   
return views, visitors, (subpub-1) return views, visitors, (subpub-1)
   
   
def get_top_level(): def get_top_level():
'''Returns the top level publishers.''' '''Returns the top level publishers.'''
return model.Session.query(model.Group).\ return model.Session.query(model.Group).\
outerjoin(model.Member, model.Member.table_id == model.Group.id and \ outerjoin(model.Member, model.Member.table_id == model.Group.id and \
model.Member.table_name == 'group' and \ model.Member.table_name == 'group' and \
model.Member.state == 'active').\ model.Member.state == 'active').\
filter(model.Member.id==None).\ filter(model.Member.id==None).\
filter(model.Group.type=='publisher').\ filter(model.Group.type=='publisher').\
order_by(model.Group.name).all() order_by(model.Group.name).all()
   
def get_children(publisher): def get_children(publisher):
'''Finds child publishers for the given publisher (object). (Not recursive)''' '''Finds child publishers for the given publisher (object). (Not recursive)'''
from ckan.model.group import HIERARCHY_CTE from ckan.model.group import HIERARCHY_CTE
return model.Session.query(model.Group).\ return model.Session.query(model.Group).\
from_statement(HIERARCHY_CTE).params(id=publisher.id, type='publisher').\ from_statement(HIERARCHY_CTE).params(id=publisher.id, type='publisher').\
all() all()
   
def go_down_tree(publisher): def go_down_tree(publisher):
'''Provided with a publisher object, it walks down the hierarchy and yields each publisher, '''Provided with a publisher object, it walks down the hierarchy and yields each publisher,
including the one you supply.''' including the one you supply.'''
yield publisher yield publisher
for child in get_children(publisher): for child in get_children(publisher):
for grandchild in go_down_tree(child): for grandchild in go_down_tree(child):
yield grandchild yield grandchild
   
def delete(period_name): def delete(period_name):
''' '''
Deletes table data for the specified period, or specify 'all' Deletes table data for the specified period, or specify 'all'
for all periods. for all periods.
''' '''
for object_type in (GA_Url, GA_Stat, GA_Publisher, GA_ReferralStat): for object_type in (GA_Url, GA_Stat, GA_Publisher, GA_ReferralStat):
q = model.Session.query(object_type) q = model.Session.query(object_type)
if period_name != 'all': if period_name != 'all':
q = q.filter_by(period_name=period_name) q = q.filter_by(period_name=period_name)
q.delete() q.delete()
model.Session.commit() model.Session.commit()
   
<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>
<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">
<select name="month"> <select name="month">
<option value='' py:attrs="{'selected': 'selected' if not c.month else None}">All months</option> <option value='' py:attrs="{'selected': 'selected' if not c.month else None}">All months</option>
   
<py:for each="val,desc in c.months"> <py:for each="val,desc in c.months">
<option value='${val}' py:attrs="{'selected': 'selected' if c.month == val else None}">${desc}</option> <option value='${val}' py:attrs="{'selected': 'selected' if c.month == val else None}">${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>
   
<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 Views</th> <th>Dataset Views</th>
<th>Dataset Visits</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>${h.link_to(publisher.title, h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport', action='read_publisher', id=publisher.name))} <td>${h.link_to(publisher.title, h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport', action='read_publisher', id=publisher.name))}
</td> </td>
  <td>${visits}</td>
<td>${views}</td> <td>${views}</td>
<td>${visits}</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>
<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">
<select name="month"> <select name="month">
<option value='' py:attrs="{'selected': 'selected' if not c.month else None}">All months</option> <option value='' py:attrs="{'selected': 'selected' if not c.month else None}">All months</option>
<py:for each="val,desc in c.months"> <py:for each="val,desc in c.months">
<option value='${val}' py:attrs="{'selected': 'selected' if c.month == val else None}">${desc}</option> <option value='${val}' py:attrs="{'selected': 'selected' if c.month == val else None}">${desc}</option>
</py:for> </py:for>
</select> </select>
<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>Views</th> <th>Views</th>
<th>Visits</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>${visits}</td>
<td>${views}</td> <td>${views}</td>
<td>${visits}</td>  
</tr> </tr>
</py:for> </py:for>
</table> </table>
   
   
</div> </div>
   
<xi:include href="../../layout.html" /> <xi:include href="../../layout.html" />
</html> </html>