#167 Refactoring controller's logic to create better Percentage graphs
#167 Refactoring controller's logic to create better Percentage graphs

import re import re
import csv import csv
import sys import sys
import json import json
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, GA_Publisher from ga_model import GA_Url, GA_Stat, GA_ReferralStat, GA_Publisher
   
log = logging.getLogger('ckanext.ga-report') log = logging.getLogger('ckanext.ga-report')
   
DOWNLOADS_AVAILABLE_FROM = '2012-12' DOWNLOADS_AVAILABLE_FROM = '2012-12'
   
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 _get_unix_epoch(strdate): def _get_unix_epoch(strdate):
from time import strptime,mktime from time import strptime,mktime
d = strptime(strdate, '%Y-%m') d = strptime(strdate, '%Y-%m')
return int(mktime(d)) return int(mktime(d))
   
def _month_details(cls, stat_key=None): def _month_details(cls, stat_key=None):
''' '''
Returns a list of all the periods for which we have data, unfortunately Returns a list of all the periods for which we have data, unfortunately
knows too much about the type of the cls being passed as GA_Url has a knows too much about the type of the cls being passed as GA_Url has a
more complex query more complex query
   
This may need extending if we add a period_name to the stats This may need extending if we add a period_name to the stats
''' '''
months = [] months = []
day = None day = None
   
q = model.Session.query(cls.period_name,cls.period_complete_day)\ q = model.Session.query(cls.period_name,cls.period_complete_day)\
.filter(cls.period_name!='All').distinct(cls.period_name) .filter(cls.period_name!='All').distinct(cls.period_name)
if stat_key: if stat_key:
q= q.filter(cls.stat_name==stat_key) q= q.filter(cls.stat_name==stat_key)
   
vals = q.order_by("period_name desc").all() vals = q.order_by("period_name desc").all()
   
if vals and vals[0][1]: if vals and vals[0][1]:
day = int(vals[0][1]) day = int(vals[0][1])
ordinal = 'th' if 11 <= day <= 13 \ ordinal = 'th' if 11 <= day <= 13 \
else {1:'st',2:'nd',3:'rd'}.get(day % 10, 'th') else {1:'st',2:'nd',3:'rd'}.get(day % 10, 'th')
day = "{day}{ordinal}".format(day=day, ordinal=ordinal) day = "{day}{ordinal}".format(day=day, ordinal=ordinal)
   
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 months, day return months, day
   
   
class GaReport(BaseController): class GaReport(BaseController):
   
def csv(self, month): def csv(self, month):
import csv import csv
   
q = model.Session.query(GA_Stat).filter(GA_Stat.stat_name!='Downloads') q = model.Session.query(GA_Stat).filter(GA_Stat.stat_name!='Downloads')
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, c.day = _month_details(GA_Stat) c.months, c.day = _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', 'Bounce rate (home page)']: if key in ['Average time on site', 'Pages per visit', 'New visits', 'Bounce rate (home page)']:
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','Bounce rate (home page)']: if key in ['New visits','Bounce rate (home page)']:
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
   
# Query historic values for sparkline rendering # Query historic values for sparkline rendering
sparkline_query = model.Session.query(GA_Stat)\ sparkline_query = model.Session.query(GA_Stat)\
.filter(GA_Stat.stat_name=='Totals')\ .filter(GA_Stat.stat_name=='Totals')\
.order_by(GA_Stat.period_name) .order_by(GA_Stat.period_name)
sparkline_data = {} sparkline_data = {}
for x in sparkline_query: for x in sparkline_query:
sparkline_data[x.key] = sparkline_data.get(x.key,[]) sparkline_data[x.key] = sparkline_data.get(x.key,[])
key, val = clean_key(x.key,float(x.value)) key, val = clean_key(x.key,float(x.value))
tooltip = '%s: %s' % (_get_month_name(x.period_name), val) tooltip = '%s: %s' % (_get_month_name(x.period_name), val)
sparkline_data[x.key].append( (tooltip,x.value) ) sparkline_data[x.key].append( (tooltip,x.value) )
# Trim the latest month, as it looks like a huge dropoff # Trim the latest month, as it looks like a huge dropoff
for key in sparkline_data: for key in sparkline_data:
sparkline_data[key] = sparkline_data[key][:-1] sparkline_data[key] = sparkline_data[key][:-1]
   
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)
sparkline = sparkline_data[e.key] sparkline = sparkline_data[e.key]
c.global_totals.append((key, val, sparkline)) c.global_totals.append((key, val, sparkline))
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))/float(len(v)) v = float(sum(v))/float(len(v))
sparkline = sparkline_data[k] sparkline = sparkline_data[k]
key, val = clean_key(k,v) key, val = clean_key(k,v)
   
c.global_totals.append((key, val, sparkline)) c.global_totals.append((key, val, sparkline))
# Sort the global totals into a more pleasant order # Sort the global totals into a more pleasant order
def sort_func(x): def sort_func(x):
key = x[0] key = x[0]
total_order = ['Total page views','Total visits','Pages per visit'] total_order = ['Total page views','Total visits','Pages per visit']
if key in total_order: if key in total_order:
return total_order.index(key) return total_order.index(key)
return 999 return 999
c.global_totals = sorted(c.global_totals, key=sort_func) c.global_totals = sorted(c.global_totals, key=sort_func)
   
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).\
order_by(GA_Stat.period_name) order_by(GA_Stat.period_name)
# Run the query on all months to gather graph data # Run the query on all months to gather graph data
graph = {} graph = {}
for stat in q: for stat in q:
graph[ stat.key ] = graph.get(stat.key,{ graph[ stat.key ] = graph.get(stat.key,{
'name':stat.key, 'name':stat.key,
'data': [] 'data': []
}) })
graph[ stat.key ]['data'].append({ graph[ stat.key ]['data'].append({
'x':_get_unix_epoch(stat.period_name), 'x':_get_unix_epoch(stat.period_name),
'y':float(stat.value) 'y':float(stat.value)
}) })
setattr(c, v+'_graph', json.dumps( _to_rickshaw(graph.values(),percentageMode=True) )) setattr(c, v+'_graph', json.dumps( _to_rickshaw(graph.values(),percentageMode=True) ))
   
# Buffer the tabular data # Buffer the tabular data
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,graph in c.global_totals if n == 'Total visits']) total = sum([x for n,x,graph 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"])
   
top_publishers, top_publishers_graph = _get_top_publishers(None) top_publishers, top_publishers_graph = _get_top_publishers(None)
   
for publisher,view,visit in top_publishers: for publisher,view,visit in top_publishers:
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", "Resource downloads", "Period Name"]) writer.writerow(["Dataset Title", "Dataset Name", "Views", "Visits", "Resource downloads", "Period Name"])
   
for package,view,visit,downloads in packages: for package,view,visit,downloads 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,
downloads, downloads,
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, c.day = _month_details(GA_Url) c.months, c.day = _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, graph_data = _get_top_publishers() c.top_publishers, graph_data = _get_top_publishers()
c.top_publishers_graph = json.dumps( _to_rickshaw(graph_data.values()) ) c.top_publishers_graph = json.dumps( _to_rickshaw(graph_data.values()) )
   
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 views''' '''Returns the datasets in order of views'''
have_download_data = True have_download_data = True
month = c.month or 'All' month = c.month or 'All'
if month != 'All': if month != 'All':
have_download_data = month >= DOWNLOADS_AVAILABLE_FROM have_download_data = month >= DOWNLOADS_AVAILABLE_FROM
   
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.pageviews::int desc') q = q.order_by('ga_url.pageviews::int desc')
top_packages = [] top_packages = []
if count == -1: if count == -1:
entries = q.all() entries = q.all()
else: else:
entries = q.limit(count) entries = q.limit(count)
   
for entry,package in entries: for entry,package in entries:
if package: if package:
# Downloads .... # Downloads ....
if have_download_data: if have_download_data:
dls = model.Session.query(GA_Stat).\ dls = model.Session.query(GA_Stat).\
filter(GA_Stat.stat_name=='Downloads').\ filter(GA_Stat.stat_name=='Downloads').\
filter(GA_Stat.key==package.name) filter(GA_Stat.key==package.name)