Refactoring and tidying Rickshaw code. Placeholder graphs inserted. Front page sparklines are sensibly sorted.
--- a/.gitignore
+++ b/.gitignore
@@ -1,6 +1,7 @@
*.py[co]
*.py~
.gitignore
+ckan.log
# Packages
*.egg
--- a/README.rst
+++ b/README.rst
@@ -32,6 +32,7 @@
googleanalytics.id = UA-1010101-1
googleanalytics.account = Account name (e.g. data.gov.uk, see top level item at https://www.google.com/analytics)
+ googleanalytics.token.filepath = ~/pyenv/token.dat
ga-report.period = monthly
ga-report.bounce_url = /
@@ -82,13 +83,17 @@
$ paster getauthtoken --config=../ckan/development.ini
+Now ensure you reference the correct path to your token.dat in your CKAN config file (e.g. development.ini)::
+
+ googleanalytics.token.filepath = ~/pyenv/token.dat
+
Tutorial
--------
Download some GA data and store it in CKAN's database. (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) and specifying the name of your auth file (token.dat by default) from the previous step::
- $ paster loadanalytics token.dat latest --config=../ckan/development.ini
+ $ paster loadanalytics latest --config=../ckan/development.ini
The value after the token file is how much data you want to retrieve, this can be
--- a/ckanext/ga_report/command.py
+++ b/ckanext/ga_report/command.py
@@ -1,5 +1,8 @@
import logging
import datetime
+import os
+
+from pylons import config
from ckan.lib.cli import CkanCommand
# No other CKAN imports allowed until _load_config is run,
@@ -53,25 +56,52 @@
self.args[0] if self.args
else 'credentials.json')
+class FixTimePeriods(CkanCommand):
+ """
+ Fixes the 'All' records for GA_Urls
+
+ It is possible that older urls that haven't recently been visited
+ do not have All records. This command will traverse through those
+ records and generate valid All records for them.
+ """
+ summary = __doc__.split('\n')[0]
+ usage = __doc__
+ max_args = 0
+ min_args = 0
+
+ def __init__(self, name):
+ super(FixTimePeriods, self).__init__(name)
+
+ def command(self):
+ import ckan.model as model
+ from ga_model import post_update_url_stats
+ self._load_config()
+ model.Session.remove()
+ model.Session.configure(bind=model.meta.engine)
+
+ log = logging.getLogger('ckanext.ga_report')
+
+ log.info("Updating 'All' records for old URLs")
+ post_update_url_stats()
+ log.info("Processing complete")
+
+
class LoadAnalytics(CkanCommand):
"""Get data from Google Analytics API and save it
in the ga_model
- Usage: paster loadanalytics <tokenfile> <time-period>
+ Usage: paster loadanalytics <time-period>
- Where <tokenfile> is the name of the auth token file from
- the getauthtoken step.
-
- And where <time-period> is:
+ Where <time-period> is:
all - data for all time
latest - (default) just the 'latest' data
YYYY-MM - just data for the specific month
"""
summary = __doc__.split('\n')[0]
usage = __doc__
- max_args = 2
- min_args = 1
+ max_args = 1
+ min_args = 0
def __init__(self, name):
super(LoadAnalytics, self).__init__(name)
@@ -92,19 +122,25 @@
from download_analytics import DownloadAnalytics
from ga_auth import (init_service, get_profile_id)
+ ga_token_filepath = os.path.expanduser(config.get('googleanalytics.token.filepath', ''))
+ if not ga_token_filepath:
+ print 'ERROR: In the CKAN config you need to specify the filepath of the ' \
+ 'Google Analytics token file under key: googleanalytics.token.filepath'
+ return
+
try:
- svc = init_service(self.args[0], None)
+ svc = init_service(ga_token_filepath, None)
except TypeError:
print ('Have you correctly run the getauthtoken task and '
- 'specified the correct token file?')
+ 'specified the correct token file in the CKAN config under '
+ '"googleanalytics.token.filepath"?')
return
downloader = DownloadAnalytics(svc, profile_id=get_profile_id(svc),
delete_first=self.options.delete_first,
skip_url_stats=self.options.skip_url_stats)
- time_period = self.args[1] if self.args and len(self.args) > 1 \
- else 'latest'
+ time_period = self.args[0] if self.args else 'latest'
if time_period == 'all':
downloader.all_()
elif time_period == 'latest':
--- a/ckanext/ga_report/controller.py
+++ b/ckanext/ga_report/controller.py
@@ -1,6 +1,7 @@
import re
import csv
import sys
+import json
import logging
import operator
import collections
@@ -13,6 +14,7 @@
log = logging.getLogger('ckanext.ga-report')
+DOWNLOADS_AVAILABLE_FROM = '2012-12'
def _get_month_name(strdate):
import calendar
@@ -20,14 +22,39 @@
d = strptime(strdate, '%Y-%m')
return '%s %s' % (calendar.month_name[d.tm_mon], d.tm_year)
-
-def _month_details(cls):
- '''Returns a list of all the month names'''
+def _get_unix_epoch(strdate):
+ from time import strptime,mktime
+ d = strptime(strdate, '%Y-%m')
+ return int(mktime(d))
+
+def _month_details(cls, stat_key=None):
+ '''
+ 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
+ more complex query
+
+ This may need extending if we add a period_name to the stats
+ '''
months = []
- vals = model.Session.query(cls.period_name).filter(cls.period_name!='All').distinct().all()
+ day = None
+
+ q = model.Session.query(cls.period_name,cls.period_complete_day)\
+ .filter(cls.period_name!='All').distinct(cls.period_name)
+ if stat_key:
+ q= q.filter(cls.stat_name==stat_key)
+
+ vals = q.order_by("period_name desc").all()
+
+ if vals and vals[0][1]:
+ day = int(vals[0][1])
+ ordinal = 'th' if 11 <= day <= 13 \
+ else {1:'st',2:'nd',3:'rd'}.get(day % 10, 'th')
+ day = "{day}{ordinal}".format(day=day, ordinal=ordinal)
+
for m in vals:
months.append( (m[0], _get_month_name(m[0])))
- return sorted(months, key=operator.itemgetter(0), reverse=True)
+
+ return months, day
class GaReport(BaseController):
@@ -35,7 +62,7 @@
def csv(self, month):
import csv
- q = model.Session.query(GA_Stat)
+ q = model.Session.query(GA_Stat).filter(GA_Stat.stat_name!='Downloads')
if month != 'all':
q = q.filter(GA_Stat.period_name==month)
entries = q.order_by('GA_Stat.period_name, GA_Stat.stat_name, GA_Stat.key').all()
@@ -52,11 +79,12 @@
entry.key.encode('utf-8'),
entry.value.encode('utf-8')])
+
def index(self):
# Get the month details by fetching distinct values and determining the
# month names from the values.
- c.months = _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
c.month_desc = 'all months'
@@ -84,11 +112,26 @@
return key, val
+ # Query historic values for sparkline rendering
+ sparkline_query = model.Session.query(GA_Stat)\
+ .filter(GA_Stat.stat_name=='Totals')\
+ .order_by(GA_Stat.period_name)
+ sparkline_data = {}
+ for x in sparkline_query:
+ sparkline_data[x.key] = sparkline_data.get(x.key,[])
+ key, val = clean_key(x.key,float(x.value))
+ tooltip = '%s: %s' % (_get_month_name(x.period_name), val)
+ sparkline_data[x.key].append( (tooltip,x.value) )
+ # Trim the latest month, as it looks like a huge dropoff
+ for key in sparkline_data:
+ sparkline_data[key] = sparkline_data[key][:-1]
+
c.global_totals = []
if c.month:
for e in entries:
key, val = clean_key(e.key, e.value)
- c.global_totals.append((key, val))
+ sparkline = sparkline_data[e.key]
+ c.global_totals.append((key, val, sparkline))
else:
d = collections.defaultdict(list)
for e in entries:
@@ -97,11 +140,19 @@
if k in ['Total page views', 'Total visits']:
v = sum(v)
else:
- v = float(sum(v))/len(v)
+ v = float(sum(v))/float(len(v))
+ sparkline = sparkline_data[k]
key, val = clean_key(k,v)
- c.global_totals.append((key, val))
- c.global_totals = sorted(c.global_totals, key=operator.itemgetter(0))
+ c.global_totals.append((key, val, sparkline))
+ # Sort the global totals into a more pleasant order
+ def sort_func(x):
+ key = x[0]
+ total_order = ['Total page views','Total visits','Pages per visit']
+ if key in total_order:
+ return total_order.index(key)
+ return 999
+ c.global_totals = sorted(c.global_totals, key=sort_func)
keys = {
'Browser versions': 'browser_versions',
@@ -138,7 +189,29 @@
for k, v in keys.iteritems():
q = model.Session.query(GA_Stat).\
- filter(GA_Stat.stat_name==k)
+ filter(GA_Stat.stat_name==k).\
+ order_by(GA_Stat.period_name)
+ # Run the query on all months to gather graph data
+ series = {}
+ x_axis = set()
+ for stat in q:
+ x_val = _get_unix_epoch(stat.period_name)
+ series[ stat.key ] = series.get(stat.key,{})
+ series[ stat.key ][x_val] = float(stat.value)
+ x_axis.add(x_val)
+ # Common x-axis for all series. Exclude this month (incomplete data)
+ x_axis = sorted(list(x_axis))[:-1]
+ # Buffer a rickshaw dataset from the series
+ def create_graph(series_name, series_data):
+ return {
+ 'name':series_name,
+ 'data':[ {'x':x,'y':series_data.get(x,0)} for x in x_axis ]
+ }
+ rickshaw = [ create_graph(name,data) for name, data in series.items() ]
+ rickshaw = sorted(rickshaw,key=lambda x:x['data'][-1]['y'])
+ setattr(c, v+'_graph', json.dumps(rickshaw))
+
+ # Buffer the tabular data
if c.month:
entries = []
q = q.filter(GA_Stat.period_name==c.month).\
@@ -155,7 +228,7 @@
# 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'])
+ total = sum([x for n,x,graph 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 ])
@@ -206,13 +279,14 @@
str('attachment; filename=datasets_%s_%s.csv' % (c.publisher_name, month,))
writer = csv.writer(response)
- writer.writerow(["Dataset Title", "Dataset Name", "Views", "Visits", "Period Name"])
-
- for package,view,visit in packages:
+ writer.writerow(["Dataset Title", "Dataset Name", "Views", "Visits", "Resource downloads", "Period Name"])
+
+ for package,view,visit,downloads in packages:
writer.writerow([package.title.encode('utf-8'),
package.name.encode('utf-8'),
view,
visit,
+ downloads,
month])
def publishers(self):
@@ -220,7 +294,7 @@
# Get the month details by fetching distinct values and determining the
# month names from the values.
- c.months = _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
c.month = request.params.get('month', '')
@@ -233,10 +307,10 @@
def _get_packages(self, publisher=None, count=-1):
'''Returns the datasets in order of views'''
- if count == -1:
- count = sys.maxint
-
+ have_download_data = True
month = c.month or 'All'
+ if month != 'All':
+ have_download_data = month >= DOWNLOADS_AVAILABLE_FROM
q = model.Session.query(GA_Url,model.Package)\
.filter(model.Package.name==GA_Url.package_id)\
@@ -246,9 +320,25 @@
q = q.filter(GA_Url.period_name==month)
q = q.order_by('ga_url.pageviews::int desc')
top_packages = []
- for entry,package in q.limit(count):
+ if count == -1:
+ entries = q.all()
+ else:
+ entries = q.limit(count)
+
+ for entry,package in entries:
if package:
- top_packages.append((package, entry.pageviews, entry.visits))
+ # Downloads ....
+ if have_download_data:
+ dls = model.Session.query(GA_Stat).\
+ filter(GA_Stat.stat_name=='Downloads').\
+ filter(GA_Stat.key==package.name)
+ if month != 'All': # Fetch everything unless the month is specific
+ dls = dls.filter(GA_Stat.period_name==month)
+
+ downloads = sum(int(d.value) for d in dls.all())
+ else:
+ downloads = 'No data'
+ top_packages.append((package, entry.pageviews, entry.visits, downloads))
else:
log.warning('Could not find package associated package')
@@ -278,7 +368,7 @@
# Get the month details by fetching distinct values and determining the
# month names from the values.
- c.months = _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
c.month = request.params.get('month', '')
--- a/ckanext/ga_report/download_analytics.py
+++ b/ckanext/ga_report/download_analytics.py
@@ -13,6 +13,7 @@
FORMAT_MONTH = '%Y-%m'
MIN_VIEWS = 50
MIN_VISITS = 20
+MIN_DOWNLOADS = 10
class DownloadAnalytics(object):
'''Downloads and stores analytics info'''
@@ -122,11 +123,15 @@
log.info('Storing publisher views (%i rows)', len(data.get('url')))
self.store(period_name, period_complete_day, data,)
+ # Make sure the All records are correct.
+ ga_model.post_update_url_stats()
+
log.info('Aggregating datasets by publisher')
ga_model.update_publisher_stats(period_name) # about 30 seconds.
+
log.info('Downloading and storing analytics for site-wide stats')
- self.sitewide_stats( period_name )
+ self.sitewide_stats( period_name, period_complete_day )
log.info('Downloading and storing analytics for social networks')
self.update_social_info(period_name, start_date, end_date)
@@ -153,7 +158,8 @@
data = collections.defaultdict(list)
rows = results.get('rows',[])
for row in rows:
- data[_normalize_url(row[0])].append( (row[1], int(row[2]),) )
+ url = _normalize_url('http:/' + row[0])
+ data[url].append( (row[1], int(row[2]),) )
ga_model.update_social(period_name, data)
@@ -178,6 +184,7 @@
end_date=end_date).execute()
packages = []
+ log.info("There are %d results" % results['totalResults'])
for entry in results.get('rows'):
(loc,pageviews,visits) = entry
url = _normalize_url('http:/' + loc) # strips off domain e.g. www.data.gov.uk or data.gov.uk
@@ -194,7 +201,7 @@
if 'url' in data:
ga_model.update_url_stats(period_name, period_complete_day, data['url'])
- def sitewide_stats(self, period_name):
+ def sitewide_stats(self, period_name, period_complete_day):
import calendar
year, month = period_name.split('-')
_, last_day_of_month = calendar.monthrange(int(year), int(month))
@@ -202,10 +209,10 @@
start_date = '%s-01' % period_name
end_date = '%s-%s' % (period_name, last_day_of_month)
funcs = ['_totals_stats', '_social_stats', '_os_stats',
- '_locale_stats', '_browser_stats', '_mobile_stats']
+ '_locale_stats', '_browser_stats', '_mobile_stats', '_download_stats']
for f in funcs:
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, period_complete_day)
def _get_results(result_data, f):
data = {}
@@ -214,7 +221,7 @@
data[key] = data.get(key,0) + result[1]
return data
- def _totals_stats(self, start_date, end_date, period_name):
+ def _totals_stats(self, start_date, end_date, period_name, period_complete_day):
""" Fetches distinct totals, total pageviews etc """
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
@@ -224,7 +231,8 @@
max_results=10000,
end_date=end_date).execute()
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]},
+ period_complete_day)
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
@@ -239,7 +247,7 @@
'New visits': result_data[0][2],
'Total visits': result_data[0][3],
}
- ga_model.update_sitewide_stats(period_name, "Totals", data)
+ ga_model.update_sitewide_stats(period_name, "Totals", data, period_complete_day)
# Bounces from / or another configurable page.
path = '/%s%s' % (config.get('googleanalytics.account'),
@@ -248,7 +256,7 @@
ids='ga:' + self.profile_id,
filters='ga:pagePath==%s' % (path,),
start_date=start_date,
- metrics='ga:bounces,ga:pageviews',
+ metrics='ga:visitBounceRate',
dimensions='ga:pagePath',
max_results=10000,
end_date=end_date).execute()
@@ -258,13 +266,14 @@
path, result_data)
return
results = result_data[0]
- bounces, total = [float(x) for x in result_data[0][1:]]
- pct = 100 * bounces/total
- log.info('%d bounces from %d total == %s', bounces, total, pct)
- ga_model.update_sitewide_stats(period_name, "Totals", {'Bounce rate (home page)': pct})
-
-
- def _locale_stats(self, start_date, end_date, period_name):
+ bounces = float(results[1])
+ # visitBounceRate is already a %
+ log.info('Google reports visitBounceRate as %s', bounces)
+ ga_model.update_sitewide_stats(period_name, "Totals", {'Bounce rate (home page)': float(bounces)},
+ period_complete_day)
+
+
+ def _locale_stats(self, start_date, end_date, period_name, period_complete_day):
""" Fetches stats about language and country """
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
@@ -279,16 +288,72 @@
for result in result_data:
data[result[0]] = data.get(result[0], 0) + int(result[2])
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, period_complete_day)
data = {}
for result in result_data:
data[result[1]] = data.get(result[1], 0) + int(result[2])
self._filter_out_long_tail(data, MIN_VIEWS)
- ga_model.update_sitewide_stats(period_name, "Country", data)
-
-
- def _social_stats(self, start_date, end_date, period_name):
+ ga_model.update_sitewide_stats(period_name, "Country", data, period_complete_day)
+
+
+ def _download_stats(self, start_date, end_date, period_name, period_complete_day):
+ """ Fetches stats about language and country """
+ import ckan.model as model
+
+ data = {}
+
+ results = self.service.data().ga().get(
+ ids='ga:' + self.profile_id,
+ start_date=start_date,
+ filters='ga:eventAction==download',
+ metrics='ga:totalEvents',
+ sort='-ga:totalEvents',
+ dimensions="ga:eventLabel",
+ max_results=10000,
+ end_date=end_date).execute()
+ result_data = results.get('rows')
+ if not result_data:
+ # We may not have data for this time period, so we need to bail
+ # early.
+ log.info("There is no download data for this time period")
+ return
+
+ def process_result_data(result_data, cached=False):
+ for result in result_data:
+ url = result[0].strip()
+
+ # Get package id associated with the resource that has this URL.
+ q = model.Session.query(model.Resource)
+ if cached:
+ r = q.filter(model.Resource.cache_url.like("%s%%" % url)).first()
+ else:
+ r = q.filter(model.Resource.url.like("%s%%" % url)).first()
+
+ package_name = r.resource_group.package.name if r else ""
+ if package_name:
+ data[package_name] = data.get(package_name, 0) + int(result[1])
+ else:
+ log.warning(u"Could not find resource for URL: {url}".format(url=url))
+ continue
+
+ process_result_data(results.get('rows'))
+
+ results = self.service.data().ga().get(
+ ids='ga:' + self.profile_id,
+ start_date=start_date,
+ filters='ga:eventAction==download-cache',
+ metrics='ga:totalEvents',
+ sort='-ga:totalEvents',
+ dimensions="ga:eventLabel",
+ max_results=10000,
+ end_date=end_date).execute()
+ process_result_data(results.get('rows'), cached=False)
+
+ self._filter_out_long_tail(data, MIN_DOWNLOADS)
+ ga_model.update_sitewide_stats(period_name, "Downloads", data, period_complete_day)
+
+ def _social_stats(self, start_date, end_date, period_name, period_complete_day):
""" Finds out which social sites people are referred from """
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
@@ -304,10 +369,10 @@
if not result[0] == '(not set)':
data[result[0]] = data.get(result[0], 0) + int(result[2])
self._filter_out_long_tail(data, 3)
- ga_model.update_sitewide_stats(period_name, "Social sources", data)
-
-
- def _os_stats(self, start_date, end_date, period_name):
+ ga_model.update_sitewide_stats(period_name, "Social sources", data, period_complete_day)
+
+
+ def _os_stats(self, start_date, end_date, period_name, period_complete_day):
""" Operating system stats """
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
@@ -322,17 +387,17 @@
for result in result_data:
data[result[0]] = data.get(result[0], 0) + int(result[2])
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, period_complete_day)
data = {}
for result in result_data:
if int(result[2]) >= MIN_VIEWS:
key = "%s %s" % (result[0],result[1])
data[key] = result[2]
- ga_model.update_sitewide_stats(period_name, "Operating Systems versions", data)
-
-
- def _browser_stats(self, start_date, end_date, period_name):
+ ga_model.update_sitewide_stats(period_name, "Operating Systems versions", data, period_complete_day)
+
+
+ def _browser_stats(self, start_date, end_date, period_name, period_complete_day):
""" Information about browsers and browser versions """
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
@@ -349,14 +414,14 @@
for result in result_data:
data[result[0]] = data.get(result[0], 0) + int(result[2])
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, period_complete_day)
data = {}
for result in result_data:
key = "%s %s" % (result[0], self._filter_browser_version(result[0], result[1]))
data[key] = data.get(key, 0) + int(result[2])
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, period_complete_day)
@classmethod
def _filter_browser_version(cls, browser, version_str):
@@ -380,7 +445,7 @@
ver = ver[0] + ver[1] + 'X' * num_hidden_digits
return ver
- def _mobile_stats(self, start_date, end_date, period_name):
+ def _mobile_stats(self, start_date, end_date, period_name, period_complete_day):
""" Info about mobile devices """
results = self.service.data().ga().get(
@@ -397,13 +462,13 @@
for result in result_data:
data[result[0]] = data.get(result[0], 0) + int(result[2])
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, period_complete_day)
data = {}
for result in result_data:
data[result[1]] = data.get(result[1], 0) + int(result[2])
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, period_complete_day)
@classmethod
def _filter_out_long_tail(cls, data, threshold=10):
--- a/ckanext/ga_report/ga_model.py
+++ b/ckanext/ga_report/ga_model.py
@@ -47,6 +47,7 @@
Column('id', types.UnicodeText, primary_key=True,
default=make_uuid),
Column('period_name', types.UnicodeText),
+ Column('period_complete_day', types.UnicodeText),
Column('stat_name', types.UnicodeText),
Column('key', types.UnicodeText),
Column('value', types.UnicodeText), )
@@ -134,7 +135,7 @@
return None, publisher_match.groups()[0]
return None, None
-def update_sitewide_stats(period_name, stat_name, data):
+def update_sitewide_stats(period_name, stat_name, data, period_complete_day):
for k,v in data.iteritems():
item = model.Session.query(GA_Stat).\
filter(GA_Stat.period_name==period_name).\
@@ -144,11 +145,13 @@
item.period_name = period_name
item.key = k
item.value = v
+ item.period_complete_day = period_complete_day
model.Session.add(item)
else:
# create the row
values = {'id': make_uuid(),
'period_name': period_name,
+ 'period_complete_day': period_complete_day,
'key': k,
'value': v,
'stat_name': stat_name
@@ -158,10 +161,56 @@
def pre_update_url_stats(period_name):
+ log.debug("Deleting '%s' records" % 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()
+
+ count = model.Session.query(GA_Url).\
+ filter(GA_Url.period_name == 'All').count()
+ log.debug("Deleting %d 'All' records" % count)
+ count = model.Session.query(GA_Url).\
+ filter(GA_Url.period_name == 'All').delete()
+ log.debug("Deleted %d 'All' records" % count)
+
+ model.Session.flush()
+ model.Session.commit()
+ model.repo.commit_and_remove()
+
+def post_update_url_stats():
+
+ """ Check the distinct url field in ga_url and make sure
+ it has an All record. If not then create one.
+
+ After running this then every URL should have an All
+ record regardless of whether the URL has an entry for
+ the month being currently processed.
+ """
+ query = """select url, pageviews::int, visits::int
+ from ga_url
+ where url not in (select url from ga_url where period_name ='All')"""
+ connection = model.Session.connection()
+ res = connection.execute(query)
+
+ views, visits = {}, {}
+ # url, views, visits
+ for row in res:
+ views[row[0]] = views.get(row[0], 0) + row[1]
+ visits[row[0]] = visits.get(row[0], 0) + row[2]
+
+ for key in views.keys():
+ package, publisher = _get_package_and_publisher(key)
+
+ values = {'id': make_uuid(),
+ 'period_name': "All",
+ 'period_complete_day': 0,
+ 'url': key,
+ 'pageviews': views[key],
+ 'visits': visits[key],
+ 'department_id': publisher,
+ 'package_id': publisher
+ }
+ model.Session.add(GA_Url(**values))
+ model.Session.commit()
def update_url_stats(period_name, period_complete_day, url_data):
@@ -213,8 +262,8 @@
'period_name': 'All',
'period_complete_day': 0,
'url': url,
- 'pageviews': sum([int(e.pageviews) for e in entries]) + old_pageviews,
- 'visits': sum([int(e.visits or 0) for e in entries]) + old_visits,
+ 'pageviews': sum([int(e.pageviews) for e in entries]) + int(old_pageviews),
+ 'visits': sum([int(e.visits or 0) for e in entries]) + int(old_visits),
'department_id': publisher,
'package_id': package
}
@@ -340,10 +389,10 @@
'''
for object_type in (GA_Url, GA_Stat, GA_Publisher, GA_ReferralStat):
q = model.Session.query(object_type)
- if period_name != 'all':
+ if period_name != 'All':
q = q.filter_by(period_name=period_name)
q.delete()
- model.Session.commit()
+ model.repo.commit_and_remove()
def get_score_for_dataset(dataset_name):
'''
--- a/ckanext/ga_report/helpers.py
+++ b/ckanext/ga_report/helpers.py
@@ -50,9 +50,12 @@
dataset = model.Package.get(ga_url.url[len('/dataset/'):])
if dataset and not dataset.state == 'active':
dataset = None
- count += 1
- if count > 10:
- break
+ # When testing, it is possible that top datasets are not available
+ # so only go round this loop a few times before falling back on
+ # a random dataset.
+ count += 1
+ if count > 10:
+ break
if not dataset:
# fallback
dataset = model.Session.query(model.Package)\