ckanext-ga-report | ckanext-ga-report |
================= | ================= |
**Status:** Development | **Status:** Development |
**CKAN Version:** 1.7.1+ | **CKAN Version:** 1.7.1+ |
Overview | Overview |
-------- | -------- |
For creating detailed reports of CKAN analytics, including totals per group. | For creating detailed reports of CKAN analytics, including totals per group. |
Whereas ckanext-googleanalytics focusses on providing page view stats a recent period and for all time (aimed at end users), ckanext-ga-report is more interested in building regular periodic reports (more for site managers to monitor). | Whereas ckanext-googleanalytics focusses on providing page view stats a recent period and for all time (aimed at end users), ckanext-ga-report is more interested in building regular periodic reports (more for site managers to monitor). |
Contents of this extension: | Contents of this extension: |
* Use the CLI tool to download Google Analytics data for each time period into this extension's database tables | * Use the CLI tool to download Google Analytics data for each time period into this extension's database tables |
* Users can view the data as web page reports | * Users can view the data as web page reports |
Installation | Installation |
------------ | ------------ |
1. Activate you CKAN python environment and install this extension's software:: | 1. Activate you CKAN python environment and install this extension's software:: |
$ pyenv/bin/activate | $ pyenv/bin/activate |
$ pip install -e git+https://github.com/datagovuk/ckanext-ga-report.git#egg=ckanext-ga-report | $ pip install -e git+https://github.com/datagovuk/ckanext-ga-report.git#egg=ckanext-ga-report |
2. Ensure you development.ini (or similar) contains the info about your Google Analytics account and configuration:: | 2. Ensure you development.ini (or similar) contains the info about your Google Analytics account and configuration:: |
googleanalytics.id = UA-1010101-1 | googleanalytics.id = UA-1010101-1 |
googleanalytics.account = Account name (i.e. data.gov.uk, see top level item at https://www.google.com/analytics) | googleanalytics.account = Account name (e.g. data.gov.uk, see top level item at https://www.google.com/analytics) |
ga-report.period = monthly | ga-report.period = monthly |
ga-report.bounce_url = /data | |
Note that your credentials will be readable by system administrators on your server. Rather than use sensitive account details, it is suggested you give access to the GA account to a new Google account that you create just for this purpose. | The ga-report.bounce_url specifies the path to use when calculating bounces. For DGU this is /data |
but you may want to set this to /. | |
3. Set up this extension's database tables using a paster command. (Ensure your CKAN pyenv is still activated, run the command from ``src/ckanext-ga-report``, alter the ``--config`` option to point to your site config file):: | 3. Set up this extension's database tables using a paster command. (Ensure your CKAN pyenv is still activated, run the command from ``src/ckanext-ga-report``, alter the ``--config`` option to point to your site config file):: |
$ paster initdb --config=../ckan/development.ini | $ paster initdb --config=../ckan/development.ini |
4. Enable the extension in your CKAN config file by adding it to ``ckan.plugins``:: | 4. Enable the extension in your CKAN config file by adding it to ``ckan.plugins``:: |
ckan.plugins = ga-report | ckan.plugins = ga-report |
Problem shooting | |
---------------- | |
* ``(ProgrammingError) relation "ga_url" does not exist`` | |
This means that the ``paster initdb`` step has not been run successfully. Refer to the installation instructions for this extension. | |
Authorization | Authorization |
-------------- | -------------- |
Before you can access the data, you need to set up the OAUTH details which you can do by following the `instructions <https://developers.google.com/analytics/resources/tutorials/hello-analytics-api>`_ the outcome of which will be a file called credentials.json which should look like credentials.json.template with the relevant fields completed. These steps are below for convenience: | Before you can access the data, you need to set up the OAUTH details which you can do by following the `instructions <https://developers.google.com/analytics/resources/tutorials/hello-analytics-api>`_ the outcome of which will be a file called credentials.json which should look like credentials.json.template with the relevant fields completed. These steps are below for convenience: |
1. Visit the `Google APIs Console <https://code.google.com/apis/console>`_ | 1. Visit the `Google APIs Console <https://code.google.com/apis/console>`_ |
2. Sign-in and create a project or use an existing project. | 2. Sign-in and create a project or use an existing project. |
3. In the `Services pane <https://code.google.com/apis/console#:services>`_ , activate Analytics API for your project. If prompted, read and accept the terms of service. | 3. In the `Services pane <https://code.google.com/apis/console#:services>`_ , activate Analytics API for your project. If prompted, read and accept the terms of service. |
4. Go to the `API Access pane <https://code.google.com/apis/console/#:access>`_ | 4. Go to the `API Access pane <https://code.google.com/apis/console/#:access>`_ |
5. Click Create an OAuth 2.0 client ID.... | 5. Click Create an OAuth 2.0 client ID.... |
6. Fill out the Branding Information fields and click Next. | 6. Fill out the Branding Information fields and click Next. |
7. In Client ID Settings, set Application type to Installed application. | 7. In Client ID Settings, set Application type to Installed application. |
8. Click Create client ID | 8. Click Create client ID |
9. The details you need below are Client ID, Client secret, and Redirect URIs | 9. The details you need below are Client ID, Client secret, and Redirect URIs |
Once you have set up your credentials.json file you can generate an oauth token file by using the | Once you have set up your credentials.json file you can generate an oauth token file by using the |
following command, which will store your oauth token in a file called token.dat once you have finished | following command, which will store your oauth token in a file called token.dat once you have finished |
giving permission in the browser:: | giving permission in the browser:: |
$ paster getauthtoken --config=../ckan/development.ini | $ paster getauthtoken --config=../ckan/development.ini |
Tutorial | Tutorial |
-------- | -------- |
Download some GA data and store it in CKAN's db. (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:: | 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 token.dat latest --config=../ckan/development.ini |
The value after the token file is how much data you want to retrieve, this can be | The value after the token file is how much data you want to retrieve, this can be |
* **all** - data for all time (since 2010) | * **all** - data for all time (since 2010) |
* **latest** - (default) just the 'latest' data | * **latest** - (default) just the 'latest' data |
* **YYYY-MM-DD** - just data for all time periods going back to (and including) this date | * **YYYY-MM-DD** - just data for all time periods going back to (and including) this date |
Software Licence | Software Licence |
================ | ================ |
This software is developed by Cabinet Office. It is Crown Copyright and opened up under the Open Government Licence (OGL) (which is compatible with Creative Commons Attibution License). | This software is developed by Cabinet Office. It is Crown Copyright and opened up under the Open Government Licence (OGL) (which is compatible with Creative Commons Attibution License). |
OGL terms: http://www.nationalarchives.gov.uk/doc/open-government-licence/ | OGL terms: http://www.nationalarchives.gov.uk/doc/open-government-licence/ |
import logging | import logging |
import datetime | import datetime |
from ckan.lib.cli import CkanCommand | from ckan.lib.cli import CkanCommand |
# No other CKAN imports allowed until _load_config is run, | # No other CKAN imports allowed until _load_config is run, |
# or logging is disabled | # or logging is disabled |
class InitDB(CkanCommand): | class InitDB(CkanCommand): |
"""Initialise the extension's database tables | """Initialise the extension's database tables |
""" | """ |
summary = __doc__.split('\n')[0] | summary = __doc__.split('\n')[0] |
usage = __doc__ | usage = __doc__ |
max_args = 0 | max_args = 0 |
min_args = 0 | min_args = 0 |
def command(self): | def command(self): |
self._load_config() | self._load_config() |
import ckan.model as model | import ckan.model as model |
model.Session.remove() | model.Session.remove() |
model.Session.configure(bind=model.meta.engine) | model.Session.configure(bind=model.meta.engine) |
log = logging.getLogger('ckanext.ga-report') | log = logging.getLogger('ckanext.ga-report') |
import ga_model | import ga_model |
ga_model.init_tables() | ga_model.init_tables() |
log.info("DB tables are setup") | log.info("DB tables are setup") |
class GetAuthToken(CkanCommand): | class GetAuthToken(CkanCommand): |
""" Get's the Google auth token | """ Get's the Google auth token |
Usage: paster getauthtoken <credentials_file> | Usage: paster getauthtoken <credentials_file> |
Where <credentials_file> is the file name containing the details | Where <credentials_file> is the file name containing the details |
for the service (obtained from https://code.google.com/apis/console). | for the service (obtained from https://code.google.com/apis/console). |
By default this is set to credentials.json | By default this is set to credentials.json |
""" | """ |
summary = __doc__.split('\n')[0] | summary = __doc__.split('\n')[0] |
usage = __doc__ | usage = __doc__ |
max_args = 0 | max_args = 0 |
min_args = 0 | min_args = 0 |
def command(self): | def command(self): |
""" | """ |
In this case we don't want a valid service, but rather just to | In this case we don't want a valid service, but rather just to |
force the user through the auth flow. We allow this to complete to | force the user through the auth flow. We allow this to complete to |
act as a form of verification instead of just getting the token and | act as a form of verification instead of just getting the token and |
assuming it is correct. | assuming it is correct. |
""" | """ |
from ga_auth import init_service | from ga_auth import init_service |
init_service('token.dat', | init_service('token.dat', |
self.args[0] if self.args | self.args[0] if self.args |
else 'credentials.json') | else 'credentials.json') |
class LoadAnalytics(CkanCommand): | class LoadAnalytics(CkanCommand): |
"""Get data from Google Analytics API and save it | """Get data from Google Analytics API and save it |
in the ga_model | in the ga_model |
Usage: paster loadanalytics <tokenfile> <time-period> | Usage: paster loadanalytics <tokenfile> <time-period> |
Where <tokenfile> is the name of the auth token file from | Where <tokenfile> is the name of the auth token file from |
the getauthtoken step. | the getauthtoken step. |
And where <time-period> is: | And where <time-period> is: |
all - data for all time | all - data for all time |
latest - (default) just the 'latest' data | latest - (default) just the 'latest' data |
YYYY-MM - just data for the specific month | YYYY-MM - just data for the specific month |
""" | """ |
summary = __doc__.split('\n')[0] | summary = __doc__.split('\n')[0] |
usage = __doc__ | usage = __doc__ |
max_args = 2 | max_args = 2 |
min_args = 1 | min_args = 1 |
def __init__(self, name): | |
super(LoadAnalytics, self).__init__(name) | |
self.parser.add_option('-d', '--delete-first', | |
action='store_true', | |
default=False, | |
dest='delete_first', | |
help='Delete data for the period first') | |
def command(self): | def command(self): |
self._load_config() | self._load_config() |
from download_analytics import DownloadAnalytics | from download_analytics import DownloadAnalytics |
from ga_auth import (init_service, get_profile_id) | from ga_auth import (init_service, get_profile_id) |
try: | try: |
svc = init_service(self.args[0], None) | svc = init_service(self.args[0], None) |
except TypeError: | except TypeError: |
print ('Have you correctly run the getauthtoken task and ' | print ('Have you correctly run the getauthtoken task and ' |
'specified the correct file here') | 'specified the correct token file?') |
return | return |
downloader = DownloadAnalytics(svc, profile_id=get_profile_id(svc)) | downloader = DownloadAnalytics(svc, profile_id=get_profile_id(svc), |
delete_first=self.options.delete_first) | |
time_period = self.args[1] if self.args and len(self.args) > 1 \ | time_period = self.args[1] if self.args and len(self.args) > 1 \ |
else 'latest' | else 'latest' |
if time_period == 'all': | if time_period == 'all': |
downloader.all_() | downloader.all_() |
elif time_period == 'latest': | elif time_period == 'latest': |
downloader.latest() | downloader.latest() |
else: | else: |
# The month to use | # The month to use |
for_date = datetime.datetime.strptime(time_period, '%Y-%m') | for_date = datetime.datetime.strptime(time_period, '%Y-%m') |
downloader.specific_month(for_date) | downloader.specific_month(for_date) |
import re | |
import csv | |
import sys | |
import logging | import logging |
import operator | import operator |
from ckan.lib.base import BaseController, c, render, request, response | import collections |
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 | from ga_model import GA_Url, GA_Stat, GA_ReferralStat |
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): |
months = [] | months = [] |
vals = model.Session.query(cls.period_name).distinct().all() | vals = model.Session.query(cls.period_name).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 |
entries = model.Session.query(GA_Stat).\ | q = model.Session.query(GA_Stat) |
filter(GA_Stat.period_name==month).\ | if month != 'all': |
order_by('GA_Stat.stat_name, GA_Stat.key').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() | |
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,)) | |
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 = request.params.get('month', c.months[0][0] if c.months else '') | c.month_desc = 'all months' |
c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) | c.month = request.params.get('month', '') |
if c.month: | |
entries = model.Session.query(GA_Stat).\ | c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) |
filter(GA_Stat.stat_name=='Totals').\ | |
filter(GA_Stat.period_name==c.month).\ | q = model.Session.query(GA_Stat).\ |
order_by('ga_stat.key').all() | filter(GA_Stat.stat_name=='Totals') |
c.global_totals = [(s.key, s.value) for s in entries ] | if c.month: |
q = q.filter(GA_Stat.period_name==c.month) | |
entries = q.order_by('ga_stat.key').all() | |
def clean_key(key, val): | |
if key in ['Average time on site', 'Pages per visit', 'New visits', 'Bounces']: | |
val = "%.2f" % round(float(val), 2) | |
if key == 'Average time on site': | |
mins, secs = divmod(float(val), 60) | |
hours, mins = divmod(mins, 60) | |
val = '%02d:%02d:%02d (%s seconds) ' % (hours, mins, secs, val) | |
if key in ['New visits','Bounces']: | |
val = "%s%%" % val | |
if key in ['Total page views', 'Total visits']: | |
val = int(val) | |
return key, val | |
c.global_totals = [] | |
if c.month: | |
for e in entries: | |
key, val = clean_key(e.key, e.value) | |
c.global_totals.append((key, val)) | |
else: | |
d = collections.defaultdict(list) | |
for e in entries: | |
d[e.key].append(float(e.value)) | |
for k, v in d.iteritems(): | |
if k in ['Total page views', 'Total visits']: | |
v = sum(v) | |
else: | |
v = float(sum(v))/len(v) | |
key, val = clean_key(k,v) | |
c.global_totals.append((key, val)) | |
c.global_totals = sorted(c.global_totals, key=operator.itemgetter(0)) | |
keys = { | keys = { |
'Browser versions': 'browsers', | 'Browser versions': 'browser_versions', |
'Operating Systems versions': 'os', | 'Browsers': 'browsers', |
'Operating Systems versions': 'os_versions', | |
'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): | |
return (name[:length] + '..') if len(name) > 60 else name | |
def fill_out_url(url): | |
import urlparse | |
return urlparse.urljoin(g.site_url, url) | |
c.social_referrer_totals, c.social_referrers = [], [] | |
q = model.Session.query(GA_ReferralStat) | |
q = q.filter(GA_ReferralStat.period_name==c.month) if c.month else q | |
q = q.order_by('ga_referrer.count::int desc') | |
for entry in q.all(): | |
c.social_referrers.append((shorten_name(entry.url), fill_out_url(entry.url), | |
entry.source,entry.count)) | |
q = model.Session.query(GA_ReferralStat.url, | |
func.sum(GA_ReferralStat.count).label('count')) | |
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) | |
for entry in q.all(): | |
c.social_referrer_totals.append((shorten_name(entry[0]), fill_out_url(entry[0]),'', | |
entry[1])) | |
for k, v in keys.iteritems(): | for k, v in keys.iteritems(): |
entries = model.Session.query(GA_Stat).\ | q = model.Session.query(GA_Stat).\ |
filter(GA_Stat.stat_name==k).\ | filter(GA_Stat.stat_name==k) |
filter(GA_Stat.period_name==c.month).\ | if c.month: |
order_by('ga_stat.value::int desc').all() | entries = [] |
setattr(c, v, [(s.key, s.value) for s in entries ]) | q = q.filter(GA_Stat.period_name==c.month).\ |
order_by('ga_stat.value::int desc') | |
d = collections.defaultdict(int) | |
for e in q.all(): | |
d[e.key] += int(e.value) | |
entries = [] | |
for key, val in d.iteritems(): | |
entries.append((key,val,)) | |
entries = sorted(entries, key=operator.itemgetter(1), reverse=True) | |
# Get the total for each set of values and then set the value as | |
# a percentage of the total | |
if k == 'Social sources': | |
total = sum([x for n,x in c.global_totals if n == 'Total visits']) | |
else: | |
total = sum([num for _,num in entries]) | |
setattr(c, v, [(k,_percent(v,total)) for k,v in entries ]) | |
return render('ga_report/site/index.html') | return render('ga_report/site/index.html') |
class GaPublisherReport(BaseController): | class GaDatasetReport(BaseController): |
""" | """ |
Displays the pageview and visit count for specific publishers based on | Displays the pageview and visit count for datasets |
the datasets associated with the publisher. | with options to filter by publisher and time period. |
""" | """ |
def publisher_csv(self, month): | |
def index(self): | ''' |
Returns a CSV of each publisher with the total number of dataset | |
views & visits. | |
''' | |
c.month = month if not month == 'all' else '' | |
response.headers['Content-Type'] = "text/csv; charset=utf-8" | |
response.headers['Content-Disposition'] = str('attachment; filename=publishers_%s.csv' % (month,)) | |
writer = csv.writer(response) | |
writer.writerow(["Publisher Title", "Publisher Name", "Views", "Visits", "Period Name"]) | |
for publisher,view,visit in _get_top_publishers(None): | |
writer.writerow([publisher.title.encode('utf-8'), | |
publisher.name.encode('utf-8'), | |
view, | |
visit, | |
month]) | |
def dataset_csv(self, id='all', month='all'): | |
''' | |
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 month: The time period, or 'all' | |
''' | |
c.month = month if not month == 'all' else '' | |
if id != 'all': | |
c.publisher = model.Group.get(id) | |
if not c.publisher: | |
abort(404, 'A publisher with that name could not be found') | |
packages = self._get_packages(c.publisher) | |
response.headers['Content-Type'] = "text/csv; charset=utf-8" | |
response.headers['Content-Disposition'] = \ | |
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([package.title.encode('utf-8'), | |
package.name.encode('utf-8'), | |
view, | |
visit, | |
month]) | |
def publishers(self): | |
'''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.months[0][0] if c.months else '') | c.month = request.params.get('month', '') |
c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) | c.month_desc = 'all months' |
# and not url like '/publisher/%%' | if c.month: |
connection = model.Session.connection() | c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) |
q = """ | |
select department_id, sum(pageviews::int) views, sum(visitors::int) visits | c.top_publishers = _get_top_publishers() |
from ga_url | |
where department_id <> '' | |
and period_name=%s | |
group by department_id order by views desc limit 20; | |
""" | |
c.top_publishers = [] | |
res = connection.execute(q, c.month) | |
for row in res: | |
c.top_publishers.append((model.Group.get(row[0]), row[1], row[2])) | |
return render('ga_report/publisher/index.html') | return render('ga_report/publisher/index.html') |
def _get_packages(self, publisher=None, count=-1): | |
def read(self, id): | '''Returns the datasets in order of visits''' |
c.publisher = model.Group.get(id) | if count == -1: |
count = sys.maxint | |
q = model.Session.query(GA_Url)\ | |
.filter(GA_Url.url.like('/dataset/%')) | |
if publisher: | |
q = q.filter(GA_Url.department_id==publisher.name) | |
if c.month: | |
q = q.filter(GA_Url.period_name==c.month) | |
q = q.order_by('ga_url.visitors::int desc') | |
if c.month: | |
top_packages = [] | |
for entry in q.limit(count): | |
package_name = entry.url[len('/dataset/'):] | |
p = model.Package.get(package_name) | |
if p: | |
top_packages.append((p, entry.pageviews, entry.visitors)) | |
else: | |
log.warning('Could not find package "%s"', package_name) | |
else: | |
ds = {} | |
for entry in q: | |
if len(ds) >= count: | |
break | |
package_name = entry.url[len('/dataset/'):] | |
p = model.Package.get(package_name) | |
if p: | |
if not p in ds: | |
ds[p] = {'views': 0, 'visits': 0} | |
ds[p]['views'] = ds[p]['views'] + int(entry.pageviews) | |
ds[p]['visits'] = ds[p]['visits'] + int(entry.visitors) | |
else: | |
log.warning('Could not find package "%s"', package_name) | |
results = [] | |
for k, v in ds.iteritems(): | |
results.append((k,v['views'],v['visits'])) | |
top_packages = sorted(results, key=operator.itemgetter(1), reverse=True) | |
return top_packages | |
def read(self): | |
''' | |
Lists the most popular datasets across all publishers | |
''' | |
return self.read_publisher(None) | |
def read_publisher(self, id): | |
''' | |
Lists the most popular datasets for a publisher (or across all publishers) | |
''' | |
count = 20 | |
c.publishers = _get_publishers() | |
id = request.params.get('publisher', id) | |
if id and id != 'all': | |
c.publisher = model.Group.get(id) | |
if not c.publisher: | |
abort(404, 'A publisher with that name could not be found') | |
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.months[0][0] if c.months else '') | c.month = request.params.get('month', '') |
c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) | if not c.month: |
c.month_desc = 'all months' | |
entry = model.Session.query(GA_Url).\ | else: |
filter(GA_Url.url=='/publisher/%s' % c.publisher.name).\ | c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) |
filter(GA_Url.period_name==c.month).first() | |
c.publisher_page_views = entry.pageviews if entry else 0 | c.publisher_page_views = 0 |
q = model.Session.query(GA_Url).\ | |
entries = model.Session.query(GA_Url).\ | filter(GA_Url.url=='/publisher/%s' % c.publisher_name) |
filter(GA_Url.department_id==c.publisher.name).\ | if c.month: |
filter(GA_Url.period_name==c.month).\ | entry = q.filter(GA_Url.period_name==c.month).first() |
order_by('ga_url.pageviews::int desc')[:20] | c.publisher_page_views = entry.pageviews if entry else 0 |
for entry in entries: | else: |
if entry.url.startswith('/dataset/'): | for e in q.all(): |
p = model.Package.get(entry.url[len('/dataset/'):]) | c.publisher_page_views = c.publisher_page_views + int(e.pageviews) |
c.top_packages.append((p,entry.pageviews,entry.visitors)) | |
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): | |
''' | |
Returns a list of the top 20 publishers by dataset visits. | |
(The number to show can be varied with 'limit') | |
''' | |
connection = model.Session.connection() | |
q = """ | |
select department_id, sum(pageviews::int) views, sum(visitors::int) visits | |
from ga_url | |
where department_id <> ''""" | |
if c.month: | |
q = q + """ | |
and period_name=%s | |
""" | |
q = q + """ | |
group by department_id order by visits desc | |
""" | |
if limit: | |
q = q + " limit %s;" % (limit) | |
# Add this back (before and period_name =%s) if you want to ignore publisher | |
# homepage views | |
# and not url like '/publisher/%%' | |
top_publishers = [] | |
res = connection.execute(q, c.month) | |
for row in res: | |
g = model.Group.get(row[0]) | |
if g: | |
top_publishers.append((g, row[1], row[2])) | |
return top_publishers | |
def _get_publishers(): | |
''' | |
Returns a list of all publishers. Each item is a tuple: | |
(names, title) | |
''' | |
publishers = [] | |
for pub in model.Session.query(model.Group).\ | |
filter(model.Group.type=='publisher').\ | |
filter(model.Group.state=='active').\ | |
order_by(model.Group.name): | |
publishers.append((pub.name, pub.title)) | |
return publishers | |
def _percent(num, total): | |
p = 100 * float(num)/float(total) | |
return "%.2f%%" % round(p, 2) | |
import os | import os |
import logging | import logging |
import datetime | import datetime |
import collections | |
from pylons import config | from pylons import config |
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_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): | 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 | |
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: |
if self.delete_first: | |
log.info('Deleting existing Analytics for period "%s"', | |
period_name) | |
ga_model.delete(period_name) | |
log.info('Downloading Analytics for period "%s" (%s - %s)', | log.info('Downloading Analytics for 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')) |
data = self.download(start_date, end_date, '~/dataset/[a-z0-9-_]+') | data = self.download(start_date, end_date, '~/dataset/[a-z0-9-_]+') |
log.info('Storing Dataset Analytics for period "%s"', | log.info('Storing Dataset Analytics for period "%s"', |
self.get_full_period_name(period_name, period_complete_day)) | self.get_full_period_name(period_name, period_complete_day)) |
self.store(period_name, period_complete_day, data, ) | self.store(period_name, period_complete_day, data, ) |
data = self.download(start_date, end_date, '~/publisher/[a-z0-9-_]+') | data = self.download(start_date, end_date, '~/publisher/[a-z0-9-_]+') |
log.info('Storing Publisher Analytics for period "%s"', | log.info('Storing Publisher Analytics for period "%s"', |
self.get_full_period_name(period_name, period_complete_day)) | self.get_full_period_name(period_name, period_complete_day)) |
self.store(period_name, period_complete_day, data,) | self.store(period_name, period_complete_day, data,) |
ga_model.update_publisher_stats(period_name) # about 30 seconds. | ga_model.update_publisher_stats(period_name) # about 30 seconds. |
self.sitewide_stats( period_name ) | self.sitewide_stats( period_name ) |
self.update_social_info(period_name, start_date, end_date) | |
def update_social_info(self, period_name, start_date, end_date): | |
start_date = start_date.strftime('%Y-%m-%d') | |
end_date = end_date.strftime('%Y-%m-%d') | |
query = 'ga:hasSocialSourceReferral=~Yes$' | |
metrics = 'ga:entrances' | |
sort = '-ga:entrances' | |
# Supported query params at | |
# https://developers.google.com/analytics/devguides/reporting/core/v3/reference | |
results = self.service.data().ga().get( | |
ids='ga:' + self.profile_id, | |
filters=query, | |
start_date=start_date, | |
metrics=metrics, | |
sort=sort, | |
dimensions="ga:landingPagePath,ga:socialNetwork", | |
max_results=10000, | |
end_date=end_date).execute() | |
data = collections.defaultdict(list) | |
rows = results.get('rows',[]) | |
for row in rows: | |
from ga_model import _normalize_url | |
data[_normalize_url(row[0])].append( (row[1], int(row[2]),) ) | |
ga_model.update_social(period_name, data) | |
def download(self, start_date, end_date, path='~/dataset/[a-z0-9-_]+'): | def download(self, start_date, end_date, path='~/dataset/[a-z0-9-_]+'): |
'''Get data from GA for a given time period''' | '''Get data from GA for a given time period''' |
start_date = start_date.strftime('%Y-%m-%d') | start_date = start_date.strftime('%Y-%m-%d') |
end_date = end_date.strftime('%Y-%m-%d') | end_date = end_date.strftime('%Y-%m-%d') |
query = 'ga:pagePath=%s$' % path | query = 'ga:pagePath=%s$' % path |
metrics = 'ga:uniquePageviews, ga:visitors' | metrics = 'ga:uniquePageviews, ga:visitors' |
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, |