From: David Read Date: Wed, 07 Nov 2012 13:38:29 +0000 Subject: Tidy logging. X-Git-Url: https://maxious.lambdacomplex.org/git/?p=ckanext-ga-report.git&a=commitdiff&h=753e746cceb78d2cb57df91505b36e76fa4ad38e --- Tidy logging. --- --- a/README.rst +++ b/README.rst @@ -31,10 +31,12 @@ 2. Ensure you development.ini (or similar) contains the info about your Google Analytics account and configuration:: 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.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):: @@ -43,6 +45,12 @@ 4. Enable the extension in your CKAN config file by adding it to ``ckan.plugins``:: 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 @@ -79,7 +87,7 @@ 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 --- a/ckanext/ga_report/command.py +++ b/ckanext/ga_report/command.py @@ -73,6 +73,14 @@ max_args = 2 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): self._load_config() @@ -83,10 +91,11 @@ svc = init_service(self.args[0], None) except TypeError: print ('Have you correctly run the getauthtoken task and ' - 'specified the correct file here') + 'specified the correct token file?') 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 \ else 'latest' --- a/ckanext/ga_report/controller.py +++ b/ckanext/ga_report/controller.py @@ -22,8 +22,9 @@ def _month_details(cls): + '''Returns a list of all the month names''' months = [] - vals = model.Session.query(cls.period_name).distinct().all() + vals = model.Session.query(cls.period_name).filter(cls.period_name!='All').distinct().all() for m in vals: months.append( (m[0], _get_month_name(m[0]))) return sorted(months, key=operator.itemgetter(0), reverse=True) @@ -58,7 +59,7 @@ c.months = _month_details(GA_Stat) # Work out which month to show, based on query params of the first item - c.month_desc = 'all time' + c.month_desc = 'all months' c.month = request.params.get('month', '') if c.month: c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) @@ -70,15 +71,15 @@ entries = q.order_by('ga_stat.key').all() def clean_key(key, val): - if key in ['Average time on site', 'Pages per visit', 'New visits']: + if key in ['Average time on site', 'Pages per visit', 'New visits', 'Bounces']: val = "%.2f" % round(float(val), 2) if key == 'Average time on site': mins, secs = divmod(float(val), 60) hours, mins = divmod(mins, 60) val = '%02d:%02d:%02d (%s seconds) ' % (hours, mins, secs, val) - if key == 'New visits': + if key in ['New visits','Bounces']: val = "%s%%" % val - if key in ['Bounces', 'Total page views', 'Total visits']: + if key in ['Total page views', 'Total visits']: val = int(val) return key, val @@ -93,11 +94,12 @@ for e in entries: d[e.key].append(float(e.value)) for k, v in d.iteritems(): - if k in ['Bounces', 'Total page views', 'Total visits']: + if k in ['Total page views', 'Total visits']: v = sum(v) else: v = float(sum(v))/len(v) key, val = clean_key(k,v) + c.global_totals.append((key, val)) c.global_totals = sorted(c.global_totals, key=operator.itemgetter(0)) @@ -134,29 +136,7 @@ c.social_referrer_totals.append((shorten_name(entry[0]), fill_out_url(entry[0]),'', entry[1])) - - browser_version_re = re.compile("(.*)\((.*)\)") for k, v in keys.iteritems(): - - def clean_field(key): - if k != 'Browser versions': - return key - m = browser_version_re.match(key) - browser = m.groups()[0].strip() - ver = m.groups()[1] - parts = ver.split('.') - if len(parts) > 1: - if parts[1][0] == '0': - ver = parts[0] - else: - ver = "%s.%s" % (parts[0],parts[1]) - if browser in ['Safari','Android Browser']: # Special case complex version nums - ver = parts[0] - if len(ver) > 2: - ver = "%s%sX" % (ver[0], ver[1]) - - return "%s (%s)" % (browser, ver,) - q = model.Session.query(GA_Stat).\ filter(GA_Stat.stat_name==k) if c.month: @@ -166,15 +146,11 @@ d = collections.defaultdict(int) for e in q.all(): - d[clean_field(e.key)] += int(e.value) + 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) - - def percent(num, total): - p = 100 * float(num)/float(total) - return "%.2f%%" % round(p, 2) # Get the total for each set of values and then set the value as # a percentage of the total @@ -182,57 +158,65 @@ total = sum([x for n,x in c.global_totals if n == 'Total visits']) else: total = sum([num for _,num in entries]) - setattr(c, v, [(k,percent(v,total)) for k,v in entries ]) + setattr(c, v, [(k,_percent(v,total)) for k,v in entries ]) return render('ga_report/site/index.html') -class GaPublisherReport(BaseController): +class GaDatasetReport(BaseController): """ - Displays the pageview and visit count for specific publishers based on - the datasets associated with the publisher. + Displays the pageview and visit count for datasets + with options to filter by publisher and time period. """ - def csv(self, month): - - c.month = month if not month =='all' else '' + def publisher_csv(self, month): + ''' + 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", "Views", "Visits", "Period Name"]) - - for publisher,view,visit in self._get_publishers(None): + 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 publisher_csv(self, id, month): - - c.month = month if not month =='all' else '' - c.publisher = model.Group.get(id) - if not c.publisher: - abort(404, 'A publisher with that name could not be found') + 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=%s_%s.csv' % (c.publisher.name, month,)) + str('attachment; filename=datasets_%s_%s.csv' % (c.publisher_name, month,)) writer = csv.writer(response) - writer.writerow(["Publisher", "Views", "Visits", "Period Name"]) + 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 index(self): + 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 # month names from the values. @@ -240,84 +224,58 @@ # Work out which month to show, based on query params of the first item c.month = request.params.get('month', '') - c.month_desc = 'all time' + c.month_desc = 'all months' if c.month: c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) - c.top_publishers = self._get_publishers() + c.top_publishers = _get_top_publishers() return render('ga_report/publisher/index.html') - def _get_publishers(self, limit=20): - 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 views 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_packages(self, publisher, count=-1): + def _get_packages(self, publisher=None, count=-1): + '''Returns the datasets in order of visits''' if count == -1: count = sys.maxint + month = c.month or 'All' + + q = model.Session.query(GA_Url,model.Package)\ + .filter(model.Package.name==GA_Url.package_id)\ + .filter(GA_Url.url.like('/dataset/%')) + if publisher: + q = q.filter(GA_Url.department_id==publisher.name) + q = q.filter(GA_Url.period_name==month) + q = q.order_by('ga_url.visitors::int desc') top_packages = [] - q = model.Session.query(GA_Url).\ - filter(GA_Url.department_id==publisher.name).\ - filter(GA_Url.url.like('/dataset/%')) - if c.month: - q = q.filter(GA_Url.period_name==c.month) - q = q.order_by('ga_url.pageviews::int desc') - - if c.month: - for entry in q[:count]: - p = model.Package.get(entry.url[len('/dataset/'):]) - top_packages.append((p,entry.pageviews,entry.visitors)) - else: - ds = {} - for entry in q.all(): - if len(ds) >= count: - break - p = model.Package.get(entry.url[len('/dataset/'):]) - 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) - - results = [] - for k, v in ds.iteritems(): - results.append((k,v['views'],v['visits'])) - - top_packages = sorted(results, key=operator.itemgetter(1), reverse=True) + + for entry,package in q.limit(count): + if package: + top_packages.append((package, entry.pageviews, entry.visitors)) + else: + log.warning('Could not find package associated package') + return top_packages - - def read(self, id): + 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.publisher = model.Group.get(id) - if not c.publisher: - abort(404, 'A publisher with that name could not be found') + 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 # Get the month details by fetching distinct values and determining the @@ -327,21 +285,68 @@ # Work out which month to show, based on query params of the first item c.month = request.params.get('month', '') if not c.month: - c.month_desc = 'all time' + c.month_desc = 'all months' else: c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) + month = c.mnth or 'All' c.publisher_page_views = 0 q = model.Session.query(GA_Url).\ - filter(GA_Url.url=='/publisher/%s' % c.publisher.name) - if c.month: - entry = q.filter(GA_Url.period_name==c.month).first() - c.publisher_page_views = entry.pageviews if entry else 0 - else: - for e in q.all(): - c.publisher_page_views = c.publisher_page_views + int(e.pageviews) + filter(GA_Url.url=='/publisher/%s' % c.publisher_name) + entry = q.filter(GA_Url.period_name==c.month).first() + c.publisher_page_views = entry.pageviews if entry else 0 c.top_packages = self._get_packages(c.publisher, 20) 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) + --- a/ckanext/ga_report/download_analytics.py +++ b/ckanext/ga_report/download_analytics.py @@ -11,15 +11,17 @@ log = logging.getLogger('ckanext.ga-report') FORMAT_MONTH = '%Y-%m' +MIN_VIEWS = 50 +MIN_VISITS = 20 class DownloadAnalytics(object): '''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.service = service self.profile_id = profile_id - + self.delete_first = delete_first def specific_month(self, date): import calendar @@ -90,23 +92,34 @@ def download_and_store(self, periods): for period_name, period_complete_day, start_date, end_date in periods: - log.info('Downloading Analytics for period "%s" (%s - %s)', + log.info('Period "%s" (%s - %s)', self.get_full_period_name(period_name, period_complete_day), - start_date.strftime('%Y %m %d'), - end_date.strftime('%Y %m %d')) + start_date.strftime('%Y-%m-%d'), + end_date.strftime('%Y-%m-%d')) + + if self.delete_first: + log.info('Deleting existing Analytics for this period "%s"', + period_name) + ga_model.delete(period_name) + + log.info('Downloading analytics for dataset views') data = self.download(start_date, end_date, '~/dataset/[a-z0-9-_]+') - log.info('Storing Dataset Analytics for period "%s"', - self.get_full_period_name(period_name, period_complete_day)) + + log.info('Storing dataset views (%i rows)', len(data.get('url'))) self.store(period_name, period_complete_day, data, ) + log.info('Downloading analytics for publisher views') data = self.download(start_date, end_date, '~/publisher/[a-z0-9-_]+') - log.info('Storing Publisher Analytics for period "%s"', - self.get_full_period_name(period_name, period_complete_day)) + log.info('Storing publisher views (%i rows)', len(data.get('url'))) self.store(period_name, period_complete_day, data,) + 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 ) + log.info('Downloading and storing analytics for social networks') self.update_social_info(period_name, start_date, end_date) def update_social_info(self, period_name, start_date, end_date): @@ -140,7 +153,7 @@ start_date = start_date.strftime('%Y-%m-%d') end_date = end_date.strftime('%Y-%m-%d') query = 'ga:pagePath=%s$' % path - metrics = 'ga:uniquePageviews, ga:visitors' + metrics = 'ga:uniquePageviews, ga:visits' sort = '-ga:uniquePageviews' # Supported query params at @@ -154,11 +167,6 @@ dimensions="ga:pagePath", max_results=10000, end_date=end_date).execute() - - if os.getenv('DEBUG'): - import pprint - pprint.pprint(results) - print 'Total results: %s' % results.get('totalResults') packages = [] for entry in results.get('rows'): @@ -177,12 +185,10 @@ start_date = '%s-01' % period_name end_date = '%s-%s' % (period_name, last_day_of_month) - print 'Sitewide_stats for %s (%s -> %s)' % (period_name, start_date, end_date) - funcs = ['_totals_stats', '_social_stats', '_os_stats', '_locale_stats', '_browser_stats', '_mobile_stats'] for f in funcs: - print ' + Fetching %s stats' % f.split('_')[1] + log.info('Downloading analytics for %s' % f.split('_')[1]) getattr(self, f)(start_date, end_date, period_name) def _get_results(result_data, f): @@ -207,19 +213,38 @@ results = self.service.data().ga().get( ids='ga:' + self.profile_id, start_date=start_date, - metrics='ga:pageviewsPerVisit,ga:bounces,ga:avgTimeOnSite,ga:percentNewVisits,ga:visitors', + metrics='ga:pageviewsPerVisit,ga:avgTimeOnSite,ga:percentNewVisits,ga:visits', max_results=10000, end_date=end_date).execute() result_data = results.get('rows') data = { 'Pages per visit': result_data[0][0], - 'Bounces': result_data[0][1], - 'Average time on site': result_data[0][2], - 'New visits': result_data[0][3], - 'Total visits': result_data[0][4], + 'Average time on site': result_data[0][1], + 'New visits': result_data[0][2], + 'Total visits': result_data[0][3], } ga_model.update_sitewide_stats(period_name, "Totals", data) + # Bounces from /data. This url is specified in configuration because + # for DGU we don't want /. + path = config.get('ga-report.bounce_url','/') + print path + results = self.service.data().ga().get( + ids='ga:' + self.profile_id, + filters='ga:pagePath=~%s$' % (path,), + start_date=start_date, + metrics='ga:bounces,ga:uniquePageviews', + dimensions='ga:pagePath', + max_results=10000, + end_date=end_date).execute() + result_data = results.get('rows') + for results in result_data: + if results[0] == path: + bounce, total = [float(x) for x in results[1:]] + pct = 100 * bounce/total + print "%d bounces from %d total == %s" % (bounce, total, pct) + ga_model.update_sitewide_stats(period_name, "Totals", {'Bounces': pct}) + def _locale_stats(self, start_date, end_date, period_name): """ Fetches stats about language and country """ @@ -235,11 +260,13 @@ data = {} 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) 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) @@ -254,13 +281,11 @@ max_results=10000, end_date=end_date).execute() result_data = results.get('rows') - twitter_links = [] data = {} for result in result_data: if not result[0] == '(not set)': data[result[0]] = data.get(result[0], 0) + int(result[2]) - if result[0] == 'Twitter': - twitter_links.append(result[1]) + self._filter_out_long_tail(data, 3) ga_model.update_sitewide_stats(period_name, "Social sources", data) @@ -278,12 +303,14 @@ data = {} 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) data = {} for result in result_data: - key = "%s (%s)" % (result[0],result[1]) - data[key] = result[2] + 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) @@ -298,17 +325,42 @@ max_results=10000, end_date=end_date).execute() result_data = results.get('rows') + # e.g. [u'Firefox', u'19.0', u'20'] + data = {} 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) data = {} for result in result_data: - key = "%s (%s)" % (result[0], result[1]) - data[key] = result[2] + 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) + @classmethod + def _filter_browser_version(cls, browser, version_str): + ''' + 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. + This is helpful when viewing stats and good to protect privacy. + ''' + ver = version_str + parts = ver.split('.') + if len(parts) > 1: + if parts[1][0] == '0': + ver = parts[0] + else: + ver = "%s" % (parts[0]) + # Special case complex version nums + if browser in ['Safari', 'Android Browser']: + ver = parts[0] + if len(ver) > 2: + num_hidden_digits = len(ver) - 2 + ver = ver[0] + ver[1] + 'X' * num_hidden_digits + return ver def _mobile_stats(self, start_date, end_date, period_name): """ Info about mobile devices """ @@ -326,10 +378,23 @@ data = {} 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) 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) + @classmethod + def _filter_out_long_tail(cls, data, threshold=10): + ''' + Given data which is a frequency distribution, filter out + results which are below a threshold count. This is good to protect + privacy. + ''' + for key, value in data.items(): + if value < threshold: + del data[key] + --- a/ckanext/ga_report/ga_auth.py +++ b/ckanext/ga_report/ga_auth.py @@ -53,7 +53,11 @@ return None accountName = config.get('googleanalytics.account') + if not accountName: + raise Exception('googleanalytics.account needs to be configured') webPropertyId = config.get('googleanalytics.id') + if not webPropertyId: + raise Exception('googleanalytics.id needs to be configured') for acc in accounts.get('items'): if acc.get('name') == accountName: accountId = acc.get('id') --- a/ckanext/ga_report/ga_model.py +++ b/ckanext/ga_report/ga_model.py @@ -1,10 +1,10 @@ import re import uuid -from sqlalchemy import Table, Column, MetaData +from sqlalchemy import Table, Column, MetaData, ForeignKey from sqlalchemy import types from sqlalchemy.sql import select -from sqlalchemy.orm import mapper +from sqlalchemy.orm import mapper, relation from sqlalchemy import func import ckan.model as model @@ -14,8 +14,6 @@ return unicode(uuid.uuid4()) metadata = MetaData() - - class GA_Url(object): @@ -32,6 +30,7 @@ Column('visitors', types.UnicodeText), Column('url', types.UnicodeText), Column('department_id', types.UnicodeText), + Column('package_id', types.UnicodeText), ) mapper(GA_Url, url_table) @@ -163,6 +162,10 @@ 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).\ @@ -172,6 +175,7 @@ item.pageviews = views item.visitors = visitors item.department_id = department_id + item.package_id = package model.Session.add(item) else: # create the row @@ -181,9 +185,31 @@ 'url': url, 'pageviews': views, 'visitors': visitors, - 'department_id': department_id + '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).\ + filter(GA_Url.period_name != "All").\ + filter(GA_Url.url==url).all() + values = {'id': make_uuid(), + 'period_name': "All", + 'period_complete_day': "0", + 'url': url, + 'pageviews': sum([int(x.pageviews) for x in items]), + 'visitors': sum([int(x.visitors) for x in items]), + 'department_id': department_id, + 'package_id': package + } + model.Session.add(GA_Url(**values)) + model.Session.commit() @@ -295,3 +321,15 @@ for grandchild in go_down_tree(child): yield grandchild +def delete(period_name): + ''' + Deletes table data for the specified period, or specify 'all' + for all periods. + ''' + for object_type in (GA_Url, GA_Stat, GA_Publisher, GA_ReferralStat): + q = model.Session.query(object_type) + if period_name != 'all': + q = q.filter_by(period_name=period_name) + q.delete() + model.Session.commit() + --- a/ckanext/ga_report/helpers.py +++ b/ckanext/ga_report/helpers.py @@ -1,17 +1,99 @@ import logging import operator + import ckan.lib.base as base import ckan.model as model +from ckan.logic import get_action +from ckanext.ga_report.ga_model import GA_Url, GA_Publisher +from ckanext.ga_report.controller import _get_publishers _log = logging.getLogger(__name__) +def popular_datasets(count=10): + import random + + publisher = None + publishers = _get_publishers(30) + total = len(publishers) + while not publisher or not datasets: + rand = random.randrange(0, total) + publisher = publishers[rand][0] + if not publisher.state == 'active': + publisher = None + continue + datasets = _datasets_for_publisher(publisher, 10)[:count] + + ctx = { + 'datasets': datasets, + 'publisher': publisher + } + return base.render_snippet('ga_report/ga_popular_datasets.html', **ctx) + +def single_popular_dataset(top=20): + '''Returns a random dataset from the most popular ones. + + :param top: the number of top datasets to select from + ''' + import random + + top_datasets = model.Session.query(GA_Url).\ + filter(GA_Url.url.like('/dataset/%')).\ + order_by('ga_url.pageviews::int desc') + num_top_datasets = top_datasets.count() + + dataset = None + if num_top_datasets: + count = 0 + while not dataset: + rand = random.randrange(0, min(top, num_top_datasets)) + ga_url = top_datasets[rand] + dataset = model.Package.get(ga_url.url[len('/dataset/'):]) + if dataset and not dataset.state == 'active': + dataset = None + count += 1 + if count > 10: + break + if not dataset: + # fallback + dataset = model.Session.query(model.Package)\ + .filter_by(state='active').first() + if not dataset: + return None + dataset_dict = get_action('package_show')({'model': model, + 'session': model.Session}, + {'id':dataset.id}) + return dataset_dict + +def single_popular_dataset_html(top=20): + dataset_dict = single_popular_dataset(top) + groups = package.get('groups', []) + publishers = [ g for g in groups if g.get('type') == 'publisher' ] + publisher = publishers[0] if publishers else {'name':'', 'title': ''} + context = { + 'dataset': dataset_dict, + 'publisher': publisher_dict + } + return base.render_snippet('ga_report/ga_popular_single.html', **context) + + def most_popular_datasets(publisher, count=20): - from ckanext.ga_report.ga_model import GA_Url if not publisher: _log.error("No valid publisher passed to 'most_popular_datasets'") return "" + results = _datasets_for_publisher(publisher, count) + + ctx = { + 'dataset_count': len(results), + 'datasets': results, + + 'publisher': publisher + } + + return base.render_snippet('ga_report/publisher/popular.html', **ctx) + +def _datasets_for_publisher(publisher, count): datasets = {} entries = model.Session.query(GA_Url).\ filter(GA_Url.department_id==publisher.name).\ @@ -29,14 +111,5 @@ for k, v in datasets.iteritems(): results.append((k,v['views'],v['visits'])) - results = sorted(results, key=operator.itemgetter(1), reverse=True) + return sorted(results, key=operator.itemgetter(1), reverse=True) - ctx = { - 'dataset_count': len(datasets), - 'datasets': results, - - 'publisher': publisher - } - - return base.render_snippet('ga_report/publisher/popular.html', **ctx) - --- a/ckanext/ga_report/plugin.py +++ b/ckanext/ga_report/plugin.py @@ -2,6 +2,10 @@ import ckan.lib.helpers as h import ckan.plugins as p from ckan.plugins import implements, toolkit + +from ckanext.ga_report.helpers import (most_popular_datasets, + popular_datasets, + single_popular_dataset) log = logging.getLogger('ckanext.ga-report') @@ -19,33 +23,15 @@ A dictionary of extra helpers that will be available to provide ga report info to templates. """ - from ckanext.ga_report.helpers import most_popular_datasets return { 'ga_report_installed': lambda: True, + 'popular_datasets': popular_datasets, 'most_popular_datasets': most_popular_datasets, + 'single_popular_dataset': single_popular_dataset } def after_map(self, map): - map.connect( - '/data/site-usage/publisher', - controller='ckanext.ga_report.controller:GaPublisherReport', - action='index' - ) - map.connect( - '/data/site-usage/publisher_{month}.csv', - controller='ckanext.ga_report.controller:GaPublisherReport', - action='csv' - ) - map.connect( - '/data/site-usage/publisher/{id}_{month}.csv', - controller='ckanext.ga_report.controller:GaPublisherReport', - action='publisher_csv' - ) - map.connect( - '/data/site-usage/publisher/{id}', - controller='ckanext.ga_report.controller:GaPublisherReport', - action='read' - ) + # GaReport map.connect( '/data/site-usage', controller='ckanext.ga_report.controller:GaReport', @@ -56,6 +42,33 @@ controller='ckanext.ga_report.controller:GaReport', action='csv' ) + + # GaDatasetReport + map.connect( + '/data/site-usage/publisher', + controller='ckanext.ga_report.controller:GaDatasetReport', + action='publishers' + ) + map.connect( + '/data/site-usage/publishers_{month}.csv', + controller='ckanext.ga_report.controller:GaDatasetReport', + action='publisher_csv' + ) + map.connect( + '/data/site-usage/dataset/datasets_{id}_{month}.csv', + controller='ckanext.ga_report.controller:GaDatasetReport', + action='dataset_csv' + ) + map.connect( + '/data/site-usage/dataset', + controller='ckanext.ga_report.controller:GaDatasetReport', + action='read' + ) + map.connect( + '/data/site-usage/dataset/{id}', + controller='ckanext.ga_report.controller:GaDatasetReport', + action='read_publisher' + ) return map --- a/ckanext/ga_report/report_model.py +++ /dev/null --- /dev/null +++ b/ckanext/ga_report/templates/ga_report/ga_popular_datasets.html @@ -1,1 +1,27 @@ + + + + + + + + --- /dev/null +++ b/ckanext/ga_report/templates/ga_report/ga_popular_single.html @@ -1,1 +1,31 @@ + + + + + + + + + + --- a/ckanext/ga_report/templates/ga_report/ga_util.html +++ b/ckanext/ga_report/templates/ga_report/ga_util.html @@ -5,15 +5,6 @@ xmlns:xi="http://www.w3.org/2001/XInclude" py:strip="" > - - - - - - - -
${title}
- @@ -45,19 +36,18 @@
-
+ --- /dev/null +++ b/ckanext/ga_report/templates/ga_report/notes.html @@ -1,1 +1,17 @@ + +
  • +

    Notes

    +
      +
    • "Views" is the number of sessions during which the page was viewed one or more times (technically known as "unique pageviews").
    • +
    • "Visits" is the number of unique user visits to a page, counted once for each visitor for each session.
    • + +
    • These usage statistics are confined to users with javascript enabled, which excludes web crawlers and API calls.
    • +
    • The results are not shown when the number of views/visits is tiny. Where these relate to site pages, results are available in full in the CSV download. Where these relate to users' web browser information, results are not disclosed, for privacy reasons.
    • +
    +
  • + + --- a/ckanext/ga_report/templates/ga_report/publisher/index.html +++ b/ckanext/ga_report/templates/ga_report/publisher/index.html @@ -5,43 +5,36 @@ - Publisher Analytics for ${g.site_title} + Usage by Publisher
  • -

    Publishers

    -

    - Dataset views records the number of times a specific dataset page has been viewed. Visits records the number of unique site visits. -

    -

    - Note: this data does not include API calls. +

    Download

    +

    + Download as CSV

  • -
  • -

    Download

    -

    - Download as CSV
    -

    -
  • +
    +

    Site Usage

    - ${usage_nav('Publishers', None)} + ${usage_nav('Publishers')} -
    +
    - +
    @@ -49,11 +42,11 @@ Publisher Dataset Views - Visits + Dataset Visits - ${h.link_to(publisher.title, h.url_for(controller='ckanext.ga_report.controller:GaPublisherReport', action='read', id=publisher.name))} + ${h.link_to(publisher.title, h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport', action='read_publisher', id=publisher.name))} ${views} ${visits} --- a/ckanext/ga_report/templates/ga_report/publisher/popular.html +++ b/ckanext/ga_report/templates/ga_report/publisher/popular.html @@ -3,7 +3,7 @@ xmlns:xi="http://www.w3.org/2001/XInclude" py:strip=""> -

    We do not currently have analytics data for ${publisher.title}

    +

    We do not currently have usage data for ${publisher.title}

    @@ -15,7 +15,7 @@
    -

    ${h.link_to("More analytics for " + publisher.title, h.url_for(controller='ckanext.ga_report.controller:GaPublisherReport',action='read',id=publisher.name))}

    +

    ${h.link_to("More usage data for " + publisher.title, h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='read_publisher',id=publisher.name))}

    --- a/ckanext/ga_report/templates/ga_report/publisher/read.html +++ b/ckanext/ga_report/templates/ga_report/publisher/read.html @@ -5,46 +5,46 @@ - Analytics for ${g.site_title} + Usage by Dataset
  • -

    Publishers

    -

    - Dataset views records the number of times a specific dataset page has been viewed. Visits records the number of unique site visits. -

    -

    - Note: this data does not include API calls. +

    Download

    +

    + Download as CSV

  • - -
  • -

    Download

    -

    - Download as CSV
    -

    -
  • +
    -

    Site Usage

    +

    Site Usage

    - ${usage_nav(c.publisher.title, c.publisher)} + ${usage_nav('Datasets')} -
    -
    + +
    - -
    - + + +
    + - +

    ${c.publisher.title}

    + +

    No page views in this period

    +
    @@ -58,7 +58,7 @@ -
    Dataset Views${visits}
    +
    --- a/ckanext/ga_report/templates/ga_report/site/index.html +++ b/ckanext/ga_report/templates/ga_report/site/index.html @@ -9,34 +9,29 @@
  • -

    Site-wide

    -

    - Note: this data does not include API calls and some values have been rounded up to 2 decimal places. Where there are a large number of browser versions they have been grouped together. +

    Download

    +

    + Download as CSV

  • -
  • -

    Download

    -

    - Download as CSV
    -

    -
  • +

    Site Usage

    - ${usage_nav('Site-wide', None)} + ${usage_nav('Site-wide')}
    - +
    @@ -88,6 +83,7 @@
    +

    Note: Where a browser has a large number of versions, these have been grouped together.

    ${stat_table(c.browser_versions)}
    @@ -100,11 +96,11 @@ ${stat_table(c.os_versions)}
    -

    Number of visits to urls referred from social networks

    +

    Number of visits that were referred from social networks

    ${social_table(c.social_referrer_totals)}
    -

    Percentage of visits referred from these social networks

    +

    Percentage of visits that were referred from these social networks

    ${stat_table(c.social_networks, 'Visits')}
    --- /dev/null +++ b/ckanext/ga_report/tests/test_download.py @@ -1,1 +1,30 @@ +from nose.tools import assert_equal +from ckanext.ga_report.download_analytics import DownloadAnalytics + +_filter_browser_version = DownloadAnalytics._filter_browser_version + +class TestBrowserVersionFilter: + def test_chrome(self): + assert_equal(_filter_browser_version('Chrome', u'6.0.472.0'), '6') + def test_firefox(self): + assert_equal(_filter_browser_version('Firefox', u'16.1'), '16') + def test_safari(self): + assert_equal(_filter_browser_version('Safari', u'534.55.3'), '53X') + assert_equal(_filter_browser_version('Safari', u'1534.55.3'), '15XX') + def test_ie(self): + assert_equal(_filter_browser_version('Internet Explorer', u'8.0'), '8') + def test_opera_mini(self): + assert_equal(_filter_browser_version('Opera Mini', u'6.5.27431'), '6') + def test_opera(self): + assert_equal(_filter_browser_version('Opera', u'11.60'), '11') + +class TestDownloadAnalytics: + def test_filter_out_long_tail(self): + data = {'Firefox': 100, + 'Obscure Browser': 5, + 'Chrome': 150} + DownloadAnalytics._filter_out_long_tail(data, 10) + assert_equal(data, {'Firefox': 100, + 'Chrome': 150}) + --- a/setup.py +++ b/setup.py @@ -26,7 +26,7 @@ entry_points=\ """ [ckan.plugins] - # Add plugins here, eg + # Add plugins here ga-report=ckanext.ga_report.plugin:GAReportPlugin [paste.paster_command]