From: David Read Date: Wed, 23 Jan 2013 12:18:26 +0000 Subject: [xs] #219 Improved punctuation. X-Git-Url: http://maxious.lambdacomplex.org/git/?p=ckanext-ga-report.git&a=commitdiff&h=c25c4f885692e2e176fc124354187176cb4236c8 --- [xs] #219 Improved punctuation. --- --- a/.gitignore +++ b/.gitignore @@ -1,6 +1,7 @@ *.py[co] *.py~ .gitignore +ckan.log # Packages *.egg --- a/ckanext/ga_report/command.py +++ b/ckanext/ga_report/command.py @@ -23,7 +23,7 @@ import ckan.model as model model.Session.remove() model.Session.configure(bind=model.meta.engine) - log = logging.getLogger('ckanext.ga-report') + log = logging.getLogger('ckanext.ga_report') import ga_model ga_model.init_tables() @@ -55,6 +55,36 @@ init_service('token.dat', 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): --- 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 +189,22 @@ 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 + graph = {} + for stat in q: + graph[ stat.key ] = graph.get(stat.key,{ + 'name':stat.key, + 'data': [] + }) + graph[ stat.key ]['data'].append({ + 'x':_get_unix_epoch(stat.period_name), + 'y':float(stat.value) + }) + setattr(c, v+'_graph', json.dumps( _to_rickshaw(graph.values(),percentageMode=True) )) + + # Buffer the tabular data if c.month: entries = [] q = q.filter(GA_Stat.period_name==c.month).\ @@ -155,7 +221,7 @@ # 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 ]) @@ -180,7 +246,9 @@ writer = csv.writer(response) writer.writerow(["Publisher Title", "Publisher Name", "Views", "Visits", "Period Name"]) - for publisher,view,visit in _get_top_publishers(None): + top_publishers, top_publishers_graph = _get_top_publishers(None) + + for publisher,view,visit in top_publishers: writer.writerow([publisher.title.encode('utf-8'), publisher.name.encode('utf-8'), view, @@ -206,13 +274,14 @@ 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 +289,7 @@ # 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', '') @@ -228,15 +297,17 @@ if c.month: c.month_desc = ''.join([m[1] for m in c.months if m[0]==c.month]) - c.top_publishers = _get_top_publishers() + c.top_publishers, graph_data = _get_top_publishers() + c.top_publishers_graph = json.dumps( _to_rickshaw(graph_data.values()) ) + return render('ga_report/publisher/index.html') 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 +317,26 @@ 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 = 0 + for x in dls: + downloads += int(x.value) + else: + downloads = 'No data' + top_packages.append((package, entry.pageviews, entry.visits, downloads)) else: log.warning('Could not find package associated package') @@ -278,7 +366,7 @@ # 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', '') @@ -296,7 +384,78 @@ c.top_packages = self._get_packages(c.publisher, 20) + # Graph query + top_package_names = [ x[0].name for x in c.top_packages ] + graph_query = model.Session.query(GA_Url,model.Package)\ + .filter(model.Package.name==GA_Url.package_id)\ + .filter(GA_Url.url.like('/dataset/%'))\ + .filter(GA_Url.package_id.in_(top_package_names)) + graph_data = {} + for entry,package in graph_query: + if not package: continue + if entry.period_name=='All': continue + graph_data[package.id] = graph_data.get(package.id,{ + 'name':package.title, + 'data':[] + }) + graph_data[package.id]['data'].append({ + 'x':_get_unix_epoch(entry.period_name), + 'y':int(entry.pageviews), + }) + + c.graph_data = json.dumps( _to_rickshaw(graph_data.values()) ) + return render('ga_report/publisher/read.html') + +def _to_rickshaw(data, percentageMode=False): + if data==[]: + return data + # Create a consistent x-axis + num_points = [ len(package['data']) for package in data ] + ideal_index = num_points.index( max(num_points) ) + x_axis = [ point['x'] for point in data[ideal_index]['data'] ] + for package in data: + xs = [ point['x'] for point in package['data'] ] + assert set(xs).issubset( set(x_axis) ), (xs, x_axis) + # Zero pad any missing values + for x in set(x_axis).difference(set(xs)): + package['data'].append( {'x':x, 'y':0} ) + assert len(package['data'])==len(x_axis), (len(package['data']),len(x_axis),package['data'],x_axis,set(x_axis).difference(set(xs))) + if percentageMode: + # Transform data into percentage stacks + totals = {} + for x in x_axis: + for package in data: + for point in package['data']: + totals[ point['x'] ] = totals.get(point['x'],0) + point['y'] + # Roll insignificant series into a catch-all + THRESHOLD = 0.01 + significant_series = [] + for package in data: + for point in package['data']: + fraction = float(point['y']) / totals[point['x']] + if fraction>THRESHOLD and not (package in significant_series): + significant_series.append(package) + temp = {} + for package in data: + if package in significant_series: continue + for point in package['data']: + temp[point['x']] = temp.get(point['x'],0) + point['y'] + catch_all = { 'name':'Other','data': [ {'x':x,'y':y} for x,y in temp.items() ] } + # Roll insignificant series into one + data = significant_series + data.append(catch_all) + # Turn each point into a percentage + for package in data: + for point in package['data']: + point['y'] = (point['y']*100) / totals[point['x']] + # Sort the points + for package in data: + package['data'] = sorted( package['data'], key=lambda x:x['x'] ) + # Strip the latest month's incomplete analytics + package['data'] = package['data'][:-1] + return data + def _get_top_publishers(limit=20): ''' @@ -319,11 +478,35 @@ top_publishers = [] res = connection.execute(q, month) + department_ids = [] for row in res: g = model.Group.get(row[0]) if g: + department_ids.append(row[0]) top_publishers.append((g, row[1], row[2])) - return top_publishers + + graph = {} + if limit is not None: + # Query for a history graph of these publishers + q = model.Session.query( + GA_Url.department_id, + GA_Url.period_name, + func.sum(cast(GA_Url.pageviews,sqlalchemy.types.INT)))\ + .filter( GA_Url.department_id.in_(department_ids) )\ + .filter( GA_Url.period_name!='All' )\ + .filter( GA_Url.url.like('/dataset/%') )\ + .filter( GA_Url.package_id!='' )\ + .group_by( GA_Url.department_id, GA_Url.period_name ) + for dept_id,period_name,views in q: + graph[dept_id] = graph.get( dept_id, { + 'name' : model.Group.get(dept_id).title, + 'data' : [] + }) + graph[dept_id]['data'].append({ + 'x': _get_unix_epoch(period_name), + 'y': views + }) + return top_publishers, graph def _get_publishers(): --- 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''' @@ -31,6 +32,11 @@ first_of_this_month = datetime.datetime(date.year, date.month, 1) _, 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) + # if this is the latest month, note that it is only up until today + now = datetime.datetime.now() + if now.year == date.year and now.month == date.month: + last_day_of_month = now.day + last_of_this_month = now periods = ((date.strftime(FORMAT_MONTH), last_day_of_month, first_of_this_month, last_of_this_month),) @@ -122,11 +128,15 @@ 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') + # Make sure the All records are correct. + ga_model.post_update_url_stats() + + log.info('Associating datasets with their 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 +163,8 @@ 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 +189,7 @@ 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 +206,7 @@ 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 +214,10 @@ 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 +226,7 @@ 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 +236,8 @@ 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 +252,7 @@ '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 +261,7 @@ 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 +271,14 @@ 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 +293,84 @@ 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 data downloads """ + 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): + progress_total = len(result_data) + progress_count = 0 + resources_not_matched = [] + for result in result_data: + progress_count += 1 + if progress_count % 100 == 0: + log.debug('.. %d/%d done so far', progress_count, progress_total) + + 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: + resources_not_matched.append(url) + continue + if resources_not_matched: + log.debug('Could not match %i or %i resource URLs to datasets. e.g. %r', + len(resources_not_matched), progress_total, resources_not_matched[:3]) + + log.info('Associating downloads of resource URLs with their respective datasets') + 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() + log.info('Associating downloads of cache resource URLs with their respective datasets') + 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 +386,10 @@ 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 +404,17 @@ 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 +431,14 @@ 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 +462,7 @@ 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 +479,13 @@ 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 +161,64 @@ def pre_update_url_stats(period_name): - model.Session.query(GA_Url).\ - filter(GA_Url.period_name==period_name).delete() - model.Session.query(GA_Url).\ - filter(GA_Url.period_name=='All').delete() + q = model.Session.query(GA_Url).\ + filter(GA_Url.period_name==period_name) + log.debug("Deleting %d '%s' records" % (q.count(), period_name)) + q.delete() + + q = model.Session.query(GA_Url).\ + filter(GA_Url.period_name == 'All') + log.debug("Deleting %d 'All' records..." % q.count()) + q.delete() + + model.Session.flush() + model.Session.commit() + model.repo.commit_and_remove() + log.debug('...done') + +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. + """ + log.debug('Post-processing "All" records...') + 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] + + progress_total = len(views.keys()) + progress_count = 0 + for key in views.keys(): + progress_count += 1 + if progress_count % 100 == 0: + log.debug('.. %d/%d done so far', progress_count, progress_total) + + 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() + log.debug('..done') def update_url_stats(period_name, period_complete_day, url_data): @@ -170,9 +227,14 @@ stores them in GA_Url under the period and recalculates the totals for the 'All' period. ''' + progress_total = len(progress_data) + progress_count = 0 for url, views, visits in url_data: + progress_count += 1 + if progress_count % 100 == 0: + log.debug('.. %d/%d done so far', progress_count, progress_total) + package, publisher = _get_package_and_publisher(url) - item = model.Session.query(GA_Url).\ filter(GA_Url.period_name==period_name).\ @@ -213,8 +275,8 @@ '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 +402,10 @@ ''' 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 @@ -106,6 +106,10 @@ for entry in entries: if len(datasets) < count: p = model.Package.get(entry.url[len('/dataset/'):]) + if not p: + _log.warning("Could not find Package for {url}".format(url=entry.url)) + continue + if not p in datasets: datasets[p] = {'views':0, 'visits': 0} datasets[p]['views'] = datasets[p]['views'] + int(entry.pageviews) --- a/ckanext/ga_report/plugin.py +++ b/ckanext/ga_report/plugin.py @@ -42,6 +42,16 @@ controller='ckanext.ga_report.controller:GaReport', action='csv' ) + map.connect( + '/data/site-usage/downloads', + controller='ckanext.ga_report.controller:GaReport', + action='downloads' + ) + map.connect( + '/data/site-usage/downloads_{month}.csv', + controller='ckanext.ga_report.controller:GaReport', + action='csv_downloads' + ) # GaDatasetReport map.connect( --- /dev/null +++ b/ckanext/ga_report/public/css/ga_report.css @@ -1,1 +1,41 @@ +.table-condensed td.sparkline-cell { + padding: 1px 0 0 0; + width: 108px; + text-align: center; +} +.rickshaw_chart_container { + position: relative; + height: 350px; + margin: 0 auto 20px auto; +} +.rickshaw_chart { + position: absolute; + left: 40px; + width: 500px; + top: 0; + bottom: 0; +} +.rickshaw_legend { + position: absolute; + right: 0; + top: 0; + margin-left: 15px; + padding: 0 5px; + background: transparent; + max-width: 150px; + overflow: hidden; + background: rgba(0,0,0,0.05); + border-radius:5px; +} +.rickshaw_y_axis { + position: absolute; + top: 0; + bottom: 0; + width: 40px; +} +.rickshaw_legend .label { + background: transparent !important; + color: #000000 !important; + font-weight: normal !important; +} --- /dev/null +++ b/ckanext/ga_report/public/scripts/ckanext_ga_reports.js @@ -1,1 +1,42 @@ +var CKAN = CKAN || {}; +CKAN.GA_Reports = {}; + +CKAN.GA_Reports.render_rickshaw = function( css_name, data, mode, colorscheme ) { + var palette = new Rickshaw.Color.Palette( { scheme: colorscheme } ); + $.each(data, function(i, object) { + object['color'] = palette.color(); + }); + + var graphElement = document.querySelector("#chart_"+css_name); + + var graph = new Rickshaw.Graph( { + element: document.querySelector("#chart_"+css_name), + renderer: mode, + series: data , + height: 328 + }); + var x_axis = new Rickshaw.Graph.Axis.Time( { graph: graph } ); + var y_axis = new Rickshaw.Graph.Axis.Y( { + graph: graph, + orientation: 'left', + tickFormat: Rickshaw.Fixtures.Number.formatKMBT, + element: document.getElementById('y_axis_'+css_name), + } ); + var legend = new Rickshaw.Graph.Legend( { + element: document.querySelector('#legend_'+css_name), + graph: graph + } ); + var hoverDetail = new Rickshaw.Graph.HoverDetail( { + graph: graph, + formatter: function(series, x, y) { + var date = '' + new Date(x * 1000).toUTCString() + ''; + var swatch = ''; + var content = swatch + series.name + ": " + parseInt(y) + '
' + date; + return content; + } + } ); + graph.render(); +}; + + --- /dev/null +++ b/ckanext/ga_report/public/scripts/vendor/d3.layout.min.js @@ -1,1 +1,1 @@ - +(function(){function a(a){var b=a.source,d=a.target,e=c(b,d),f=[b];while(b!==e)b=b.parent,f.push(b);var g=f.length;while(d!==e)f.splice(g,0,d),d=d.parent;return f}function b(a){var b=[],c=a.parent;while(c!=null)b.push(a),a=c,c=c.parent;return b.push(a),b}function c(a,c){if(a===c)return a;var d=b(a),e=b(c),f=d.pop(),g=e.pop(),h=null;while(f===g)h=f,f=d.pop(),g=e.pop();return h}function g(a){a.fixed|=2}function h(a){a!==f&&(a.fixed&=1)}function i(){j(),f.fixed&=1,e=f=null}function j(){f.px+=d3.event.dx,f.py+=d3.event.dy,e.resume()}function k(a,b,c){var d=0,e=0;a.charge=0;if(!a.leaf){var f=a.nodes,g=f.length,h=-1,i;while(++hd&&(c=b,d=e);return c}function u(a){return a.reduce(v,0)}function v(a,b){return a+b[1]}function w(a,b){return x(a,Math.ceil(Math.log(b.length)/Math.LN2+1))}function x(a,b){var c=-1,d=+a[0],e=(a[1]-d)/b,f=[];while(++c<=b)f[c]=e*c+d;return f}function y(a){return[d3.min(a),d3.max(a)]}function z(a,b){return a.sort=d3.rebind(a,b.sort),a.children=d3.rebind(a,b.children),a.links=D,a.value=d3.rebind(a,b.value),a.nodes=function(b){return E=!0,(a.nodes=a)(b)},a}function A(a){return a.children}function B(a){return a.value}function C(a,b){return b.value-a.value}function D(a){return d3.merge(a.map(function(a){return(a.children||[]).map(function(b){return{source:a,target:b}})}))}function F(a,b){return a.value-b.value}function G(a,b){var c=a._pack_next;a._pack_next=b,b._pack_prev=a,b._pack_next=c,c._pack_prev=b}function H(a,b){a._pack_next=b,b._pack_prev=a}function I(a,b){var c=b.x-a.x,d=b.y-a.y,e=a.r+b.r;return e*e-c*c-d*d>.001}function J(a){function l(a){b=Math.min(a.x-a.r,b),c=Math.max(a.x+a.r,c),d=Math.min(a.y-a.r,d),e=Math.max(a.y+a.r,e)}var b=Infinity,c=-Infinity,d=Infinity,e=-Infinity,f=a.length,g,h,i,j,k;a.forEach(K),g=a[0],g.x=-g.r,g.y=0,l(g);if(f>1){h=a[1],h.x=h.r,h.y=0,l(h);if(f>2){i=a[2],O(g,h,i),l(i),G(g,i),g._pack_prev=i,G(i,h),h=g._pack_next;for(var m=3;m0?(H(g,j),h=j,m--):(H(j,h),g=j,m--)}}}var q=(b+c)/2,r=(d+e)/2,s=0;for(var m=0;m0&&(a=d)}return a}function X(a,b){return a.x-b.x}function Y(a,b){return b.x-a.x}function Z(a,b){return a.depth-b.depth}function $(a,b){function c(a,d){var e=a.children;if(e&&(i=e.length)){var f,g=null,h=-1,i;while(++h=0)f=d[e]._tree,f.prelim+=b,f.mod+=b,b+=f.shift+(c+=f.change)}function ba(a,b,c){a=a._tree,b=b._tree;var d=c/(b.number-a.number);a.change+=d,b.change-=d,b.shift+=c,b.prelim+=c,b.mod+=c}function bb(a,b,c){return a._tree.ancestor.parent==b.parent?a._tree.ancestor:c}function bc(a){return{x:a.x,y:a.y,dx:a.dx,dy:a.dy}}function bd(a,b){var c=a.x+b[3],d=a.y+b[0],e=a.dx-b[1]-b[3],f=a.dy-b[0]-b[2];return e<0&&(c+=e/2,e=0),f<0&&(d+=f/2,f=0),{x:c,y:d,dx:e,dy:f}}d3.layout={},d3.layout.bundle=function(){return function(b){var c=[],d=-1,e=b.length;while(++de&&(e=h),d.push(h)}for(g=0;g=i[0]&&o<=i[1]&&(k=g[d3.bisect(j,o,1,m)-1],k.y+=n,k.push(e[f]));return g}var a=!0,b=Number,c=y,d=w;return e.value=function(a){return arguments.length?(b=a,e):b},e.range=function(a){return arguments.length?(c=d3.functor(a),e):c},e.bins=function(a){return arguments.length?(d=typeof a=="number"?function(b){return x(b,a)}:d3.functor(a),e):d},e.frequency=function(b){return arguments.length?(a=!!b,e):a},e},d3.layout.hierarchy=function(){function e(f,h,i){var j=b.call(g,f,h),k=E?f:{data:f};k.depth=h,i.push(k);if(j&&(m=j.length)){var l=-1,m,n=k.children=[],o=0,p=h+1;while(++l0&&(ba(bb(g,a,d),a,m),i+=m,j+=m),k+=g._tree.mod,i+=e._tree.mod,l+=h._tree.mod,j+=f._tree.mod;g&&!V(f)&&(f._tree.thread=g,f._tree.mod+=k-j),e&&!U(h)&&(h._tree.thread=e,h._tree.mod+=i-l,d=a)}return d}var f=a.call(this,d,e),g=f[0];$(g,function(a,b){a._tree={ancestor:a,prelim:0,mod:0,change:0,shift:0,number:b?b._tree.number+1:0}}),h(g),i(g,-g._tree.prelim);var k=W(g,Y),l=W(g,X),m=W(g,Z),n=k.x-b(k,l)/2,o=l.x+b(l,k)/2,p=m.depth||1;return $(g,function(a){a.x=(a.x-n)/(o-n)*c[0],a.y=a.depth/p*c[1],delete a._tree}),f}var a=d3.layout.hierarchy().sort(null).value(null),b=T,c=[1,1];return d.separation=function(a){return arguments.length?(b=a,d):b},d.size=function(a){return arguments.length?(c=a,d):c},z(d,a)},d3.layout.treemap=function(){function i(a,b){var c=-1,d=a.length,e,f;while(++c0)d.push(g=f[o-1]),d.area+=g.area,(k=l(d,n))<=h?(f.pop(),h=k):(d.area-=d.pop().area,m(d,n,c,!1),n=Math.min(c.dx,c.dy),d.length=d.area=0,h=Infinity);d.length&&(m(d,n,c,!0),d.length=d.area=0),b.forEach(j)}}function k(a){var b=a.children;if(b&&b.length){var c=e(a),d=b.slice(),f,g=[];i(d,c.dx*c.dy/a.value),g.area=0;while(f=d.pop())g.push(f),g.area+=f.area,f.z!=null&&(m(g,f.z?c.dx:c.dy,c,!d.length),g.length=g.area=0);b.forEach(k)}}function l(a,b){var c=a.area,d,e=0,f=Infinity,g=-1,i=a.length;while(++ge&&(e=d)}return c*=c,b*=b,c?Math.max(b*e*h/c,c/(b*f*h)):Infinity}function m(a,c,d,e){var f=-1,g=a.length,h=d.x,i=d.y,j=c?b(a.area/c):0,k;if(c==d.dx){if(e||j>d.dy)j=j?d.dy:0;while(++fd.dx)j=j?d.dx:0;while(++f= 0 ? value.substring(i) : (i = value.length, ""), t = []; + while (i > 0) t.push(value.substring(i -= 3, i + 3)); + return t.reverse().join(",") + f; + } + function d3_formatPrefix(d, i) { + var k = Math.pow(10, Math.abs(8 - i) * 3); + return { + scale: i > 8 ? function(d) { + return d / k; + } : function(d) { + return d * k; + }, + symbol: d + }; + } + function d3_ease_clamp(f) { + return function(t) { + return t <= 0 ? 0 : t >= 1 ? 1 : f(t); + }; + } + function d3_ease_reverse(f) { + return function(t) { + return 1 - f(1 - t); + }; + } + function d3_ease_reflect(f) { + return function(t) { + return .5 * (t < .5 ? f(2 * t) : 2 - f(2 - 2 * t)); + }; + } + function d3_ease_identity(t) { + return t; + } + function d3_ease_poly(e) { + return function(t) { + return Math.pow(t, e); + }; + } + function d3_ease_sin(t) { + return 1 - Math.cos(t * Math.PI / 2); + } + function d3_ease_exp(t) { + return Math.pow(2, 10 * (t - 1)); + } + function d3_ease_circle(t) { + return 1 - Math.sqrt(1 - t * t); + } + function d3_ease_elastic(a, p) { + var s; + if (arguments.length < 2) p = .45; + if (arguments.length < 1) { + a = 1; + s = p / 4; + } else s = p / (2 * Math.PI) * Math.asin(1 / a); + return function(t) { + return 1 + a * Math.pow(2, 10 * -t) * Math.sin((t - s) * 2 * Math.PI / p); + }; + } + function d3_ease_back(s) { + if (!s) s = 1.70158; + return function(t) { + return t * t * ((s + 1) * t - s); + }; + } + function d3_ease_bounce(t) { + return t < 1 / 2.75 ? 7.5625 * t * t : t < 2 / 2.75 ? 7.5625 * (t -= 1.5 / 2.75) * t + .75 : t < 2.5 / 2.75 ? 7.5625 * (t -= 2.25 / 2.75) * t + .9375 : 7.5625 * (t -= 2.625 / 2.75) * t + .984375; + } + function d3_eventCancel() { + d3.event.stopPropagation(); + d3.event.preventDefault(); + } + function d3_eventSource() { + var e = d3.event, s; + while (s = e.sourceEvent) e = s; + return e; + } + function d3_eventDispatch(target) { + var dispatch = new d3_dispatch, i = 0, n = arguments.length; + while (++i < n) dispatch[arguments[i]] = d3_dispatch_event(dispatch); + dispatch.of = function(thiz, argumentz) { + return function(e1) { + try { + var e0 = e1.sourceEvent = d3.event; + e1.target = target; + d3.event = e1; + dispatch[e1.type].apply(thiz, argumentz); + } finally { + d3.event = e0; + } + }; + }; + return dispatch; + } + function d3_transform(m) { + var r0 = [ m.a, m.b ], r1 = [ m.c, m.d ], kx = d3_transformNormalize(r0), kz = d3_transformDot(r0, r1), ky = d3_transformNormalize(d3_transformCombine(r1, r0, -kz)) || 0; + if (r0[0] * r1[1] < r1[0] * r0[1]) { + r0[0] *= -1; + r0[1] *= -1; + kx *= -1; + kz *= -1; + } + this.rotate = (kx ? Math.atan2(r0[1], r0[0]) : Math.atan2(-r1[0], r1[1])) * d3_transformDegrees; + this.translate = [ m.e, m.f ]; + this.scale = [ kx, ky ]; + this.skew = ky ? Math.atan2(kz, ky) * d3_transformDegrees : 0; + } + function d3_transformDot(a, b) { + return a[0] * b[0] + a[1] * b[1]; + } + function d3_transformNormalize(a) { + var k = Math.sqrt(d3_transformDot(a, a)); + if (k) { + a[0] /= k; + a[1] /= k; + } + return k; + } + function d3_transformCombine(a, b, k) { + a[0] += k * b[0]; + a[1] += k * b[1]; + return a; + } + function d3_interpolateByName(name) { + return name == "transform" ? d3.interpolateTransform : d3.interpolate; + } + function d3_uninterpolateNumber(a, b) { + b = b - (a = +a) ? 1 / (b - a) : 0; + return function(x) { + return (x - a) * b; + }; + } + function d3_uninterpolateClamp(a, b) { + b = b - (a = +a) ? 1 / (b - a) : 0; + return function(x) { + return Math.max(0, Math.min(1, (x - a) * b)); + }; + } + function d3_Color() {} + function d3_rgb(r, g, b) { + return new d3_Rgb(r, g, b); + } + function d3_Rgb(r, g, b) { + this.r = r; + this.g = g; + this.b = b; + } + function d3_rgb_hex(v) { + return v < 16 ? "0" + Math.max(0, v).toString(16) : Math.min(255, v).toString(16); + } + function d3_rgb_parse(format, rgb, hsl) { + var r = 0, g = 0, b = 0, m1, m2, name; + m1 = /([a-z]+)\((.*)\)/i.exec(format); + if (m1) { + m2 = m1[2].split(","); + switch (m1[1]) { + case "hsl": + { + return hsl(parseFloat(m2[0]), parseFloat(m2[1]) / 100, parseFloat(m2[2]) / 100); + } + case "rgb": + { + return rgb(d3_rgb_parseNumber(m2[0]), d3_rgb_parseNumber(m2[1]), d3_rgb_parseNumber(m2[2])); + } + } + } + if (name = d3_rgb_names.get(format)) return rgb(name.r, name.g, name.b); + if (format != null && format.charAt(0) === "#")