Make sure that the url normalization for social info takes the correct data
[ckanext-ga-report.git] / ckanext / ga_report / download_analytics.py
blob:a/ckanext/ga_report/download_analytics.py -> blob:b/ckanext/ga_report/download_analytics.py
--- a/ckanext/ga_report/download_analytics.py
+++ b/ckanext/ga_report/download_analytics.py
@@ -1,7 +1,12 @@
 import os
 import logging
 import datetime
+import httplib
+import urllib
 import collections
+import requests
+import json
+import re
 from pylons import config
 from ga_model import _normalize_url
 import ga_model
@@ -13,17 +18,19 @@
 FORMAT_MONTH = '%Y-%m'
 MIN_VIEWS = 50
 MIN_VISITS = 20
+MIN_DOWNLOADS = 10
 
 class DownloadAnalytics(object):
     '''Downloads and stores analytics info'''
 
-    def __init__(self, service=None, profile_id=None, delete_first=False,
+    def __init__(self, service=None, token=None, profile_id=None, delete_first=False,
                  skip_url_stats=False):
         self.period = config['ga-report.period']
         self.service = service
         self.profile_id = profile_id
         self.delete_first = delete_first
         self.skip_url_stats = skip_url_stats
+        self.token = token
 
     def specific_month(self, date):
         import calendar
@@ -31,6 +38,11 @@
         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),)
@@ -111,19 +123,23 @@
                 accountName = config.get('googleanalytics.account')
 
                 log.info('Downloading analytics for dataset views')
-                data = self.download(start_date, end_date, '~/%s/dataset/[a-z0-9-_]+' % accountName)
+                data = self.download(start_date, end_date, '~^/dataset/[a-z0-9-_]+')
 
                 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, '~/%s/publisher/[a-z0-9-_]+' % accountName)
+                data = self.download(start_date, end_date, '~^/organization/[a-z0-9-_]+')
 
                 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, period_complete_day )
@@ -139,21 +155,31 @@
         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()
+        try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            args = dict(ids='ga:' + self.profile_id,
+                       filters=query,
+                       metrics=metrics,
+                       sort=sort,
+                       dimensions="ga:landingPagePath,ga:socialNetwork",
+                       max_results=10000)
+
+            args['start-date'] = start_date
+            args['end-date'] = end_date
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
+
         data = collections.defaultdict(list)
         rows = results.get('rows',[])
         for row in rows:
-            url = _normalize_url('http:/' + row[0])
+            url = row[0]
             data[url].append( (row[1], int(row[2]),) )
         ga_model.update_social(period_name, data)
 
@@ -168,22 +194,34 @@
 
         # 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:pagePath",
-                                 max_results=10000,
-                                 end_date=end_date).execute()
+	# https://ga-dev-tools.appspot.com/explorer/
+        try:
+            args = {}
+            args["sort"] = "-ga:pageviews"
+            args["max-results"] = 100000
+            args["dimensions"] = "ga:pagePath"
+            args["start-date"] = start_date
+            args["end-date"] = end_date
+            args["metrics"] = metrics
+            args["ids"] = "ga:" + self.profile_id
+            args["filters"] = query
+            args["alt"] = "json"
+            print args
+            results = self._get_json(args)
+
+        except Exception, e:
+            log.exception(e)
+            return dict(url=[])
 
         packages = []
-        for entry in results.get('rows'):
+        log.info("There are %d results" % results['totalResults'])
+	if results['totalResults'] > 0:
+          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
-
-            if not url.startswith('/dataset/') and not url.startswith('/publisher/'):
+            #url = _normalize_url('http:/' + loc) # strips off domain e.g. www.data.gov.uk or data.gov.uk
+            url = loc
+	    #print url
+            if not url.startswith('/dataset/') and not url.startswith('/organization/'):
                 # filter out strays like:
                 # /data/user/login?came_from=http://data.gov.uk/dataset/os-code-point-open
                 # /403.html?page=/about&from=http://data.gov.uk/publisher/planning-inspectorate
@@ -203,7 +241,7 @@
         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, period_complete_day)
@@ -215,25 +253,78 @@
             data[key] = data.get(key,0) + result[1]
         return data
 
+    def _get_json(self, params, prev_fail=False):
+        ga_token_filepath = os.path.expanduser(config.get('googleanalytics.token.filepath', ''))
+        if not ga_token_filepath:
+            print 'ERROR: In the CKAN config you need to specify the filepath of the ' \
+                'Google Analytics token file under key: googleanalytics.token.filepath'
+            return
+
+        log.info("Trying to refresh our OAuth token")
+        try:
+            from ga_auth import init_service
+            self.token, svc = init_service(ga_token_filepath, None)
+            log.info("OAuth token refreshed")
+        except Exception, auth_exception:
+            log.error("Oauth refresh failed")
+            log.exception(auth_exception)
+            return
+
+        try:
+            headers = {'authorization': 'Bearer ' + self.token}
+            r = requests.get("https://www.googleapis.com/analytics/v3/data/ga", params=params, headers=headers)
+            if r.status_code != 200:
+                log.info("STATUS: %s" % (r.status_code,))
+                log.info("CONTENT: %s" % (r.content,))
+                raise Exception("Request with params: %s failed" % params)
+
+            return json.loads(r.content)
+        except Exception, e:
+              log.exception(e)
+
+        return dict(url=[])
+
     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,
-                                 start_date=start_date,
-                                 metrics='ga:pageviews',
-                                 sort='-ga:pageviews',
-                                 max_results=10000,
-                                 end_date=end_date).execute()
+        try:
+            args = {}
+            args["max-results"] = 100000
+            args["start-date"] = start_date
+            args["end-date"] = end_date
+            args["ids"] = "ga:" + self.profile_id
+
+            args["metrics"] = "ga:pageviews"
+            args["sort"] = "-ga:pageviews"
+            args["alt"] = "json"
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
         result_data = results.get('rows')
         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,
-                                 start_date=start_date,
-                                 metrics='ga:pageviewsPerVisit,ga:avgTimeOnSite,ga:percentNewVisits,ga:visits',
-                                 max_results=10000,
-                                 end_date=end_date).execute()
+        try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            args = {}
+            args["max-results"] = 100000
+            args["start-date"] = start_date
+            args["end-date"] = end_date
+            args["ids"] = "ga:" + self.profile_id
+
+            args["metrics"] = "ga:pageviewsPerVisit,ga:avgTimeOnSite,ga:percentNewVisits,ga:visits"
+            args["alt"] = "json"
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
         result_data = results.get('rows')
         data = {
             'Pages per visit': result_data[0][0],
@@ -244,39 +335,66 @@
         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'),
-                          config.get('ga-report.bounce_url', '/'))
-        results = self.service.data().ga().get(
-                                 ids='ga:' + self.profile_id,
-                                 filters='ga:pagePath==%s' % (path,),
-                                 start_date=start_date,
-                                 metrics='ga:bounces,ga:pageviews',
-                                 dimensions='ga:pagePath',
-                                 max_results=10000,
-                                 end_date=end_date).execute()
+        path = '/' #% (config.get('googleanalytics.account'),                          config.get('ga-report.bounce_url', '/'))
+
+        try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            args = {}
+            args["max-results"] = 100000
+            args["start-date"] = start_date
+            args["end-date"] = end_date
+            args["ids"] = "ga:" + self.profile_id
+
+            args["filters"] = 'ga:pagePath==%s' % (path,)
+            args["dimensions"] = 'ga:pagePath'
+            args["metrics"] = "ga:visitBounceRate"
+            args["alt"] = "json"
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
         result_data = results.get('rows')
         if not result_data or len(result_data) != 1:
             log.error('Could not pinpoint the bounces for path: %s. Got results: %r',
                       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},
+        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,
-                                 start_date=start_date,
-                                 metrics='ga:pageviews',
-                                 sort='-ga:pageviews',
-                                 dimensions="ga:language,ga:country",
-                                 max_results=10000,
-                                 end_date=end_date).execute()
+
+        try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            args = {}
+            args["max-results"] = 100000
+            args["start-date"] = start_date
+            args["end-date"] = end_date
+            args["ids"] = "ga:" + self.profile_id
+
+            args["dimensions"] = "ga:language,ga:country"
+            args["metrics"] = "ga:pageviews"
+            args["sort"] = "-ga:pageviews"
+            args["alt"] = "json"
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
         result_data = results.get('rows')
         data = {}
         for result in result_data:
@@ -291,16 +409,124 @@
         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 = {}
+
+        try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            args = {}
+            args["max-results"] = 100000
+            args["start-date"] = start_date
+            args["end-date"] = end_date
+            args["ids"] = "ga:" + self.profile_id
+
+            args["filters"] = 'ga:eventAction==Download'
+            args["dimensions"] = "ga:eventLabel"
+            args["metrics"] = "ga:totalEvents"
+            args["alt"] = "json"
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
+        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 = urllib.unquote(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()
+		
+		# new style internal download links
+		if re.search('(?:/resource/)(.*)(?:/download/)',url):
+		    resource_id = re.search('(?:/resource/)(.*)(?:/download/)',url)
+                    r = q.filter(model.Resource.id.like("%s%%" % resource_id.group(1))).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'))
+
+        '''try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            args = dict( ids='ga:' + self.profile_id,
+                         filters='ga:eventAction==download-cache',
+                         metrics='ga:totalEvents',
+                         sort='-ga:totalEvents',
+                         dimensions="ga:eventLabel",
+                         max_results=10000)
+            args['start-date'] = start_date
+            args['end-date'] = end_date
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
+        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,
-                                 start_date=start_date,
-                                 metrics='ga:pageviews',
-                                 sort='-ga:pageviews',
-                                 dimensions="ga:socialNetwork,ga:referralPath",
-                                 max_results=10000,
-                                 end_date=end_date).execute()
+
+        try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            args = dict( ids='ga:' + self.profile_id,
+                         metrics='ga:pageviews',
+                         sort='-ga:pageviews',
+                         dimensions="ga:socialNetwork,ga:referralPath",
+                         max_results=10000)
+            args['start-date'] = start_date
+            args['end-date'] = end_date
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
         result_data = results.get('rows')
         data = {}
         for result in result_data:
@@ -312,14 +538,24 @@
 
     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,
-                                 start_date=start_date,
-                                 metrics='ga:pageviews',
-                                 sort='-ga:pageviews',
-                                 dimensions="ga:operatingSystem,ga:operatingSystemVersion",
-                                 max_results=10000,
-                                 end_date=end_date).execute()
+        try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            args = dict( ids='ga:' + self.profile_id,
+                         metrics='ga:pageviews',
+                         sort='-ga:pageviews',
+                         dimensions="ga:operatingSystem,ga:operatingSystemVersion",
+                         max_results=10000)
+            args['start-date'] = start_date
+            args['end-date'] = end_date
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
         result_data = results.get('rows')
         data = {}
         for result in result_data:
@@ -337,14 +573,27 @@
 
     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,
-                                 start_date=start_date,
-                                 metrics='ga:pageviews',
-                                 sort='-ga:pageviews',
-                                 dimensions="ga:browser,ga:browserVersion",
-                                 max_results=10000,
-                                 end_date=end_date).execute()
+
+        try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            args = dict( ids='ga:' + self.profile_id,
+                         metrics='ga:pageviews',
+                         sort='-ga:pageviews',
+                         dimensions="ga:browser,ga:browserVersion",
+                         max_results=10000)
+
+            args['start-date'] = start_date
+            args['end-date'] = end_date
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
+
         result_data = results.get('rows')
         # e.g. [u'Firefox', u'19.0', u'20']
 
@@ -386,14 +635,24 @@
     def _mobile_stats(self, start_date, end_date, period_name, period_complete_day):
         """ Info about mobile devices """
 
-        results = self.service.data().ga().get(
-                                 ids='ga:' + self.profile_id,
-                                 start_date=start_date,
-                                 metrics='ga:pageviews',
-                                 sort='-ga:pageviews',
-                                 dimensions="ga:mobileDeviceBranding, ga:mobileDeviceInfo",
-                                 max_results=10000,
-                                 end_date=end_date).execute()
+        try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            args = dict( ids='ga:' + self.profile_id,
+                         metrics='ga:pageviews',
+                         sort='-ga:pageviews',
+                         dimensions="ga:mobileDeviceBranding, ga:mobileDeviceInfo",
+                         max_results=10000)
+            args['start-date'] = start_date
+            args['end-date'] = end_date
+
+            results = self._get_json(args)
+        except Exception, e:
+            log.exception(e)
+            results = dict(url=[])
+
 
         result_data = results.get('rows')
         data = {}