Implements downloads counts (for dataset resources) and fixes an issue with 'All' records.
[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 +1,10 @@
 import os
 import logging
 import datetime
+import httplib
 import collections
+import requests
+import json
 from pylons import config
 from ga_model import _normalize_url
 import ga_model
@@ -18,13 +21,14 @@
 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
@@ -32,6 +36,11 @@
         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),)
@@ -126,7 +135,7 @@
                 # Make sure the All records are correct.
                 ga_model.post_update_url_stats()
 
-                log.info('Aggregating datasets by publisher')
+                log.info('Associating datasets with their publisher')
                 ga_model.update_publisher_stats(period_name) # about 30 seconds.
 
 
@@ -144,17 +153,27 @@
         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:
@@ -173,15 +192,32 @@
 
         # 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()
+        try:
+            # Because of issues of invalid responses, we are going to make these requests
+            # ourselves.
+            headers = {'authorization': 'Bearer ' + self.token}
+
+            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"
+
+            r = requests.get("https://www.googleapis.com/analytics/v3/data/ga", params=args, headers=headers)
+            if r.status_code != 200:
+              raise Exception("Request with params: %s failed" % args)
+
+            results = json.loads(r.content)
+            print len(results.keys())
+        except Exception, e:
+            log.exception(e)
+            #return dict(url=[])
+            raise e
 
         packages = []
         log.info("There are %d results" % results['totalResults'])
@@ -221,25 +257,77 @@
             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
+
+        try:
+            log.info("Trying to refresh our OAuth token")
+            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],
@@ -252,14 +340,28 @@
         # 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:visitBounceRate',
-                                 dimensions='ga:pagePath',
-                                 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["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',
@@ -275,14 +377,28 @@
 
     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:
@@ -298,20 +414,32 @@
 
 
     def _download_stats(self, start_date, end_date, period_name, period_complete_day):
-        """ Fetches stats about language and country """
+        """ 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()
+        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
@@ -320,7 +448,14 @@
             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.
@@ -334,20 +469,35 @@
                 if package_name:
                     data[package_name] = data.get(package_name, 0) + int(result[1])
                 else:
-                    log.warning(u"Could not find resource for URL: {url}".format(url=url))
+                    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()
+        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)
@@ -355,14 +505,25 @@
 
     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:
@@ -374,14 +535,24 @@
 
     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:
@@ -399,14 +570,27 @@
 
     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']
 
@@ -448,14 +632,24 @@
     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 = {}