Merge pull request #10 from edx/csv

Added support for CSV output

.pylintrc

@@ -62,7 +62,10 @@ disable=
 # E1103: maybe no member
     E1103,
 # C0111: missing docstring (handled by pep257)
-    C0111
+    C0111,
+
+# We can decide if names are invalid on our own
+    invalid-name,
 
 [REPORTS]
 

README.rst

@@ -2,7 +2,9 @@
 edX Analytics API Client
 ========================
 
-The edX Analytics API Client (henceforth, client) allows users to retrieve data from the edX data warehouse.
+The edX Analytics API Client (henceforth, client) allows users to retrieve data from the edX data warehouse. Currently,
+the client supports retrieving course activity and enrollment data. By default, all data is returned in the JSON format.
+Enrollment data may also be retrieved in the CSV format by changing the data_format argument.
 
 Testing
 =======

analyticsclient/client.py

@@ -2,6 +2,7 @@ import logging
 
 import requests
 import requests.exceptions
+from analyticsclient import data_format as DF
 
 from analyticsclient.course import Course
 from analyticsclient.exceptions import ClientError, InvalidRequestError, NotFoundError, TimeoutError
@@ -36,7 +37,7 @@ class Client(object):
         self.status = Status(self)
         self.courses = lambda course_id: Course(self, course_id)
 
-    def get(self, resource, timeout=None):
+    def get(self, resource, timeout=None, data_format=DF.JSON):
         """
         Retrieve the data for a resource.
 
@@ -45,13 +46,17 @@ class Client(object):
             resource (str): Path in the form of slash separated strings.
             timeout (float): Continue to attempt to retrieve a resource for this many seconds before giving up and
                 raising an error.
+            data_format (str): Format in which data should be returned
 
-        Returns: A structure consisting of simple python types (dict, list, int, str etc).
+        Returns: API response data in specified data_format
 
         Raises: ClientError if the resource cannot be retrieved for any reason.
 
         """
-        response = self._request(resource, timeout=timeout)
+        response = self._request(resource, timeout=timeout, data_format=data_format)
+
+        if data_format == DF.CSV:
+            return response.text
 
         try:
             return response.json()
@@ -82,13 +87,18 @@ class Client(object):
             return False
 
     # pylint: disable=no-member
-    def _request(self, resource, timeout=None):
+    def _request(self, resource, timeout=None, data_format=DF.JSON):
         if timeout is None:
             timeout = self.timeout
 
+        accept_format = 'application/json'
+        if data_format == DF.CSV:
+            accept_format = 'text/csv'
+
         headers = {
-            'Accept': 'application/json',
+            'Accept': accept_format,
         }
+
         if self.auth_token:
             headers['Authorization'] = 'Token ' + self.auth_token
 
@@ -114,7 +124,7 @@ class Client(object):
             return response
 
         except requests.exceptions.Timeout:
-            message = "Response from {0} exceeded timeout of {1}s."
+            message = "Response from {0} exceeded timeout of {1}s.".format(resource, timeout)
             log.exception(message)
             raise TimeoutError(message)
 

analyticsclient/course.py

 import urllib
-import analyticsclient.activity_type as at
+import analyticsclient.activity_type as AT
+import analyticsclient.data_format as DF
 
 
 class Course(object):
@@ -18,7 +19,7 @@ class Course(object):
         self.client = client
         self.course_id = unicode(course_id)
 
-    def enrollment(self, demographic=None, start_date=None, end_date=None):
+    def enrollment(self, demographic=None, start_date=None, end_date=None, data_format=DF.JSON):
         """
         Get course enrollment data.
 
@@ -33,6 +34,7 @@ class Course(object):
             demographic (str): Demographic by which enrollment data should be grouped.
             start_date (str): Minimum date for returned enrollment data
             end_date (str): Maxmimum date for returned enrollment data
+            data_format (str): Format in which data should be returned
         """
         path = 'courses/{0}/enrollment/'.format(self.course_id)
         if demographic:
@@ -49,13 +51,15 @@ class Course(object):
         if querystring:
             path += '?{0}'.format(querystring)
 
-        return self.client.get(path)
+        return self.client.get(path, data_format=data_format)
 
-    def recent_activity(self, activity_type=at.ANY):
+    def recent_activity(self, activity_type=AT.ANY, data_format=DF.JSON):
         """
         Get the recent course activity.
 
         Arguments:
             activity_type (str): The type of recent activity to return. Defaults to ANY.
+            data_format (str): Format in which data should be returned
         """
-        return self.client.get('courses/{0}/recent_activity/?activity_type={1}'.format(self.course_id, activity_type))
+        path = 'courses/{0}/recent_activity/?activity_type={1}'.format(self.course_id, activity_type)
+        return self.client.get(path, data_format=data_format)

analyticsclient/data_format.py

+CSV = 'csv'
+JSON = 'json'

analyticsclient/tests/test_client.py

@@ -3,7 +3,9 @@ import json
 import httpretty
 import mock
 import requests.exceptions
+from testfixtures import log_capture
 
+from analyticsclient import data_format
 from analyticsclient.client import Client
 from analyticsclient.exceptions import ClientError, TimeoutError
 from analyticsclient.tests import ClientTestCase
@@ -18,6 +20,7 @@ class ClientTests(ClientTestCase):
 
     def tearDown(self):
         httpretty.disable()
+        httpretty.reset()
 
     def test_has_resource(self):
         httpretty.register_uri(httpretty.GET, self.test_url, body='')
@@ -58,14 +61,38 @@ class ClientTests(ClientTestCase):
 
     # pylint: disable=protected-access
     @mock.patch('requests.get', side_effect=requests.exceptions.Timeout)
-    def test_request_timeout(self, mock_get):
+    @log_capture()
+    def test_request_timeout(self, mock_get, lc):
         url = self.test_url
         timeout = None
-        self.assertRaises(TimeoutError, self.client._request, self.test_endpoint, timeout=timeout)
         headers = {'Accept': 'application/json'}
+
+        self.assertRaises(TimeoutError, self.client._request, self.test_endpoint, timeout=timeout)
+        msg = 'Response from {0} exceeded timeout of {1}s.'.format(self.test_endpoint, self.client.timeout)
+        lc.check(('analyticsclient.client', 'ERROR', msg))
+        lc.clear()
         mock_get.assert_called_once_with(url, headers=headers, timeout=self.client.timeout)
         mock_get.reset_mock()
 
         timeout = 10
         self.assertRaises(TimeoutError, self.client._request, self.test_endpoint, timeout=timeout)
         mock_get.assert_called_once_with(url, headers=headers, timeout=timeout)
+        msg = 'Response from {0} exceeded timeout of {1}s.'.format(self.test_endpoint, timeout)
+        lc.check(('analyticsclient.client', 'ERROR', msg))
+
+    def test_request_format(self):
+        httpretty.register_uri(httpretty.GET, self.test_url, body='{}')
+
+        response = self.client.get(self.test_endpoint)
+        self.assertEquals(httpretty.last_request().headers['Accept'], 'application/json')
+        self.assertDictEqual(response, {})
+
+        httpretty.register_uri(httpretty.GET, self.test_url, body='not-json')
+        response = self.client.get(self.test_endpoint, data_format=data_format.CSV)
+        self.assertEquals(httpretty.last_request().headers['Accept'], 'text/csv')
+        self.assertEqual(response, 'not-json')
+
+        httpretty.register_uri(httpretty.GET, self.test_url, body='{}')
+        response = self.client.get(self.test_endpoint, data_format=data_format.JSON)
+        self.assertEquals(httpretty.last_request().headers['Accept'], 'application/json')
+        self.assertDictEqual(response, {})

analyticsclient/tests/test_course.py

@@ -3,6 +3,7 @@ import json
 import httpretty
 import re
 from analyticsclient import activity_type as at
+from analyticsclient import data_format
 from analyticsclient import demographic as demo
 from analyticsclient.exceptions import NotFoundError, InvalidRequestError
 
@@ -91,3 +92,15 @@ class CoursesTests(ClientTestCase):
         self.assertCorrectEnrollmentUrl(self.course, demo.EDUCATION)
         self.assertCorrectEnrollmentUrl(self.course, demo.GENDER)
         self.assertCorrectEnrollmentUrl(self.course, demo.LOCATION)
+
+    def test_enrollment_data_format(self):
+        uri = self.get_api_url('courses/{0}/enrollment/'.format(self.course.course_id))
+
+        httpretty.register_uri(httpretty.GET, uri, body='{}')
+
+        self.course.enrollment()
+        self.assertEquals(httpretty.last_request().headers['Accept'], 'application/json')
+
+        httpretty.register_uri(httpretty.GET, uri, body='not-json')
+        self.course.enrollment(data_format=data_format.CSV)
+        self.assertEquals(httpretty.last_request().headers['Accept'], 'text/csv')

requirements.txt

@@ -8,3 +8,4 @@ pep8==1.5.7
 pylint==1.2.1
 pep257==0.3.2
 mock==1.0.1
+testfixtures==4.0.0