Merge branch 'standardize-dir'

penn/directory.py

@@ -1,5 +1,6 @@
 from os import path
 from .base import WrapperBase
+from nameparser import HumanName
 
 
 BASE_URL = "https://esb.isc-seo.upenn.edu/8091/open_data/"
@@ -20,17 +21,44 @@ class Directory(WrapperBase):
       >>> from penn import Directory
       >>> d = Directory('MY_USERNAME_TOKEN', 'MY_PASSWORD_TOKEN')
     """
-
-    def search(self, params):
+    @staticmethod
+    def standardize(res):
+        # Standardize name
+        name = HumanName(res['list_name'])
+        name.capitalize()
+        res['list_name'] = str(name)
+        if 'detail_name' in res:
+            dname = HumanName(res['detail_name'])
+            dname.capitalize()
+            res['detail_name'] = str(dname)
+        # Lowercase email
+        if 'list_email' in res:
+            res['list_email'] = res['list_email'].lower()
+        # Remove `Faculty - ` from affiliation
+        res['list_affiliation'] = res['list_affiliation'].replace('Faculty - ', '')
+        return res
+
+
+    def search(self, params, standardize=False):
         """Get a list of person objects for the given search params.
 
         :param params: Dictionary specifying the query parameters
+        :param standardize: Whether to standardize names and other features,
+            currently disabled for backwards compatibility. Currently
+            standardizes names, lowercases emails, and removes faculty label
+            from affiliation.
 
         >>> people = d.search({'first_name': 'tobias', 'last_name': 'funke'})
         """
-        return self._request(ENDPOINTS['SEARCH'], params)
-
-    def detail_search(self, params):
+        resp = self._request(ENDPOINTS['SEARCH'], params)
+        if not standardize:
+            return resp
+        # Standardization logic
+        for res in resp['result_data']:
+            res = self.standardize(res)
+        return resp
+
+    def detail_search(self, params, standardize=False):
         """Get a detailed list of person objects for the given search params.
 
         :param params:
@@ -43,7 +71,8 @@ def detail_search(self, params):
         result_data = []
         for person in response['result_data']:
             try:
-                detail = self.person_details(person['person_id'])
+                detail = self.person_details(person['person_id'],
+                                             standardize=standardize)
             except ValueError:
                 pass
             else:
@@ -52,12 +81,15 @@ def detail_search(self, params):
         response['result_data'] = result_data
         return response
 
-    def person_details(self, person_id):
+    def person_details(self, person_id, standardize=False):
         """Get a detailed person object
 
         :param person_id:
             String corresponding to the person's id.
 
         >>> instructor = d.person('jhs878sfd03b38b0d463b16320b5e438')
         """
-        return self._request(path.join(ENDPOINTS['DETAILS'], person_id))
+        resp = self._request(path.join(ENDPOINTS['DETAILS'], person_id))
+        if standardize:
+            resp['result_data'] = [self.standardize(res) for res in resp['result_data']]
+        return resp

requirements.txt

@@ -2,3 +2,4 @@ nose==1.3.4
 requests==2.4.3
 beautifulsoup4==4.4.1
 html5lib==0.999
+nameparser==0.4.0

tests/directory_test.py

@@ -15,3 +15,42 @@ def test_lastname_search(self):
         person = self.dir.detail_search({'last_name':'Domingoes'})['result_data']
         self.assertEquals(len(person), 1)
         self.assertEquals(person[0]['result_data'][0]['detail_name'], "ADAM W DOMINGOES")
+
+    def test_lastname_search_standardized(self):
+        person = self.dir.detail_search({'last_name':'Domingoes'}, standardize=True)['result_data']
+        self.assertEquals(len(person), 1)
+        self.assertEquals(person[0]['result_data'][0]['detail_name'], "Adam W Domingoes")
+
+    def test_person_id(self):
+        # Alex Wissmann's person id
+        details = self.dir.person_details('041cd6e739387e24db2483785b87b082')['result_data']
+        self.assertEquals(details[0]['detail_name'], "ADAM W DOMINGOES")
+
+    def test_person_id_standardized(self):
+        # Alex Wissmann's person id
+        details = self.dir.person_details('041cd6e739387e24db2483785b87b082', True)['result_data']
+        self.assertEquals(details[0]['detail_name'], "Adam W Domingoes")
+
+    def test_faculty_name_not_standardized(self):
+        fac = self.dir.search({'first_name': 'kostas'})
+        self.assertEquals(fac['result_data'][0]['list_name'], "DANIILIDIS, KONSTANTINOS ")
+
+    def test_faculty_name_standardized(self):
+        fac = self.dir.search({'first_name': 'kostas'}, standardize=True)
+        self.assertEquals(fac['result_data'][0]['list_name'], "Konstantinos Daniilidis")
+
+    def test_email_not_standardized(self):
+        email = self.dir.search({'first_name': 'amy', 'last_name': 'gallagher'})
+        self.assertEqual(email['result_data'][0]['list_email'], '[email protected]')
+
+    def test_email_standardized(self):
+        email = self.dir.search({'first_name': 'amy', 'last_name': 'gallagher'}, standardize=True)
+        self.assertEqual(email['result_data'][0]['list_email'], '[email protected]')
+
+    def test_afl_not_standardized(self):
+        afl = self.dir.search({'first_name': 'kostas'})
+        self.assertEqual(afl['result_data'][0]['list_affiliation'], 'Faculty - ASSOC PROFESSOR')
+
+    def test_afl_standardized(self):
+        afl = self.dir.search({'first_name': 'kostas'}, standardize=True)
+        self.assertEqual(afl['result_data'][0]['list_affiliation'], 'ASSOC PROFESSOR')