Add validation checks in Jurisdiction constructor

clarify/jurisdiction.py

@@ -7,6 +7,9 @@
 import lxml.html
 from lxml.cssselect import CSSSelector
 
+CLARITY_RESULTS_HOSTNAME = "results.enr.clarityelections.com"
+SUPPORTED_LEVELS = ['state', 'county', 'city', 'precinct']
+
 
 class Jurisdiction(object):
 
@@ -20,12 +23,40 @@ def __init__(self, url, level, name=''):
         """
         To create an instance, pass a Clarity results URL for the top-level
         political jurisdiction (a state, for example), and the corresponding
-        level in lowercase ("state" or "county").
+        level in lowercase ("state", "county", "city", or "precinct").
         """
 
+        # Preliminary check for any type which is not string-like
+        if url == None:
+            raise TypeError('Invalid url parameter')
+        try:
+            # Attempt to convert to str
+            url = str(url)
+        except TypeError as type_error:
+            raise type_error
+        except ValueError as value_error:
+            raise value_error
+        except Exception as error:
+            # If converting to str failed, raise type error
+            raise TypeError('Invalid url parameter')
+        if type(url) != str:
+            raise TypeError('Invalid url parameter')
+        # if url is an HTTP URL to Clarity Election Results
+        if len(url) >= 40 and url[0:40] == 'http://' + CLARITY_RESULTS_HOSTNAME + '/':
+            # Replace HTTP with HTTPS; retain the string after the URL origin
+            url = 'https://' + CLARITY_RESULTS_HOSTNAME + '/' + url[40]
+        # if url is not in the allowed list of supported origins
+        if len(url) < 41 or url[0:41] != 'https://' + CLARITY_RESULTS_HOSTNAME + '/':
+            raise ValueError('Unsupported url origin')
         self.url = url
         self.parsed_url = self._parse_url()
         self.state = self._get_state_from_url()
+
+        if type(level) != str:
+            raise TypeError('Invalid level parameter')
+        level = level.lower()
+        if level not in SUPPORTED_LEVELS:
+            raise ValueError('Unsupported level')
         self.level = level
         self.name = name
         self.summary_url = self._get_summary_url()

tests/test_jurisdiction.py

@@ -145,6 +145,19 @@
 COUNTY_URL_RE = re.compile(r'https://results.enr.clarityelections.com/(?P<state>[A-Z]{2})/(?P<county>[A-Za-z\.]+)/(?P<page_id>\d+)/(?P<page_id_2>\d+)/')
 
 
+# non-string object which has a type(str(Stringlike)) === str
+class Stringlike(object):
+    def __init__(self, url):
+        self.url = url
+    def __str__(self):
+        return self.url
+
+# non-string object which has a type(str(Stringlike)) !== str
+class NotStringlike(object):
+    def __init__(self, url):
+        # do something but don't implement __str__ and don't store url
+        self.noturl = 'noturl'
+
 def mock_county_response_callback(req):
     m = COUNTY_REDIRECT_URL_RE.match(req.url)
     assert m is not None
@@ -166,11 +179,87 @@ def mock_subjurisdiction_redirect_page_meta(page_id):
 
 
 class TestJurisdiction(TestCase):
+    # Test the constructor
     def test_construct(self):
+        """
+        A Jurisdiction with a valid, supported URL and level string should create a class instance and not raise an Exception.
+        """
         url = 'https://results.enr.clarityelections.com/KY/15261/30235/en/summary.html'
         jurisdiction = Jurisdiction(url=url, level='state')
         self.assertEqual(jurisdiction.url, url)
 
+    def test_construct_valid_county(self):
+        """
+        A Jurisdiction with a valid, supported URL and level string should create a class instance and not raise an Exception.
+        """
+        url = 'https://results.enr.clarityelections.com/KY/15261/30235/en/summary.html'
+        jurisdiction = Jurisdiction(url=url, level='county')
+        self.assertEqual(jurisdiction.url, url)
+
+    def test_construct_valid_city(self):
+        """
+        A Jurisdiction with a valid, supported URL and level string should create a class instance and not raise an Exception.
+        """
+        url = 'https://results.enr.clarityelections.com/KY/15261/30235/en/summary.html'
+        jurisdiction = Jurisdiction(url=url, level='city')
+        self.assertEqual(jurisdiction.url, url)
+
+    def test_construct_valid_precinct(self):
+        """
+        A Jurisdiction with a valid, supported URL and level string should create a class instance and not raise an Exception.
+        """
+        url = 'https://results.enr.clarityelections.com/KY/15261/30235/en/summary.html'
+        jurisdiction = Jurisdiction(url=url, level='precinct')
+        self.assertEqual(jurisdiction.url, url)
+
+    def test_construct_valid_url_stringlike(self):
+        """
+        A Jurisdiction with a valid, supported URL and level string should create a class instance and not raise an Exception.
+        """
+        url = Stringlike('https://results.enr.clarityelections.com/KY/15261/30235/en/summary.html')
+        jurisdiction = Jurisdiction(url=url, level='state')
+        self.assertEqual(jurisdiction.url, str(url))
+
+    def test_construct_invalid_url_notstringlike(self):
+        """
+        A Jurisdiction with a valid, supported URL and level string should raise a TypeError.
+        """
+        url = NotStringlike('https://results.enr.clarityelections.com/KY/15261/30235/en/summary.html')
+        with self.assertRaises(ValueError):
+            Jurisdiction(url=url, level='state')
+
+    def test_construct_invalid_url_type(self):
+        """
+        A Jurisdiction with an invalid URL data type should raise an Exception.
+        """
+        with self.assertRaises(TypeError):
+            Jurisdiction(url=None, level='state')
+
+    def test_construct_invalid_hostname(self):
+        """
+        A Jurisdiction with an unsupported hostname in the URL should raise an Exception.
+        """
+        url = 'https://bad.hostname/KY/15261/30235/en/summary.html'
+        with self.assertRaises(ValueError):
+            Jurisdiction(url=url, level='state')
+
+    def test_construct_no_level(self):
+        """
+        A Jurisdiction with a valid, supported hostname but no level in the URL should raise an Exception.
+        """
+        url = 'https://results.enr.clarityelections.com/KY/15261/30235/en/summary.html'
+        with self.assertRaises(TypeError):
+            Jurisdiction(url=url)
+
+    def test_construct_invalid_level_str(self):
+        """
+        A Jurisdiction with a valid, supported hostname but an invalid level in the URL should raise an Exception.
+        """
+        url = 'https://results.enr.clarityelections.com/KY/15261/30235/en/summary.html'
+        with self.assertRaises(ValueError):
+            Jurisdiction(url=url, level='foo')
+
+
     @responses.activate
     def test_get_subjurisdictions_state(self):
         # subjurisdiction url path is in script tag