Fix election date parsing bug (#566)

Refactor the election date parsing to both DRY up the code, add tests,
simplify string regex to just expect a `(.+)` pattern for date string.

Author: karpet

ksvotes/services/nvris_client.py

@@ -3,6 +3,7 @@
 import os
 import re
 from ksvotes.services.form_filler_service import FormFillerService
+from ksvotes.utils import parse_election_date as parse_election_date_util
 import logging
 
 logger = logging.getLogger(__name__)
@@ -73,15 +74,13 @@ def marshall_payload(self, flavor, **kwargs):
         return payload
 
     def parse_election_date(self, election):
-        pattern = r"(Prim\w+|General) \((.+)\)"
-        m = re.search(pattern, election)
-        if m:
-            return m.group(2)
-        else:
+        date = parse_election_date_util(election)
+        if not date:
             logger.error(
                 "%s No match for election '%s'" % (self.registrant.session_id, election)
             )
             return "(none)"
+        return date.strftime("%m/%d/%Y")
 
     def normalize_unit(self, unit):
         return re.sub(r"^(#|apt.? |apartment )", "", unit, flags=re.IGNORECASE)

ksvotes/tests/test_utils.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
-from ksvotes.utils import zip_code_matches
+from ksvotes.utils import zip_code_matches, parse_election_date
+from datetime import datetime
 from test_plus import TestCase
 from django.conf import settings
 
@@ -12,12 +13,22 @@ def update_session(self, registrant):
         self.client.cookies[settings.SESSION_COOKIE_NAME] = session.session_key
         return session
 
+    def test_zip_code_matches(self):
+        sosrec = {"Address": "123 Main St #456 Wichita, KS, 12345-9999"}
 
-def test_zip_code_matches():
-    sosrec = {"Address": "123 Main St #456 Wichita, KS, 12345-9999"}
+        self.assertTrue(zip_code_matches(sosrec, 12345))
+        self.assertTrue(zip_code_matches(sosrec, "12345"))
+        self.assertFalse(zip_code_matches(sosrec, 98765))
+        self.assertFalse(zip_code_matches(sosrec, 9999))
+        self.assertFalse(zip_code_matches(sosrec, "myzip"))
 
-    assert zip_code_matches(sosrec, 12345) == True
-    assert zip_code_matches(sosrec, "12345") == True
-    assert zip_code_matches(sosrec, 98765) == False
-    assert zip_code_matches(sosrec, 9999) == False
-    assert zip_code_matches(sosrec, "myzip") == False
+    def test_parse_election_date(self):
+        dates = {
+            "Presidential Preference Primary (March 19, 2024)": datetime(2024, 3, 19),
+            "General (November 5, 2024)": datetime(2024, 11, 5),
+            "Special Election (March 3, 2026)": datetime(2026, 3, 3),
+            "Elecciones Especiales (3 de Marzo de 2026)": datetime(2026, 3, 3),
+        }
+        for election, dt in dates.items():
+            d = parse_election_date(election)
+            self.assertEqual(dt, d, f"parsed {election}")

ksvotes/utils.py

@@ -161,13 +161,11 @@ def construct_county_choices(default):
 
 
 def parse_election_date(election):
-    pattern = (
-        r"(Primary|Primaria|General|Special Election|Elecciones Especiales) \((.+)\)"
-    )
+    pattern = r".+ \((.+)\)"
     m = re.match(pattern, str(election))
     if not m:
         return None
-    date = m.group(2)
+    date = m.group(1)
     return dateparser.parse(date)