build prospect intelligence platform MVP

2026-09-02 17:38:50 +02:00
commit 44be4efc82
29 changed files with 973 additions and 0 deletions
@@ -0,0 +1,46 @@
+import unittest
+
+from app.domain import (
+    classify_website,
+    deduplicate_businesses,
+    normalize_business,
+    normalize_domain,
+    score_business,
+    is_suppressed,
+)
+
+
+class DomainTests(unittest.TestCase):
+    def test_normalization_canonicalizes_contact_and_domain(self):
+        value = normalize_business({"name": "  Acme  ", "website": "HTTPS://WWW.Acme.co.za/path", "email": " SALES@ACME.CO.ZA "})
+        self.assertEqual(value["name"], "Acme")
+        self.assertEqual(value["website_domain"], "acme.co.za")
+        self.assertEqual(value["email"], "sales@acme.co.za")
+
+    def test_website_classification_is_conservative(self):
+        self.assertEqual(classify_website("https://acme.co.za"), "business_site")
+        self.assertEqual(classify_website("https://www.facebook.com/acme"), "social_profile")
+        self.assertEqual(classify_website(""), "missing")
+
+    def test_scoring_is_versioned_and_explains_factors(self):
+        result = score_business({"name": "Acme", "website_domain": "acme.co.za", "email": "a@acme.co.za", "phone": "123"})
+        self.assertEqual(result["score_version"], "mvp-1")
+        self.assertGreaterEqual(result["score"], 70)
+        self.assertIn("business_site", result["factors"])
+
+    def test_suppression_matches_email_domain_or_phone(self):
+        business = {"email": "person@example.com", "website_domain": "example.com", "phone": "+27123456789"}
+        self.assertTrue(is_suppressed(business, [{"kind": "domain", "value": "example.com"}]))
+        self.assertTrue(is_suppressed(business, [{"kind": "email", "value": "person@example.com"}]))
+        self.assertTrue(is_suppressed(business, [{"kind": "phone", "value": "+27123456789"}]))
+        self.assertFalse(is_suppressed(business, [{"kind": "email", "value": "other@example.com"}]))
+
+    def test_deduplication_prefers_richer_record(self):
+        rows = [{"name": "Acme", "website": "https://acme.com"}, {"name": " acme ", "website": "https://www.acme.com", "email": "a@acme.com"}]
+        result = deduplicate_businesses(rows)
+        self.assertEqual(len(result), 1)
+        self.assertEqual(result[0]["email"], "a@acme.com")
+
+
+if __name__ == "__main__":
+    unittest.main()