rebuild source discovery workflows

2026-09-04 20:52:37 +02:00
parent 3a9b553440
commit 6aefbdc1f3
18 changed files with 853 additions and 106 deletions
@@ -0,0 +1,59 @@
+import os
+import unittest
+from unittest.mock import patch
+
+from app.sources import GoogleBrowserSearchBlocked, GoogleBrowserSearchSource
+
+
+class GoogleBrowserSearchSourceTests(unittest.TestCase):
+    def setUp(self):
+        self.old_enabled = os.environ.pop("GOOGLE_BROWSER_SEARCH_ENABLED", None)
+
+    def tearDown(self):
+        GoogleBrowserSearchSource._last_request_at = None
+        if self.old_enabled is None:
+            os.environ.pop("GOOGLE_BROWSER_SEARCH_ENABLED", None)
+        else:
+            os.environ["GOOGLE_BROWSER_SEARCH_ENABLED"] = self.old_enabled
+
+    def test_disabled_feature_blocks_without_network_io(self):
+        source = GoogleBrowserSearchSource()
+        with patch("app.sources.urlopen") as network:
+            with self.assertRaises(GoogleBrowserSearchBlocked) as raised:
+                source.discover(
+                    {"approved": True, "public_access": True, "terms_accepted": True, "rate_limit": 6},
+                    criteria={"keywords": ["solar installers"], "city": "Cape Town"},
+                    limits={"max_records": 5},
+                )
+        self.assertEqual(raised.exception.code, "GOOGLE_BROWSER_BLOCKED")
+        self.assertEqual(raised.exception.reason, "feature_disabled")
+        network.assert_not_called()
+
+    def test_enabled_fetch_builds_query_from_criteria_and_parses_visible_result_links(self):
+        os.environ["GOOGLE_BROWSER_SEARCH_ENABLED"] = "true"
+
+        class Response:
+            def read(self, _limit):
+                return b'<html><body><a href="https://acme.example/about"><h3>Acme Solar</h3></a><a href="https://www.google.com/preferences"><h3>Settings</h3></a></body></html>'
+
+            def __enter__(self):
+                return self
+
+            def __exit__(self, *_):
+                return False
+
+        with patch("app.sources.urlopen", return_value=Response()) as network:
+            page = GoogleBrowserSearchSource().discover(
+                {"approved": True, "public_access": True, "terms_accepted": True, "rate_limit": 6},
+                criteria={"keywords": ["solar"], "city": "Cape Town"},
+                limits={"max_records": 4},
+            )
+        self.assertEqual(page.records, [{"name": "Acme Solar", "website": "https://acme.example/about", "email": "", "phone": "", "description": "Public Google search result", "location": ""}])
+        request = network.call_args.args[0]
+        self.assertIn("q=solar+Cape+Town", request.full_url)
+        self.assertNotIn("query", request.full_url)
+        self.assertTrue(page.metadata["public_html_only"])
+
+
+if __name__ == "__main__":
+    unittest.main()