Filter placeholder records from RescueGroups (e.g. "More Dogs Soon!")

Some rescues publish placeholder entries pointing users at their
website rather than a real adoptable animal. These were getting picked
up and posted to social. Skip records whose name matches an entry in
adoption_sources/rescue_groups_blocklist.json (case-insensitive).

Closes #94

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: MilesMorel

adoption_sources/rescue_groups.py

@@ -18,6 +18,15 @@
 
 logger = logging.getLogger(__name__)
 
+_blocklist_path = __file__.replace(".py", "_blocklist.json")
+with open(_blocklist_path) as _f:
+    _blocklist = json.loads(_f.read())
+
+# Substrings (case-insensitive) that mark a record as a placeholder rather than
+# an actual adoptable pet. Some rescues publish entries like "More Dogs Soon!"
+# to point users at their website; those should never be posted.
+PLACEHOLDER_NAME_SUBSTRINGS: tuple[str, ...] = tuple(s.lower() for s in _blocklist["name_substrings"])
+
 
 class SourceRescueGroups(PetSource):
     """
@@ -104,8 +113,12 @@ def fetch_pets(self) -> Iterator[AdoptablePet]:
 
         for animal in data:
             pet = self._parse_animal(animal, orgs_by_id)
-            if pet:
-                yield pet
+            if not pet:
+                continue
+            if self._is_placeholder_name(pet.name):
+                logger.info(f"Skipping placeholder record: {pet.name!r}")
+                continue
+            yield pet
 
     def _parse_animal(self, animal: dict, orgs_by_id: dict) -> AdoptablePet | None:
         """Parse a single animal record from the API response."""
@@ -163,6 +176,10 @@ def _parse_animal(self, animal: dict, orgs_by_id: dict) -> AdoptablePet | None:
             logger.warning(f"Failed to parse animal {animal.get('id', 'unknown')}: {e}")
             return None
 
+    def _is_placeholder_name(self, name: str) -> bool:
+        lowered = (name or "").lower()
+        return any(needle in lowered for needle in PLACEHOLDER_NAME_SUBSTRINGS)
+
     def _clean_name(self, name: str) -> str:
         """
         Clean up pet name by removing promotional text.

adoption_sources/rescue_groups_blocklist.json

@@ -0,0 +1,5 @@
+{
+  "name_substrings": [
+    "more dogs soon"
+  ]
+}

tests/test_rescue_groups.py

@@ -58,5 +58,18 @@ def test_falls_back_to_org_url_when_neither_pet_nor_org_has_adoption_url(self):
         self.assertEqual(pet.adoption_url, "https://org.example.com")
 
 
+class PlaceholderNameTests(unittest.TestCase):
+    def setUp(self):
+        self.source = SourceRescueGroups(api_key="dummy")
+
+    def test_more_dogs_soon_is_placeholder(self):
+        self.assertTrue(self.source._is_placeholder_name("More Dogs Soon!"))
+        self.assertTrue(self.source._is_placeholder_name("more dogs soon"))
+
+    def test_real_pet_name_is_not_placeholder(self):
+        self.assertFalse(self.source._is_placeholder_name("Pippin"))
+        self.assertFalse(self.source._is_placeholder_name("Buddy"))
+
+
 if __name__ == "__main__":
     unittest.main()