feat: improve KitchenOwl import parsing and tagging

home/doris-kitchen/tests/test_tagging.py

@@ -0,0 +1,121 @@
+import sys
+import unittest
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parents[1]
+sys.path.insert(0, str(ROOT))
+
+from app.services.tagging import curate_recipe_tags
+from app.services.recipe_importer import normalize_recipe_from_url
+
+
+class TaggingTests(unittest.TestCase):
+    def test_tater_tots_drop_boilerplate_and_gain_meaningful_tags(self):
+        payload = normalize_recipe_from_url(
+            'https://www.joshuaweissman.com/recipes/the-best-tater-tots-ever-recipe'
+        )
+
+        self.assertNotIn('the-best-tater-tots-ever', payload['tags'])
+        self.assertNotIn('chef-joshua-weissman-recipe', payload['tags'])
+        self.assertIn('potato', payload['tags'])
+        self.assertIn('tater-tots', payload['tags'])
+        self.assertIn('side-dish', payload['tags'])
+        self.assertNotIn('beef', payload['tags'])
+
+    def test_halal_chicken_rice_drops_boilerplate_and_infers_cuisine(self):
+        payload = normalize_recipe_from_url(
+            'https://www.joshuaweissman.com/recipes/halal-guys-chicken-over-rice-but-better-recipe'
+        )
+
+        self.assertNotIn('halal-guys-chicken-over-rice-but-better', payload['tags'])
+        self.assertNotIn('chef-joshua-weissman-recipe', payload['tags'])
+        self.assertIn('middle-eastern', payload['tags'])
+        self.assertIn('chicken', payload['tags'])
+        self.assertIn('rice', payload['tags'])
+        self.assertNotIn('skillet', payload['tags'])
+
+    def test_existing_machine_tags_can_be_replaced_when_recipe_content_disagrees(self):
+        tags = curate_recipe_tags(
+            name='The Best Tater Tots Ever',
+            description='Homemade tater tots worth making. Fry until crisp.',
+            items=[
+                {'name': 'Russet potatoes', 'description': 'peeled', 'optional': False},
+                {'name': 'beef tallow', 'description': 'for frying', 'optional': False},
+            ],
+            source='https://www.joshuaweissman.com/recipes/the-best-tater-tots-ever-recipe',
+            existing_tags=['beef', 'main-course', 'savory'],
+        )
+
+        self.assertNotIn('beef', tags)
+        self.assertNotIn('main-course', tags)
+        self.assertIn('side-dish', tags)
+        self.assertIn('savory', tags)
+
+    def test_existing_human_curated_tags_are_preserved(self):
+        tags = curate_recipe_tags(
+            name='Simple Roast Chicken',
+            description='A chicken dinner with herbs.',
+            items=[{'name': 'chicken', 'description': '', 'optional': False}],
+            source='https://example.com/roast-chicken',
+            existing_tags=['family-favorite', 'weeknight'],
+        )
+
+        self.assertIn('family-favorite', tags)
+        self.assertIn('weeknight', tags)
+
+    def test_mango_sticky_rice_is_treated_as_dessert_not_savory_main(self):
+        tags = curate_recipe_tags(
+            name='Perfect Mango Sticky Rice At Home',
+            description='Mango sticky rice is something everyone can, and should, make.',
+            items=[
+                {'name': 'glutinous rice', 'description': '1.5 cups', 'optional': False},
+                {'name': 'full-fat coconut milk', 'description': '1 can', 'optional': False},
+                {'name': 'mangos', 'description': '2', 'optional': False},
+                {'name': 'granulated sugar', 'description': '1/2 cup', 'optional': False},
+            ],
+            source='https://www.joshuaweissman.com/recipes/perfect-mango-sticky-rice-recipe',
+            existing_tags=['chef-joshua-weissman-recipe', 'main-course', 'savory'],
+        )
+
+        self.assertIn('dessert', tags)
+        self.assertIn('sweet', tags)
+        self.assertIn('thai', tags)
+        self.assertNotIn('main-course', tags)
+        self.assertNotIn('savory', tags)
+
+    def test_description_only_chicken_reference_does_not_force_chicken_tag(self):
+        tags = curate_recipe_tags(
+            name='Quick Asian Beef Ramen Noodles',
+            description='Try the chicken version next time if you want a variation.',
+            items=[
+                {'name': 'ground beef', 'description': '200g', 'optional': False},
+                {'name': 'ramen noodles', 'description': '2 packs', 'optional': False},
+                {'name': 'soy sauce', 'description': '1 tbsp', 'optional': False},
+            ],
+            source='https://www.recipetineats.com/asian-beef-and-noodles/',
+            existing_tags=['asian', 'noodles'],
+        )
+
+        self.assertIn('beef', tags)
+        self.assertNotIn('chicken', tags)
+
+    def test_pasta_chips_are_snack_not_main_course(self):
+        tags = curate_recipe_tags(
+            name='Air Fryer Pasta Chips',
+            description='This could be one of the greatest homemade snacks of all time.',
+            items=[
+                {'name': 'farfalle pasta', 'description': '8 oz', 'optional': False},
+                {'name': 'pecorino romano', 'description': 'to taste', 'optional': False},
+                {'name': 'olive oil', 'description': 'drizzle', 'optional': False},
+            ],
+            source='https://www.joshuaweissman.com/recipes/air-fryer-pasta-chips-recipe',
+            existing_tags=['italian', 'main-course', 'pasta', 'savory'],
+        )
+
+        self.assertIn('snack', tags)
+        self.assertIn('pasta', tags)
+        self.assertNotIn('main-course', tags)
+
+
+if __name__ == '__main__':
+    unittest.main()