munch-ease-backend/scripts/recipe_urls.py
2025-11-04 18:31:13 +11:00

37 lines
1.7 KiB
Python

"""
Central list of recipe URLs used by manual harnesses and snapshot tools.
Keep this list focused on stable, public pages across a variety of sites.
"""
URLS = [
# ld+json: YES (Recipe present under @graph)
"https://cookieandkate.com/favorite-fried-eggs-recipe/",
# ld+json: UNKNOWN (blocked/SSL issues in our environment)
"https://www.taste.com.au/recipes/classic-chewy-brownie/f0960a83-1fa9-4e3e-b616-a995182613e0",
# ld+json: PRESENT but MALFORMED (first ld+json is Recipe but not valid JSON)
"https://www.inspiredtaste.net/24412/cocoa-brownies-recipe/",
# ld+json: YES (Recipe present)
"https://www.simplyrecipes.com/recipes/french_toast/",
# Additional diverse sources
# ld+json: YES (Recipe present)
"https://pinchofyum.com/lemon-chicken-piccata-with-grilled-bread",
# ld+json: YES (Recipe present)
"https://www.bbcgoodfood.com/recipes/chicken-tikka-masala",
# ld+json: YES (Recipe present)
"https://www.recipetineats.com/beef-stroganoff/",
# ld+json: YES (Recipe present)
"https://sallysbakingaddiction.com/triple-chocolate-layer-cake/",
# ld+json: YES (Recipe present)
"https://downshiftology.com/recipes/shakshuka/",
# ld+json: UNKNOWN (403 blocked by CDN)
"https://damndelicious.net/2025/08/01/corn-salsa/",
# ld+json: UNKNOWN (403 blocked by CDN)
"https://www.thekitchn.com/chicken-tamale-pie-23752740",
# ld+json: YES (Recipe present)
"https://www.simplyrecipes.com/recipes/perfect_guacamole/",
# ld+json: YES (Recipe present)
"https://www.kingarthurbaking.com/recipes/banana-bread-recipe",
# ld+json: YES (Recipe present; multiple ld+json, only the first has Recipe)
"https://tasty.co/recipe/one-pot-garlic-parmesan-pasta",
]