Updates get_yields util functionality to handle additional metrics (#963

)

recipe_scrapers/_utils.py

@@ -33,6 +33,32 @@
 
 SERVE_REGEX_TO = re.compile(r"\d+(\s+to\s+|-)\d+", flags=re.I | re.X)
 
+RECIPE_YIELD_TYPES = (
+    ("dozen", "dozens"),
+    ("batch", "batches"),
+    ("cake", "cakes"),
+    ("sandwich", "sandwiches"),
+    ("bun", "buns"),
+    ("cookie", "cookies"),
+    ("muffin", "muffins"),
+    ("cupcake", "cupcakes"),
+    ("loaf", "loaves"),
+    ("pie", "pies"),
+    ("cup", "cups"),
+    ("pint", "pints"),
+    ("gallon", "gallons"),
+    ("ounce", "ounces"),
+    ("pound", "pounds"),
+    ("gram", "grams"),
+    ("liter", "liters"),
+    ("piece", "pieces"),
+    ("layer", "layers"),
+    ("scoop", "scoops"),
+    ("bar", "bars"),
+    ("patty", "patties"),
+    # ... add more types as needed, in (singular, plural) format ...
+)
+
 
 def get_minutes(element, return_zero_on_not_found=False):  # noqa: C901: TODO
     if element is None:
@@ -114,12 +140,16 @@ def get_yields(element):
     """
     Will return a string of servings or items, if the recipe is for number of items and not servings
     the method will return the string "x item(s)" where x is the quantity.
+    Returns a string of servings or items. If the recipe is for a number of items (not servings),
+    it returns "x item(s)" where x is the quantity. This function handles cases where the yield is in dozens,
+    such as "4 dozen cookies", returning "4 dozen" instead of "4 servings". Additionally
+    accommodates yields specified in batches (e.g., "2 batches of brownies"), returning the yield as stated.
     :param element: Should be BeautifulSoup.TAG, in some cases not feasible and will then be text.
     :return: The number of servings or items.
+    :return: The number of servings, items, dozen, batches, etc...
     """
     if element is None:
         raise ElementNotFoundInHtml(element)
-
     if isinstance(element, str):
         serve_text = element
     else:
@@ -129,14 +159,16 @@ def get_yields(element):
         serve_text = serve_text.split(SERVE_REGEX_TO.split(serve_text, 2)[1], 2)[1]
 
     matched = SERVE_REGEX_NUMBER.search(serve_text).groupdict().get("items") or 0
-    servings = "{} serving{}".format(matched, "" if int(matched) == 1 else "s")
+    serve_text_lower = serve_text.lower()
+
+    for singular, plural in RECIPE_YIELD_TYPES:
+        if singular in serve_text_lower:
+            return "{} {}".format(matched, singular if int(matched) == 1 else plural)
 
     if SERVE_REGEX_ITEMS.search(serve_text) is not None:
-        # This assumes if object(s), like sandwiches, it is 1 person.
-        # Issue: "Makes one 9-inch pie, (realsimple-testcase, gives "9 items")
-        servings = "{} item{}".format(matched, "" if int(matched) == 1 else "s")
+        return "{} item{}".format(matched, "" if int(matched) == 1 else "s")
 
-    return servings
+    return "{} serving{}".format(matched, "" if int(matched) == 1 else "s")
 
 
 def normalize_string(string):

tests/test_data/hersheyland.com/herseyland.json

@@ -62,5 +62,5 @@
   "site_name": null,
   "title": "HERSHEY'S \"Perfectly Chocolate\" Chocolate Cake Recipe | Hersheyland",
   "total_time": 45,
-  "yields": "1 serving"
+  "yields": "1 cake"
 }

tests/test_data/kingarthurbaking.com/kingarthur.json

@@ -70,5 +70,5 @@
   "site_name": "King Arthur Baking",
   "title": "Spiced Rye Ginger Cookies",
   "total_time": 35,
-  "yields": "22 items"
+  "yields": "22 cookies"
 }

tests/test_data/myplate.gov/usdamyplate.json

@@ -70,5 +70,5 @@
   "site_name": null,
   "title": "Fabulous Fig Bars",
   "total_time": 75,
-  "yields": "24 servings"
+  "yields": "24 bars"
 }

tests/test_data/myrecipes.com/myrecipes.json

@@ -35,5 +35,5 @@
   "site_name": "MyRecipes",
   "title": "Cacio e Pepe",
   "total_time": 20.0,
-  "yields": "2 items"
+  "yields": "2 cups"
 }

tests/test_data/ohsheglows.com/ohsheglows.json

@@ -52,5 +52,5 @@
   "site_name": "Oh She Glows",
   "title": "Obsession-Worthy Peanut Butter Cookie Ice Cream",
   "total_time": 22,
-  "yields": "8 servings"
+  "yields": "8 cups"
 }

tests/test_data/paninihappy.com/paninihappy.json

@@ -40,5 +40,5 @@
   "site_name": null,
   "title": "Grilled Mac & Cheese with BBQ Pulled Pork",
   "total_time": 30,
-  "yields": "4 items"
+  "yields": "4 sandwiches"
 }