usapears: implement redundant-nutrient-value filtering

recipe_scrapers/usapears.py

@@ -1,4 +1,7 @@
+import re
+
 from ._abstract import AbstractScraper
+from ._exceptions import ElementNotFoundInHtml
 from ._utils import get_minutes, normalize_string
 
 
@@ -25,6 +28,31 @@ def ingredients(self):
             for paragraph in ingredient_elements
         ]
 
+    def nutrients(self):
+        container = self.soup.find("ul", {"itemprop": "nutrition"})
+        if not container:
+            raise ElementNotFoundInHtml("Could not find nutritional info container")
+
+        results = {}
+        redundant_pattern = r"<strong>(.+)[:] </strong>"
+        for item in container.find_all("li", {"itemprop": True}):
+            nutrient = item["itemprop"]
+            content = "".join(str(elem) for elem in item.children)
+            if re.match(redundant_pattern, content):
+                content = re.sub(redundant_pattern, "", content)
+            results[nutrient] = content
+
+        corrections = {
+            "carbohydrates": "carbohydrateContent",
+            "protein": "proteinContent",
+            "fat": "fatContent",
+        }
+        for mistake, correction in corrections.items():
+            if mistake in results:
+                results[correction] = results.pop(mistake)
+
+        return results
+
     def ratings(self):
         try:
             ratings = self.schema.ratings()

tests/test_data/usapears.org/usapears.json

@@ -23,14 +23,11 @@
   "yields": "6 servings",
   "description": "This simple, tasty pear recipe was created by Chef Jamie Lauren of Absinthe Brasserie and Bar in San Francisco and cookbook author Mollie Katzen. If you can’t find Bosc pears at your local grocery store, red or green Anjou pears also work well.",
   "total_time": 25,
-  "ratings": null,
-  "ratings_count": 4,
   "nutrients": {
-    "servingSize": "Serving Size: 1 Pear",
-    "calories": "Calories: 230",
-    "carbohydrateContent": "Carbohydrate: 34g",
-    "proteinContent": "Protein: 5g",
-    "fiberContent": "Dietary Fiber: 7g"
+    "servingSize": "1 Pear",
+    "calories": "230",
+    "carbohydrateContent": "34g",
+    "proteinContent": "5g"
   },
   "image": "https://usapears.org/wp-content/uploads/2014/10/Sauteed-Bosc-Pears1.jpg"
 }