Merge ca20dc6 into c6887ca

README.rst

@@ -100,6 +100,7 @@ Scrapers available for:
 - `https://aflavorjournal.com/ <https://aflavorjournal.com/>`_
 - `https://ah.nl/ <https://ah.nl/>`_
 - `https://akispetretzikis.com/ <https://akispetretzikis.com/>`_
+- `https://aldi.com.au/ <https://aldi.com.au/>`_
 - `https://alexandracooks.com/ <https://alexandracooks.com/>`_
 - `https://alittlebityummy.com/ <https://alittlebityummy.com/>`_
 - `https://allrecipes.com/ <https://allrecipes.com/>`_

recipe_scrapers/__init__.py

@@ -17,6 +17,7 @@
 from .aflavorjournal import AFlavorJournal
 from .akispetretzikis import AkisPetretzikis
 from .albertheijn import AlbertHeijn
+from .aldi import Aldi
 from .alexandracooks import AlexandraCooks
 from .alittlebityummy import ALittleBitYummy
 from .allrecipes import AllRecipes
@@ -364,6 +365,7 @@
     AfghanKitchenRecipes.host(): AfghanKitchenRecipes,
     AkisPetretzikis.host(): AkisPetretzikis,
     AlbertHeijn.host(): AlbertHeijn,
+    Aldi.host(): Aldi,
     AlexandraCooks.host(): AlexandraCooks,
     AllRecipes.host(): AllRecipes,
     AllTheHealthyThings.host(): AllTheHealthyThings,

recipe_scrapers/aldi.py

@@ -0,0 +1,68 @@
+# mypy: allow-untyped-defs
+import re
+
+from ._abstract import AbstractScraper
+from ._utils import get_minutes, get_yields
+
+
+class Aldi(AbstractScraper):
+    @classmethod
+    def host(cls):
+        return "aldi.com.au"
+
+    def site_name(self):
+        return "Aldi"
+
+    def author(self):
+        return self.soup.find("meta", {"name": "author"}).get("content")
+
+    def title(self):
+        return self.soup.find("h1").text
+
+    def category(self):
+        element = self.soup.select_one("a.tab-nav--link.dropdown--list--link.m-active")
+        if element:
+            title = element.text
+            recipe_position = title.find(" Recipe")
+            if recipe_position != -1:
+                return title[:recipe_position]
+
+    def prep_time(self):
+        return get_minutes(self._get_value(re.compile("prep", re.IGNORECASE)))
+
+    def cook_time(self):
+        return get_minutes(self._get_value(re.compile("cook", re.IGNORECASE)))
+
+    def total_time(self):
+        return (self.prep_time() or 0) + (self.cook_time() or 0)
+
+    def yields(self):
+        value = self._get_value(re.compile("(makes)|(serves)", re.IGNORECASE))
+        return get_yields(str(value))
+
+    def image(self):
+        figure = self.soup.find("figure", {"class": "csc-textpic-image csc-textpic-last"})
+        if figure:
+            image = figure.find("img")
+            if image:
+                return image.get("src")
+        return None
+
+    def ingredients(self):
+        h2 = self.soup.find("h2", string=re.compile("Ingredients"))
+        list_element = h2.find_next_sibling("ul")
+        ingredients = []
+        for li in list_element.find_all("li"):
+            ingredients.append(li.text.strip())
+        return ingredients
+
+    def instructions(self):
+        list_element = self.soup.find("ol")
+        return "\n".join(li.text.strip() for li in list_element.find_all("li"))
+
+    def _get_value(self, label):
+        label_element = self.soup.find("b", string=label)
+        if label_element:
+            parts = [sibling.strip() for sibling in label_element.find_next_siblings(string=True) if sibling.strip()]
+            return " ".join(parts)
+        return None

tests/test_data/aldi.com.au/aldi.json

@@ -0,0 +1,40 @@
+{
+  "author": "ALDI AU",
+  "canonical_url": "https://www.aldi.com.au/recipes/breakfast-recipes/overloaded-cheese-toastie-recipe/",
+  "host": "aldi.com.au",
+  "image": "https://www.aldi.com.au/fileadmin/_processed_/8/b/csm_1000340_WK23_PD_455x315__OVERLOADED-CHEESE-TOASTIE__1_RET_4506d46579.jpg",
+  "category": "Breakfast",
+  "ingredients": [
+    "1 tbsp Remano pesto",
+    "1 slice Westacre tasty cheese",
+    "3 slices tomato",
+    "2 slices sourdough bread",
+    "Handful of The Fresh Salad Co baby spinach"
+  ],
+  "ingredient_groups": [
+    {
+      "purpose": null,
+      "ingredients": [
+        "1 tbsp Remano pesto",
+        "1 slice Westacre tasty cheese",
+        "3 slices tomato",
+        "2 slices sourdough bread",
+        "Handful of The Fresh Salad Co baby spinach"
+      ]
+    }
+  ],
+  "instructions": "Preheat a sandwich press.\nSpread pesto on one slice of bread and layer with cheese, tomato and spinach and close the sandwich with the remaining slice of bread.\nPlace the sandwich in the press and cook for 3-4 minutes or until the cheese melts and the sandwich is golden brown.\nAllow to cool slightly before serving.",
+  "instructions_list": [
+    "Preheat a sandwich press.",
+    "Spread pesto on one slice of bread and layer with cheese, tomato and spinach and close the sandwich with the remaining slice of bread.",
+    "Place the sandwich in the press and cook for 3-4 minutes or until the cheese melts and the sandwich is golden brown.",
+    "Allow to cool slightly before serving."
+  ],
+  "language": "en-AU",
+  "site_name": "Aldi",
+  "title": "Overloaded Cheese Toastie Recipe",
+  "prep_time": 5,
+  "cook_time": 5,
+  "total_time": 10,
+  "yields": "1 serving"
+}