Adding canonical_url for all scrapers & associated updates: batch two (…

…#902)

README.rst

@@ -165,7 +165,6 @@ Scrapers available for:
 - `https://forktospoon.com/ <https://forktospoon.com/>`_
 - `https://fredriksfika.allas.se/ <https://fredriksfika.allas.se/>`_
 - `https://www.750g.com <https://www.750g.com>`_
-- `https://geniuskitchen.com/ <https://geniuskitchen.com>`_
 - `https://www.gesund-aktiv.com/ <https://www.gesund-aktiv.com>`_
 - `https://giallozafferano.it/ <https://giallozafferano.it>`_
 - `https://gimmesomeoven.com/ <https://www.gimmesomeoven.com/>`_
@@ -182,7 +181,6 @@ Scrapers available for:
 - `https://www.hassanchef.com/ <https://www.hassanchef.com/>`_
 - `https://headbangerskitchen.com/ <https://www.headbangerskitchen.com/>`_
 - `https://www.heb.com/ <https://www.heb.com/recipe/landing>`_
-- `https://heinzbrasil.com.br/ <https://heinzbrasil.com.br>`_
 - `https://hellofresh.com/ <https://hellofresh.com>`_
 - `https://hellofresh.co.uk/ <https://hellofresh.co.uk>`_
 - `https://www.hellofresh.de/ <https://www.hellofresh.de/>`_

recipe_scrapers/__init__.py

@@ -87,7 +87,6 @@
 from .forktospoon import ForkToSpoon
 from .franzoesischkochen import FranzoesischKochen
 from .g750g import G750g
-from .geniuskitchen import GeniusKitchen
 from .gesundaktiv import GesundAktiv
 from .giallozafferano import GialloZafferano
 from .gimmesomeoven import GimmeSomeOven
@@ -104,7 +103,6 @@
 from .hassanchef import HassanChef
 from .headbangerskitchen import HeadbangersKitchen
 from .heb import HEB
-from .heinzbrasil import HeinzBrasil
 from .hellofresh import HelloFresh
 from .herseyland import HerseyLand
 from .homechef import HomeChef
@@ -370,7 +368,6 @@
     ForksOverKnives.host(): ForksOverKnives,
     FranzoesischKochen.host(): FranzoesischKochen,
     G750g.host(): G750g,
-    GeniusKitchen.host(): GeniusKitchen,
     GialloZafferano.host(): GialloZafferano,
     GimmeSomeOven.host(): GimmeSomeOven,
     Globo.host(): Globo,
@@ -386,7 +383,6 @@
     HandleTheHeat.host(): HandleTheHeat,
     HassanChef.host(): HassanChef,
     HeadbangersKitchen.host(): HeadbangersKitchen,
-    HeinzBrasil.host(): HeinzBrasil,
     HelloFresh.host(): HelloFresh,
     HelloFresh.host(domain="at"): HelloFresh,
     HelloFresh.host(domain="be"): HelloFresh,

recipe_scrapers/greatbritishchefs.py

@@ -14,9 +14,6 @@ def title(self):
     def total_time(self):
         return self.schema.total_time()
 
-    def yields(self):
-        return self.schema.yields()
-
     def image(self):
         return self.schema.image()
 

recipe_scrapers/heb.py

@@ -9,27 +9,30 @@ def host(cls, domain="com"):
         return f"heb.{domain}"
 
     def title(self):
-        return self.soup.find("h1", {"class": "title"}).get_text()
+        title_tag = self.soup.find("h1", {"data-qe-id": "recipeTitle"})
+        return title_tag.get_text()
 
     def total_time(self):
-        minutes_tag = self.soup.find("div", {"itemprop": "totalTime"})
-        return get_minutes(minutes_tag.parent.get_text())
+        total_time_tag = self.soup.find("span", {"data-qe-id": "recipeTotalTime"})
+        return get_minutes(total_time_tag.get_text())
 
     def yields(self):
-        yields_tag = self.soup.find("div", {"itemprop": "recipeYield"})
-        return get_yields(yields_tag.parent.get_text())
+        yields_tag = self.soup.find("p", {"data-qe-id": "recipeServingSize"})
+        return get_yields(yields_tag.get_text())
 
     def ingredients(self):
-        ingredients_container = self.soup.find(class_="ingredientswrapper")
-        ingredients = ingredients_container.findAll("div", {"class": "recipestepstxt"})
+        ingredients_container = self.soup.find(
+            "div", {"data-qe-id": "recipeIngredientsContainer"}
+        )
+        ingredients = ingredients_container.findAll("li")
 
         return [normalize_string(ingredient.get_text()) for ingredient in ingredients]
 
     def _instructions_list(self):
-        instructions_container = self.soup.find("div", {"class": "instructions"})
-        instructions = instructions_container.findAll(
-            "span", {"class": "instructiontxt"}
+        instructions_container = self.soup.find(
+            "div", {"data-qe-id": "recipeInstructionsContainer"}
         )
+        instructions = instructions_container.findAll("li")
         return [
             normalize_string(instruction.get_text()) for instruction in instructions
         ]

recipe_scrapers/joyfoodsunshine.py

@@ -9,7 +9,7 @@ def host(cls):
         return "joyfoodsunshine.com"
 
     def author(self):
-        return self.soup.find("span", {"class": "entry-author-name"}).get_text()
+        return self.schema.author()
 
     def title(self):
         return self.soup.find("h2", {"class": "wprm-recipe-name"}).get_text()
@@ -26,8 +26,6 @@ def yields(self):
             )
         )
 
-        return self.schema.ingredients()
-
     def image(self):
         # span class = wprm-recipe-image find src
         return self.schema.image()

recipe_scrapers/latelierderoxane.py

@@ -19,7 +19,8 @@ def title(self):
 
     def description(self):
         div = self.soup.find("div", {"class": "bloc_chapeau bloc_blog"})
-        return div.find("p").get_text()
+        cleaned_description = div.find("p").get_text()
+        return normalize_string(cleaned_description)
 
     def total_time(self):
         return get_minutes(self.get_bloc_temps_value_by_index(0))

recipe_scrapers/lecker.py

@@ -1,6 +1,7 @@
 # mypy: disallow_untyped_defs=False
 from ._abstract import AbstractScraper
 from ._exceptions import SchemaOrgException
+from ._utils import normalize_string
 
 
 class Lecker(AbstractScraper):
@@ -85,4 +86,5 @@ def cuisine(self):
             return None
 
     def description(self):
-        return self.schema.description()
+        cleaned_description = self.schema.description()
+        return normalize_string(cleaned_description)