updating scrapers

RyanNoelk · Jul 22, 2017 · 5a8786f · 5a8786f
1 parent 01fff9c
commit 5a8786f
Show file tree

Hide file tree

Showing 3 changed files with 37 additions and 17 deletions.
diff --git a/example.py b/example.py
@@ -5,7 +5,7 @@
 from recipe_scrapers import scrap_me
 
 try:
-    scrap_me = scrap_me('http://allrecipes.com/recipe/213742/meatball-nirvana/?internalSource=staff%20pick&referringId=80&referringContentType=recipe%20hub&clickId=cardslot%205')
+    scrap_me = scrap_me('https://www.budgetbytes.com/2017/07/slow-cooker-sesame-beef/')
     pprint(scrap_me.data())
 except KeyError:
     print "Website is not supported."
diff --git a/recipe_scrapers/budgetbytes.py b/recipe_scrapers/budgetbytes.py
@@ -22,10 +22,13 @@ def total_time(self):
         }
 
     def servings(self):
-        return self.soup.find('span', {'itemprop': 'recipeYield'}).get_text()
+        try:
+            return self.soup.find('span', {'itemprop': 'recipeYield'}).get_text()
+        except:
+            return ''
 
     def ingredients(self):
-        ingredients_html = self.soup.findAll('li', {'class': 'ingredient'})
+        ingredients_html = self.soup.findAll('li', {'class': 'wprm-recipe-ingredient'})
         ingredients = []
 
         for ingredient in ingredients_html:
@@ -49,7 +52,7 @@ def ingredients(self):
         return ingredients
 
     def instructions(self):
-        instructions_html = self.soup.findAll('li', {'class': 'instruction'})
+        instructions_html = self.soup.findAll('li', {'class': 'wprm-recipe-instruction'})
 
         return [
             normalize_string(instruction.get_text())
@@ -61,4 +64,7 @@ def description(self):
         return li[0].get_text()
 
     def image(self):
-        return self.soup.find('img', {'itemprop': 'image'})["src"]
+        try:
+            return self.soup.find('img', {'itemprop': 'image'})["src"]
+        except:
+            return ''
diff --git a/recipe_scrapers/budgetbytesv2.py b/recipe_scrapers/budgetbytesv2.py
@@ -16,19 +16,27 @@ def title(self):
         return self.soup.find('h1').get_text()
 
     def total_time(self):
-        return {
-            'prep-time': self.soup.find(
+        try:
+            prep = self.soup.find(
                 'span',
                 {'class': 'wprm-recipe-prep_time-minutes'}
-            ).get_text(),
-            'cook-time': self.soup.find(
+            ).get_text()
+        except AttributeError:
+            prep = '0'
+        try:
+            cook = self.soup.find(
                 'span',
                 {'class': 'wprm-recipe-cook_time-minutes'}
             ).get_text()
-        }
+        except AttributeError:
+            cook = '0'
+        return {prep, cook}
 
     def servings(self):
-        return self.soup.find('span', {'itemprop': 'recipeYield'}).get_text().split(' ', 1)[0]
+        try:
+            return self.soup.find('span', {'itemprop': 'recipeYield'}).get_text().split(' ', 1)[0]
+        except:
+            return ''
 
     def ingredients(self):
         ingredients_html = self.soup.findAll('li', {'class': 'wprm-recipe-ingredient'})
@@ -52,14 +60,14 @@ def ingredients(self):
                 }
             except AttributeError:
                 ingredient_dict = {
-                    'title': ingredient.find(
+                    'title': normalize_string(ingredient.find(
                         'span',
                         {'class': 'wprm-recipe-ingredient-name'}
-                    ).get_text()
+                    ).get_text())
                 }
             except:
                 ingredient_dict = {
-                    'title': ingredient
+                    'title': normalize_string(ingredient.get_text())
                 }
 
             ingredients.append(ingredient_dict)
@@ -75,8 +83,14 @@ def instructions(self):
         ]
 
     def description(self):
-        li = self.soup.find('article', {'class': 'post'}).findAll('p')
-        return li[0].get_text()
+        try:
+            li = self.soup.find('article', {'class': 'post'}).findAll('p')
+            return li[0].get_text()
+        except:
+            return ''
 
     def image(self):
-        return self.soup.find('img', {'class': 'alignnone'})["src"]
+        try:
+            return self.soup.find('img', {'class': 'alignnone'})["src"]
+        except:
+            return ''