Added serving-size (#61)

* Added servings method to abstract class.
Added get_servings to _utils.

* Tests: made test-data-html-file relative from each test-file (Makes it possible for IDE or single unittest run).
Added serving-test to all except All recipes BR, The vintage mixer and What's gaby cooking (do not have servings in test-cases), will return 0.

* Added method-comment to _utils.get_servings.

* Consistency in code structure code.

* Changed servings to yields.
yields-method will now return a string of "x item(s)" or "x serving(s)", will default toward servings when found a number, will default toward an empty string if it defaults of missing content.

* Added yields-method to README

README.md

@@ -17,6 +17,7 @@ then:
 
     scraper.title()
     scraper.total_time()
+    scraper.yields()
     scraper.ingredients()
     scraper.instructions()
     scraper.links()

recipe_scrapers/_abstract.py

@@ -24,6 +24,7 @@ def __getattribute__(self, name):
             decorated_methods = [
                 'title',
                 'total_time',
+                'yields',
                 'instructions',
                 'ingredients',
                 'links'
@@ -32,6 +33,8 @@ def __getattribute__(self, name):
                 to_return = ''
             if name == 'total_time':
                 to_return = 0
+            if name == 'yields':
+                to_return = ''
             if name == 'ingredients':
                 to_return = []
             if name == 'links':
@@ -73,6 +76,10 @@ def total_time(self):
         """ total time it takes to preparate the recipe in minutes """
         raise NotImplementedError("This should be implemented.")
 
+    def yields(self):
+        """ The number of servings or items in the recipe """
+        raise NotImplementedError("This should be implemented.")
+
     def ingredients(self):
         raise NotImplementedError("This should be implemented.")
 

recipe_scrapers/_utils.py

@@ -7,6 +7,18 @@
     r'(\D*(?P<hours>\d+)\s*(hours|hrs|hr|h|Hours|H))?(\D*(?P<minutes>\d+)\s*(minutes|mins|min|m|Minutes|M))?'
 )
 
+SERV_REGEX_NUMBER = re.compile(
+    r'(\D*(?P<items>\d+)?\D*)'
+)
+
+SERV_REGEX_ITEMS = re.compile(
+    r'\bsandwiches\b |\btacquitos\b | \bmakes\b', flags=re.I | re.X
+)
+
+SERV_REGEX_TO = re.compile(
+    r'\d+(\s+to\s+|-)\d+', flags=re.I | re.X
+)
+
 
 def get_minutes(element):
     try:
@@ -23,6 +35,38 @@ def get_minutes(element):
         return 0
 
 
+def get_yields(element):
+    """
+    Will return a string of servings or items, if the receipt is for number of items and not servings
+    the method will return the string "x item(s)" where x is the quantity.
+    :param element: Should be BeautifulSoup.TAG, in some cases not feasible and will then be text.
+    :return: The number of servings or items.
+    """
+    try:
+
+        if isinstance(element, str):
+            tstring = element
+        else:
+            tstring = element.get_text()
+
+        if SERV_REGEX_TO.search(tstring):
+            tstring = tstring.split(SERV_REGEX_TO.split(tstring)[1])[1]
+
+        matched = SERV_REGEX_NUMBER.search(tstring).groupdict().get('items') or 0
+        servings = "{} serving(s)".format(matched)
+
+        if SERV_REGEX_ITEMS.search(tstring) 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(s)".format(matched)
+
+        return servings
+
+    except AttributeError as e:  # if dom_element not found or no matched
+        print("get_serving_numbers error {}".format(e))
+        return ''
+
+
 def normalize_string(string):
     return re.sub(
         r'\s+', ' ',

recipe_scrapers/allrecipes.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
 
 
 class AllRecipes(AbstractScraper):
@@ -17,6 +17,12 @@ def total_time(self):
             {'class': 'ready-in-time'})
         )
 
+    def yields(self):
+        return get_yields(self.soup.find(
+            'meta',
+            {'id': 'metaRecipeServings', 'itemprop': 'recipeYield'}).get("content")
+                          )
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'li',

recipe_scrapers/allrecipesbr.py

@@ -30,6 +30,9 @@ def ingredients(self):
             for ingredient in ingredients
         ]
 
+    def yields(self):
+        return 0
+
     def instructions(self):
         instructions = self.soup.find(
             'ol', {'itemprop': 'recipeInstructions'}

recipe_scrapers/bbcfood.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
 
 
 class BBCFood(AbstractScraper):
@@ -24,6 +24,12 @@ def total_time(self):
             )
         ])
 
+    def yields(self):
+        return get_yields(self.soup.find(
+            'p',
+            {'class': 'recipe-metadata__serving'})
+        )
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'li',

recipe_scrapers/bbcgoodfood.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
 
 
 class BBCGoodFood(AbstractScraper):
@@ -24,6 +24,12 @@ def total_time(self):
             ).find('span'))
         ])
 
+    def yields(self):
+        return get_yields(self.soup.find(
+                'span',
+                {'class': 'recipe-details__text', 'itemprop': 'recipeYield'}
+            ))
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'li',

recipe_scrapers/bonappetit.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import normalize_string
+from ._utils import normalize_string, get_yields
 
 
 class BonAppetit(AbstractScraper):
@@ -17,6 +17,14 @@ def title(self):
     def total_time(self):
         return 0
 
+    def yields(self):
+        return get_yields(
+            self.soup.find(
+                'span',
+                {'class': "recipe__header__servings recipe__header__servings--basically"}
+            ).find('span')
+        )
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'li',

recipe_scrapers/closetcooking.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
 
 
 class ClosetCooking(AbstractScraper):
@@ -17,6 +17,10 @@ def title(self):
     def total_time(self):
         return get_minutes(self.soup.find(itemprop='totalTime').parent)
 
+
+    def yields(self):
+        return get_yields(self.soup.find(itemprop='recipeYield').parent)
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'li',

recipe_scrapers/cookstr.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
 
 
 class Cookstr(AbstractScraper):
@@ -26,6 +26,18 @@ def total_time(self):
                 total_time += get_minutes(time.parent.parent)
         return total_time
 
+    def yields(self):
+        sections = self.soup.findAll(
+            'span',
+            {'class': 'attrLabel'}
+        )
+        total_serves = 0
+        for section in sections:
+            serves = section.find(text='Serves')
+            if serves:
+                total_serves += get_yields(serves.parent.parent)
+        return total_serves
+
     def ingredients(self):
         ingredients = self.soup.find(
             'div',

recipe_scrapers/epicurious.py

@@ -2,7 +2,7 @@
 
 
 from ._abstract import AbstractScraper
-from ._utils import normalize_string
+from ._utils import normalize_string, get_yields
 
 
 class Epicurious(AbstractScraper):
@@ -21,6 +21,12 @@ def title(self):
     def total_time(self):
         return 0
 
+    def yields(self):
+        return get_yields(self.soup.find(
+            'dd',
+            {'itemprop': 'recipeYield'}
+        ))
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'li',

recipe_scrapers/finedininglovers.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
 
 
 class FineDiningLovers(AbstractScraper):
@@ -20,6 +20,13 @@ def total_time(self):
             {'itemprop': 'prepTime'})
         )
 
+
+    def yields(self):
+        return get_yields(self.soup.find(
+            'span',
+            {'itemprop': 'recipeYield'})
+        )
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'li',

recipe_scrapers/foodnetwork.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
 
 
 class FoodNetwork(AbstractScraper):
@@ -17,6 +17,13 @@ def total_time(self):
             {'class': 'm-RecipeInfo__a-Description--Total'})
         )
 
+    def yields(self):
+        return get_yields(self.soup.find(
+                'ul',
+                {'class': 'o-RecipeInfo__m-Yield'}
+            ).find('span', {'class': 'o-RecipeInfo__a-Description'})
+                          )
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'p',

recipe_scrapers/foodrepublic.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
 
 
 class FoodRepublic(AbstractScraper):
@@ -24,6 +24,12 @@ def total_time(self):
             )
         ])
 
+    def yields(self):
+        return get_yields(self.soup.find(
+            'span',
+            {'itemprop': 'recipeYield'}
+        ))
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'li',

recipe_scrapers/geniuskitchen.py

@@ -1,5 +1,6 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
+
 
 class GeniusKitchen(AbstractScraper):
 
@@ -16,6 +17,13 @@ def total_time(self):
             {'class': 'time'})
         )
 
+    def yields(self):
+        return get_yields(self.soup.find(
+            'td',
+            {'class': 'servings'}
+        ).find('span', {'class': 'count'})
+                          )
+
 
     def ingredients(self):
         ingredients = []

recipe_scrapers/giallozafferano.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
 import json
 
 
@@ -25,6 +25,12 @@ def total_time(self):
             )
         ])
 
+    def yields(self):
+        return get_yields(self.soup.find(
+                'li',
+                {'class': 'yield'})
+            )
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'dd',

recipe_scrapers/hellofresh.py

@@ -20,6 +20,9 @@ def total_time(self):
             {'data-translation-id': "recipe-detail.preparation-time"}
         ).parent.parent)
 
+    def yields(self):
+        return 0
+
     def ingredients(self):
         ingredients_container = self.soup.find(
             'div',

recipe_scrapers/hundredandonecookbooks.py

@@ -17,6 +17,9 @@ def total_time(self):
             {'class': 'preptime'})
         )
 
+    def yields(self):
+        return 0
+
     def ingredients(self):
         ingredients = self.soup.find(
             'div',

recipe_scrapers/inspiralized.py

@@ -1,5 +1,5 @@
 from ._abstract import AbstractScraper
-from ._utils import get_minutes, normalize_string
+from ._utils import get_minutes, normalize_string, get_yields
 
 
 class Inspiralized(AbstractScraper):
@@ -17,6 +17,12 @@ def total_time(self):
             {'itemprop': 'totalTime'})
         )
 
+    def yields(self):
+        return get_yields(self.soup.find(
+            'span',
+            {'itemprop': 'recipeYield'})
+        )
+
     def ingredients(self):
         ingredients = self.soup.findAll(
             'li',