def __init__(self, comp_ing_dict, edge_dict, all_recipes, match_dict=None):
"""
initializes recommender
:param comp_ing_dict: compound : ingredient dict
:param edge_dict: flavor network edges
:param all_recipes: recipe dict
:param make_match_dict: bool for building match lookup (for performance)
:return:
"""
self.ing_matcher = IngredientMatcher(comp_ing_dict)
self.comp_ing_dict = self.ing_matcher.sanitized_comp_ing_dict
cocktail_ing_lists = all_recipes.values()
self.uniq_ing_list = list(set([i.strip().lower() for sublist in cocktail_ing_lists for i in sublist if i]))
self.edge_dict = edge_dict
if not match_dict:
self.match_dict = self.ing_matcher.make_match_dict(self.uniq_ing_list)
else:
self.match_dict = match_dict
self.recipe_splitter = re.compile(',\s')
class FlavorRecommender(object):
"""
recommends a new ingredient based on the flavor network
"""
def __init__(self, comp_ing_dict, edge_dict, all_recipes, match_dict=None):
"""
initializes recommender
:param comp_ing_dict: compound : ingredient dict
:param edge_dict: flavor network edges
:param all_recipes: recipe dict
:param make_match_dict: bool for building match lookup (for performance)
:return:
"""
self.ing_matcher = IngredientMatcher(comp_ing_dict)
self.comp_ing_dict = self.ing_matcher.sanitized_comp_ing_dict
cocktail_ing_lists = all_recipes.values()
self.uniq_ing_list = list(set([i.strip().lower() for sublist in cocktail_ing_lists for i in sublist if i]))
self.edge_dict = edge_dict
if not match_dict:
self.match_dict = self.ing_matcher.make_match_dict(self.uniq_ing_list)
else:
self.match_dict = match_dict
self.recipe_splitter = re.compile(',\s')
def make_rec(self, drink_recipe, flavor_threshold):
"""
makes a single recommendation
:param drink_recipe: user supplied recipe
:flavor_threshold min number of shared compounds to consider
:return: suggested ingredient
"""
recipe_ing_list = self.recipe_splitter.split(drink_recipe)
potential_suggestions = []
for ingredient in self.uniq_ing_list:
for recipe_ingredient in recipe_ing_list:
pair = (ingredient, recipe_ingredient)
try:
score = compute_single_avg_weight(pair, self.edge_dict, self.match_dict)
if score > flavor_threshold and pair[0] != pair[1]:
potential_suggestions.append(pair)
except KeyError:
pass
if potential_suggestions:
return random.sample(potential_suggestions, min(len(potential_suggestions), 4))
else:
return 'no suggestion'
def order_ingredients(suggestions, n_suggestions):
"""
assures suggested ingredients are in the form suggestion : original ingredient,
only needed for alternative suggestion methods (not used here)
:param suggestions: all possible suggestions
:param n_suggestions number of suggestions to return
:return: suggestions in the proper order
"""
ordered_suggestions = []
for ipair in suggestions:
if ipair[0] in drink_recipe:
ordered_suggestions.append((ipair[1], ipair[0]))
else:
ordered_suggestions.append(ipair)
return ordered_suggestions