def get_recipe_details(best_recipe_combo, user_profile_data):
profile_init = rootprofile.UserProfile(user_profile_data)
recipe_init = recipes.Recipes(profile_init)
# get ingredients from the recipe list
recipe_details = []
i = 0
for rec_idx in best_recipe_combo:
recipe_details.append(recipe_init.recipe_clean[rec_idx])
# recipe_details = recipe_details.update(rec_idx=recipe_init.recipe_clean[rec_idx])
# print(recipe_details)
return recipe_details
def get_shopping_list(best_recipe_combo, user_profile_data):
profile_init = rootprofile.UserProfile(user_profile_data)
recipe_init = recipes.Recipes(profile_init)
# get ingredients from the recipe list
ingredient_list = []
for rec_idx in best_recipe_combo:
ingredient_list = ingredient_list + recipe_init.recipe_clean[rec_idx][
'ingredients']
while '' in ingredient_list:
ingredient_list.remove('')
return (ingredient_list)
def get_recipe_list(session, user):
user_profile_data = pd.read_json(session['data'])
# Run Recipe Recomendation Alg
profile_init = rootprofile.UserProfile(user_profile_data)
recipe_init = recipes.Recipes(profile_init)
# Check for any recipes to ignore
ignore_list = get_user_ignore_responses(session, user)
# number of meal preferences
if 'meals_per_week' not in user_profile_data.keys():
meals_per_week = 6 # Default
elif user_profile_data.meals_per_week.values[0] == '':
meals_per_week = 6 # Default - User did not specify in survey
else:
meals_per_week = int(user_profile_data.meals_per_week.values[0])
# Algroithm initiation settings
num_generations = 10
amount_per_population = 30
amount_parents_mating = 10
# Run GA Recipe/Meal Plan Optimization
# (macro_list=profile_init.macro_label_list)
GA = models.GA(macro_list=profile_init.macro_label_list)
label_of_weights = GA.labels
weekly_diet_amount = (GA.user_df[GA.macro_labels] / 3.0) * meals_per_week
best_recipe_combo, weekly_diet_amount = GA.AMGA(
num_generations, meals_per_week, amount_per_population,
amount_parents_mating, weekly_diet_amount, ignore_list)
# user_profile_data['list_keys'] = [best_recipe_combo]
# user_profile_data.list_keys = best_recipe_combo
user_profile_data['plan_exists'] = True
recipe_names = []
for rec_idx in best_recipe_combo:
recipe_names.append(recipe_init.recipe_clean[rec_idx]['name'])
# user_profile_data.recipe_names = recipe_names
# user_profile_data['recipe_names'] = [recipe_names]
user_meal_plan = pd.DataFrame(data={
'recipe_id': best_recipe_combo,
'recipe_name': recipe_names
})
session['user_meal_plan'] = user_meal_plan.to_json()
df_ingredient_NDB = get_ingredient_NDB_number(session, best_recipe_combo)
# session['df_ingredient_NDB'] = df_ingredient_NDB.to_json()
return best_recipe_combo, weekly_diet_amount, user_profile_data, df_ingredient_NDB
def create_recipe_rec_df(session, user_meal_plan):
user_profile_data = pd.read_json(session['data'])
#// START: initialize for df creation to plot bar/radar plots
profile_init = rootprofile.UserProfile(user_profile_data)
recipe_init = recipes.Recipes(profile_init)
list_keys = user_meal_plan['recipe_id'].get_values()
#// END
#// START: loop through each recipe and convert into mathmatical nutrition space
# return:
# df_list -> for stacked barplot graph of each recipe individual and what is inside the recipe
# df_summed_list -> for radar and barplot graph of meal plan
# recipe_id_list -> used book keeping/linking
# name_list -> used to help name and make images pretty
df_list = []
df_summed_list = []
recipe_id_list = []
name_list = []
recipe_itr = 0
for recipe in list_keys:
# print("*******************************", "Recipe Itr: ", recipe_itr, "Out of: ", len(list_keys) - 1, recipe)
try:
temp_recipe_df = recipe_init.recipe_list_to_conversion_factor_list(
recipe)
multiplier_normalizer = profile_init.profile_macro_filtered_df[
'calories'].get_values()[0] / 3.0
multiplier_normalizer = multiplier_normalizer / temp_recipe_df[
'Energy (kcal)'].sum()
for column in [profile_init.macro_list + profile_init.micro_list]:
temp_recipe_df[
column] = temp_recipe_df[column] * multiplier_normalizer
df_list.append(temp_recipe_df)
df_summed_list.append(
temp_recipe_df.loc[:, profile_init.macro_list +
profile_init.micro_list].sum().to_frame())
name_list.append(recipe_init.recipe_clean[recipe]['name'])
recipe_id_list.append(recipe)
except:
print("FAILED, Recipe concatenation...RECIPE=", recipe)
recipe_itr += 1
df = pd.concat(df_summed_list, axis=1)
df = df.T.reset_index(drop=True)
se = pd.Series(name_list)
df['recipe_name'] = se.values
se2 = pd.Series(recipe_id_list)
df['recipe_id'] = se2.values
return df, df_list, df_summed_list, profile_init, name_list
def run_master_ingredient_sub(user_profile_data):
profile_init = rootprofile.UserProfile(user_profile_data)
recipe_init = recipes.Recipes(profile_init)
research_init = research.Research(profile_init)
# Run Subsitution option for Ingredients
recipe_itr = 0
df_list = []
df_summed_list = []
recipe_id_list = []
name_list = []
if 'list_keys' not in user_profile_data.keys():
# If no recipes exist for user create a meal plan
# if len(best_recipe_combo) == 0:
print("****** Running get recipes ****")
best_recipe_combo, weekly_diet_amount, user_profile_data = get_recipe_list(
user_profile_data, recipe_init)
for recipe in user_profile_data.list_keys:
print(recipe)
temp_recipe_df = recipe_init.recipe_list_to_conversion_factor_list(
recipe)
df_list.append(temp_recipe_df)
df_summed_list.append(
temp_recipe_df.loc[:, profile_init.macro_list +
profile_init.micro_list].sum().to_frame())
name_list.append(recipe_init.recipe_clean[recipe]['name'])
recipe_id_list.append(recipe)
recipe_itr += 1
# SMASH Files together for optimization mapp
df = pd.concat(df_summed_list, axis=1)
df = df.T.reset_index(drop=True)
se = pd.Series(name_list)
df['recipe_name'] = se.values
se2 = pd.Series(recipe_id_list)
df['recipe_id'] = se2.values
# Create Recipe Visuals for Display
recipe_visuals(df_list, df, profile_init, name_list)
return df, df_list
def get_ingredient_NDB_number(session, best_recipe_combo):
# Returns Ingredits in recipes to dataframe
# Save ingredients to Session Data
profile_init = rootprofile.UserProfile(pd.read_json(session['data']))
recipe_init = recipes.Recipes(profile_init)
df_ingredient_NDB = pd.DataFrame()
for recipe_id in best_recipe_combo:
recipe_data = recipe_init.recipe_list_to_conversion_factor_list(
recipe_id)[['Description', 'NDB_NO']]
df_ingredient_NDBi = pd.DataFrame(recipe_data)
df_ingredient_NDBi['recipe_id'] = recipe_id
if df_ingredient_NDB.empty:
df_ingredient_NDB = df_ingredient_NDBi
else:
df_ingredient_NDB = pd.concat(
[df_ingredient_NDB, df_ingredient_NDBi])
df_ingredient_NDB = df_ingredient_NDB.reset_index()
# print(df_ingredient_NDB.recipe_id.unique())
return (df_ingredient_NDB)
def get_recipe_list(user_profile_data, user):
# Run Recipe Recomendation Alg
profile_init = rootprofile.UserProfile(user_profile_data)
recipe_init = recipes.Recipes(profile_init)
# Check for any recipes to ignore
user_profile_data = get_user_ignore_responses(user_profile_data, user)
ignore_list = user_profile_data.ignore_list
# number of meal preferences
if 'meals_per_week' not in user_profile_data.keys():
meals_per_week = 6 # Default
elif user_profile_data.meals_per_week.values[0] == '':
meals_per_week = 6 # Default - User did not specify in survey
else:
meals_per_week = int(user_profile_data.meals_per_week.values[0])
# Algroithm initiation settings
num_generations = 10
amount_per_population = 30
amount_parents_mating = 10
# Run GA Recipe/Meal Plan Optimization
GA = models.GA()
label_of_weights = GA.labels
weekly_diet_amount = (GA.user_df[GA.macro_labels] / 3.0) * meals_per_week
best_recipe_combo, weekly_diet_amount = GA.AMGA(
num_generations, meals_per_week, amount_per_population,
amount_parents_mating, weekly_diet_amount, ignore_list)
user_profile_data.list_keys = best_recipe_combo
recipe_names = []
for rec_idx in best_recipe_combo:
recipe_names.append(recipe_init.recipe_clean[rec_idx]['name'])
user_profile_data.recipe_names = recipe_names
return best_recipe_combo, weekly_diet_amount, user_profile_data
def get_pantry_suggetsions(user_profile_data,
ingredient_list,
num_suggested_recipes=5):
# Input: list of pantry items and nutriotial Goals
# Output: List of x recipe ID
# initlize user needs:
profile_init = rootprofile.UserProfile(user_profile_data)
recipe_init = recipes.Recipes(profile_init)
model_init = models.Models(recipe_init)
## Virtual Pantry to recipe suggestion
# ingredient_list = ['12 large egg', '12 oz mayonnaise', '12 oz BBQ Sauce', '24 oz mustard', '6 skinless chicken breast',
# '12 salmon burgers', '12 pita', '2 zucchini', '4 onions', '1 pound mushrooms',
# '12 cups lettuce', '24 beer', '1 whole duck', '4 oranges', '2 potatoes', '2 red peppers',
# '4 pounds white rice', '48 oz peanut butter', '6 sausage', '4 pounds ham', '1 cauliflower',
# '6 sticks butter', '12 cups sugar', '12 brown sugar', 'water', '1 whole lemon', '0.25 cup lemon juice',
# '10 whole apples', '1 tomato', '0.25 cup Lime juice', '1 bottle rum', '12 egg whites', '1 whole garlic']
pantry_list = []
pantry_tag_list = []
for ingredient in ingredient_list:
ingredient_filtered = model_init.transform_data_for_tokenizer(
ingredient)
tokens = model_init.tokenizer.texts_to_matrix([ingredient_filtered])
prediction_key = str(
model_init.loaded_model.predict(
np.array(tokens)).argsort()[0][-1:][::-1][0])
for item in model_init.NDB_tag_unique_unique_dict.items():
if item[1] == int(prediction_key):
tag_key = item[0]
break
prediction_value = recipe_init.nutrition_init.NDB_NO_lookup(
tag_key, filter_list=['Description']).get_values()[0][0]
pantry_tag_list.append(tag_key)
recipe_id_list = []
percent_overlap_list = []
ingredient_pantry_list = []
recipe_full_list = []
for recipe in recipe_init.recipe_clean.keys():
temp_recipe_list = []
itr = 0
while itr < len(recipe_init.recipe_clean[recipe]["NDB_NO_tags"]):
try:
if recipe_init.nutrition_init.NDB_NO_lookup(
recipe_init.recipe_clean[recipe]["NDB_NO_tags"][itr],
filter_list=['Category']).get_values()[0][0] not in [
'Fats_and_Oils', 'Spices_and_Herbs'
]:
temp_recipe_list.append(
recipe_init.recipe_clean[recipe]["NDB_NO_tags"][itr])
except:
pass
itr += 1
overlap_list = list(
set(temp_recipe_list).intersection(pantry_tag_list))
amount_overlap = len(overlap_list)
amount_recipe = len(temp_recipe_list)
recipe_id_list.append(recipe)
try:
percent_overlap_list.append(amount_overlap / amount_recipe)
ingredient_pantry_list.append(overlap_list)
recipe_full_list.append(temp_recipe_list)
except:
percent_overlap_list.append(0.0)
ingredient_pantry_list.append([])
recipe_full_list.append([])
df = pd.DataFrame({
'recipe_id': recipe_id_list,
'percent_overlap': percent_overlap_list,
'ingredient_pantry_list': ingredient_pantry_list,
'recipe_full_list': recipe_full_list
})
df = df.sort_values(by='percent_overlap', ascending=False).reset_index()[[
'recipe_id', 'percent_overlap', 'ingredient_pantry_list',
'recipe_full_list'
]]
# create list of recipe ids
recipe_ids = df.recipe_id.values[:num_suggested_recipes]
return recipe_ids
def get_single_ingredient_replacement(session, ingredientSubForm, recipe_id,
nbd_no):
# Single food replacement based on macros
# Get intial information on user
profile_init = rootprofile.UserProfile(pd.read_json(session['data']))
recipe_init = recipes.Recipes(profile_init)
research_init = research.Research(profile_init)
ingredient_list = recipe_init.recipe_clean[recipe_id]['ingredients']
# temp_recipe_dict = {}
# temp_recipe_dict[recipe_id] = recipe_init.recipe_clean[recipe_id].copy()
if ingredientSubForm.ingredientSub.data is not '':
replacement_key_dict = {
1: 'Baked',
2: 'Beef',
3: 'Beverages',
4: 'Breakfast_Cereals',
5: 'Cereal_Grains_and_Pasta',
6: 'Dairy_and_Egg',
7: 'Fats_and_Oils',
8: 'Finfish_and_Shellfish',
9: 'Fruits_and_Fruit_Juices',
10: 'Lamb_Veal_and_Game',
11: 'Legumes_and_Legume',
12: 'Nut_and_Seed',
13: 'Pork',
14: 'Poultry',
15: 'Sausages_and_Luncheon_Meats',
16: 'Soups_Sauces_and_Gravies',
17: 'Spices_and_Herbs',
18: 'Sweets',
19: 'Vegetables_and_Vegetable'
}
# split_raw_return = raw_input_return.split(",")
# "05097":8,"44005":7
# TODO: improve to allow user to input multiple replacements
# master_tag_list = []
# replace_list = []
# for replacement_ndb_tag in range(len(ingredientSubForm.ingredientSub.data)):
# replacement_ndb_tag = replacement.split(":")[0].lstrip(" ")
# replacement_category_key = int(replacement.split(":")[-1].strip("'").strip('"'))
# Get the replacement_ndb_tag from the entry form
# replacement_ndb_tag = ingredientSubForm.ingredientSub.data
replacement_ndb_tag = nbd_no
replacement_category_key = ingredientSubForm.foodType.data
tag_list, potential_switches = research_init.macro_space_distance_top_n(
4, replacement_ndb_tag,
[replacement_key_dict[int(replacement_category_key)]])
# Verify one of top three option is not the same as input
for i, tag in enumerate(tag_list):
if tag == replacement_ndb_tag:
tag_list.remove(tag)
potential_switches.remove(potential_switches[i])
switch_df = pd.DataFrame(data={
'tags': tag_list[:3],
"potential_switches": potential_switches[:3]
})
session['potential_switches'] = potential_switches
# # Get Visuals Data
# print("******Ingredeint Sub Visual****")
# # Split User input into the item to replace and type, format: ['"44005":7']
# master_tag_list = []
# replace_list = []
#
# temp_recipe_dict = {}
# temp_recipe_dict[recipe_id] = recipe_init.recipe_clean[recipe_id].copy()
#
# new_recipe_dict = recipe_init.recipe_alternitive_create(replacement_ndb_tag, tag_list, temp_recipe_dict)
# replace_list.append(replacement_ndb_tag)
# master_tag_list.append(tag_list)
#
# # Create an iterable list. Change Masetr tage list from:
# # master_tag_list = [['"04042"', '"04618"', '"04545"']] to
# # iterable_list = [('"04042"',), ('"04618"',), ('"04545"',)]
# iterable_list = list(itertools.product(*master_tag_list))
# new_recipe_dict = recipe_init.recipe_alternitive_iter_create(replace_list, iterable_list, temp_recipe_dict)
# temp = recipe_init.recipe_list_to_conversion_factor_list(recipe_id)
# df_list = []
# name_list = []
# for recipe in new_recipe_dict.keys():
# temp_recipe_df = recipe_init.recipe_list_to_conversion_factor_list(recipe, dict=new_recipe_dict)
# df_list.append(temp_recipe_df)
# name_list.append(new_recipe_dict[recipe]['name'])
# print('*____ Visuals ___*')
# # print("df_list", df_list)
# # print("profile_init", profile_init)
# # print('name_list', name_list)
# visualizations.Plots(df_list, profile_init).bar_plot_recipe(name_list, 'test_replacement_barplot', session)
# # print("visual 1")
# # visualizations.Plots(df_list, profile_init).radar_plot_recipe(name_list, 'test_replacement_radar_plot')
# # print("visual 2")
return switch_df, potential_switches[:3]
