In [2]:
from yummly import Client
import json
import requests
import pandas as pd
import numpy as np
import re # need to import regex
In [3]:
# API call for the first 500 BB recipes labeled as such only!
header= {'X-Yummly-App-ID':'79663a75', 'X-Yummly-App-Key':'02b233108f476f3110e0f65437c4d6dd'}
url='http://api.yummly.com/v1/api/recipes?'
parameters={
'allowedCourse[]':'course^course-Breakfast and Brunch',
'excludedCourse[]': ['course^course-Main Dishes','course^course-Appetizers', 'course^course-Salads', 'course^course-Lunch',
'course^course-Side Dishes','course^course-Desserts','course^course-Breads',
'course^course-Soups', 'course^course-Beverages', 'course^course-Condiments and Sauces',
'course^course-Cocktails', 'course^course-Snacks'],
'maxResult': 501,
'start': 499
}
response=requests.get(url, headers = header, params = parameters)
In [4]:
response.status_code
Out[4]:
In [5]:
BB=response.json()
print type(BB)
print BB.keys()
In [6]:
#only interrested in the information under matches.
print len(BB['matches'])
print type(BB['matches'])
print BB['matches'][0].keys()
In [7]:
#checkout one recipe
BB_matches=BB['matches']
BB_matches[0]
Out[7]:
In [8]:
#import previous list of recipes collected
df=pd.read_csv('BB_main.csv')
BB_ids=df.id
print BB_ids[0]
BB1_ids=[]
for recipe in BB_matches:
BB1_ids.append(recipe['id'])
print BB1_ids[0]
#check if there are dupplicate recipes
[i for i, j in zip(BB_ids, BB1_ids) if i == j]
Out[8]:
In [10]:
#forming lists to create dataframes of the features we want.
main_list = []
ingredients_list = []
attributes_list = []
for food in BB_matches:
_d1 = {}
_d1['id'] = food['id']
_d1['rating'] = food['rating']
_d1['recipeName'] = food['recipeName']
_d1['sourceDisplayName'] = food['sourceDisplayName']
main_list.append(_d1)
_d2 = {}
_d2['id'] =food['id']
_d2['course'] = 'Breakfast and Brunch'
_d2['ingredient_list'] = food['ingredients']
for i in food['ingredients']:
i = i.lower() # additional code to conver to lowercase
i = re.sub(r'\d+%\s', '', i) # additional code to remove 1%, 2%, etc
i = re.sub(r'\xae', '', i) # remove '\xae' characters
i = re.sub(r'shredded\s', '', i)
i = re.sub(r'chopped\s', '', i)
i = re.sub(r'diced\s', '', i)
i = re.sub(r'crumbled\s', '', i)
i = re.sub(r'fresh\s', '', i)
i = re.sub(r'grated\s', '', i)
i = re.sub(r'fat free\s', '', i)
i = re.sub(r'boneless\s', '', i)
i = re.sub(r'boneless skinless\s', '', i)
i = re.sub(r'minced\s', '', i)
i = re.sub(r'sliced\s', '', i)
i = re.sub(r'(?!ground beef)ground ', '', i)
i = re.sub(r'^dried\s', '', i)
i = re.sub(r'^cooked\s', '', i)
_d2[i] = 1
ingredients_list.append(_d2)
_d3 = {}
_d3['id'] = food['id']
for k, v in food['attributes'].items():
for i in v:
_d3[i] = 1
attributes_list.append(_d3)
flavors_dict = {}
for food in BB_matches:
flavors_dict[food.get('id')] = food.get('flavors')
In [66]:
# read in dictionary for course and cuisine and create list of possible values for each
cuisine_df = pd.read_csv('cuisine_headers.csv', names=['cuisine'])
cuisine_list= cuisine_df.cuisine
In [67]:
#create dictionary of cuisine and course for each recipe
cuisine_dict={}
for food in BB_matches:
cuisine_dict[food.get('id')]= food['attributes'].get('cuisine')
_cuisines= {}
for k, v in cuisine_dict.iteritems():
cuisine_val = {}
for course in cuisine_list:
try:
if course in v :
cuisine_val[course] = 1
else:
cuisine_val[course] = 0
except TypeError:
cuisine_val[course] = 0
_cuisines[k] = cuisine_val
In [68]:
# second api call to get other features for each recipe
key_id= '_app_id=79663a75&_app_key=02b233108f476f3110e0f65437c4d6dd'
url='http://api.yummly.com/v1/api/recipe/'
In [69]:
# retrieve other features for all recipes
def get_recipe(_id):
response = requests.get(url + _id + '?' + key_id)
return response.json()
recipes=[]
for _id in BB1_ids :
recipes.append(get_recipe(_id))
In [19]:
response.status_code
Out[19]:
In [20]:
print len(recipes)
print recipes[1].keys()
In [21]:
#for each recipe create a new dictionary of selected attributes and append into a list
recipe_details=[]
for recipe in recipes:
_dict={}
#import pdb; pdb.set_trace()
_dict['id']=recipe['id']
_dict['ingredientCount']= len(recipe['ingredientLines'])
_dict['numberOfServings']= recipe['numberOfServings']
if 'prepTimeInSeconds' in recipe.keys():
_dict['prepTimeInSeconds']= recipe['prepTimeInSeconds']
else:
_dict['prepTimeInSeconds']= None
if 'cookTimeInSeconds' in recipe.keys():
_dict['cookTimeInSeconds']= recipe['cookTimeInSeconds']
else:
_dict['cookTimeInSeconds']= None
_dict['totalTimeInSeconds']=recipe['totalTimeInSeconds']
recipe_details.append(_dict)
In [11]:
#create dataframes, arrange column index and save into csv
# df_main = pd.DataFrame(main_list)
# df_main.to_csv('BB_main_1.csv', encoding ='utf-8')
df_ingredients = pd.DataFrame(ingredients_list)
df_ingredients = df_ingredients.fillna(0)
cols = list(df_ingredients)
cols.insert(0, cols.pop(cols.index('id')))
cols.insert(1, cols.pop(cols.index('course')))
df_ingredients= df_ingredients.ix[:,cols]
df_ingredients.to_csv('BB_ingredients_1.csv', encoding ='utf-8')
# df_attributes = pd.DataFrame(attributes_list)
# df_attributes = df_attributes.fillna(0)
# cols = list(df_attributes)
# cols.insert(0, cols.pop(cols.index('id')))
# df_attributes = df_attributes.ix[:,cols]
# df_attributes.to_csv('BB_attributes_1.csv')
# df_flavors = pd.DataFrame(flavors_dict).transpose()
# df_flavors.reset_index(level=0, inplace=True)
# df_flavors=df_flavors.rename(columns = {'index':'id'})
# df_flavors.to_csv('BB_flavors_1.csv')
# df_cuisines = pd.DataFrame(_cuisines).transpose()
# df_cuisines.reset_index(level=0, inplace=True)
# df_cuisines=df_cuisines.rename(columns = {'index':'id'})
# df_cuisines.to_csv('BB_cuisines_1.csv')
# df_details=pd.DataFrame(recipe_details)
# cols = list(df_details)
# cols.insert(0, cols.pop(cols.index('id')))
# df_details=df_details.ix[:,cols]
# df_details.to_csv('BB_details_1.csv')
In [72]:
df_ingredients.head()
Out[72]:
In [123]:
BB = pd.read_csv('BB_ingredients.csv')
In [124]:
BB.head()
Out[124]:
In [101]:
data=pd.concat([BB, df_ingredients],join='outer', axis=0, ignore_index=True)
In [104]:
type(df_ingredients)
Out[104]:
In [102]:
data
Out[102]:
In [111]:
for i in df_ingredients.columns:
if 'English' in i:
print i
In [ ]: