gpn23-recipes/1_clean_json.py

import json
import re

def normalize_ingredients(recipe):
    if 'recipeIngredient' in recipe:
        recipe['recipeIngredient'] = [
            ingredient.replace('g / Gramm', 'g') for ingredient in recipe['recipeIngredient']
        ]
        recipe['recipeIngredient'] = [
            ingredient.replace('kg / Kilogramm', 'kg') for ingredient in recipe['recipeIngredient']
        ]
        recipe['recipeIngredient'] = [
            ingredient.replace('.0 ', ' ') for ingredient in recipe['recipeIngredient']
        ]
        recipe['recipeIngredient'] = [
            re.sub("^0 (g|kg|Milliliter|None) ", "", ingredient) for ingredient in recipe['recipeIngredient']
        ]
        recipe['recipeIngredient'] = [
            ingredient \
                for ingredient in recipe['recipeIngredient'] \
                if ingredient != "None"
        ]
    return recipe


def normalize_instructions(recipe):
    if 'recipeInstructions' in recipe:
        recipe['recipeInstructions'] = [
           instruction \
                for instruction in recipe['recipeInstructions'] \
                if instruction.get('text')
        ]
        if not recipe['recipeInstructions']:
            del recipe['recipeInstructions']
    return recipe


def normalize_recipe(recipe):
    cleaned_recipe = recipe
    cleaned_recipe = normalize_ingredients(cleaned_recipe)
    cleaned_recipe = normalize_instructions(cleaned_recipe)
    return cleaned_recipe


# TODO