gpn23-recipes/1_clean_json.py

import json
import re
import os

from config import *

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: functions oben sind so nicht mehr funktional


def check_recipe(recipe) -> list:
    md = []
    if 'steps' in recipe:
        steps = recipe['steps']
    else:
        steps = []
        md.append("No steps?")
    for i, step in enumerate(steps, 1):
        md += check_ingredients(step)
    return md

def normalize_amount(a):
    if a == round(a):
        return round(a)
    else:
        return a

def check_ingredients(step):
    md = []
    for ingredient in step['ingredients']:
        i_amount = ingredient["amount"]
        i_amount = normalize_amount(i_amount)

        if not ingredient.get("unit"):
            i_unit = ""
            if i_amount:
                md.append("- Amount but no unit?")
        else:
            i_unit = ingredient["unit"]["name"]

        if i_unit and i_unit not in VALID_UNITS:
            md.append(f"- Invalid unit {i_unit}")

        if not ingredient.get("food"):
            md.append("- No food element in ingredient?")
            continue
    return md


def make_link(recipe):
    return f"[{recipe["name"]}]({TANDOOR_URL + "/view/recipe/" + str(recipe["id"])})"

def main():

    recipes = []
    for json_file in os.listdir(OUTDIR_JSON):
        with open(os.path.join(OUTDIR_JSON, json_file), 'r', encoding='utf-8') as f:
            data = json.load(f)
        recipes.append(data)

    for recipe in recipes:
        md = check_recipe(recipe)
        if not md:
            print(f"## 💚 {make_link(recipe)}")
        else:
            print(f"## 💔 {make_link(recipe)}")
            for line in md:
                print(line)

if __name__ == "__main__":
    main()