gpn23-recipes/1_clean_json.py

import json
import re
import os

from config import *


def clean_recipe(recipe):
    # remove useless ever-changing fields
    for keyword in recipe["keywords"]:
        keyword.pop("updated_at", None)

    # remove buggy food properties
    recipe.pop("food_properties", None)
    for step in recipe["steps"]:
        for ingredient in step["ingredients"]:
            ingredient["food"].pop("properties", None)

    # sort conversions
    recipe.pop("food_properties", None)
    for step in recipe["steps"]:
        for ingredient in step["ingredients"]:
            ingredient["conversions"] = sorted(ingredient["conversions"], key=lambda c: c["unit"])

    return recipe


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():

    # Clean

    for json_file in os.listdir(OUTDIR_JSON):
        with open(os.path.join(OUTDIR_JSON, json_file), "r", encoding="utf-8") as f:
            recipe = json.load(f)
        recipe = clean_recipe(recipe)
        with open(os.path.join(OUTDIR_JSON, json_file), "w", encoding="utf-8") as f:
            json.dump(recipe, f, sort_keys=True, indent=2)

    # Read all and check

    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()