jsoncompare.py

from __future__ import annotations

import json
import sys
from pathlib import Path
from typing import Any
from urllib.request import urlopen

TYPE = "TYPE"
PATH = "PATH"
VALUE = "VALUE"


class Diff:
    def __init__(self, first: Any, second: Any, with_values: bool = False):
        self.difference: list[tuple[str, str]] = []
        self.check(first, second, with_values=with_values)

    def check(self, first: Any, second: Any, path: str = "", with_values: bool = False) -> None:
        if with_values and second is not None and not isinstance(first, type(second)):
            self.save_diff(f"{path} - {type(first).__name__}, {type(second).__name__}", TYPE)
            return

        if isinstance(first, dict):
            for key, value in first.items():
                new_path = key if not path else f"{path}.{key}"
                if isinstance(second, dict) and key in second:
                    self.check(value, second[key], path=new_path, with_values=with_values)
                else:
                    self.save_diff(new_path, PATH)
            return

        if isinstance(first, list):
            for index, item in enumerate(first):
                new_path = f"{path}[{index}]"
                sec = second[index] if isinstance(second, list) and index < len(second) else None
                if sec is None and second is not None:
                    self.save_diff(f"{new_path} - {type(item).__name__}", TYPE)
                self.check(item, sec, path=new_path, with_values=with_values)
            return

        if with_values and second is not None and first != second:
            self.save_diff(f"{path} - {first} | {second}", VALUE)

    def save_diff(self, diff_message: str, diff_type: str) -> None:
        item = (diff_type, diff_message)
        if item not in self.difference:
            self.difference.append(item)


def get_content_from_uri(uri: str) -> bytes:
    with urlopen(uri) as response:
        return response.read()


def get_content_from_file(file_path: str) -> str:
    return Path(file_path).read_text(encoding="utf-8")


def get_content(location: str | dict[str, Any]) -> Any:
    if isinstance(location, dict):
        return location
    content = get_content_from_uri(location) if location.startswith("http") else get_content_from_file(location)
    return json.loads(content)


def compare(location1: str, location2: str) -> list[dict[str, str]]:
    json1 = get_content(location1)
    json2 = get_content(location2)
    diff1 = Diff(json1, json2, True).difference
    diff2 = Diff(json2, json1, False).difference
    diffs: list[dict[str, str]] = []

    for diff_type, message in diff1:
        mapped = "CHANGED"
        if diff_type == PATH:
            mapped = "REMOVED"
        diffs.append({"type": mapped, "message": message})

    for _, message in diff2:
        diffs.append({"type": "ADDED", "message": message})

    return diffs


if __name__ == "__main__":
    if len(sys.argv) != 3:
        sys.exit("Usage: python jsoncompare.py <file-or-url-1> <file-or-url-2>")

    location1 = sys.argv[1]
    location2 = sys.argv[2]
    diffs = compare(location1, location2)
    if diffs:
        print(f"\nFound differences comparing {location1} and {location2}")
    for diff in diffs:
        print(f"{diff['type']}: {diff['message']}")