Find duplicates in your bookmarks

find_dup_bookmarks.py

# -*- coding: utf-8 -*-

"""
Find duplicates in your bookmarks.

Usage:
    python find_dup_bookmarks.py /path/to/your/Bookmarks

Chrome saves your bookmarks in a json file.
For macOS, it is `/Users/{your user name}/Libaray/Application Support/Chrome/Default/Bookmarks`.
For other OS, use google to find out.
"""

import sys
import json


def get_urls_with_paths(bookmark: dict, parent_path: str, res: dict[str, list[str]]):
    if bookmark["type"] == "folder":
        for item in bookmark["children"]:
            get_urls_with_paths(item, parent_path + "/" + bookmark["name"], res)
    elif bookmark["type"] == "url":
        url = bookmark["url"]
        path = parent_path + "/" + bookmark["name"] + ""
        if url in res:
            res[url].append(path)
        else:
            res[url] = [path]
    else:
        print(f"unknown type: {bookmark['type']}")


def print_duplicated_items(items: list[tuple[str, list[str]]]):
    unique = len(items)
    total = sum(len(v) for _, v in items)
    if unique == 0:
        print("Congrats! You have no duplication in your bookmarks!")
    else:
        print(
            "There are {} duplicated items and it will go down to {} after deduplication. Please reivew:".format(
                total, unique
            )
        )
        for n1, (url, paths) in enumerate(items):
            print("[{}] {}".format(n1 + 1, url))
            for n2, path in enumerate(paths):
                print("\t{}) {}".format(n2 + 1, path))


def main(path):
    with open(path) as f:
        bookmarks = json.load(f)

    res = {}
    for bookmark in bookmarks["roots"].values():
        get_urls_with_paths(bookmark, "", res)

    duplicated = list(filter(lambda x: len(x[1]) > 1, res.items()))
    print_duplicated_items(duplicated)


if __name__ == "__main__":
    # you can use argparse
    # I just want a simple script... so no error checking
    path = sys.argv[1]
    main(path)