douban.py

import requests

API = "https://api.wmdb.tv/api/v1/top?type={movie_type}&skip=0&limit={limit}"


def get_top_movies(movie_type, limit=250):
    r = requests.get(API.format(movie_type=movie_type, limit=limit))
    if not r.ok:
        raise Exception(f"Can not get movies for {movie_type} type")
    return r.json()


def main():
    douban_movies = get_top_movies("Douban")
    douban_movies_imdb_ids = [d.get("imdbId") for d in douban_movies]
    douban_movies_ids = [d.get("doubanId") for d in douban_movies]
    imdb_movies = get_top_movies("Imdb")
    imdb_movies_douban_ids = [i.get("doubanId") for i in imdb_movies]
    imdb_movies_ids = [i.get("imdbId") for i in imdb_movies]
    print(len(set(douban_movies_ids) - set(imdb_movies_douban_ids)))
    print(len(set(imdb_movies_ids) - set(douban_movies_imdb_ids)))
    i = 0
    j = 0
    same_rating_list = []
    close_rating_list = []
    max_diff_list = []
    max_diff = 0
    for d in douban_movies:
        if d.get("imdbId") in imdb_movies_ids:
            if d.get("imdbRating") == d.get("doubanRating"):
                i += 1
                same_rating_list.append(
                    {
                        "name": d.get("data")[0]["name"],
                        "doubanRating": d.get("doubanRating"),
                        "imdbRating": d.get("imdbRating"),
                    }
                )
            else:
                diff = float(d.get("imdbRating")) - float(d.get("doubanRating"))
                abs_diff = round(abs(diff), 2)
                print(abs_diff)
                max_diff = max(abs_diff, max_diff)
                if abs_diff >= 1.0:
                    max_diff_list.append(
                        {
                            "name": d.get("data")[0]["name"],
                            "doubanRating": d.get("doubanRating"),
                            "imdbRating": d.get("imdbRating"),
                        }
                    )
                if abs_diff <= 0.2:
                    close_rating_list.append(
                        {
                            "name": d.get("data")[0]["name"],
                            "doubanRating": d.get("doubanRating"),
                            "imdbRating": d.get("imdbRating"),
                        }
                    )
                    j += 1
    # print(len(same_rating_list), len(close_rating_list), len(max_diff_list))
    # print(f"Total {i} in same rating")
    # print(f"Total {j} in diff 0.1 rating")
    print("Have Same Rating:", same_rating_list)
    print("Have close Rating:", close_rating_list)
    print("Have different Rating:", max_diff_list)


if __name__ == "__main__":
    main()