def delete_triple(triples: list,
                  complete_triples: List) -> List:

    new_triples = []
    triples = set(triples)
    for x in tqdm.tqdm(complete_triples, ncols=80):
        if x in triples:
            continue
        new_triples.append(x)

    return list(new_triples)


def fb5m_delete_triples_explicit(triples:List[Tuple], save_folder: str = "/home/qzheng/data/kb_bias/subgraph/ypd_fb5m") -> None:
    complete_triples = load_fb5m_triples()
    new_triples = delete_triple(triples, complete_triples)

    rel_id_dict = {}
    ent_id_dict = {}
    for x in tqdm.tqdm(new_triples, ncols=100):
        sub = x[0]
        rel = x[1]
        obj = x[2]
        if (obj.startswith("m.") or obj.startswith("g.")) is False: continue
        if sub not in ent_id_dict:
            ent_id_dict[sub] = len(ent_id_dict)

        if obj not in ent_id_dict:
            ent_id_dict[obj] = len(ent_id_dict)

        if rel not in rel_id_dict:
            rel_id_dict[rel] = len(rel_id_dict)

    if os.path.isdir(save_folder) is False:
        os.system("mkdir " + save_folder)

    with open(save_folder + "relation2id.txt", "w") as f1:
        f1.write(str(len(rel_id_dict)) + '\n')
        for x in tqdm.tqdm(rel_id_dict, ncols=100):
            f1.write(str(x) + "\t" + str(rel_id_dict[x]) + "\n")

    print("entitiy num {}".format(len(ent_id_dict)))
    with open(save_folder + "entity2id.txt", "w") as f1:
        f1.write(str(len(ent_id_dict)) + '\n')
        for x in tqdm.tqdm(ent_id_dict, ncols=100):
            f1.write(str(x) + "\t" + str(ent_id_dict[x]) + "\n")

    with open(save_folder + "train2id.txt", "w") as f1:
        f1.write(str(len(new_triples)) + '\n')
        random.shuffle(new_triples)
        for x in tqdm.tqdm(new_triples, ncols=100):
            sub = x[0]
            rel = x[1]
            obj = x[2]
            f1.write(str(ent_id_dict[sub]) + " " + str(ent_id_dict[obj]) + " " + str(rel_id_dict[rel]) + "\n")