Function main

tools/tools.py:332–355

()

                print(f"处理进度: {progress}")

def main():
    # infile = '/data/H-RAG/H_RAG/data/kg/wikidata-20240101-all.json'
    # outfile = '/data/H-RAG/H_RAG/data/kg/extracted.json'
    # data1 = "/data/H-RAG/H_RAG/data/kg/triples.txt"
    # 示例用法
    # input_path = "CN-DBpediaMention2Entity.txt"  # 替换成你的已排序输入文件路径
    # temp1 = "temp1.txt"
    # temp2 = "temp2.txt"
    # output_path = "CN-DBpedia_pass.txt"  # 替换成你想要输出的文件路径
    # remove_duplicates('triple_data/wiki_entities_sorted.txt', output_file = 'triple_data/wiki_entities.txt')
    # remove_duplicates('triple_data/wiki_aliases_sorted.txt', output_file = 'triple_data/wiki_aliases.txt')
    # remove_duplicates('triple_data/wiki_descriptions_sorted.txt', output_file = 'triple_data/wiki_descriptions.txt')
    # str_full_to_half_width(temp1, temp2)
    # remove_duplicates(temp2, output_path)
    # file_split('CN-DBpedia_pass.txt', 4)
    # print_file('/data/H-RAG/H_RAG/data/kg/triples.txt', 1000)
    # count_lines('/data/H-RAG/H_RAG/data/kg/triples.txt')
    # get_entities_from_triples("triples.txt", "triples_entities.txt", ['>', '<'])
    # get_entities_from_wikidata('/data/H-RAG/H_RAG/data/kg/extracted.json', 'wiki_entities.txt', 'wiki_aliases.txt', 'wiki_descriptions.txt')
    # yago-beyond-wikipedia, yago-facts, yago-schema, yago-taxonomy, yago-schema
    # get_ttl('/data/H-RAG/H_RAG/data/kg/yago-schema.ttl', 'yago_wiki_triples.txt')
    # json_pass(infile, outfile)
    # remove_text_in_brackets('triple_data/dbpedia_entities.txt', 'triple_data/dbpedia_entities_clean.txt', True)
    remove_duplicates('triple_data/temp.txt', 'triple_data/dbpedia_entities_clean.txt')


if __name__ == "__main__":