wikipedia.py 文件源码

def download(corpus_dir, out_file, lang, date=DEFAULT_DATE):
    data_dir = corpus_dir
    if exists(join(data_dir, out_file)):
        return

    url = DUMP_URL.format(lang=lang, date=date)
    local("mkdir -p {dir}".format(dir=data_dir))
    with lcd(data_dir):
        local("wget {}".format(url))
        local("mv {} {}".format(DUMP_FILE.format(lang=lang, date=date), out_file))


# @task