中国古诗词语料

古诗词语料可以用来进行相关古诗词方面的语言理解和信息挖掘~

#! -*- coding: utf-8 -*-
import os json sqlite3

def check_json（f）:
    filepath = os.path.join（‘./json‘ f）
    with open（filepath） as file:
        try:
            _ = json.loads（file.read（））
            return True
        except:
            assert False u“校验（%s）失败“ % f


def test_json（）:
    “““
        测试古诗JSON文件是否有效
    “““
    map（check_json os.listdir（‘./json‘））



def test_sqlite（）:
    “““
        测试ci数据库文件是否有效
    “““
    conn = sqlite3.connect（‘./ci/ci.db‘）

    c = conn.cursor（）

    c.execute（“SELECT name FROM sqlite_master WHERE type=‘table‘“）

    tables = c.fetchall（）
    
    assert len（tables） == 2 u“Sqlite文件异常“

 属性            大小     日期    时间   名称
----------- ---------  ---------- -----  ----
     目录           0  2017-10-26 01:33  chinese-poetry-master\
     文件          95  2017-10-26 01:33  chinese-poetry-master\.travis.yml
     文件        1076  2017-10-26 01:33  chinese-poetry-master\LICENSE
     文件        4943  2017-10-26 01:33  chinese-poetry-master\README.md
     目录           0  2017-10-26 01:33  chinese-poetry-master\ci\
     文件         574  2017-10-26 01:33  chinese-poetry-master\ci\README.md
     文件     5779456  2017-10-26 01:33  chinese-poetry-master\ci\ci.db
     目录           0  2017-10-26 01:33  chinese-poetry-master\images\
     文件       26002  2017-10-26 01:33  chinese-poetry-master\images\WechatIMG1.jpeg
     文件      200735  2017-10-26 01:33  chinese-poetry-master\images\ci_author_topK.png
     文件     1521043  2017-10-26 01:33  chinese-poetry-master\images\ci_rhythmic_topK.png
     文件      180750  2017-10-26 01:33  chinese-poetry-master\images\ci_words_topK.png
     文件      495776  2017-10-26 01:33  chinese-poetry-master\images\full-tang-poetry.png
     文件      253331  2017-10-26 01:33  chinese-poetry-master\images\song_author_topK.png
     文件      221835  2017-10-26 01:33  chinese-poetry-master\images\song_text_topK.png
     文件      247971  2017-10-26 01:33  chinese-poetry-master\images\tang_author_topK.png
     文件      241416  2017-10-26 01:33  chinese-poetry-master\images\tang_text_topK.png
     目录           0  2017-10-26 01:33  chinese-poetry-master\json\
     文件     3084922  2017-10-26 01:33  chinese-poetry-master\json\authors.song.json
     文件      672785  2017-10-26 01:33  chinese-poetry-master\json\authors.tang.json
     文件      533535  2017-10-26 01:33  chinese-poetry-master\json\poet.song.0.json
     文件      496150  2017-10-26 01:33  chinese-poetry-master\json\poet.song.1000.json
     文件      610051  2017-10-26 01:33  chinese-poetry-master\json\poet.song.10000.json
     文件      617841  2017-10-26 01:33  chinese-poetry-master\json\poet.song.100000.json
     文件      612167  2017-10-26 01:33  chinese-poetry-master\json\poet.song.101000.json
     文件      568337  2017-10-26 01:33  chinese-poetry-master\json\poet.song.102000.json
     文件      528829  2017-10-26 01:33  chinese-poetry-master\json\poet.song.103000.json
     文件      602579  2017-10-26 01:33  chinese-poetry-master\json\poet.song.104000.json
     文件      708665  2017-10-26 01:33  chinese-poetry-master\json\poet.song.105000.json
     文件      440301  2017-10-26 01:33  chinese-poetry-master\json\poet.song.106000.json
     文件      576187  2017-10-26 01:33  chinese-poetry-master\json\poet.song.107000.json
............此处省略304个文件信息