data_quality_server/tables/fields/content.py

"""
正文字段检查
"""


class ContentChecker(object):
    """
    正文检查
    """
    errors_tables = {
        "01": {"name": "正文长度异常",
               "code": "01",
               "rules": {
                   "01": {
                       "name": "低于200字",
                       "code": "01",
                       "checkFn": "lt200",
                   },
               },
               },
        "02": {"name": "中文占比异常",
               "code": "02",
               "rules": {
                   "01": {
                       "name": "中文字符占比低于30%",
                       "code": "01",
                       "checkFn": "chineseCharLt30",
                   },
               },
               },
    }

    def __init__(self):
        pass

    @staticmethod
    def lt200(content: str) -> bool:
        return len(content) < 200

    @staticmethod
    def chineseCharLt30(content: str) -> bool:
        chinese_char_count = sum(list(map(lambda x: 1 if '\u4e00' <= x <= '\u9fef' else 0, content)))
        return float(chinese_char_count) / float(len(content)) < 0.3