使用 pandas 模式的 Python 列验证

Question

我正在尝试使用 PandasSchema 验证我的 DataFrame 库。我一直在验证某些列，例如：

1.ip_address- 应包含格式为 1.1.1.1 的 ip 地址，或者如果任何其他值应引发错误，则应为 null。 2.initial_date- 格式 yyyy-mm-dd h:m:s 或 mm-dd-yyyy h:m:s 等。 3.customertype 应该是 in['type1','type2','type3'] 其他任何东西都会引发错误。 4.客户满意=是/否或空白 5.customerid 不应超过 5 个字符，例如 cus01,cus02 6.time 应为 %:%: 格式或 h:m:s 格式，否则会引发异常。

from pandas_schema import Column, Schema
def check_string(sr):
    try:
        str(sr)
    except InvalidOperation:
        return False
    return True
def check_datetime(self,dec):
        try:
            datetime.datetime.strptime(dec, self.date_format)
            return True
        except:
            return False
def check_int(num):
    try:
        int(num)
    except ValueError:
        return False
    return True
    

string_validation=[CustomElementValidation(lambda x: check_string(x).str.len()>5 ,'Field Correct')]
int_validation = [CustomElementValidation(lambda i: check_int(i), 'is not integer')]
contain_validation = [CustomElementValidation(lambda y: check_string(y) not in['type1','type2','type3'], 'Filed is correct')]
date_time_validation=[CustomElementValidation(lambda dt: check_datetime(dt).strptime('%m/%d/%Y %H:%M %p'),'is not a date
 time')]
null_validation = [CustomElementValidation(lambda d: d is not np.nan, 'this field cannot be null')]

schema = Schema([
                 Column('CompanyID', string_validation + null_validation),
                 Column('initialdate', date_time_validation),
                 Column('customertype', contain_validation),
                 Column('ip', string_validation),
                 Column('customersatisfied', string_validation)])
errors = schema.validate(combined_df)
errors_index_rows = [e.row for e in errors]
pd.DataFrame({'col':errors}).to_csv('errors.csv')

Answer 1

我刚刚查看了 PandasShema 的文档，并且大多数（如果不是全部）您正在寻找它的开箱即用功能。看看：

• InListValidation
• IsDtypeValidation
• DateFormatValidation
• MatchesPatternValidation

作为解决您的问题的快速尝试，类似的方法应该可以：

from pandas_schema.validation import (
    InListValidation
    ,IsDtypeValidation
    ,DateFormatValidation
    ,MatchesPatternValidation
)

schema = Schema([
    # Match a string of length between 1 and 5
    Column('CompanyID', [MatchesPatternValidation(r".{1,5}")]),

    # Match a date-like string of ISO 8601 format (https://www.iso.org/iso-8601-date-and-time-format.html)
    Column('initialdate', [DateFormatValidation("%Y-%m-%d %H:%M:%S")], allow_empty=True),
    
    # Match only strings in the following list
    Column('customertype', [InListValidation(["type1", "type2", "type3"])]),

    # Match an IP address RegEx (https://www.oreilly.com/library/view/regular-expressions-cookbook/9780596802837/ch07s16.html)
    Column('ip', [MatchesPatternValidation(r"(?:[0-9]{1,3}\.){3}[0-9]{1,3}")]),

    # Match only strings in the following list    
    Column('customersatisfied', [InListValidation(["yes", "no"])], allow_empty=True)
])