使用 Cerberus 进行依赖关系验证

Dependencies validation using Cerberus

我正在使用 Cerberus 验证 CSV 文件,但正在为我假设的一些基本逻辑而苦苦挣扎

场景:

一个 CSV 文件有 2 列。仅当 Column 1 有值时,Column 2 才需要有值。如果 Column 1 为空,则 Column 2 也应为空。

我认为这将是编写的最直接的规则之一,但到目前为止,一切都没有按预期工作。

下面是使用 python 个字典的相同逻辑。

from cerberus import Validator
v = Validator()

schema = {
    "col1": {"required": False},
    "col2": {"required": True, "dependencies": "col1"},
}

document = {
    "col1": "a",
    "col2": ""
}

v.validate(document, schema)  # This responds with True!? Why?
v.errors
{}

我本以为 Column 2 会出现错误,因为已经提供了 Column 1,但这里的结果是 True,意思是没有错误

我检查了 raised issues on github 但似乎找不到任何明显的解决方案。

Note
The evaluation of this rule (dependencies) does not consider any constraints defined with the required rule.

无论 "required" 是什么:

from cerberus import Validator
v = Validator()

document = {
    "col1": "a",
    "col2": ""
}

schema = {
    "col1": {"required": False},
    "col2": {"required": True, "dependencies": "col1"},
}

print(v.validate(document, schema))  # True
print(v.errors) # {}

schema = {
    "col1": {"required": True},
    "col2": {"required": True, "dependencies": "col1"},
}


print(v.validate(document, schema))  # True
print(v.errors)  # {}

schema = {
    "col1": {"required": True},
    "col2": {"required": False, "dependencies": "col1"},
}


print(v.validate(document, schema))  # True
print(v.errors)  # {}

http://docs.python-cerberus.org/en/stable/validation-rules.html#dependencies


更新:

针对您的情况“如果 col1 中有值,则强制要求 col2 的解决方案。”。
要应用复杂的规则 - 创建自定义 Validator,如下所示:

from cerberus import Validator


class MyValidator(Validator):
    def _validate_depends_on_col1(self, depends_on_col1, field, value):
        """ Test if a field value is set depending on `col1` field value.
        """
        if depends_on_col1 and self.document.get('col1', None) and not value:
            self._error(field, f"`{field}` cannot be empty given that `col1` has a value")


v = MyValidator()

schema = {
    "col1": {"required": False},
    "col2": {"required": True, "depends_on_col1": True},
}

print(v.validate({"col1": "a", "col2": ""}, schema))  # False
print(v.errors) # {'col2': ['`col2` cannot be empty given that `col1` has a value']}

print(v.validate({"col1": "", "col2": ""}, schema))  # True
print(v.errors) # {}

print(v.validate({"col1": 0, "col2": "aaa"}, schema))  # True
print(v.errors) # {}

请注意,您需要 运行 约定哪些列 col1 值应被视为空值(以调整自定义验证器规则)。


指定"dependancy"字段名称的扩展版本:

class MyValidator(Validator):
    def _validate_depends_on_col(self, col_name, field, value):
        """ Test if a field value is set depending on `col_name` field value.
        """
        if col_name and self.document.get(col_name, None) and not value:
            self._error(field, f"`{field}` cannot be empty given that `{col_name}` has a value")


v = MyValidator()

document = {"col1": "a", "col2": ""}

schema = {
    "col1": {"required": False},
    "col2": {"required": True, "depends_on_col": "col1"},
}

http://docs.python-cerberus.org/en/stable/customize.html

假设您将 csv 输入转换为文档列表,您可以先对文档进行预处理以删除空的 col2 字段:

for document in documents:
    if not document["col2"]:
        document.pop("col2")

那么这个模式就可以完成工作:

{"col1": {
    "oneof": [
        {"empty": True},
        {"empty": False, "dependencies": "col2"}
    ]
}}

请注意,dependenciesrequired 规则不考虑字段的值,而只考虑文档中字段的存在。