Cerberus 依赖项如何引用文档中较高的字段?

How can a Cerberus dependency reference a field higher up in the document?

我正在尝试为具有引用文档中较高字段的依赖项的文档创建架构。例如:

document = {
    'packages': {
        'some-package': {'version': 1}
    },
    'build-steps': {
        'needs-some-package': {'foo': 'bar'},
        'other-thing': {'funky': 'stuff'}
    }
}

我在这里苦苦挣扎的是加强 build-steps.needs-some-package 和 packages.some-package 之间的依赖关系。只要构建步骤包含 "needs-some-package",包就必须包含 "some-package".

当 "needs-some-package" 不存在时,不需要 "some-package"。所以这个文件也应该有效。

other_document = {
    'packages': {
        'other-package': {'version': 1}
    },
    'build-steps': {
        'other-thing': {'funky': 'stuff'}
    }
}

在合适的位置具有依赖关系的模式是

schema = {
    'packages': {
        'type': 'dict',
        'valueschema': {
            'type': 'dict'
        }
    },
    'build-steps': {
        'type': 'dict',
        'schema': {
            'needs-some-package': {
                'type': 'dict',
                'dependencies': 'packages.some-package'
            },
            'other-thing': {
                'type': 'dict'
            }
        }
    }
}

但这不起作用,因为 Cerberus 似乎正在 "build-steps" 下的子文档中寻找 "packages"。有什么办法可以上文档树吗?或者引用与文档根相关的内容?

描述的问题已在版本 1.0.2 中得到解决:

When a subdocument is processed the lookup for a field in question starts at the level of that document. In order to address the processed document as root level, the declaration has to start with a ^. An occurance of two initial carets (^^) is interpreted as a literal, single ^ with no special meaning.

示例代码:

import cerberus

schema = {
    'packages': {
        'type': 'dict',
        'valueschema': {
            'type': 'dict'
        }
    },
    'build-steps': {
        'type': 'dict',
        'schema': {
            'needs-some-package': {
                'type': 'dict',
                'dependencies': '^packages.some-package'
            },
            'other-thing': {
                'type': 'dict'
            }
        }
    }
}

document = {
    'packages': {
        'some-package': {'version': 1}
    },
    'build-steps': {
        'needs-some-package': {'foo': 'bar'},
        'other-thing': {'funky': 'stuff'}
    }
}

validator = cerberus.Validator(schema)
print(validator.validate(document))