False positive "Input should be a valid string" when empty string

[skewty]

This Code

from pydantic_xml import RootXmlModel

class SenderAddress(RootXmlModel[str], tag="address"):
    pass

class SenderName(RootXmlModel[str], tag="name"):
    pass

class SenderLocation(RootXmlModel[str], tag="location"):
    pass

class SenderDataDef(RootXmlModel, tag="senderdata"):
    root: list[tuple[SenderAddress | None, SenderName | None, SenderLocation | None]]

SenderDataDef.from_xml(b"""<?xml version="1.0" encoding="UTF-8"?>
<senderdata>
    <address>400</address>
    <name></name>                <!-- this empty string causes the issue -->
    <address>401</address>
    <name>Bob Andersen</name>
    <address>402</address>
    <name>Hansi</name>
    <address>403</address>
    <name>George Lucas</name>
    <address>404</address>
    <name>Michael Jensen</name>
    <address>406</address>
    <name>406</name>
    <address>407</address>
    <name>Fenger</name>
    <address>408</address>
    <name>408</name>
    <address>410</address>
    <name>410</name>
</senderdata>
""")

Raises

pydantic_core._pydantic_core.ValidationError: 1 validation error for SenderDataDef
0.1
  [line -1]: Input should be a valid string [type=string_type, input_value=PydanticUndefined, input_type=PydanticUndefinedType]

Also Tried

I tried to use:

from pydantic import model_validator

    @model_validator(mode="before")
    @classmethod
    def _accept_empty_strings(cls, data: Any) -> Any:
        # catch the None and change it to "" here
        return data

but it doesn't run until after the error above.

Any ideas? (I believe this is an lxml "feature")

[dapper91]

@skewty Hi,

The text of an empty element is parsed by the xml parser as None, not as an empty string:

>>> import lxml.etree
>>> root = lxml.etree.fromstring('<root><element></element></root>')
>>> root.find('element').text is None
True

but in your case SenderName is a non-optional string.

Therefore you should make it Optional:

class SenderName(RootXmlModel[Optional[str]], tag="name"):
    root: Optional[str] = None

or set a default value if you desire it to be a string anyway:

class SenderName(RootXmlModel[str], tag="name"):
    root: str = ''

[skewty]

Tried out your suggestion and came across this issue which was recently reported:

https://youtrack.jetbrains.com/issue/PY-70544/False-positive-UnionType-not-accepted-but-Optional-is.

Also, I see you are adding type hints to both the generic and root.

I saw, I believe in the docs, RootXmlModel being used without double typing as such:

class SenderName(RootXmlModel[Optional[str]], tag="name"):
    pass

Is providing the type for root field necessary?

This did the trick and solves my issues / makes it through my unit tests:

class SenderName(RootXmlModel[str], tag="name"):
    root: str = ''

class SenderDataDef(RootXmlModel, tag="senderdata"):
    root: list[tuple[SenderAddress | None, SenderName | None, SenderLocation | None]]

def test_request_system_info_many_from_xml():
    input_xml = Path("request-systeminfo-many.xml").read_bytes()
    request = models.SystemInfoRequest.from_xml(input_xml)
    assert len(request.sender_data) == 9
    assert request.sender_data[0].address == "400"
    assert request.sender_data[0].name == ""
    assert request.sender_data[-1].address == "410"
    assert request.sender_data[-1].name == "410"
    output_xml = request.to_xml(skip_empty=False)  # this must be false
    assert request == models.SystemInfoRequest.from_xml(output_xml)

Big thanks!

[dapper91]

I saw, I believe in the docs, RootXmlModel being used without double typing as such:

class SenderName(RootXmlModel[Optional[str]], tag="name"):
    pass

Is providing the type for root field necessary?

Yes, you have to provide a type for the root field, but you may omit it for the class itself to get rid of double typing. So this should work too:

class SenderName(RootXmlModel, tag="name"):
    root: Optional[str] = None