Consider adding a class_name_generator option

[loopj]

Currently, we have support for configuring an element_name_generator and attribute_name_generator function in models metadata or in XmlContext.

This is fantastic for reducing boilerplate when generating element names based on field names of a class.

In the following examples, I'm using PEP8 style field names (snake_case) and PEP8 style class names (PascalCase), and generating XML which uses PascalCase for element names.

xml_context = XmlContext(element_name_generator=pascal_case)
serializer = XmlSerializer(context=xml_context)

@dataclass
class Something:
    my_field: int

serializer.render(Something(my_field=123))

This generates the expected XML:

<Something>
    <MyField>123</MyField>
</Something>

Since element_name_generator is also used for generating the local_name of a class when none is explicitly provided, that causes some interesting behavior:

xml_context = XmlContext(element_name_generator=pascal_case)
serializer = XmlSerializer(context=xml_context)

@dataclass
class IPAddress:
    val: str

serializer.render(IPAddress(val="192.168.1.1"))

In this case, the (already pascal case) class name is being run through the pascal_case element_name_generator, which is clobbering the case of the class from IPAddress to Ipaddress:

<Ipaddress>
    <Val>192.168.1.1</Val>
</Ipaddress>

I'm aware that this can be solved by specifying explicit names in either the field metadata or the class metadata, but I think it would be nice to have a way to specify a class_name_generator in the XmlContext to avoid this issue.

For example:

context = XmlContext(
    element_name_generator=pascal_case,
    class_name_generator=return_input,
)

serializer = XmlSerializer(context=context)

Under the hood we could have class_name_generator default to element_name_generator if not explicitly provided, to maintain the current behavior.

Let me know what you think, and I can try to whip up a PR if you're interested in supporting this.

[loopj]

For reference, my current workaround is to define the following:

def preserving_pascal_case(name: str) -> str:
    """Convert a field/class name to PascalCase, preserving existing PascalCase names."""
    if "_" in name or name.islower():
        return pascal_case(name)
    else:
        return name