api("com.thoughtworks.qdox:qdox:2.0.3")
        api("com.uwyn:jhighlight:1.0")
        api("com.vladsch.flexmark:flexmark-all:0.34.60") {
            because("Higher versions tested are either incompatible (0.62.2) or bring additional unwanted dependencies (0.36.8)")
        }
        api("org.apache.pdfbox:pdfbox:2.0.24") {
            because("Flexmark 0.34.60 brings in a vulnerable version of pdfbox")
        }

# This is the same as the official TensorFlow builds.
# See https://developer.nvidia.com/cuda-gpus#compute
# `compute_XY` enables PTX embedding in addition to SASS. PTX
# is forward compatible beyond the current compute capability major
# release while SASS is only forward compatible inside the current
# major release. Example: sm_80 kernels can run on sm_89 GPUs but
# not on sm_90 GPUs. compute_80 kernels though can also run on sm_90 GPUs.

        Any next read or write will be done from that position.

        To be awaitable, compatible with async, this is run in threadpool.
        """
        return await super().seek(offset)

    async def close(self) -> None:
        """
        Close the file.

        To be awaitable, compatible with async, this is run in threadpool.
        """
        return await super().close()

from pydantic import BaseModel
from pydantic.version import VERSION as PYDANTIC_VERSION
from starlette.datastructures import UploadFile
from typing_extensions import get_args, get_origin

# Copy from Pydantic v2, compatible with v1
if sys.version_info < (3, 10):
    WithArgsTypes: tuple[Any, ...] = (typing._GenericAlias, types.GenericAlias)  # type: ignore[attr-defined]
else:
    WithArgsTypes: tuple[Any, ...] = (

    You can also use it yourself, for example to convert objects before saving them
    in a database that supports only JSON.

    Read more about it in the
    [FastAPI docs for JSON Compatible Encoder](https://fastapi.tiangolo.com/tutorial/encoder/).
    """
    custom_encoder = custom_encoder or {}
    if custom_encoder:
        if type(obj) in custom_encoder:

# GraphQL { #graphql }

As **FastAPI** is based on the **ASGI** standard, it's very easy to integrate any **GraphQL** library also compatible with ASGI.

You can combine normal FastAPI *path operations* with GraphQL on the same application.

/// tip

**GraphQL** solves some very specific use cases.

It has **advantages** and **disadvantages** when compared to common **web APIs**.

# GraphQL { #graphql }

Como **FastAPI** se basa en el estándar **ASGI**, es muy fácil integrar cualquier paquete de **GraphQL** que también sea compatible con ASGI.

Puedes combinar las *path operations* normales de FastAPI con GraphQL en la misma aplicación.

/// tip | Consejo

**GraphQL** resuelve algunos casos de uso muy específicos.

Tiene **ventajas** y **desventajas** en comparación con las **APIs web** comunes.

        runBinaryCompatibilityCheckWithFailure(v1, v2) {
            block()
        }

    private
    fun CheckResult.assertBinaryCompatible() {
        assertTrue(richReport.toAssertionMessage("Expected to be compatible but the check failed"), isBinaryCompatible)
    }

    private
    fun CheckResult.assertNotBinaryCompatible() {

@lru_cache
def get_cached_model_fields(model: type[BaseModel]) -> list[ModelField]:
    return get_model_fields(model)  # type: ignore[return-value]


# Duplicate of several schema functions from Pydantic v1 to make them compatible with
# Pydantic v2 and allow mixing the models

TypeModelOrEnum = Union[type["BaseModel"], type[Enum]]
TypeModelSet = set[TypeModelOrEnum]


def normalize_name(name: str) -> str:

        * `Path`
        * `Query`
        * ...as these are compatible with the WebSockets protocol (e.g. `Body` is not).
    * [Updated documentation for WebSockets](https://fastapi.tiangolo.com/advanced/websockets/).
    * PR [#178](https://github.com/tiangolo/fastapi/pull/178) by [@jekirl](https://github.com/jekirl).

* Upgrade the compatible version of Pydantic to `0.26.0`.