)
    return model_fields


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


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

TypeModelOrEnum = type["BaseModel"] | type[Enum]
TypeModelSet = set[TypeModelOrEnum]

// (case-sensitive) and will fail otherwise.
// Further, some clients require that the ETag value is a double-quoted
// string. Therefore, this package provides dedicated functions for
// adding and extracting the ETag to/from HTTP headers.
package etag

import (
	"bytes"
	"crypto/hmac"
	"crypto/md5"
	"encoding/base64"
	"encoding/hex"
	"errors"
	"fmt"
	"net/http"

      debugData = source.readByteString(opaqueDataLength.toLong())
    }
    handler.goAway(lastStreamId, errorCode, debugData)
  }

  /** Unlike other `readXxx()` functions, this one must log the frame before returning. */
  @Throws(IOException::class)
  private fun readWindowUpdate(
    handler: Handler,
    length: Int,
    flags: Int,
    streamId: Int,
  ) {

Let's continue from the previous example. We wanted to **annotate the function with one type**, but we wanted to be able to return from the function something that actually includes **more data**.

We want FastAPI to keep **filtering** the data using the response model. So that even though the function returns more data, the response will only include the fields declared in the response model.

      vspace, target_tensor_ids, source_tensor_ids, sources_that_are_targets,
      output_gradients, result, /*build_default_zeros_grads*/ false));
  return absl::OkStatus();
}

// Helper functions which delegate to `AbstractOperation`, update
// the state of the ForwardOperation and call the tape as appropriate.
// These APIs are mainly to facilitate testing and are subject to change.
namespace internal {

import java.util.Iterator;
import java.util.Map;
import java.util.Set;
import java.util.SortedMap;
import java.util.Spliterator;
import java.util.Spliterators;
import java.util.function.BiFunction;
import java.util.function.BinaryOperator;
import java.util.function.Function;
import java.util.stream.Collector;
import java.util.stream.Collectors;
import org.jspecify.annotations.Nullable;

/**

// TFE_Execute by creating an op with the same name as the function.
TF_CAPI_EXPORT extern void TFE_ContextAddFunction(TFE_Context* ctx,
                                                  TF_Function* function,
                                                  TF_Status* status);

// Removes a function from the context. Once removed, you can no longer
// TFE_Execute it or TFE_Execute any TFE_Op which has it as an attribute or any

/// warning

你大概可以跳過這段。

這裡是有關 FastAPI 內部技術細節。

如果你有相當多的技術背景（例如協程、執行緒、阻塞等），並且對 FastAPI 如何處理 `async def` 與常規 `def` 感到好奇，請繼續閱讀。

///

### 路徑操作函式 { #path-operation-functions }

當你使用 `def` 而不是 `async def` 宣告*路徑操作函式*時，該函式會在外部的執行緒池（threadpool）中執行，然後等待結果，而不是直接呼叫（因為這樣會阻塞伺服器）。

/// warning | 警告

你可以跳过这里。

这些都是 FastAPI 如何在内部工作的技术细节。

如果你有相当多的技术知识（协程、线程、阻塞等），并且对 FastAPI 如何处理 `async def` 与常规 `def` 感到好奇，请继续。

///

### 路径操作函数 { #path-operation-functions }

当你使用 `def` 而不是 `async def` 来声明一个*路径操作函数*时，它运行在外部的线程池中并等待其结果，而不是直接调用（因为它会阻塞服务器）。

// name of the function handler passed (this handler must be a method of
// `adminAPIHandlers`).
//
// - sets up call to send AuditLog
//
// While this is a middleware function (i.e. it takes a handler function and
// returns one), due to flags being passed based on required conditions, it is
// done per-"handler function registration" in the router.
//