```Python hl_lines="30-31" title="app/routers/items.py"
{!../../docs_src/bigger_applications/app/routers/items.py!}
```

/// tip

最后的这个路径操作将包含标签的组合：`["items"，"custom"]`。

并且在文档中也会有两个响应，一个用于 `404`，一个用于 `403`。

///

## `FastAPI` 主体

现在，让我们来看看位于 `app/main.py` 的模块。

在这里你导入并使用 `FastAPI` 类。

这将是你的应用程序中将所有内容联结在一起的主文件。

并且由于你的大部分逻辑现在都存在于其自己的特定模块中，因此主文件的内容将非常简单。

			// whether you also have the s3:ListBucket
			// permission.
			// * If you have the s3:ListBucket permission
			//   on the bucket, Amazon S3 will return an
			//   HTTP status code 404 ("no such key")
			//   error.
			// * if you don’t have the s3:ListBucket
			//   permission, Amazon S3 will return an HTTP
			//   status code 403 ("access denied") error.`

- The 404 request handler for the GCE Ingress load balancer now exports prometheus metrics, including:

  - `http_404_request_total` (the number of 404 requests handled)
  - `http_404_request_duration_ms` (the amount of time the server took to respond in ms)

// This operation is useful to determine if a bucket exists.
// The operation returns a 200 OK if the bucket exists and you
// have permission to access it. Otherwise, the operation might
// return responses such as 404 Not Found and 403 Forbidden.
func (api objectAPIHandlers) HeadBucketHandler(w http.ResponseWriter, r *http.Request) {
	ctx := newContext(r, w, "HeadBucket")

	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))

		},

		// Test case - 8.
		// Test case with non-existent source file.
		// Case for the purpose of failing `api.ObjectAPI.GetObjectInfo`.
		// Expecting the response status code to http.StatusNotFound (404).
		{
			bucketName:       bucketName,
			uploadID:         uploadID,
			copySourceHeader: url.QueryEscape(SlashSeparator + bucketName + SlashSeparator + "non-existent-object"),

	go stats.trackEWMA()
}

type proxyResult struct {
	Proxy bool
	Err   error
}

// get Reader from replication target if active-active replication is in place and
// this node returns a 404
func proxyGetToReplicationTarget(ctx context.Context, bucket, object string, rs *HTTPRangeSpec, h http.Header, opts ObjectOptions, proxyTargets *madmin.BucketTargets) (gr *GetObjectReader, proxy proxyResult, err error) {

    Nullability was previously ambiguous and lenient but now the compiler will
    enforce strict null checks.

 *  New: The response message is now non-null. This is the "Not Found" in the
    status line "HTTP 404 Not Found". If you are building responses
    programmatically (with `new Response.Builder()`) you must now always supply
    a message. An empty string `""` is permitted. This value was never null on

    deprecated=True,
)
router2_default = APIRouter()
router4_override = APIRouter(
    prefix="/level4",
    tags=["level4a", "level4b"],
    dependencies=[Depends(dep4)],
    responses={
        404: {"description": "Client error level 4"},
        504: {"description": "Server error level 4"},
    },
    default_response_class=ResponseLevel4,
    callbacks=callback_router4.routes,
    deprecated=True,
)

		// of an object to check for object existence. For versioned buckets, MinIO's non-recursive
		// call will report top level prefixes in deleted state, whereas spark/hadoop interpret this as non-empty
		// and throw a 404 exception. This is especially a problem for spark jobs overwriting the same partition
		// repeatedly. This workaround recursively lists the top 3 entries including delete markers to reflect the

    assertCached(true, 307)
    assertCached(true, 308)
    assertCached(false, 400)
    assertCached(false, 401)
    assertCached(false, 402)
    assertCached(false, 403)
    assertCached(true, 404)
    assertCached(true, 405)
    assertCached(false, 406)
    assertCached(false, 408)
    assertCached(false, 409)
    // the HTTP spec permits caching 410s, but the RI doesn't.
    assertCached(true, 410)