nv_ingest_api.util.service_clients.rest package#

Submodules#

nv_ingest_api.util.service_clients.rest.rest_client module#

class nv_ingest_api.util.service_clients.rest.rest_client.RestClient(

host: str,

port: int,

max_retries: int = 0,

max_backoff: int = 32,

default_connect_timeout: float = 300.0,

default_read_timeout: float | None = None,

http_allocator: Callable[[], Any] | None = None,

**kwargs,

)[source]#

Bases: MessageBrokerClientBase

A client for interfacing with an HTTP endpoint (e.g., nv-ingest), providing mechanisms for sending and receiving messages with retry logic using the requests library by default, but allowing a custom HTTP client allocator.

Extends MessageBrokerClientBase for interface compatibility.

fetch_message( job_id: str, timeout: float | Tuple[float, float] | None = None, ) → ResponseSchema[source]#

Fetches a job result message from the server’s fetch endpoint.

Handles retries for connection errors and non-terminal HTTP errors based on the max_retries configuration. Specific HTTP statuses are treated as immediate failures (terminal) or as job not ready (HTTP 202).

Parameters:

job_id (str) – The server-assigned identifier of the job to fetch.
timeout (float or tuple of float, optional) – Specific timeout override for this request.

Returns:

response_code = 0: Success (HTTP 200) with the job result.
response_code = 1: Terminal failure (e.g., 404, 400, 5xx, or max retries exceeded).
response_code = 2: Job not ready (HTTP 202).

Return type:

ResponseSchema

Raises:

TypeError – If the configured client does not support the required HTTP GET method.

get_client() → Any[source]#

Returns the underlying HTTP client instance.

Returns:: The active HTTP client instance.
Return type:: Any

property max_retries: int#

Maximum number of retry attempts configured for operations.

Returns:: The maximum number of retries.
Return type:: int

perform_retry_backoff(existing_retries: int) → int[source]#

Performs exponential backoff sleep if retries are permitted.

Calculates the delay using exponential backoff (2^existing_retries) capped by self._max_backoff. Sleeps for the calculated delay if the number of existing_retries is less than max_retries.

Parameters:: existing_retries (int) – The number of retries already attempted for the current operation.
Returns:: The incremented retry count (existing_retries + 1).
Return type:: int
Raises:: RuntimeError – If existing_retries is greater than or equal to max_retries (when max_retries > 0).

ping() → ResponseSchema[source]#

Checks if the HTTP server endpoint is responsive using an HTTP GET request.

Returns:: An object encapsulating the outcome: - response_code = 0 indicates success (HTTP status code < 400). - response_code = 1 indicates failure, with details in response_reason.
Return type:: ResponseSchema

submit_message( channel_name: str, message: str, for_nv_ingest: bool = False, timeout: float | Tuple[float, float] | None = None, ) → ResponseSchema[source]#

Submits a job message payload to the server’s submit endpoint.

Handles retries for connection errors and non-terminal HTTP errors based on the max_retries configuration. Specific HTTP statuses are treated as immediate failures.