requester-kit 0.5.1

Async HTTP connector toolkit with retries, testing helpers, and Prometheus metrics.

requester-kit

Async HTTP connector toolkit with retries, typed responses, and optional Prometheus metrics.

Install

uv add requester-kit

Metrics extra:

uv add --extra metrics requester-kit

Quickstart

from uuid import UUID

from pydantic import BaseModel

from requester_kit import BaseRequesterKit, RequesterKitResponse


class UserInfo(BaseModel):
    id: UUID
    name: str


class UsersAPI(BaseRequesterKit):
    async def get_user_info(self, user_id: UUID) -> RequesterKitResponse[UserInfo]:
        return await self.get(f"/users/{user_id}", response_model=UserInfo)


async def run():
    users = UsersAPI(base_url="https://api.example.com")
    response = await users.get_user_info(UUID("00000000-0000-0000-0000-000000000000"))
    if response.is_ok and response.parsed_data:
        print(response.parsed_data.name)

Retries and logging

from requester_kit import BaseRequesterKit
from requester_kit.types import LoggerSettings, RetrySettings

client = BaseRequesterKit(
    base_url="https://api.example.com",
    retryer_settings=RetrySettings(
        retries=3,
        delay=0.2,
        increment=0.1,
        custom_status_codes={429},
    ),
    logger_settings=LoggerSettings(
        log_error_for_4xx=False,
        log_error_for_5xx=True,
    ),
)

Prometheus metrics

Pass enable_prometheus_metrics=True to BaseRequesterKit to track HTTP request duration. Each HTTP call records a Histogram named requester_kit_request_duration_seconds with labels: method (for example UsersAPI.get_user_info), status_code, status_class, and attempt. This provides request count and timing via the standard _count and _sum series.

Errors are counted in requester_kit_request_errors_total with labels: method, status_code, error_type (http_status or http_error), and attempt.

Payload sizes are recorded in Histograms: requester_kit_request_payload_bytes and requester_kit_response_bytes with the same labels as requester_kit_request_duration_seconds.

from requester_kit import BaseRequesterKit

client = BaseRequesterKit(
    base_url="https://api.example.com",
    enable_prometheus_metrics=True,
)

Expose metrics in FastAPI:

from fastapi import FastAPI, Response
from prometheus_client import CONTENT_TYPE_LATEST, generate_latest

app = FastAPI()

@app.get("/metrics")
def metrics():
    return Response(generate_latest(), media_type=CONTENT_TYPE_LATEST)

Expose metrics using prometheus-fastapi-instrumentator:

from fastapi import FastAPI
from prometheus_fastapi_instrumentator import Instrumentator

app = FastAPI()

Instrumentator().instrument(app).expose(app, endpoint="/metrics")

Why this works: BaseRequesterKit writes metrics to the default Prometheus registry, and both generate_latest() and prometheus-fastapi-instrumentator expose that same registry, so your HTTP client metrics appear alongside your app metrics on /metrics.

Testing

Use pytest-mock to stub out connector methods:

from unittest import mock

from pydantic import BaseModel

from requester_kit import BaseRequesterKit
from requester_kit.types import RequesterKitResponse


class UserInfo(BaseModel):
    id: str
    name: str


class UsersAPI(BaseRequesterKit):
    async def get_user_info(self, user_id: str) -> RequesterKitResponse[UserInfo]:
        return await self.get(f"/users/{user_id}", response_model=UserInfo)


async def test_get_user_info(mocker):
    mocker.patch.object(
        UsersAPI,
        "get_user_info",
        new=mock.AsyncMock(
            return_value=RequesterKitResponse(
                status_code=200,
                is_ok=True,
                parsed_data=UserInfo(id="1", name="Ada"),
            )
        ),
    )

    client = UsersAPI(base_url="https://api.example.com")
    response = await client.get_user_info("1")

    assert response.is_ok