Python SDK

Thin, typed, retry-aware Python client — sync + async. MIT licensed.

Install

pip install rlbwhisk

Basic use

from rlbwhisk import Client

c = Client()  # anonymous — 100 req/hour per IP
# or: Client(api_key="rlbwhisk_live_...")
# or: set RLBWHISK_API_KEY in your environment

# Manual mode — structured findings
result = c.check("Ths is a smple sentnce.", language="en")
for m in result.misspellings:
    print(m.word, "->", m.suggestions[:3])

# Auto mode — get corrected text
fixed = c.fix("Ths is a smple sentnce.", language="en")
print(fixed.corrected_text)
# "This is a simple sentence."

Async

import asyncio
from rlbwhisk import AsyncClient

async def main():
    c = AsyncClient()
    r = await c.check("Ths is a test.", language="en")
    print(r.misspellings)

asyncio.run(main())

Error handling

from rlbwhisk import (
    Client,
    WhiskError,
    InvalidApiKey,
    RateLimitExceeded,
    InvalidLanguage,
    InternalError,
)

try:
    r = c.check("hi", language="en")
except InvalidApiKey:
    print("Bad key — generate a fresh one at /whisk/keys/")
except RateLimitExceeded as e:
    print(f"Retry after {e.retry_after}s (request {e.request_id})")
except InvalidLanguage:
    print("Language not supported")
except InternalError as e:
    print(f"Server error, quote request_id={e.request_id} in support")
except WhiskError as e:
    print(f"[{e.code}] {e.message}")

The client auto-retries 429 Too Many Requests up to max_retries=3 times, respecting the Retry-After header.

Options

c = Client(
    api_key="rlbwhisk_live_...",   # or RLBWHISK_API_KEY env
    base_url="https://reglegbrief.com/whisk/api/v1",   # default
    timeout=30.0,                   # seconds
    max_retries=3,
)

Methods

Data classes

@dataclass
class Misspelling:
    word: str            # lowercased
    offset: int          # character offset in source text
    length: int
    suggestions: list[str]
    kind: str            # "spell" | "repeat"

@dataclass
class CheckResponse:
    language: str
    total_words: int
    misspellings: list[Misspelling]

@dataclass
class Change:
    offset: int
    before: str
    after: str
    kind: str

@dataclass
class FixResponse:
    language: str
    total_words: int
    corrected_text: str
    changes: list[Change]
    unresolved: list[Misspelling]   # findings with no suggestion

Source

MIT licensed. Repo: github.com/rlbwhisk/rlbwhisk-python