PaddlePaddle/PaddleOCR · critical · AuthError

Token is required. Set PADDLEOCR_ACCESS_TOKEN or pass token=

Error message

Token is required. Set PADDLEOCR_ACCESS_TOKEN or pass token=.

What it means

AuthError raised by AsyncPaddleOCRClient.__init__ when neither a token argument nor the PADDLEOCR_ACCESS_TOKEN environment variable is present. The async API client requires an access token before any request can be made, so construction fails fast.

Source

Thrown at paddleocr/_api_client/async_client.py:62

class AsyncPaddleOCRClient:
    """Async client for PaddleOCR API.

    Supports asyncio.gather for concurrent job submission and polling.
    """

    def __init__(
        self,
        token: Optional[str] = None,
        base_url: Optional[str] = None,
        request_timeout: float = 300.0,
        poll_timeout: float = 600.0,
        timeout: Optional[float] = None,
        client_platform: Optional[str] = None,
    ):
        self._token = token or os.environ.get("PADDLEOCR_ACCESS_TOKEN", "")
        if not self._token:
            raise AuthError(
                "Token is required. Set PADDLEOCR_ACCESS_TOKEN or pass token=."
            )
        resolved_base_url = (
            base_url or os.environ.get("PADDLEOCR_BASE_URL") or DEFAULT_BASE_URL
        )
        if timeout is not None:
            request_timeout = timeout
            poll_timeout = timeout
        self._http = AsyncHTTPClient(
            self._token,
            resolved_base_url,
            request_timeout,
            client_platform=client_platform,
        )
        self._poller = AsyncPoller(self._http, max_wait_time=poll_timeout)

    async def __aenter__(self):
        await self._http.__aenter__()

View on GitHub (pinned to 2661c7c0ef)

Solutions

1. Export PADDLEOCR_ACCESS_TOKEN before starting the process
2. Pass the token explicitly: AsyncPaddleOCRClient(token='...')
3. Load the token from your secret store (e.g. dotenv, vault) at startup and pass it explicitly rather than relying on the environment

Example fix

# before
client = AsyncPaddleOCRClient()
# after
client = AsyncPaddleOCRClient(token=os.environ['PADDLEOCR_TOKEN'])

import os

token = os.environ.get('PADDLEOCR_ACCESS_TOKEN') or load_from_secret_store()
if not token:
    raise RuntimeError('missing PaddleOCR access token; configure secret store')

from paddleocr._api_client.errors import AuthError

try:
    client = AsyncPaddleOCRClient(token=token)
except AuthError:
    # fail the deployment check, do not retry
    raise

When it happens

Trigger: Instantiating the async client with no token= and no PADDLEOCR_ACCESS_TOKEN in the environment (e.g. fresh shell, CI without env secrets, .env file not loaded).

Common situations: Deploying to CI/new machines where the env var was never exported; typos in the variable name (PADDLEOCR_ACCESS_TOKEN vs PADDLE_OCR_ACCESS_TOKEN); running examples copied from docs without inserting credentials.

Related errors

• Token is required. Set PADDLEOCR_ACCESS_TOKEN or pass token=
• PaddleOCR source adapter is not configured.
• worker mode does not support a custom fetch implementation.
• Conflicting values provided for ${label}: ${aliases.join(",
• ${warnings.join(" ")} (pipeline warnings promoted to errors

AI-assisted analysis of PaddlePaddle/PaddleOCR@2661c7c0ef (2026-08-14). Data as JSON: /api/errors/e7e3d7f933c0deb7. Report an issue: GitHub.