How-to guides

How-tos are problem-shaped recipes. Each one answers a single, narrowly scoped question — "how do I do X?" — and assumes you already know the basics. If you are new, work through the tutorials first.

Available recipes

• OCR a folder of PDFs concurrently — AsyncClient
• asyncio.Semaphore + gather(..., return_exceptions=True) for a bounded-concurrency ingestion pipeline.
• Configure retries — retry only on 503, change backoff, respect Retry-After.
• Non-Latin PDFs — works out of the box; see this page only if you want to override the bundled font.
• Use a custom httpx.Client — mTLS, proxies, connection limits, custom CA bundles.
• Batch with partial failures — keep successful results when one item in a batch fails, using BatchResponse.iter_results.

For exhaustive parameter lists see the API reference; for conceptual background see Explanation.