Module ocr 

OCR backend trait and implementations.

OCR Backend Extension

The OcrBackend trait is public. Callers can implement their own backends for specific use cases:

• ocrs (this crate, ocr feature): pure-Rust Tesseract-style engine. WASM-compatible. Requires external model files (.rten format).

• Mistral OCR (ocr-mistral feature): blocking HTTP adapter to Mistral’s cloud OCR API.

• Google Vision (ocr-google feature): blocking HTTP adapter to Google Cloud Vision’s images:annotate OCR endpoint.

• AWS Textract (ocr-aws feature): blocking HTTP adapter to Amazon Textract’s DetectDocumentText endpoint.

• Azure Document Intelligence (ocr-azure feature): blocking HTTP adapter to the prebuilt-read model.

• Paddle ONNX (ocr-onnx feature): adapter around pdf-ocr’s PaddleOCR engine.

• Custom: implement OcrBackend directly for proprietary or cloud-based OCR flows.

Structs

AwsTextractBackend

Blocking HTTP OCR backend for AWS Textract.

AzureDocIntelBackend

Blocking HTTP OCR backend for Azure AI Document Intelligence.

GoogleVisionBackend

Blocking HTTP OCR backend for Google Cloud Vision.

MistralOcrBackend

Blocking HTTP OCR backend for the Mistral OCR API.

OcrResult

Result of recognizing text in a single image.

OcrWord

A single recognized word.

OcrsBackend

Pure-Rust OCR backend backed by the ocrs engine (which uses ONNX models via rten).

PaddleOnnxBackend

OCR backend backed by pdf-ocr’s PaddleOCR ONNX pipeline.

Enums

OcrError

Errors returned by an OcrBackend.

Traits

OcrBackend

Pluggable OCR backend.

Functions

best_available_backend

Return the first available OCR backend in priority order.

ocr_page_default

Run OCR on a rendered page image using the default ocrs backend.