Module ocr 

Pluggable raster OCR engines (Tesseract CLI / ocrs). Pluggable raster OCR engines for table recovery.

Dependency inversion: lattice / cell bucketing consume OcrWord lists; only the recognizer implementation changes.

• TesseractCliEngine — default on native (Tesseract / RapidOCR CLI).
• [OcrsEngine] — pure-Rust RTen models behind feature ocr-ocrs.
• [HostOcrEngine] — wasm32 host callback (PP-OCRv6 etc.).

Structs

OcrWord

One recognized word with axis-aligned box in image pixel space.

TesseractCliEngine

Shells out to Tesseract (or RapidOCR via Python when EDGEPARSE_OCR_ENGINE=rapidocr).

Traits

OcrEngine

Raster OCR provider (SOLID: depend on this abstraction, not a CLI spawn).

Functions

default_engine

Default engine for the current build / env.