infrastructure 0.1.0 API
Packages
Package
Description
Development-time diagnostics for the registration pipeline: an opt-in (
--diag) subsystem
that records and visualizes each page's internal state — skew, column detection, the reference it
aligns to, and its final clamped placement — as overlay images, projection-profile plots, a
machine-readable JSONL log, and an aggregate summary.The PDF adapters:
PdfImagesCliExtractor (the
PdfImageExtractor) and PdfBoxJbig2Assembler (the Jbig2Assembler).Process and filesystem-scratch plumbing for the adapters.
The Leptonica-backed page registrar: the
PageRegistrar
adapter (LeptonicaPageRegistrar) that
deskews, detects the main column (MainColumnDetector), and scales/places each
page using the domain's projection-profile, reference and transform-planner algorithms.