Conversion Tools Engine4 min read• Updated October 2026
PDF to Structured JSON Engine
Extracts semantic document trees, text blocks, reading order, and geometry bounding boxes.
Overview
Ideal for machine learning engineers and developers building custom ingestion pipelines. Produces clean JSON with bounding box coordinates (x, y, width, height) for every block.
Key Specifications & Highlights
Reading Order Determination: Corrects multi-column text flow (e.g. newspaper or research paper format).
Heading Tag Classification: Classifies lines into h1, h2, paragraph, list, or table headers automatically.
Page Geometry Anchoring: Normalizes coordinate points to standardized 72 DPI PDF point coordinates.