V2 Coming July 2026 — Currently running V1

Engineered from the Ground Up

A visual walkthrough of how PDFox V2 transforms PDF rendering — from a canvas-locked viewer to a modular, DOM-native editing platform. Powered by DNPR — proprietary PDF 2.0 interpreter (patent pending).

V1 Architecture (PDF.js)

The V1 architecture relies on PDF.js to render pages to Canvas, then layers custom editing UI on top. The rendering engine and editing system are fundamentally separate.

V1 Rendering Pipeline

Input
PDF Binary
Uploaded file
→
Library
PDF.js
Mozilla
→
Output
Canvas
Pixel image
→
Bolt-on
Overlay Layers
Text, annotations

V1 Layer Stack

UI (Toolbar, Sidebar, Dialogs)
Editing Addons (Text overlay, Annotations, Signatures)
Canvas Renderer (PDF.js)
PDF.js Parser
PDF Binary Stream
The dashed layer is the architectural problem — editing bolted on top of viewing

V2 Architecture (PDFox.js)

The V2 architecture integrates editing directly into the rendering model. The PDF stream is interpreted into DOM elements that are natively editable by the browser. Powered by DNPR — proprietary PDF 2.0 interpreter (patent pending).

V2 Rendering Pipeline

Input
PDF Binary
Uploaded file
→
Parse
Parser
Proprietary acceleration
→
Interpret
Interpreter
Full PDF 2.0
→
Adapt
Adapter
DOM/Canvas/SVG
→
Render
Renderer
Editable DOM

V2 Layer Stack

UI (Minimal — editing is inline)
DOM Renderer (editable text, SVG graphics, DOM images)
DNPR Interpreter Core
PDF Parser (proprietary acceleration)
PDF Binary Stream
No addon layer — rendering and editing are the same system

Triple Render Output

The same core interpreter feeds three independent renderers

PDF Binary
→
Core Interpreter
→
DOM
Editable HTML elements
Canvas
Pixel-perfect preview
SVG
Scalable vector output

DNPR Capability Areas

PDFox V2 is built on DNPR (DOM-Native PDF Rendering) — a proprietary, patent-pending architecture that treats PDF as a structured object, not a flat image. Each capability area is modular and independently testable.

Parsing & Interpretation

Reads PDF 2.0 binary streams, handles incremental updates and linearized files. Processes malformed PDFs gracefully.

Foundation
DOM Rendering

Patent-pending DNPR output. PDF content rendered as live, editable elements — enabling inline editing without overlay layers.

DNPR Core
Graphics Engine

SVG path output for all PDF graphics operators. Selectable, editable vector graphics rendered directly in the DOM.

Rendering
Typography

Comprehensive font handling with advanced typography, right-to-left text support, and native browser font rendering.

Typography
Security

Enterprise-grade encryption, digital signatures with long-term validation, and full legacy format support.

Security
Compliance

PDF/A archival validation and PDF/UA accessibility compliance.

Compliance
Forms & Annotations

AcroForm field types, form filling, flattening, FDF/XFDF import/export. 11+ annotation types: highlight, underline, sticky notes, shapes, ink, redaction markup.

Editing
Editing

High-level editing capabilities: document loading, page lifecycle, undo/redo, selection management, and save operations.

Editing

Architecture Comparison

Dimension V1 (PDF.js) V2 (PDFox.js)
Render Target Canvas (pixel image) DOM + Canvas + SVG (triple output)
Editing Model Overlay layers bolted on canvas Native DOM — editing is rendering
Architecture Monolithic Modular
Acceleration Limited Full proprietary acceleration
Font Rendering Basic conversion Native browser font rendering
Graphics Model Canvas 2D API SVG paths (selectable, editable)
Text Model Invisible overlay for selection Native inline editing (click and type)
Concurrency Web Worker Web Worker (with progress)
PDF Standard PDF 1.7+ (partial 2.0 features) Full PDF 2.0 compliance
15-Page PDF Load ~1,300ms ~200ms (6.5× faster)
Infrastructure Traditional server + nginx Serverless cloud architecture
Bundle Size Full monolithic library Import only needed capabilities

Cloud Infrastructure

PDFox runs on a fully serverless AWS stack. No application servers, no databases, no persistent storage. All PDF processing happens in the user's browser — the cloud infrastructure only serves static files and handles authentication.

Secure Storage

Private storage for application assets. No public access; all requests routed through secure CDN.

Live
Global CDN

HTTPS distribution with global edge caching. Serves application assets from the nearest edge location. Cache invalidation on every deploy.

Live
Edge Authentication

Secure authentication at the CDN edge with time-based verification and session management.

Live
Serverless API

Serverless headless PDF processing. DNPR engine running server-side. Zero persistence by architecture — no disk, no database, no content logs.

Pending
GitHub Actions CI/CD

Automated build, deploy, and cache invalidation.

Live
Source Code Protection

Multi-layer source code protection.

Live

Serverless API (Pending)

Serverless headless PDF processing for automation, AI agents, and developer SDKs. DNPR engine running on a serverless runtime. Zero persistence by architecture — no disk, no database, no content logs.

Endpoint Purpose
POST /v1/extract-text Text extraction with positions and styles
POST /v1/parse Metadata extraction
POST /v1/fill-form Programmatic form filling
POST /v1/validate PDF/A and PDF/UA validation
POST /v1/merge Multi-document merge
POST /v1/split Page extraction
POST /v1/encrypt Enterprise-grade encryption
POST /v1/annotate Add annotations
Serverless function endpoint
DNPR engine layer
Zero persistence — no disk, no DB, no content logs

See What This Architecture Enables

Explore the editing features made possible by the V2 DOM-native architecture.

View V2 Features