Download

PDFix SDK

PDF Accessibility Automation API

Auto-tag, remediate, validate and convert PDFs programmatically. 4 AI engines. 100% local processing. One command from untagged PDF to certified PDF/UA.

Windows · macOS · Linux · Free Trial

# AddTags.py

# import utils to load required shared libraries
from Utils import inputPath, outputPath, stream_to_data
from pdfixsdk import *

pdfix = GetPdfix()
if pdfix is None:
    raise Exception('Pdfix Initialization fail')

doc = pdfix.OpenDoc(inputPath + "/test.pdf", "")
if doc is None:
    raise Exception('Unable to open pdf : ' + pdfix.GetError())

# load template configuration from JSON file
tmpl = doc.GetTemplate()
if not tmpl:
    raise Exception('Unable to open pdf : ' + pdfix.GetError())
↓ Download FREE
Get Started ➜
$ SDK Pricing ➜

PDF/UA

uA-1, UA-2, WCAG 2.2, Section 508, ADA, EAA

100 %

Local processing – on-premises

5 Languages

 Python, Java, C++, .NET, JS

90%

Reduction in document processing time

WHAT PDFIX SDK DOES

PDF accessibility automation, data extraction and HTML conversion – one API

A high-performance C++ PDF library that covers three enterprise workflows via unified API or CLI.

No external dependencies required.

PDF accessibility automation PDF/UA & WCAG

Auto-tag PDFs for full PDF/UA and WCAG compliance. Fix structure, headings, Alt text, tables and reading order programmatically at any scale.

PDF data extraction – text, structure, tables and metadata

Extract structured data from PDFs for analytics, AI/ML pipelines and document processing systems. Output to JSON, XML, CSV.

  • Raw text extraction: metadata, form fields, annotations, text, images, positions, colors
  • Intelligent Layout recognition: paragraphs, headings, tables, figures, headers, footers
  • Structure tags: complete tag tree, attributes, position and content
  • RegEx and pattern matching for precise data capture
  • Ready for big data, AI and machine learning pipelines

PDF to HTML conversion – 3 rendering methods

Convert PDFs to HTML for web publishing, responsive rendering and accessible web content. PDF forms become HTML forms.

  • Original Layout: fixed-layout HTML, preserves exact design
  • Responsive Layout: HTML5 + CSS3, adapts for mobile via layout recognition
  • Tags-based: semantic HTML via Derivation algorithm from tagged PDFs
  • PDF forms → HTML forms (inputs, dropdowns, checkboxes, radio)
  • Embed PDF content directly into web pages
  • “The PDFix library focuses on and excels at automatically recognizing logical structures, which are essential for Liquid Mode in Acrobat Reader and accessibility.”
    Michael Karbe, Managing Director, Actino Software GmbH  

AUTO-TAGGING IN PDFIX SDK

Auto-tag PDFs – 4 workflow methods × 4 AI engines

Two decisions to make: which workflow method fits your document and volume, and which AI engine you want to use for AI-generated templates. Both decisions are independent – start with the method, then choose the engine.

HOW TO AUTO-TAG → CHOOSE A METHOD

BASIC

NO TEMPLATE

Fastest results, zero setup. SDK detects text, reading order, tables and lists automatically.

✓ Zero setup

PREFLIGHT

AUTO-GENERATED TEMPLATE

Preflight analyzes the document and auto-generates a layout template – better heading and footer detection.

✓ Best balance

✓ Automated analysis

AI-GENERATED

INTELLIGENT TEMPLATE

AI model generates a JSON template that guides PDFix SDK. Choose from 4 AI engines below.

✓ Most intelligent

✓ Adaptive

PRE-DEFINED

JSON TEMPLATE

Define structure once for your document type – invoices, statements, reports. Apply across thousands of files.

✓ Consistent results

✓ Enterprise-grade

What is a layout template?

Template-based auto-tagging is a deterministic approach – a precise JSON file describing the layout of a specific document type. PDFix SDK applies it to every document of the same type: same rules, same output, every time.

How it works

  1. Define rules in JSON → e.g. “find bold text at 12 pt → tag as Heading 2”
  2. Use text anchors instead of hard-coded coordinates → e.g. “word INVOICE → number to the right is the invoice ID”
  3. SDK matches elements and generates a fully PDF/UA-compliant tag tree
  4. Apply the same template across hundreds or thousands of documents – no manual work

Best for

Batches from the same source → invoices, bank statements, monthly reports, forms generated by InDesign or accounting systems. Reduces per-document tagging cost by 80%+

❯ Layout Template Guide
❯ Examples on GitHub

CHOOSE AI TAGGING ENGINE – FOR AI GENERATED TEMPLATES

IBM Docling

LOCAL · FREE

On-prem · European enterprise

Data never leaves your network. Runs via Docker. Ideal for finance, healthcare and government where data sovereignty is required.

Headings · Paragraphs · Tables · Images · Multi-column

❯ Explore IBM Docling

Paddle AI

LOCAL · FREE

Highest layout accuracy · Formulas

Best for complex layouts, scientific and academic documents. Handles mathematical formulas. Runs locally via Docker.

Headings · Tables · Formulas · Math · Lists

❯ Explore Paddle AI

Amazon Textract

CLOUD · PAID AWS

AWS-native · Scanned PDFs · OCR

Best for scanned PDFs and documents with inconsistent structure. AWS-native customers and hybrid cloud workflows – connects via AWS credentials.

OCR · Scanned PDFs · Tables · Forms · Handwriting

❯ Explore Amazon Textract

OpenAI

CLOUD · API KEY

Best AI Alt text · MathML · Tables

Best results for Alt text generation, table summaries and MathML for formulas. Enterprise-grade AI understanding of images and complex tables in context.

AltText · MathML · Table summaries · Image context

❯ AltText
❯ MathML
❯ Table Summaries

DEEP DIVE → Auto-Tag PDFs at Scale: AI & Template-Based Tagging with PDFix SDK – code examples, CLI commands and video for all 4 tagging methods

COMPLIANCE AUTOMATION PIPELINE

From untagged PDF to certified PDF/UA – automated compliance pipeline

PDFix SDK covers the complete PDF/UA compliance workflow end-to-end – from raw input to validated, compliant output – in a single automated pipeline.

Input PDF

Untagged or partially tagged

Auto-tag

Docling · Paddle · Textract · OpenAI

Make Accessible

One-command fix for PDF/UA issues

Validation

ISO 14289-1 validation

Accessible PDFs

PDF/UA compliant output

Validation Capabilities

veraPDF

ISO 14289-1 PDF/UA-1 compliance. 136 Matterhorn Protocol failure conditions. Machine-readable JSON/XML output for CI/CD. veraPDF guide →

◉ PDF/UA Validation
❖ veraPDF Guide

INTEGRATION & DEPLOYMENT

Integrate PDFix SDK any way you want – 5 languages, CLI, Docker and Cl/CD

Every integration path is supported. On-premises or cloud, scripted or programmatic, desktop GUI or automated pipeline.

On-Premises

Data never leaves your network

CLI

pdfix_app binary, scriptable

Docker

All AI actions as containers

Desktop GUI

PDFix Desktop for manual runs

Cl / CD

GitHub Actions, GitLab, cron

❯ All examples on GitHub
❯ API reference docs
❯ CLI Documentation

Disclaimer: PDFix provides tools and technology to assist in making your documents accessible, but we do not guarantee 100% document accessibility. Achieving full compliance requires human checks and intervention. Please note that PDFix is a technology provider, not a service provider. The responsibility for document compliance rests with the user.

Your PDF problems solved in a few hours

Download PDFix SDK – free trial, all features

Describe your project and our experts will craft a custom solution to save you time, money and headaches. Or start immediately – download the SDK and run your first auto-tag in under 5 minutes.

⤓ Download SDK
➔ Getting started guide
✉︎ Contact us

You might also like