iText Suite 9.6: Easier WTPDF, Color Contrast Checks, and Smarter OCR!

This iText Suite release from Apryse includes key iText Core enhancements for creating accessible and reusable PDF content meeting the PDF/UA and Well-Tagged PDF (WTPDF) standards. We’ve integrated automatic PDF/UA color contrast checks during PDF/UA creation, and also introduced easier ways to create WTPDF documents – targeting both the accessibility and reuse conformance levels.

The other big news for this release is pdfOCR 5.0.0. This is a major version release which expands pdfOCR’s ML-based OCR engine with pre-trained PaddleOCR and EasyOCR ONNX model support. These models are not only highly accurate, but are also available for more languages are than the docTR ones previously supported. There’s also huge .NET performance improvements, and GPU acceleration.

WCAG Color Contrast Checks

iText Core can automatically warn when foreground/background contrast looks too low for WCAG. These checks don’t replace human review, but they highlight likely failures while developers are generating PDFs.

Its functionality is similar to that provided by the WebAIM Contrast Checker for WCAG compliance, where you can select different values for the foreground and background colors.

WCAG (Web Content Accessibility Guidelines) defines measurable contrast thresholds (e.g., 4.5:1 for normal text) for web content, and PDF/UA defers to these WCAG criteria when assessing text and graphics contrast in PDFs.

Easier WTPDF Creation

The WTPDF (Well‑Tagged PDF) standard was developed by the PDF Association’s PDF/UA and PDF Reuse Technical Working Groups to specify the use of Tagged PDF for content reuse and accessibility.

Based on PDF 2.0, its accessibility requirements match PDF/UA‑2, ensuring documents meet the same user‑focused standards.

iText has supported WTPDF since its publication in 2024, However, API improvements now make creating WTPDF documents with iText as easy as PDF/A and PDF/UA.

Non-EU Trusted List Validation

Building on the List of Trusted Lists (LOTL) support introduced in the iText Suite 9.3 release, iText Core 9.6.0 adds an alternative path for ecosystems that publish one XML trust list (no LOTL pointers).

This was achieved by adding a convenient “single file trust list” alternative implementation, useful for countries where you have a single XML trust list rather than an EU-style LOTL + pointers ecosystem.

Updated LOTL Trusted Certificates

Speaking of the EU LOTL, the European Parliament has published a new set of trusted certificates used for validation. If you're currently using iText's EU LOTL validation and rely on our public repository to access the list of certificates, you will need to update the eu-trusted-lists-resources dependency to version 1.1 before the transition period ends on 28th April.

If this dependency is not updated, throughout the transition period iText will display log messages to alert you that the resources are outdated.

Improved Digital Signature Validation and Certificate Retrieval

We introduced a single, configurable mechanism for digital signature validation to control all online data retrieval (CRL, OCSP, issuing certificates, and LOTL). This simplifies enforcing network policies and improves reliability in restricted environments.

Certificate chain building and validation is also improved to better handle complex PKI setups. Chain construction gathers all reachable certificates, including multiple CA candidates and cross-signed certificates, without enforcing constraints during building.

Validation applies RFC 5280–compliant checks on the resulting paths, focusing on correctly enforcing inherited name constraints across the entire chain.

Extended and Enhanced ONNX OCR

Ideally, every PDF would be well tagged and accessible. However, many PDFs lack proper text, requiring OCR to make content reusable. This is why we developed pdfOCR, our add-on for iText Core.

In pdfOCR 5.0.0, the ML-based ONNX OCR engine now supports PaddleOCR and EasyOCR models in addition to the docTR models previously supported. These models perform well across many use cases and support multiple languages.

PaddleOCR/EasyOCR Model Support

PaddleOCR launched in late 2020 as an ultra-lightweight OCR system from the PaddlePaddle ecosystem. It gained momentum with the 2021 release of v2.0 and multilingual models (including English) that showed promising results.

It excels in structured document handling and memory efficiency, featuring many enterprise-level functions. Current versions offer improved accuracy and capabilities, optimized inferencing, and support over 100 languages.

EasyOCR is a more beginner-friendly multilingual OCR solution compared to PaddleOCR. It offers fewer options, but is easy to set up and supports over 80 languages.

We now have a HuggingFace repository where you can download many compatible models to start quickly. You can also try alternative models, though some require conversion to ONNX format to work with pdfOCR’s ONNX engine.

GPU Acceleration

This release includes many improvements beyond new model support. Optional GPU acceleration is now enabled for pdfOCR’s ONNX engine, freeing the CPU for other tasks and boosting performance.

ONNX Runtime supports multiple execution providers for hardware acceleration, though not all are production-ready. We have tested pdfOCR only with Nvidia CUDA-enabled GPUs. For other hardware, refer to OnnxRuntime’s official docs on execution providers.

General OCR Improvements

We've improved how pdfOCR positions recognized text boxes for rotated content. This enhancement lets pdfOCR better match the original orientation and placement of text, including small-angle rotations, not just 0°, 90°, 180°, and 270°.

Support for retrieving OCR text bounding rectangles in image pixel coordinates instead of PDF coordinate space has been added, eliminating manual conversion when working at the image level.

.NET Performance Gains

For .NET, ONNX engine performance improved significantly due to optimizations. A wrapper for Java’s FloatBuffer class prevents copying large float arrays for models. Additional improvements in image handling and processing yield faster performance on both Java and .NET.

HTML to WTPDF and extended CSS coverage

We simplified converting modern web content into high-quality, accessible PDFs with the pdfHTML add-on and expanded its CSS support for consistent styling and fewer manual adjustments.

In pdfHTML 6.3.2, we incorporated the improved Well-Tagged PDF (WTPDF) creation from iText Core 9.6.0 to enable WTPDF conformance support. We also added one CSS pseudo-class, enhanced another, and made minor refinements and bug fixes.

The :has() pseudo-class matches an element when the selector inside finds a match relative to that element.

Simply put, it acts as a “parent selector,” letting you style a parent or preceding element based on its content or following elements.

The other pseudo-class is the :not() selector for the @page at-rule. You can now specify @page :not(:first) to style all pages except the first page, correcting previous incorrect behavior.

pdfHTML now uses supplier-based dependency injection, creating dependencies on demand instead of storing them. This matches iText Core’s stateless design and ensures more predictable setup across conversions.

Bug Fixes and Miscellaneous Improvements

We fixed several customer and internal issues. The MemoryLimitsAwareHandler  that limits memory use when processing PDFs now behaves differently when creating versus updating xref tables.

An issue with some Type3 fonts caused glyphs sharing Unicode mappings to render incorrectly or duplicate, affecting appearance and extraction. This was fixed.

A regression removed the /DA (default appearance) from signature fields during signing, causing Acrobat warnings and re-signing issues. The fix preserves the DA entry to avoid this.

Copying pages from some PDF/A documents with altered outline/action dictionaries caused a PdfException about objects from different PDFs. Also, PDF/A compliance checks now avoid null pointer exceptions by correctly assigning resource dictionaries.

Error handling improvements include clearer errors for a missing subType in PDF annotations and merging PDFs with syntax issues, aiding troubleshooting.

For .NET, an issue where old Newtonsoft.Json versions were pulled for .NET Framework 4.6.1 was fixed by updating Microsoft.Extensions.DependencyModel to 10.0.2. This doesn't affect .NET Standard 2.0, which uses System.Text.Json since iText Core 9.2.0.

Pull Requests

For this release, we thank schallb for their contribution to improve LocationTextExtractionStrategy by adding support for custom element and newline separators.

As the author says, they needed a way to better distinguish spaces and newlines in the extracted text from separators inserted by iText at word boundaries or new lines. This change allows overriding the separators in the formatted extracted text.

Showcase PDF

As always, we turned the iText Core release notes into a PDF using the pdfHTML add-on. This document showcases several iText features beyond HTML conversion, including digital signing, MAC protection, SVG rendering, layout options, font optimizations, and more.

For this release, we applied improvements to WTPDF creation to meet accessibility and reuse conformance levels. It also complies with the latest PDF/UA-2 standard for accessible PDFs and the PDF/A-4F archiving standard.

The source code and resources used to create it are embedded in the document. Alternatively, visit our public repository for an example .NET project that generates accessible PDFs from suitable web pages, applying correct fonts, injecting custom content, creating a table of contents, and more.

You can find full details of what’s new in iText Suite 9.6 in the release announcement. Alternatively, check the release notes on the iText Knowledge Base for more technical information, download links and related documentation/examples.