PD4ML v4 is a Java library (well, the .NET edition of PD4ML is still at v3, with the port of v4 pending), but the documents it produces are needed everywhere: in Django and Flask apps, PHP shops, ASP.NET services, Node.js backends and Rails applications. From today, all of them can use PD4ML v4 directly. Five new integrations bring its complete API – the same one the Java manual describes – to Python, PHP, .NET, Node.js and Ruby.

One install, no Java code

Each integration installs with the language’s own package manager:

• Python: pip install pd4ml (PyPI)
• PHP: composer require pd4ml/pd4ml-php (Packagist)
• Node.js: npm install pd4ml (npm)
• Ruby: gem install pd4ml (RubyGems)
• .NET: the Pd4ml.Net NuGet package from pd4ml.com/net

Then one command – pd4ml install (in PHP projects vendor/bin/pd4ml install) – downloads the most recent PD4ML release and the Bouncy Castle libraries it needs for signing from the PD4ML Maven repository, checking every file against its published checksum. The Python, Node.js and Ruby integrations even do that on their own at first use. The only other requirement is a Java runtime (8 or later) on the machine.

The same API, in your language

The familiar PD4ML calls read naturally in each language. A PDF from an HTML string in Python:

import pd4ml

pd = pd4ml.PD4ML()
pd.setPageSize(pd4ml.PageSize.A4)
pd.read_html_string("<h1>Hello, PDF</h1><p>Generated with <b>PD4ML</b>.</p>")
pd.writePDF("hello.pdf")

The same in PHP:

use Pd4ml\PD4ML;
use Pd4ml\Java;

$pd = new PD4ML();
$pd->setPageSize(Java::PageSize()->A4);
$pd->readHtmlString('<h1>Hello, PDF</h1><p>Generated with <b>PD4ML</b>.</p>');
$pd->writePDF('hello.pdf');

In C#:

using Pd4mlNet;

var pd = new PD4ML();
pd.SetPageSize(PageSize.A4);
pd.ReadHtmlString("<h1>Hello, PDF</h1><p>Generated with <b>PD4ML</b>.</p>");
pd.WritePDF("hello.pdf");

In Node.js:

const { PD4ML, PageSize } = require('pd4ml');

const pd = new PD4ML();
pd.setPageSize(PageSize.A4);
pd.readHtmlString('<h1>Hello, PDF</h1><p>Generated with <b>PD4ML</b>.</p>');
pd.writePDF('hello.pdf');

And in Ruby:

require 'pd4ml'

pd = Pd4ml::PD4ML.new
pd.set_page_size(Pd4ml::PageSize::A4)
pd.read_html_string('<h1>Hello, PDF</h1><p>Generated with <b>PD4ML</b>.</p>')
pd.write_pdf('hello.pdf')

Beyond “hello world”, every feature of PD4ML is there: PDF/A and PDF/UA, ZUGFeRD and Factur-X invoices, DOCX and RTF output, page images, headers and footers with page numbers, watermarks, bookmarks, fonts, encryption and permissions, and the PDF toolkit for merging, overlaying and signing existing documents. Each integration also adds what makes it feel at home: file paths, byte strings and streams instead of Java I/O, the language’s own error handling, callbacks as progress listeners – and a command line tool for quick conversions.

A note on .NET

PD4ML for .NET is a temporary solution. Like the other integrations, it runs the Java PD4ML in a small bridge process next to your application – so it needs a Java runtime, but in return gives .NET applications today’s PD4ML: the same features, the same output and the same releases as the Java library. The native PD4ML.NET, a fully managed port, is based on PD4ML v3; its port of v4 is pending, and when it is ready, it will take this package’s place. Until then, Pd4ml.Net is the way to use PD4ML v4 from C# and the other .NET languages, with typed classes for the whole API.

Documentation

Every integration comes with a user guide: installation, how it works, use cases with runnable code for the common tasks, tips, troubleshooting and an API reference. They are in the new Integrations section of the PD4ML documentation:

• PD4ML for Python
• PD4ML for PHP
• PD4ML for .NET
• PD4ML for Node.js
• PD4ML for Ruby

For PD4ML itself – layout, page formats, fonts, standards – the guides refer to the PD4ML Programmer’s Manual, whose API they all share. The sources of the PHP, Node.js and Ruby packages are on GitHub: pd4ml-php, pd4ml-node, pd4ml-ruby.

Availability

All five integrations are available now. They work with the current PD4ML release, and they pick up new PD4ML releases with a single pd4ml install. As with PD4ML for Java, an evaluation runs without a license, with watermarked output; your existing PD4ML license applies to them as well.

Questions, or a framework you would like to see covered? Let us know – we’d like to hear how you use PD4ML from your language.

Newsletters, catalogs, glossaries and price lists read better in narrow columns than in lines that run across the whole page. PD4ML v3 users have long relied on generateMulticolumn() for exactly that – and with PD4ML v4.0.24 and v4.1.2 the feature is back, rebuilt for the v4 rendering pipeline.

One call

PD4ML pd4ml = new PD4ML();
pd4ml.generateMulticolumn(2, 8, true);   // 2 columns per page, 8 mm gap
pd4ml.readHTML(new URL("file:newsletter.html"));
pd4ml.writePDF(fos);

The call is the same as in v3, so existing v3 code carries over unchanged.

How it works

• The document is laid out to the width of a column, and the columns of each page are filled one after another. Every page break – the ones the layout finds as well as <pd4ml:page.break> – moves on to the next column; after the last column a new page starts.
• <pd4ml:page.break multicolumn="true"/> starts a new page right away, for example for an appendix or a new chapter.
• Page headers, footers, backgrounds and watermarks belong to the page, not to a column: they span all columns, and $[page] / $[total] count pages, not columns.
• Footnotes are placed at the bottom of the column that refers to them; links and bookmarks point to the right column.
• Any page format works: three columns on landscape A4 are one setPageSize(PageSize.A4.rotate()) away.

See it in action

Two new runnable examples in the pd4ml-examples repository, each with its HTML source and the resulting PDF:

• Two-column newsletter with a page header and footer, a table, a footnote and an appendix on a new page: H008MulticolumnNewsletter.java, H008.htm, PDF result
• Three-column glossary on landscape pages: H009MulticolumnGlossary.java, H009.htm, PDF result

Documentation

• Usage Examples: Multicolumn Layout – the v4 and v3 calls side by side, with the newsletter markup
• PD4ML Programmer’s Manual, 10.2 Multi-column pages
• Release notes

Availability

Multicolumn layout is part of the PD4ML v4.0.24 and v4.1.2 snapshot builds, already available for download and from the PD4ML Maven repository. It applies to PDF output; as in v3, it is ignored together with adjustHtmlWidth() and fitPageVertically().

Questions or a layout that doesn’t come out as expected? Let us know – we’d like to see your multicolumn documents.

v4.0.23fx2 / v4.1.1 releases complete a comprehensive static-analysis review of the entire codebase using SpotBugs, an industry-standard Java code-quality and security scanner. Findings dropped from 1,457 to 562 — a 61% reduction — verified across six independent re-scan passes.

Reliability & correctness

• Resolved a thread-safety race condition in application configuration loading that could, under concurrent use, expose a partially-initialized internal object.
• Corrected multiple rounding/truncation defects in page-layout calculations (table, flex, and image positioning) that could cause sub-pixel inaccuracies.
• Fixed an off-by-one detection bug affecting web font format recognition.
• Corrected several string-comparison bugs that could cause incorrect matching under edge-case input.
• Closed gaps where errors were being silently discarded instead of logged.

Code quality & maintainability

• Brought internal data-sorting components in line with Java serialization best practices.
• Modernized number-object construction throughout the codebase (~144 call sites) to current Java standards, eliminating deprecated API usage.
• Removed 12 confirmed-dead code paths (fields written but never read), reducing attack surface and maintenance burden.
• Addressed switch-statement completeness gaps that could mask unhandled cases.
• General conformance cleanup across best-practice, internationalization, and multi-threading-safety categories.

Governance

• Every remaining finding not fixed outright was individually reviewed; those intentionally left in place are documented in-code with the reasoning, and are now machine-verifiable via standard suppression annotations rather than relying on comments alone — confirmed effective through re-scanning.
• No findings were dismissed without review; the sweep covered the tool’s full default rule set.

Net result: a measurably lower defect-density codebase with an auditable, reproducible verification trail — no open finding was closed without either a fix or a documented, reviewed rationale.