xml_gap.xml

<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE article PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN" "docbook/docbookx.dtd">
<article lang="en-US">
  <articleinfo>
    <title>Closing the XML Gap in TypeScript (for Localization Workflows)</title>
    <author>
      <firstname>Rodolfo M.</firstname>
      <surname>Raya</surname>
      <affiliation>
        <orgname><ulink url="https://maxprograms.com">Maxprograms</ulink></orgname>
        <address format="linespecific"><email>rmraya@maxprograms.com</email></address>
      </affiliation>
    </author>
    <pubdate>17 June 2026</pubdate>
    <copyright>
      <year>2026</year>
      <holder>Maxprograms</holder>
    </copyright>
    <abstract>
      <para>Describes the move from Java to TypeScript for XML processing in localization
        workflows, and the TypesXML, TypesBCP47, and TypesXLIFF libraries built to bring DTD and
        XML Schema validation for TMX, XLIFF, and SRX to Node.js.</para>
    </abstract>
  </articleinfo>

  <section id="intro">
    <title>Introduction</title>
    <para>If you've ever tried to process XML in Node.js or TypeScript, you've probably run into a
      limitation: parsing is easy, but validation against DTD or XML Schema is not.</para>
    <para>That becomes a real problem when working with localization standards like TMX, XLIFF or
      SRX, where structure and validity are not optional. They define whether your data is
      usable.</para>
    <para>This is the gap I ran into, and eventually decided to close.</para>
  </section>

  <section id="the-old-setup-java-for-xml">
    <title>The old setup: Java for XML</title>
    <para>Localization standards such as TMX, XLIFF, SRX and TBX are all XML-based. When I started
      working with them more than 20 years ago, Java was the most practical choice. Its XML
      tooling wasn't perfect, but it was reliable enough for parsing and validation.</para>
    <para>That worked well for a long time.</para>
    <para>The problem wasn't XML; it was everything around it.</para>
    <para>Java desktop UI options didn't evolve in a way that fit my needs. I used SWT for years,
      but it became less maintained, and alternatives like Swing or JavaFX didn't appeal to
      me.</para>
  </section>

  <section id="moving-to-typescript">
    <title>Moving to TypeScript</title>
    <para>Around 2018, I shifted toward building user interfaces with HTML, CSS and JavaScript.
      Eventually, TypeScript became the obvious choice because it provides a safer development
      model with static checks.</para>
    <para>That solved the UI side.</para>
    <para>But it created a new problem:</para>
    <itemizedlist>
      <listitem><para>UI in TypeScript</para></listitem>
      <listitem><para>XML processing still in Java</para></listitem>
    </itemizedlist>
    <para>Maintaining two stacks for a single workflow wasn't ideal.</para>
  </section>

  <section id="the-missing-piece-xml-validation-in-typescript">
    <title>The missing piece: XML validation in TypeScript</title>
    <para>At the time, there were XML parsers in JavaScript, but none supported:</para>
    <itemizedlist>
      <listitem><para>DTD validation</para></listitem>
      <listitem><para>XML Schema validation</para></listitem>
    </itemizedlist>
    <para>For localization formats, that's a hard requirement. Without validation, you can't
      guarantee that a TMX or XLIFF file is structurally correct.</para>
    <para>So I started building <ulink url="https://maxprograms.com/products/typesxml.html">TypesXML</ulink>,
      a native XML parser and validator for TypeScript.</para>
  </section>

  <section id="a-minimal-example">
    <title>A minimal example</title>
    <para>Here's what basic XML parsing looks like:</para>
    <programlisting xml:space="preserve">import { Catalog, DOMBuilder, SAXParser } from "typesxml";

const parser: SAXParser = new SAXParser();
const handler: DOMBuilder = new DOMBuilder();
const catalog = new Catalog('/pathTo/catalog.xml');

parser.setCatalog(catalog);
parser.setContentHandler(handler);
parser.setValidating(true);
parser.parseFile("example.xml");

const document = handler.getDocument();
console.log(document.toString());</programlisting>
    <para>The same parser can be configured to:</para>
    <itemizedlist>
      <listitem><para>validate against DTD</para></listitem>
      <listitem><para>validate against XML Schema</para></listitem>
      <listitem><para>resolve XML catalogs</para></listitem>
    </itemizedlist>
    <para>This makes it possible to process XML in Node.js with the same level of control
      traditionally associated with Java libraries.</para>
  </section>

  <section id="building-the-missing-stack">
    <title>Building the missing stack</title>
    <para>Creating a full XML stack took time:</para>
    <itemizedlist>
      <listitem><para>DOM and SAX-style parsing came first</para></listitem>
      <listitem><para>DTD validation followed</para></listitem>
      <listitem><para>XML Schema validation was the final step</para></listitem>
    </itemizedlist>
    <para>XML Schema support is significantly more complex, but it's essential for formats like
      XLIFF.</para>
  </section>

  <section id="supporting-pieces-language-data">
    <title>Supporting pieces: language data</title>
    <para>Localization workflows also depend heavily on language codes.</para>
    <para>BCP47 defines the standard, but the official data is distributed as a large text file.
      To make it easier to use in TypeScript, I built
      <ulink url="https://github.com/maxprograms-com/TypesBCP47">TypesBCP47</ulink>, a companion
      library that parses that data and exposes it in a structured way.</para>
  </section>

  <section id="proof-of-concept-rewriting-existing-tools">
    <title>Proof of concept: rewriting existing tools</title>
    <para>To validate the approach, I rewrote existing tools in TypeScript:</para>
    <itemizedlist>
      <listitem><para><ulink url="https://maxprograms.com/products/srxeditor.html">SRXEditor</ulink></para></listitem>
      <listitem><para><ulink url="https://maxprograms.com/products/tmxvalidator.html">TMXValidator</ulink></para></listitem>
    </itemizedlist>
    <para>These tools:</para>
    <itemizedlist>
      <listitem><para>process large XML files</para></listitem>
      <listitem><para>validate against official grammars</para></listitem>
      <listitem><para>maintain performance comparable to the original Java versions</para></listitem>
    </itemizedlist>
    <para>That confirmed that TypeScript can handle real-world localization workloads.</para>
  </section>

  <section id="xliff-support">
    <title>XLIFF support</title>
    <para>XLIFF uses XML Schema, which required implementing a full schema validator.</para>
    <para>With that in place, I built <ulink url="https://maxprograms.com/products/typesxliff.html">TypesXLIFF</ulink>,
      a library for parsing, generating, and validating XLIFF 2.x files.</para>
    <para>It supports:</para>
    <itemizedlist>
      <listitem><para>XML Schema validation</para></listitem>
      <listitem><para>validation against the XLIFF specification</para></listitem>
      <listitem><para>a typed object model</para></listitem>
      <listitem><para>JSON conversion</para></listitem>
    </itemizedlist>
  </section>

  <section id="what-this-enables">
    <title>What this enables</title>
    <para>With these pieces in place, it's now possible to:</para>
    <itemizedlist>
      <listitem><para>Validate TMX, XLIFF and SRX directly in Node.js</para></listitem>
      <listitem><para>Build localization pipelines entirely in TypeScript</para></listitem>
      <listitem><para>Remove the need for a separate Java backend</para></listitem>
      <listitem><para>Keep both UI and processing in the same language</para></listitem>
    </itemizedlist>
  </section>

  <section id="whats-next">
    <title>What's next</title>
    <para>TBX support is in progress, along with updates to other tools to move them fully to
      TypeScript.</para>
    <para>Some Java code still exists in older components, but new development is now centered on
      TypeScript.</para>
  </section>

  <section id="closing-thoughts">
    <title>Closing thoughts</title>
    <para>For a long time, XML processing in localization workflows effectively required
      Java.</para>
    <para>That's no longer the case.</para>
    <para>If you're working with XML-based formats in Node.js, especially in localization or
      structured data pipelines, you can now handle parsing, validation, and processing entirely
      in TypeScript.</para>
  </section>

  <section id="author">
    <title>About the author</title>
    <mediaobject>
      <imageobject>
        <imagedata fileref="images/p-rodolfo-white.jpg" />
      </imageobject>
    </mediaobject>
    <para>Rodolfo M. Raya, CTO of Maxprograms SAS, is involved in the development of localization
      standards including XLIFF, TMX, TBX, SRX and GlossML.</para>
    <para>He develops tools and libraries that implement these specifications, including the
      software and validation services available on this site.</para>
  </section>
</article>