TypesXLIFF - TypeScript Library for XLIFF Parsing, Generation and Validation

Quick example

Load an XLIFF file and read the first source segment:

import { XliffParser } from 'typesxliff';

const parser = new XliffParser();
await parser.parseFile("file.xlf");

const doc = parser.getXliffDocument();
const segment = doc?.files[0].units[0].segments[0];

console.log(segment?.source.toString());

Why TypesXLIFF

Most TypeScript tools for XLIFF either parse the XML into plain objects without a structured model, or rely on generic XML libraries that expose no XLIFF-specific semantics. TypesXLIFF sits between those extremes: it provides a fully typed class hierarchy that mirrors the XLIFF 2.x specification, so you can build, inspect, and validate documents without writing raw XML.

• Full XLIFF 2.x object model - not just parsing into plain objects
• Type-safe API for building and modifying documents programmatically
• JSON round-trip for integration with other systems
• Built on TypesXML - streaming XML parser with validation support

Use Cases

• Parse XLIFF 2.x files in Node.js or TypeScript
• Generate XLIFF documents programmatically
• Convert XLIFF to/from JSON
• Validate XLIFF structure and content
• Extract terminology and bilingual term pairs from XLIFF content
• Build localization or translation pipelines

Parsing an existing XLIFF file

import { Catalog } from "typesxml";
import { XliffParser, XliffDocument } from "typesxliff";

const parser = new XliffParser();
// Using a catalog is optional. When provided, the SAX parser can resolve grammar
// schemas and populate default attribute values declared in them.
// A sample catalog covering XLIFF 2.0, 2.1 and 2.2 is included in the catalog/ folder.
parser.setCatalog(new Catalog('/path/to/typesxliff/catalog/catalog.xml'));
parser.parseFile('/path/to/file.xlf');
const doc: XliffDocument | undefined = parser.getXliffDocument();

Building and writing an XLIFF document

import { XliffDocument, XliffFile, XliffUnit, XliffSegment, XliffSource } from "typesxliff";

const doc = new XliffDocument("2.1", "en", "es");

const file = new XliffFile("f1");
const unit = new XliffUnit("u1");
const segment = new XliffSegment("s1");
const source = new XliffSource();
source.addContent("Hello, world!");
segment.setSource(source);
unit.addItem(segment);
file.addEntry(unit);
doc.addFile(file);

doc.writeDocument('/path/to/output.xlf', true);

Documentation & Samples

• Read the Getting Started guide for a step-by-step introduction.
• Use the Parsing XLIFF Files guide for detailed parsing workflows, including catalog usage.
• Follow the Building XLIFF Documents guide to construct and serialize documents from scratch.
• See the JSON Conversion guide for round-trip XLIFF ⇄ JSON workflows.

Built with TypesXLIFF

These tools are built on TypesXLIFF and show practical uses of the library:

• xliff-validation

A command-line tool for validating XLIFF 2.x files. It uses TypesXLIFF's object model and isValid() methods to check structural and semantic constraints against the XLIFF 2.x specification, and supports custom grammar catalogs for schema resolution.

npm install -g xliff-validation
xliffvalidator -xliff file.xlf

• TypesTerms

A TypeScript library and command-line tool for extracting term candidates from XLIFF 2.x files using the YAKE! statistical algorithm. Supports monolingual extraction (source terms ranked by casing, position, and frequency) and bilingual extraction (source/target translation pairs from segments with confirmed translations).

npm install -g typesterms
termextractor -xliff file.xlf
bilingualextractor -xliff file.xlf

• XliffDeepl

A command-line tool that translates XLIFF 2.1/2.2 files using DeepL's machine translation service. It sends the input XLIFF file to DeepL, waits for the translation to finish and writes the results back to the specified XLIFF output file. Supports custom glossaries and configurable formality levels.

npm install -g xliff-deepl 
xliff-deepl translate -i file.xlf -o translated.xlf

• XliffAI

A command-line tool and library that translates XLIFF 2.x files using AI, sending a group of segments per request instead of one at a time, carrying glossary terms, existing TM matches, and a few segments of surrounding context so terminology and register stay consistent across a batch. Supports seven engines behind one interface - ChatGPT, Claude, Gemini, Ollama, Mistral, Qwen, and Z.ai. AI-proposed translations are never written into a segment's <target>; each one is added as an <mtc:match> candidate for review, leaving the input file untouched.

npm install -g xliffai
xliffai translate -i file.xlf -o translated.xlf -e chatgpt -k $OPENAI_API_KEY -m gpt-5