Package for (de)serialization of Valve's KeyValue format (VDF)

Usage no npm install needed!

<script type="module">
  import vdfParser from 'https://cdn.skypack.dev/vdf-parser';



vdf-parser is a library that can convert VDF to JSON and vice versa.

It is mostly based on rossengeorgiev/vdf-parser and includes some features inspired by RoyalBingBong/vdfplus (which lacked some features supported by the former...), but contains many new features and fixes.

Format: https://developer.valvesoftware.com/wiki/KeyValues

VDF may contain comments. However, they are not preserved during decoding.


  • Supports unquoted keys and values
  • Supports keys that appear multiple times by "arrayifying" them – whether they contain child objects or other values
  • Supports uppercase characters in keys and values (and unquoted floats in values)
  • Supports value type recognition and automatic conversion (booleans, integers, floats)
  • Supports non well formed objects (missing newlines), for example: "1" { "label" "#SFUI_WinMatchColon" "value" "#SFUI_Rounds" }
  • Supports conditionals (ex. [$WIN32||$X360])
  • Includes TypeScript types


 * Parse a VDF string into a JavaScript object
 * @param text VDF text
 * @param options Parsing options. Accepts a boolean for backwards compatibility ("types" option defaulting to true)
 * @returns Parsed object
export function parse(text: string, options?: VDFParseOptions | boolean): object;

 * Parse a JavaScript object into a VDF string
 * @param obj The object to stringify
 * @param options Parsing options. Accepts a boolean for backwards compatibility ("pretty" option defaulting to false)
 * @returns VDF string
export function stringify(obj: object, options?: VDFStringifyOptions | boolean): string;

where options are as follows:

interface VDFParseOptions {
     * Attempt to automatically convert numbers and booleans to their correct types, defaults to true
    types: boolean = true;

     * Arrayify the values if they appear multiple times.
     * Enabled by default, because Source does support multiple values with the same key (as separate entries).
     * One may want to disable it if they expect a single value and their code is not prepared for different cases.
     * In such case, the existing text value would be replaced with the new one, and existing object patched with the new values.
    arrayify: boolean = true;

     * If defined, conditionals will be taken into account while parsing the VDF.
     * Provide a list of defined conditionals without leading dollar sign and any found conditionals will be validated against this list.
     * If you provide an empty array, everything requiring any conditional defined will be dropped.
     * Conditions in VDF are processed from left to right.
     * See README and test.js for examples of usage.
    conditionals?: string[];

interface VDFStringifyOptions {
     * Add indentation to the resulting text, defaults to false
    pretty: boolean = false;

     * Indent with the following characters, defaults to a tabulator, requires "pretty" to be set to true
    indent: string = "\t";


This library has support for conditionals. For example, if you parse the following text:

"test1" "test" [$WIN32||$X360]  // preserved
"test2" "test" [!$PS3]          // preserved
"test3" "test" [!$WIN32&&!$OSX] // removed
"test4" "test" [$X360]          // removed

with the following line:

VDF.parse(text, { conditionals: ['WIN32'] });

the first two lines will be preserved and the last two ones will get removed.

The conditional strings must follow the format, be written with capitalized letters, start with a dollar sign and contain no spaces, otherwise parsing will fail. You can open an issue if you encounter any valid VDF file in the wild that does not meet these conditions.


npm install vdf-parser


const VDF = require('vdf-parser');
var parsed = VDF.parse(text);


import * as VDF from 'vdf-parser';
let parsed = VDF.parse(text);