Parse data: URLs

This package helps you parse data: URLs according to the WHATWG Fetch Standard:

constparseDataURL=require("data-urls");consttextExample=parseDataURL("data:,Hello%2C%20World!");console.log(textExample.mimeType.toString());// "text/plain;charset=US-ASCII"console.log(textExample.body);// Uint8Array(13) [ 72, 101, 108, 108, 111, 44, … ]consthtmlExample=parseDataURL("data:text/html,%3Ch1%3EHello%2C%20World!%3C%2Fh1%3E");console.log(htmlExample.mimeType.toString());// "text/html"console.log(htmlExample.body);// Uint8Array(22) [ 60, 104, 49, 62, 72, 101, … ]constpngExample=parseDataURL("data:image/png;base64,iVBORw0KGgoAAA"+"ANSUhEUgAAAAUAAAAFCAYAAACNbyblAAAAHElEQVQI12P4"+"//8/w38GIAXDIBKE0DHxgljNBAAO9TXL0Y4OHwAAAABJRU"+"5ErkJggg==");console.log(pngExample.mimeType.toString());// "image/png"console.log(pngExample.body);// Uint8Array(85) [ 137, 80, 78, 71, 13, 10, … ]

This package's main module's default export is a function that accepts a string and returns a { mimeType, body } object, or null if the result cannot be parsed as a data: URL.

• The mimeType property is an instance of whatwg-mimetype's MIMEType class.
• The body property is a Uint8Array instance.

As shown in the examples above, you can easily get a stringified version of the MIME type using its toString() method. Read on for more on getting the stringified version of the body.

To decode the body bytes of a parsed data URL, you'll need to use the charset parameter of the MIME type, if any. This contains an encoding label; there are various possible labels for a given encoding. You can use the TextDecoder API for this:

constparseDataURL=require("data-urls");constdataURL=parseDataURL(arbitraryString);// If there's no charset parameter, e.g. if `arbitraryString` is `"data:text/plain,H%C3%A9llo!"`,// then let's guess UTF-8.constencodingLabel=dataURL.mimeType.parameters.get("charset")??"utf-8";constdecoder=newTextDecoder(encodingLabel);constbodyDecoded=decoder.decode(dataURL.body);

(Note that as of the time of this writing in 2026-01, Node.js's built-in TextDecoder has many correctness bugs, so we suggest using the polyfill from the @exodus/bytes package until they are fixed.)

Using the parsed charset is quite important, since the spec requires that if no parseable MIME type is given, the default is "US-ASCII", aka windows-1252—not UTF-8, like you might asume. So for example, given an arbitraryString of "data:,H%E9llo!", the above code snippet will correctly produce a bodyDecoded of "Héllo!" by using the windows-1252 decoder, whereas if you used a UTF-8 decoder you'd get back "H�llo!".

If you are using the whatwg-url package, you may already have a "URL record" object on hand, as produced by that package's parseURL export. In that case, you can use this package's fromURLRecord export to save a bit of work:

const{ parseURL }=require("whatwg-url");constdataURLFromURLRecord=require("data-urls").fromURLRecord;consturlRecord=parseURL("data:,Hello%2C%20World!");constdataURL=dataURLFromURLRecord(urlRecord);

In practice, we expect this functionality only to be used by consumers like jsdom, which are using these packages at a very low level.