PDF image using Uint8Array throws an error

[chrishj59]

Bug Report

Description of the problem

I am using an image as the Uint8Array that is downloaded from AWS.

doc.image throws the error:
TypeError [ERR_INVALID_ARG_VALUE]: The argument 'path' must be a string or Uint8Array without null bytes. Received Uint8Array(1239071)

The buffer is Uint8Array as console.log(jpgBuffer instanceof Uint8Array) returns true;

Code sample

const jpgBuffer: Uint8Array = await response.Body.transformToByteArray();
console.log(jpgBuffer instanceof Uint8Array); // true
doc.image(jpgBuffer, 50, 45, { width: 50 });

Your environment

• pdfkit version: ^0.13.0
• Node version: 16.13.1
• Browser version (if applicable):
• Operating System: MacOs 13.3.1

[karlwir]

@chrishj59

I had the same issue when fetching an image from s3.

Had to create a buffer from the Uint8Array before handing it over to doc.image

const object: GetObjectCommandOutput | undefined = await getImageFromS3(imageName);

const byteArray: Uint8Array | undefined = object?.Body.transformToByteArray();

if (byteArray) {
    const buffer = Buffer.from(byteArray);
    doc.image(buffer);
}

[JamBalaya56562]

Good news — this is fixed on master by #1764 ("Use Uint8Array instead of Buffer").

The cause was the type check in PDFImage.open: it used Buffer.isBuffer(src), so a plain Uint8Array (e.g. the one you get back from S3) didn't match any branch and fell through to fs.readFileSync(src) — hence the ERR_INVALID_ARG_VALUE: The argument 'path' must be a string or Uint8Array without null bytes.

That check is now an instanceof test, and since Buffer is a subclass of Uint8Array it covers both:

// lib/image.js
if (src instanceof Uint8Array) {
  data = src;
} else if (src instanceof ArrayBuffer) {
  data = new Uint8Array(src);
}

(source)

I checked this against current master and a plain Uint8Array now works directly, as does a view with a non-zero byteOffset (e.g. from subarray), while Buffer and ArrayBuffer keep working:

const bytes = new Uint8Array(await response.Body.transformToByteArray());
doc.image(bytes, 50, 45, { width: 50 }); // no error

So @karlwir's Buffer.from(byteArray) workaround is no longer needed once you're on a build that includes #1764 (it isn't in a released version yet — it landed after v0.19.1).

Hope that helps!