Implement support for the `crypto.subtle.digest` operation #23

oleiade · 2022-11-11T11:05:54Z

This PR addresses #8 .

It adds an implementation for the crypto.subtle.digest operation to the extension.

import { crypto } from "k6/x/webcrypto";

export default function () {
  const d = crypto.subtle.digest("SHA-256", "Hello, world!").then((hash) => {
    const h = buf2hex(hash);
    console.log(h);
  });
}

function buf2hex(buffer) {
  return [...new Uint8Array(buffer)]
    .map((x) => x.toString(16).padStart(2, "0"))
    .join("");
}

It implements the function according to the [Web Crypto API's specification](https://w3c.github.io/webcrypto/#Crypto-method-randomUUID). Please note that this feature is part of a restricted feature set of the API.

Note that the test suite is inherited and adapted from the WPT test suite dedicated to the digest operation.

Checklist

Although not absolutely mandatory, as we have the WPT test suite testing this end2end, I'd like to add more unit tests for this.

Let me know what you think 🙇🏻

webcrypto/subtle_crypto.go

olegbespalov · 2022-11-29T07:17:35Z

webcrypto/subtle_crypto.go

+	// 3.
+	normalizedAlgorithm, err := NormalizeAlgorithm(algorithm, OperationIdentifierDigest)
+	if err != nil {
+		// "if an error occurred, return a Promise rejected with NormalizedAlgorithm"


🤔 I spotted this in WebCrypto specification, but to be honest, I'm not sure if I get this comment. Maybe I'm (and imagine like any other reader of this code) missing some context that we can put here.

What does mean rejected with NormalizedAlgorithm"?

This comment is legacy and reminded me to change the older implementation. I'll drop it.

The Web Crypto specification is very cryptic in certain regards, indeed. It took me a long time and a lot of research to determine what that meant. Most of what I've done on that front is inspired by the Deno implementation.

The algorithm normalization is essentially the step of converting an input Object or string, in a predefined format, into a (set of) internal structures that can be worked on while also ensuring to fail if any validation steps fail. It's a step that's quite easy in a dynamic language but was quite hard to do in a static language. I'm still not happy with the results, but it will do for now :)

I'll add a comment here to explain just that 👍🏻

I feel like this comment made it worse 🤣

The specification says:

Let normalizedAlgorithm be the result of [normalizing an algorithm](https://www.w3.org/TR/WebCryptoAPI/#dfn-normalize-an-algorithm), with alg set to algorithm and op set to "digest". If an error occurred, return a Promise rejected with normalizedAlgorithm.

And as the Specification does not have multiple returns normalizedAlgorithm is just the result of the operation which can be an error (which "occurred").

So from that I would recommend not having a comment as that IMO makes it worse

Suggested change

// "if an error occurred, return a Promise rejected with NormalizedAlgorithm"

webcrypto/subtle_crypto.go

imiric

Hey, great work on this, but I think you're still working on it, right? Since the test doesn't actually work for me, even after converting the data to a TypedArray, and the CI fails. It would be better if this PR was converted to draft if so.

Still, I left some comments from the first look.

webcrypto/subtle_crypto.go

examples/digest.js

imiric · 2022-12-07T12:26:49Z

webcrypto/subtle_crypto.go

+	asObject := data.ToObject(sc.vu.Runtime())
+	arrayBuffer, ok := asObject.Export().(goja.ArrayBuffer)
+	if !ok {
+		reject(fmt.Errorf("could not get ArrayBuffer from data"))


I think this is the reason for the CI failure. I get this error when running the test locally, and when I fix the example to pass a TypedArray. Does this work for you?

I think we'll need a type switch here, but I'm not sure how we can get an ArrayBuffer from a TypedArray, or even assert that data is a TypedArray instance, given that they're private in Goja. 😕 If you can't find a way, I would ask Mihail for help.

It was indeed the issue. I forgot to Get("buffer") on it. Probably a fail during a rebase 🦖
This is fixed now, and the tests pass 🎉

TypedArray is a "view" over ArrayBuffers. They are an indication for the runtime of how to group and interpret bytes in an ArrayBuffer. As such, there's no conversion needed, and one can access their underlying ArrayBuffer using the buffer property 🙇🏻

I'm familiar with TypedArrays, but I was hoping we could get their ArrayBuffer using the Goja Go API, instead of doing it in JS via the buffer property. It would be safer if we didn't rely on duck typing 😅 If that's not possible, then I suppose it's the only thing we can do.

Though your latest change breaks passing an actual ArrayBuffer instance, which should be supported according to the spec, so we should type assert on both data.ToObject(rt) and data.ToObject(rt).Get("buffer").

Also, let's check for null/undefined as well.

I see what you mean. I understand that doing it in Go is not possible now. If I remember correctly, when I experimented with it, treating the value as a goja.ArrayBuffer led to a behavior where the ArrayBuffer was empty. The solution I adopted here is based on this discussion: dop251/goja#379.

Good catch on the ArrayBuffer breaking change 👍🏻 I'll update it accordingly 🙇🏻

webcrypto/subtle_crypto.go

webcrypto/subtle_crypto_test.go

mstoykov · 2023-03-09T13:52:38Z

examples/digest.js

+  crypto.subtle.digest("SHA-256", stringToArrayBuffer("Hello, world!")).then(
+    (hash) => {
+      console.log(arrayBufferToHex(hash));
+    },
+    (err) => {
+      throw err;
+    }
+  );


nit: probably should be using the async/await syntax at this point in time.

mstoykov · 2023-03-09T15:16:07Z

webcrypto/algorithm.go

-		return Algorithm{}, NewError(0, SyntaxError, "algorithm name is not a string")
+	// 2.
+	var initialAlg Algorithm
+	switch algorithm.ExportType().Kind() {


I am not really a fan of using ExportType as IME it is really ... fragile.

Usually I would recommend actually checking the prototype of the value, but this is fine for now.

mstoykov · 2023-03-09T15:18:15Z

webcrypto/algorithm.go

 	// 6.
-	// FIXME: handle this case in later versions
 	err := NewError(0, ImplementationError, fmt.Sprintf("unsupported algorithm type: %s", desiredType))


Why not keep the comment?

This was a rebase mistake, for sure 🤦🏻

mstoykov · 2023-03-09T15:50:14Z

webcrypto/subtle_crypto.go

+	// 3.
+	normalizedAlgorithm, err := NormalizeAlgorithm(algorithm, OperationIdentifierDigest)
+	if err != nil {
+		// "if an error occurred, return a Promise rejected with NormalizedAlgorithm"


I feel like this comment made it worse 🤣

The specification says:

Let normalizedAlgorithm be the result of [normalizing an algorithm](https://www.w3.org/TR/WebCryptoAPI/#dfn-normalize-an-algorithm), with alg set to algorithm and op set to "digest". If an error occurred, return a Promise rejected with normalizedAlgorithm.

And as the Specification does not have multiple returns normalizedAlgorithm is just the result of the operation which can be an error (which "occurred").

So from that I would recommend not having a comment as that IMO makes it worse

Suggested change

// "if an error occurred, return a Promise rejected with NormalizedAlgorithm"

mstoykov

LGTM!

I have some small comments but arguably all of them can be fixed later or not at all.

oleiade · 2023-03-10T17:41:00Z

Thanks a lot for taking the time to review it @mstoykov much appreciated. I'll merge the PR for now, to move forward, and will address your comments separately 🙇🏻

oleiade self-assigned this Nov 11, 2022

oleiade added the enhancement New feature or request label Nov 11, 2022

oleiade force-pushed the feature/digest branch 4 times, most recently from 7c6483a to 4ea9992 Compare November 21, 2022 08:54

oleiade force-pushed the main branch from 3eb23b3 to 5fd1a24 Compare November 21, 2022 10:37

oleiade force-pushed the feature/digest branch 2 times, most recently from 5a4f6e8 to a38c42c Compare November 21, 2022 15:27

oleiade requested review from imiric and olegbespalov November 28, 2022 09:46

oleiade force-pushed the feature/digest branch from a38c42c to aaca416 Compare November 28, 2022 15:06

olegbespalov reviewed Nov 29, 2022

View reviewed changes

olegbespalov mentioned this pull request Nov 30, 2022

Implement support for the subtle.generateKey operation #26

Closed

imiric suggested changes Dec 7, 2022

View reviewed changes

oleiade force-pushed the feature/digest branch 9 times, most recently from ea42bd7 to b987b0a Compare December 9, 2022 09:24

Implement support for the crypto.digest operation

813000a

oleiade force-pushed the feature/digest branch from b987b0a to 813000a Compare December 9, 2022 09:31

oleiade requested a review from mstoykov March 9, 2023 13:24

mstoykov reviewed Mar 9, 2023

View reviewed changes

mstoykov approved these changes Mar 9, 2023

View reviewed changes

oleiade merged commit cacea29 into main Mar 10, 2023

oleiade deleted the feature/digest branch March 10, 2023 17:41

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Implement support for the `crypto.subtle.digest` operation #23

Implement support for the `crypto.subtle.digest` operation #23

oleiade commented Nov 11, 2022

olegbespalov Nov 29, 2022

oleiade Nov 29, 2022

mstoykov Mar 9, 2023

imiric left a comment

imiric Dec 7, 2022

oleiade Dec 8, 2022

imiric Dec 12, 2022 •

edited

oleiade Dec 14, 2022

mstoykov Mar 9, 2023

mstoykov Mar 9, 2023

mstoykov Mar 9, 2023

oleiade Mar 10, 2023

mstoykov Mar 9, 2023

mstoykov left a comment

oleiade commented Mar 10, 2023

Implement support for the crypto.subtle.digest operation #23

Implement support for the crypto.subtle.digest operation #23

Conversation

oleiade commented Nov 11, 2022

Checklist

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

imiric left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

imiric Dec 12, 2022 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mstoykov left a comment

Choose a reason for hiding this comment

oleiade commented Mar 10, 2023

Implement support for the `crypto.subtle.digest` operation #23

Implement support for the `crypto.subtle.digest` operation #23

imiric Dec 12, 2022 •

edited