crypto: use WebIDL converters in WebCryptoAPI

[panva]

It is currently the case that what works in Node.js webcrypto will most likely work in every browser's webcrypto as well.

This PR makes the other direction work as well.

Our built in validators are just too strict and intervene in cases when WebCryptoAPI through use of WebIDL should successfully coerse a value.

---

This is a semver-major PRs that contain breaking changes and should be released in the next major version. change because:

• it changes some of the error codes
• no longer accepts string values for BufferSource inputs and options

---

Example errors generated through the webcrypto specific webidl

await crypto.subtle.digest('sha-256', []);

TypeError: Failed to execute 'digest' on 'SubtleCrypto': 2nd argument is not instance of ArrayBuffer, Buffer, TypedArray, or DataView.
    at codedTypeError (node:internal/crypto/webidl:43:15)
    at makeException (node:internal/crypto/webidl:52:10)
    at converters.BufferSource (node:internal/crypto/webidl:216:11)
    at SubtleCrypto.digest (node:internal/crypto/webcrypto:79:28)
  code: 'ERR_INVALID_ARG_TYPE'

await crypto.subtle.generateKey({ name: 'HMAC' }, true, ['sign', 'verify']);

TypeError: Failed to normalize algorithm: passed algorithm can not be converted to 'HmacKeyGenParams' because 'hash' is required in 'HmacKeyGenParams'.
    at codedTypeError (node:internal/crypto/webidl:43:15)
    at makeException (node:internal/crypto/webidl:52:10)
    at Object.HmacKeyGenParams (node:internal/crypto/webidl:321:15)
    at normalizeAlgorithm (node:internal/crypto/util:316:61)
    at SubtleCrypto.generateKey (node:internal/crypto/webcrypto:116:15)
  code: 'ERR_MISSING_OPTION'

await crypto.subtle.generateKey({name:'HMAC', hash: 'SHA-256', length: {}}, false, ['sign', 'verify']);

TypeError: Failed to normalize algorithm: 'length' of 'HmacKeyGenParams' (passed algorithm) is not a finite number.
    at codedTypeError (node:internal/crypto/webidl:43:15)
    at makeException (node:internal/crypto/webidl:52:10)
    at Object.unsigned long (node:internal/crypto/webidl:118:15)
    at converter (node:internal/crypto/webidl:526:36)
    at Object.HmacKeyGenParams (node:internal/crypto/webidl:318:32)
    at normalizeAlgorithm (node:internal/crypto/util:316:61)
    at SubtleCrypto.generateKey (node:internal/crypto/webcrypto:116:15)
  code: 'ERR_INVALID_ARG_TYPE'

I would also love to hide all stack traces from all functions in lib/internal/crypto/webidl.js and also ones stemming from the normalizeAlgorithm function in lib/internal/crypto/util.js.

I've tried my best using the utils exported from lib/internal/errors.js but it lead nowhere, all it did was rename the functions in the trace but they remained visible.

If you have an idea how to do this please reach out to me on the OpenJS Foundation slack. Probably not required for the PR to land though.

---

notable-change PRs with changes that should be highlighted in changelogs. summary

WebCryptoAPI functions' arguments are now coersed and validated as per their WebIDL definitions like in other Web Crypto API implementations. This further improves interoperability with other implementations of Web Crypto API.

[nodejs-github-bot]

Review requested:

• @nodejs/crypto

[targos]

What makes this semver-major?

[aduh95]

What makes this semver-major?

Have a look at the changes in tests, it breaks some existing behavior, definitely semver-major.

[panva]

It's a little bit of everything, some errors are fixed to be correct type, some inputs are no longer incorrectly throwing because they get coersed, some inputs are now correctly throwing. At the same time we're changing some stable APIs thrown error codes.

I'll leave the semverness up to the group to decide when the PR gets ready for review (which it isn't just yet, but thank you regardless!).

[targos]

Have a look at the changes in tests, it breaks some existing behavior, definitely semver-major.

I'm not questioning the label. I'm asking for a description of the major changes. This is to help writing the changelog for it.

[panva]

Have a look at the changes in tests, it breaks some existing behavior, definitely semver-major.

I'm not questioning the label. I'm asking for a description of the major changes. This is to help writing the changelog for it.

Understood, I'll write a more detailed summary when (and if?) this lands.