crypto: add key object API

[tniessen]

Preamble: This is a huge change and I will do my best to help with reviewing it. There might still be dozens of places that need some work, but so far, everything seems to be working nicely. I summarized the motivation behind this change in #15113 (comment). There are also lots of possible discussions around this, e.g. whether key derivation should consume / produce key objects etc.

---

This commit makes multiple important changes:

1. A new key object API is introduced. The KeyObject class itself is
   not exposed to users, instead, several new APIs can be used to
   construct key objects: createSecretKey, createPrivateKey and
   createPublicKey. The new API also allows to convert between
   different key formats, and even though the API itself is not
   compatible to the WebCrypto standard in any way, it makes
   interoperability much simpler.

2. Key objects can be used instead of the raw key material in all
   relevant crypto APIs.

3. The handling of asymmetric keys has been unified and greatly
   improved. Node.js now fully supports both PEM-encoded and
   DER-encoded public and private keys.

4. Conversions between buffers and strings have been moved to native
   code for sensitive data such as symmetric keys due to security
   considerations such as zeroing temporary buffers.

5. For compatibility with older versions of the crypto API, this
   change allows to specify Buffers and strings as the "passphrase"
   option when reading or writing an encoded key. Note that this
   can result in unexpected behavior if the password contains a
   null byte.

---

cc @nodejs/crypto @nodejs/security-wg @nodejs/security

Checklist

• make -j4 test (UNIX), or vcbuild test (Windows) passes
• tests and/or benchmarks are included
• documentation is changed or added
• commit message follows commit guidelines

[nodejs-github-bot]

@tniessen build started: https://ci.nodejs.org/blue/organizations/jenkins/node-test-pull-request-lite-pipeline/detail/node-test-pull-request-lite-pipeline/1531/pipeline

[refack]

First pass, looks good.
Some guideline nits.

[TimothyGu]

The new API also allows to convert between different key formats, and even though the API itself is not compatible to the WebCrypto standard in any way, it makes interoperability much simpler.

What are the differences between this new API and the API exposed through web crypto? Could we realistically change this API to be compatible?

[tniessen]

What are the differences between this new API and the API exposed through web crypto? Could we realistically change this API to be compatible?

WebCrypto has a very different design. I originally called the new API CryptoKey but changed it to KeyObject to avoid confusion with the WebCrypto CryptoKey class. Some of the key differences:

• CryptoKey objects can be exported as raw/SPKI/PKCS#8 only and that is done using a separate function, exportKey, not using an instance method. Our options are much more complex, e.g. to support encrypted keys, and I think it makes sense for this to be an instance method.

• The type property is actually compatible to the getType() function of the KeyObject class, and if we decide to implement @addaleax suggestion and turn it into a property, it will match the WebCrypto spec. However, CryptoKey instances have three additional properties that would have a huge impact on the way people can use key objects:

  readonly attribute boolean extractable;
  readonly attribute object algorithm;
  readonly attribute object usages;
  

  We could implement them, but they don't make sense unless we adopt the whole WebCrypto API in my opinion, and as I said before, I don't think we should do that.

• Symmetric CryptoKey objects don't have a property describing the size of the underlying key, and the only way to get the type of an asymmetric key is to parse the algorithm identifier associated with the key.

---

Rebased, old HEAD was 6c92496a7cf304785162ad5cd37632cea916b22b.

[sam-github]

I really like the direction, I think its a good API. Left some comments, mostly on the API. The C++ is a lot to read through, sorry, I ran out of time. I would prefer this not be experimental. I assume it doesn't intentionally break current APIs, and it also isn't particularly complex in terms of its API, so its hard to see it generating much comment or controversy or need to change. Would be great if it can go into 11.x, get some mileage there, and be stable in 12.xx. I doubt it can make it into 10.x, its a lot of churn.

[jasnell]

Definite +1 on this moving forward. Code generally looking good but won't sign off until it's further along.

[ChALkeR]

/cc @joepie91

[tniessen]

One question has come up multiple times: Should the key object API permit X509 certificates when constructing public keys? Personally, I think it makes more sense to provide an API for certificates that allows to extract the public key as a key object because a certificate is not the same as an asymmetric key, but others might feel differently.

cc @nodejs/crypto

[benbucksch]

The new API introduced here is extremely ambiguous and confusing.
We have a "KeyObject" and a "key object", which are not the same type.
We have KeyObject type, key object, key parameter, and key property. All different.
We have key.key.key. Not kidding.