Hydrate does not call setters on the schema

[zxlin]

Bug report

Steps to Reproduce

Example closely modeled from docs on setters: https://mongoosejs.com/docs/tutorials/getters-setters.html#setters

const userSchema = new mongoose.Schema({
  email: {
    type: String,
    set: v => v.toLowerCase(), // this should be called by hydrate
  }
});

const User = mongoose.model('User', userSchema);

const object = { email: 'TEST@gmail.com' };

const user = User.hydrate(object);

console.log(user); // => { email: 'TEST@gmail.com' }

Expected Behavior

console.log(user); // should yield => { email: 'test@gmail.com' }

Basically the below 2 functions should yield the same thing, but it currently does not

console.log((new User(object).email); // => test@gmail.com
console.log((User.hydrate(object)).email); // => TEST@gmail.com

Versions

mongoose@6.2.10
Node v14.19.1
MongoDB: 4.4.6

Possible Resolutions

I understand why setters might not be called, if it was truly hydrating, then perhaps the setter is redundant, but in the case where a schema field may have a setter and getter, this does become necessary. So option 1 is to call the setter always, option 2 is a parameter option to enable calling the setter. I'm not sure what is the best approach, wanted to get some feedback here.

Thanks for the great work everyone!

[vkarpov15]

This is currently expected behavior. hydrate() is meant to hydrate a document that is loaded from MongoDB, so the assumption is that setters already ran.

We can add an option to hydrate() to allow running setters. But I'm curious what your use case is - why use User.hydrate() instead of new User()?

[zxlin]

@vkarpov15 thanks for the reply! We're currently caching an array of mongoose objects, however, the storage mechanism only allows for plain old JavaScript objects. So we want to hydrate the POJOs when we need to use it and ensure that the mongoose object isn't treated as a "new" object hence we don't use the new constructor.

[vkarpov15]

Where do the POJOs in your cache come from?

[zxlin]

Hi @vkarpov15, the POJOs are from mongoose documents, the result of .toObject()

[vkarpov15]

@zxlin so those documents should have setters run on them at some point, at least when they're created. One alternative approach is to call const doc = new Model(obj); to create a new model instance and run setters before calling toObject(). Does that work for you?

[zxlin]

Hi @vkarpov15, so the flow is something like the below, and let's assume we have a field called text and the model has this schema on the text field:

text: {
  type: String,
  set: (v) => v.toLowerCase(),
  get : (v) => v.toUpperCase(),
}

So the steps:

1. retrieve document from mongodb: const doc = await Model.findOne({ text: 'abc' }) (here doc.text === 'abc')
2. convert to POJO: const obj = doc.toObject() (at this point, obj.text === 'ABC' due to the getter)
3. store obj into cache
4. retrieve obj from cache
5. hydrate the POJO: const doc = Model.hydrate(obj)
6. here is where we run into the issue: while obj has the text field, because .hydrate doesn't call the setter, that field is not set correctly and has the wrong value. In this case, doc.text is upper case 'ABC' which is incorrect and does not match our step 1 which is the expected value of lower case 'abc'

This is the simplified example, hopefully this helps you better understand where we're encountering this issue.