Mutations
A mutation is very similar to a query. It receives a GraphQL request, decides what to do with it, and then returns some data. But because mutations can result in changes to your data, they have their own special Mutation type to differentiate them.
Mutation Steps
When talking about mutations, it’s important to distinguish between the different elements that make up the overall process. For example, let’s imagine that a user submits a form for creating a new document.
- First, this will trigger a request to your GraphQL endpoint, sent by Apollo Client. You can do this either by writing your own mutation higher-order component, or by using one of Vulcan’s pre-made mutation HoCs.
- When the GraphQL endpoint receives this request, it will look for a corresponding mutation resolver. Again you can either code your own resolver, or use Vulcan’s default mutations.
- The mutation resolver then calls a mutator. The mutator is the function that does the actual job of validating the request and mutating your data. The reason for this additional layer is that you’ll often want to mutate data for outside your GraphQL API. By extracting that logic you’re able to call the same exact mutator function whether you’re inserting a new document sent by the front-end or, say, seeding your database with content extracted from an API. As usual, Vulcan offers a set of default mutators.
- Finally, the mutator calls a database connector to perform the actual database operation. By abstracting out the database operation, we’re able to make mutators (and by extension your entire GraphQL API) database-agnostic. This means that you can switch from MongoDB to MySQL without having to modify any of the above three layers.
API
All mutations follow the “single argument” rule. In other words, they all have a single input argument which then contains one or more of the following nested arguments:
filter: afilterobject (see filtering) used to target the document to mutate.id: an alternate way to directly specify theidof the document to mutate.data: the mutation data.
The mutations then all return a data object which contains the mutated document (note that the name of that property is always data no matter the returned document’s type).
Create Mutation
Supported input arguments: data.
Generated type:
1 | createMovie(input: CreateMovieInput) : MovieMutationOutput |
Example mutation:
1 | query testCreate { |
Update Mutation
Supported input arguments: filter, id, data.
Generated type:
1 | updateMovie(input: UpdateMovieInput ) : MovieMutationOutput |
Example mutation:
1 | query testUpdate1 { |
or
1 | query testUpdate2 { |
Upsert Mutation
Supported input arguments: filter, id, data.
Generated type:
1 | upsertMovie(input: UpsertMovieInput ) : MovieMutationOutput |
Example mutation:
1 | query testUpsert { |
Delete Mutation
Supported input arguments: filter, id.
Generated type:
1 | deleteMovie(input: MovieDeleteInput) : MovieMutationOutput |
Example mutation:
1 | query testDelete1 { |
or
1 | query testDelete2 { |
Client
Hooks
useCreate
1 | import React, { useState } from 'react'; |
useUpdate
1 | import React, { useState } from 'react'; |
useDelete
TODO
Higher-Order Components
Vulcan includes three main default higher-order components to make calliing mutations from your React components easier. Note that when using the Forms module, all three mutation HoCs are automatically added for you.
withCreate
This HoC takes the following two options:
collection: the collection to operate on.fragment: specifies the data to ask for as a return value.
And passes on a createMovie (or createPost, createUser, etc.) function to the wrapped component, which takes a single data argument and returns a promise:
1 | this.props |
withUpdate
Same options as withCreate. The returned updateMovie mutation takes three arguments: filter, _id, and data:
filter: afilterinput pointing to the document to modify. See the filtering section._id: an_idused to identify a specific document (note that eitherfilteror_idshould be set).data: the fields to modify or delete (as a list of field name/value pairs with deleted fields set tonull, e.g.{title: 'My New Title', body: 'My new body', status: null}).
1 | this.props |
or
1 | this.props |
withDelete
A single collection option. The returned deleteMovie mutation takes filter and _id arguments:
1 | this.props |
withMutation
In addition to the three main mutation HoCs, The withMutation HoC provides an easy way to call a specific mutation on the server by letting you create ad-hoc mutation containers.
It takes two options:
name: the name of the mutation to call on the server (will also be the name of the prop passed to the component).args: (optional) an object containing the mutation’s arguments and types.
For example, here’s how to wrap the MyComponent component to pass it an addEmailNewsletter function as prop:
1 | const mutationOptions = { |
You would then call the function with:
1 | this.props.addEmailNewsletter({email: 'foo@bar.com'}) |
Server
Resolvers
Vulcan provides a set of default Create, Update, Upsert and Delete mutations you can use to save time:
1 | import { |
The options object can have the following properties:
typeName(String): the resolver’s type name (required).create(Boolean): whether to create acreatemutation (defaults totrue).update(Boolean): whether to create aupdatemutation (defaults totrue).upsert(Boolean): whether to create aupsertmutation (defaults totrue).delete(Boolean): whether to create adeletemutation (defaults totrue).
To learn more about what exactly the default mutations do, you can find their code here.
Custom Mutations
You can also add your own mutations resolvers using addGraphQLMutation and addGraphQLResolvers:
1 | import { addGraphQLMutation, addGraphQLResolvers } from 'meteor/vulcan:core'; |
Mutators
A mutator is the function that actually does the work of mutating data on the server. As opposed to the mutation, which is usually a fairly light function called directly through the GraphQL API, a mutator will take care of the heavy lifting, including validation, callbacks, etc., and should be written in such a way that it can be called from anywhere: a GraphQL API, a REST API, from the server directly, etc.
Default Mutators
Vulcan features three standard mutators: createMutator, updateMutator, and deleteMutator. They are in essence thin wrappers around the standard Mongo insert, update, and remove.
These mutation functions should be defined outside your GraphQL mutation resolvers, so that you’re able to call them from outside a GraphQL context (for example, to seed your database through a server script).
They take the following arguments:
collection: the collection affected.document: the document to mutate.data: (updateMutatoronly) the mutation payload.currentUser: the user performing the operation.validate: whether to validate the operation based on the current user.context: the resolver context.
If validate is set to true, these boilerplate operations will:
- Check that the current user has permission to insert/edit each field.
- Validate the document against collection schema.
- Add
userIdto document (insert only). - Run any validation callbacks (e.g.
movies.new.validate).
They will then run the mutation’s document (or the data object) through the collection’s sync callbacks (e.g. movie.create.sync), perform the operation, and finally run the async callbacks (e.g. movie.create.async).
For example, here is the Posts collection using the createMutator boilerplate mutator:
1 | createMutator({ |
Mutator Callbacks
Default mutators create the following callback hooks for every collection:
typename.operation.validate: called to validate the document or modifier.typename.operation.before: called before the database operation.typename.operation.after: called after the database operation, but before the mutator returns.typename.operation.async: called in an async manner after the mutator returns.
You can learn more about callbacks in the Callbacks section.
Custom Mutators
If you’re writing your own resolvers you can of course also write your own mutators, either by using Vulcan’s Connectors or even by accessing your database directly.
One thing to be aware of though is that by doing this you’ll bypass any callback hooks used by the default mutators, and you’ll also have to take care of your own data validation.