GraphQL如何仅在需要时使解析器获取数据？

Question

I have two GraphQL type:

type Author {
  id: String!
  name: String!
}

type Book {
  id: String!
  author: Author!
  name: String!
}

In my database, it is implemented by a foreign key inside the books table:

table authors (pseudo code)

`id` INTEGER UNSIGNED
`name` STRING

table books (pseudo code)

`id` INTEGER UNSIGNED
`author_id` INTEGER UNSIGNED REFERENCE `authors.id`
`name` STRING

So when I resolve a GraphQL query like:

query allTheBooks {
  id
  name
  author {
    id
  }
}

I dont want to do another SQL query than the one that will get the list of books, as I already have the data in the books.author_id field.

So I replace the following code (js):

Book: {
  async author(book, args, context, info) {
    // this code trigger a database SELECT query that fetch all the fields of the author row associated with the book
    return book.author().get();
  }
}

by:

Book: {
  async author(book, args, context, info) {
    return {
      id: book.author_id,
    }
  }
}

And it work well !

I only have 1 SQL query where I had 1 + N (N is the number of rows returned by the first query).

But how can I return the other fields without sending a query by field ?

I imagine that it's possible by returning an unique promise for all the fields, that will resolve with the all the content of the author row, but I wonder if there is a cleaner way...

Thank you in advance for your answer !

Answer 1

For now, I find a more or less clean way to do this:

Book: {
  async author(book, args, context, info) {
    let authorPromise = null;

    async function getAuthor() {
      if (authorPromise === null) {
        authorPromise = book.author().get();
      }

      return authorPromise;
    }

    return new Proxy({}, {
      async get({}, name) {
        if (name === "id") {
          return book.author_id;
        }

        if (name in authorFields) {
          const author = await getFile();
          return author[name];
        }
      },
    });
  }
}

It ensure me to do only one query per row, but leave me with another question about the joining of the tables, but I'll post it here :-)

Answer 2

You don't have to make 1 + N query. You can batch N queries into 1. Just check out https://github.com/facebook/dataloader .