@sanity-typed/client-mock NPM

@sanity-typed/client-mock

Mock @sanity-typed/client for local development and testing

Page Contents

Install

npm install sanity @sanity-typed/client-mock

Usage

Use createClient instead of the createClient from @sanity-typed/client and include initial documents.

product.ts:

// import { defineArrayMember, defineField, defineType } from "sanity";
import {
  defineArrayMember,
  defineField,
  defineType,
} from "@sanity-typed/types";

/** No changes using defineType, defineField, and defineArrayMember */
export const product = defineType({
  name: "product",
  type: "document",
  title: "Product",
  fields: [
    defineField({
      name: "productName",
      type: "string",
      title: "Product name",
      validation: (Rule) => Rule.required(),
    }),
    defineField({
      name: "tags",
      type: "array",
      title: "Tags for item",
      of: [
        defineArrayMember({
          type: "object",
          name: "tag",
          fields: [
            defineField({ type: "string", name: "label" }),
            defineField({ type: "string", name: "value" }),
          ],
        }),
      ],
    }),
  ],
});

sanity.config.ts:

import { deskTool } from "sanity/desk";

// import { defineConfig } from "sanity";
import { defineConfig } from "@sanity-typed/types";
import type { InferSchemaValues } from "@sanity-typed/types";

import { post } from "./schemas/post";
import { product } from "./schemas/product";

/** No changes using defineConfig */
const config = defineConfig({
  projectId: "59t1ed5o",
  dataset: "production",
  plugins: [deskTool()],
  schema: {
    types: [
      product,
      // ...
      post,
    ],
  },
});

export default config;

/** Typescript type of all types! */
export type SanityValues = InferSchemaValues<typeof config>;
/**
 *  SanityValues === {
 *    product: {
 *      _createdAt: string;
 *      _id: string;
 *      _rev: string;
 *      _type: "product";
 *      _updatedAt: string;
 *      productName: string;
 *      tags?: {
 *        _key: string;
 *        _type: "tag";
 *        label?: string;
 *        value?: string;
 *      }[];
 *    };
 *    // ... all your types!
 *  }
 */

mocked-client.ts:

import type { SanityValues } from "sanity.config";

// import { createClient } from "@sanity-typed/client";
import { createClient } from "@sanity-typed/client-mock";

// export const client = createClient({
export const client = createClient<SanityValues>({
  documents: [
    {
      _createdAt: "2011-12-15T03:57:59.213Z",
      _id: "id",
      _rev: "rev",
      _type: "product",
      _updatedAt: "2029-01-01T06:23:59.079Z",
      productName: "Mock Product",
      tags: [
        {
          _key: "key",
          _type: "tag",
          label: "Mock Tag Label",
          value: "Mock Tag Value",
        },
      ],
    },
    // ...
  ],

  // ...@sanity/client options
  projectId: "59t1ed5o",
  dataset: "production",
  useCdn: true,
  apiVersion: "2023-05-23",
});

export const makeTypedQuery = async () =>
  client.fetch('*[_type=="product"]{_id,productName,tags}');
/**
 *  makeTypedQuery() === Promise<[{
 *    _id: "id",
 *    productName: "Mock Product",
 *    tags: [
 *      {
 *        _key: "key",
 *        _type: "tag",
 *        label: "Mock Tag Label",
 *        value: "Mock Tag Value",
 *      },
 *    ],
 *  }]>
 */

Depending on your tree-shaking setup, you'll want to swap between the real client and the mock client. Additionally, using @sanity-typed/faker along with the mock client can be a great way to generate fake data.

mocks.ts:

import { base, en } from "@faker-js/faker";
import config from "sanity.config";

import { sanityConfigToFaker, sanityDocumentsFaker } from "@sanity-typed/faker";

export const getMockDataset = () => {
  const sanityFaker = sanityConfigToFaker(config, {
    faker: { locale: [en, base] },
  });
  /**
   *  typeof sanityFaker === {
   *    [type in keyof SanityValues]: () => SanityValues[type];
   *  }
   */

  const documentsFaker = sanityDocumentsFaker(config, sanityFaker);
  /**
   *  typeof documentsFaker === () => SanityValues[keyof SanityValues][]
   */

  return documentsFaker();
};

swapping-client.ts:

import type { SanityValues } from "sanity.config";

import { createClient as createLiveClient } from "@sanity-typed/client";
import { createClient as createMockClient } from "@sanity-typed/client-mock";

import { getMockDataset } from "./mocks";

const config = {
  projectId: "59t1ed5o",
  dataset: "production",
  useCdn: true,
  apiVersion: "2023-05-23",
};

export const client = process.env.VERCEL
  ? createLiveClient<SanityValues>(config)
  : createMockClient<SanityValues>({ ...config, documents: getMockDataset() });

export const makeTypedQuery = async () =>
  client.fetch('*[_type=="product"]{_id,productName,tags}');

Usage with `groqd` (actually `groq-builder`)

@scottrippey is working on an amazing typed groqd called groq-builder, a schema-aware, strongly-typed GROQ query builder with auto-completion and type-checking for your GROQ queries. When given a function, fetch will provide a GROQ builder for your use:

npm install groq-builder

client-with-groq-builder.ts:

import type { SanityValues } from "sanity.config";

import { createClient } from "@sanity-typed/client";

export const client = createClient<SanityValues>({
  projectId: "59t1ed5o",
  dataset: "production",
  useCdn: true,
  apiVersion: "2023-05-23",
});

export const makeTypedQuery = async () =>
  /** No need for createGroqBuilder, `q` is already typed! */
  client.fetch((q) =>
    q.star
      .filterByType("product")
      .project({ _id: true, productName: true, tags: true })
  );
/**
 *  typeof makeTypedQuery === () => Promise<{
 *    _id: string;
 *    productName: string;
 *    tags: {
 *      _key: string;
 *      _type: "tag";
 *      label?: string;
 *      value?: string;
 *    }[] | null;
 *  }[]>
 */

It will use the returned query and parse directly so you get typed results and runtime validation.

Deciding between using groq-builder or directly typed queries is your decision! There are pros or cons to consider:

Typescript isn't optimized for parsing strings the way @sanity-typed/groq does, which can run into strange errors. Meanwhile, a builder is typescript-first, allowing for complex structures without any issues.
Runtime validation is amazing! It was something I considered and abandoned so it's great to have a solution.
The way @sanity-typed/groq had to be written, it can't do any auto-completion in IDEs like groq-builder can. There was no way around this. Typed objects and methods are going to be superior to parsing a string. Again, typescript wasn't made for it.
There is something to be said for writing queries in their native syntax with less layers between. Writing GROQ queries directly lets you concern yourself only with their documentation, especially when issues arise.
I'm not 100% certain that groq-builder handles all GROQ operations.
groq-builder is currently in beta. You'll need to reference groqd's documentation and sometimes they don't match 1-to-1.

Considerations

GROQ Query results changes in seemingly breaking ways

Similar to parsing, evaluating groq queries will attempt to match how sanity actually evaluates queries. Again, any fixes to match that or changes to groq evaluation will likely not be considered a major change but, rather, a bug fix.

Typescript Errors in IDEs

Often you'll run into an issue where you get typescript errors in your IDE but, when building workspace (either you studio or app using types), there are no errors. This only occurs because your IDE is using a different version of typescript than the one in your workspace. A few debugging steps:

VSCode

The JavaScript and TypeScript Nightly extension (identifier ms-vscode.vscode-typescript-next) creates issues here by design. It will always attempt to use the newest version of typescript instead of your workspace's version. I ended up uninstalling it.
Check that VSCode is actually using your workspace's version even if you've defined the workspace version in .vscode/settings.json. Use TypeScript: Select TypeScript Version to explictly pick the workspace version.
Open any typescript file and you can see which version is being used in the status bar. Please check this (and provide a screenshot confirming this) before creating an issue. Spending hours debugging your issue ony to find that you're not using your workspace's version is very frustrating.

`Type instantiation is excessively deep and possibly infinite`

🚨 CHECK Typescript Errors in IDEs FIRST!!! ISSUES WILL GET CLOSED IMMEDIATELY!!! 🚨

You might run into the dreaded Type instantiation is excessively deep and possibly infinite error when writing GROQ queries. This isn't too uncommon with more complex GROQ queries. Unfortunately, this isn't a completely avoidable problem, as typescript has limits on complexity and parsing types from a string is an inherently complex problem. A set of steps for a workaround:

While not ideal, use @ts-expect-error to disable the error. You could use @ts-ignore instead, but ideally you'd like to remove the comment if a fix is released.
You still likely want manual types. Intersect the returned type with whatever is missing as a patch.
Create a PR in groq/src/specific-issues.test.ts with your issue. #642 is a great example for this. Try to reduce your query and config as much as possible. The goal is a minimal reproduction.
If a PR isn't possible, make an issue with the same content. ie, the query and config you're using. Again, reduce them as much as possible. And then, now that you've done all the work, move it into a PR instead!
I'm one person and some of these issues are quite complex. Take a stab at fixing the bug! There's a ridiculous amount of tests so it's relatively safe to try things out.

People will sometimes create a repo with their issue. Please open a PR with a minimal test instead. Without a PR there will be no tests reflecting your issue and it may appear again in a regression. Forking a github repo to make a PR is a more welcome way to contribute to an open source library.

Breaking Changes

1 to 2

No more `createClient<SanityValues>()(config)`

Removing the double function signature from createClient and renaming dataset to documents:

- const client = createClient<SanityValues>({
-   dataset: [/* ... */],
- })({
+ const client = createClient<SanityValues>({
+ documents: [/* ... */],
  dataset: "production",
  // ...
});

We no longer derive types from your config values. Most of the types weren't significant, but the main loss will be _originalId when the perspective was "previewDrafts".