ra-postgraphile

Postgraphile client for react-admin

Typescript API: https://bowlingx.github.io/ra-postgraphile/

Install

$ yarn add ra-postgraphile / npm install ra-postgraphile --save

Usage

The ra-postgraphile data provider accepts 2 arguments:

• client - The ApolloClient instance to use.

• config - optional configuration

  pgDataProvider(client, [config])

The following examples shows the basic usage:

import React, { useEffect, useState } from 'react'
import { Admin, Resource } from 'react-admin'
import { useApolloClient } from '@apollo/react-hooks'
import pgDataProvider from 'ra-postgraphile'
import { PostList, PostEdit, PostCreate } from './posts'
import { CommentList, CommentEdit, CommentCreate } from './posts'

const App = () => {
  const [dataProvider, setDataProvider] = useState(null)
  const client = useApolloClient()

  useEffect(() => {
    ;(async () => {
      const dataProvider = await pgDataProvider(client)
      setDataProvider(() => dataProvider)
    })()
  }, [])

  return (
    dataProvider && (
      
        
        
      
    )
  )
}

export default App

Limitations

The project has been tested only with the following plugins enabled on postgraphile: There is limited support for the postgis plugin and not all input/query types are properly mapped.

const PgSimplifyInflectorPlugin = require('@graphile-contrib/pg-simplify-inflector')
const PgConnectionFilterPlugin = require('postgraphile-plugin-connection-filter')

Please see src/__test_utils/QueryRunner.ts for a minimal example setup.

For full-text search capabilities, the following plugin is also required:

const PostGraphileFulltextFilterPlugin = require('postgraphile-plugin-fulltext-filter')

Configuration

You can pass an optional configuration object:

const pgDataProviderConfig: ProviderOptions = {
  typeMap: {
    YourType: {
      expand: true,
    },
  },
}

• typeMap - allows you to configure complex types and how they should be handled. Please see (src/defaultTypeConfig.ts) for a default mapping. Your config will be merged with the defaults.

TypeConfig options

The following can be configured

Please see (src/types.ts) for detailed types of TypeConfig.

Supported Apollo client versions

If you require support for apollo-client v2 please use version <6, otherwise use >=6.

I recommend to upgrade to @apollo/client v3 packages because the core graphql data provider requires version 3 in it's latest versions.

Supported concepts

ra-postgraphile works on graphql types that are exposed by postgraphile. We currently support the following constructs to query data:

• Tables
• Views
• Functions with custom return type.

Please see here for an example schema.

Primary Keys

react-admin requires each resource to be identified by a unique id. If your resource does not have an id field, we alias the id field to your primaryKey. All types processed by ra-postgraphile require a primary key.

Custom filters

The library will use the postgraphile-plugin-connection-filter filter property by default to apply filters specified (e.g. for ``).

Default Behaviour

Multiple fields will then be merged with and and send to postgraphile e.g.:

{
  "and": [
    {
      "field": {
        "includes": "value"
      }
    }
  ]
}

To customize a filter, you have to provide a parse and format function to the actual input Component. You can then customize the filter operation:

Text Example

import type { FilterSpec } from 'ra-postgraphile'

const startsWithInsensitive = {
  parse: (value: string): FilterSpec => {
    return {
      operator: 'startsWithInsensitive',
      value,
    }
  },
  format: (v: FilterSpec) => v?.value || '',
}

const Filters = (props: Record) => (
  
)

This will then be transformed to

{
  "and": [
    {
      "field": {
        "startsWithInsensitive": "value"
      }
    }
  ]
}

Nested Filters example

It's also possible to define multiple filters per field to create nested filters

const filterThatFiltersMultipleFields = {
  parse: (value: string) => ({
    value: value
      ? [
          {
            firstKey: { every: { someReferenceId: { equalTo: value } } },
            secondKey: { every: { someReferenceId: { equalTo: value } } },
          },
        ]
      : {},
    key: 'or' /* can be or, and etc... */,
  }),
  format: (value: any) => value?.value?.[0]?.firstKey?.every?.someReferenceId?.equalTo,
}

This will be transformed to a filter like

{
  "and": [
    {
      "or": [
        {
          "firstKey": {
            "every": {
              "someReferenceId": {
                "equalTo": "value"
              }
            }
          },
          "secondKey": {
            "every": {
              "someReferenceId": {
                "equalTo": "value"
              }
            }
          }
        }
      ]
    }
  ]
}

Contribution

• Contribution is very welcome 😃.
• Please create your commit messages based on semantic-release syntax and semantics (e.g. properly mark Breaking changes etc.). This let's us automatically create release notes and releases to NPM.

Development

We are using yarn for package management.

To run all tests you have to start the dependent postgres container with docker-compose up -d.