Filtering Schema

index

Introduction

The implementation of a GraphQL Service requires a certain structure to follow for Slickgrid-Universal to work correctly (it will fail if your GraphQL Schema is any different than what is shown below).

Implementation

For the implementation in your code, refer to the GraphQL Service section.

filterBy

The filtering uses filterBy with a structure which we think is flexible enough. The query will have a filterBy argument with an array of filter properties:

  • filterBy: array of filter object(s) (see below)

    • field: field name to filter

    • value: search filter value

    • operator: a GraphQL enum (server side) that can have 1 of these choices:

      • LT, LE, GT, GE, NE, EQ, Contains, Not_Contains, StartsWith, EndsWith, IN, NIN

        • Contains is the default and will be used (by the grid) when operator is not provided

        • IN and NIN (alias to NOT_IN) are mainly used for multi-select filtering

Note: the filterBy order is following the order of how the filter objects were entered in the array.

For example, a filter that would search for a firstName that starts with "John"

  • matches: "John", "Johnny", ...

users (first: 20, offset: 10, filterBy: [{field: firstName, operator: StartsWith, value: 'John'}]) {
  totalCount
  nodes {
    name
    firstName
    lastName
    gender
  }
}

Complex Objects

Dealing with complex objects are a little bit more involving. Because of some limitation with our GraphQL for .Net implementation, we decided to leave field as regular strings and keep the dot notation within the string. For that behavior to work, a new keepArgumentFieldDoubleQuotes property was added that can be passed to the GraphQL initOptions() function. For example, given a complex object field (defined in the Column Definition) that is field: "billing.street" will give this GraphQL query (if you have keepArgumentFieldDoubleQuotes set to True).

Grid Definition example

import { GraphqlService, GraphqlPaginatedResult, GraphqlServiceApi } from '@slickgrid-universal/graphql';

export class Example {
  columnDefinitions: Column[];
  gridOptions: GridOption;
  dataset: any[];

  prepareDatagrid(private graphqlService: GraphqlService) {
    this.columnDefinitions = [
      { id: 'name', name: 'Name', field: 'name', filterable: true, sortable: true },
      { id: 'company', name: 'Company', field: 'company', filterable: true },
      { id: 'billingStreet', name: 'Billing Address Street', field: 'billing.address.street', filterable: true, sortable: true },
      { id: 'billingZip', name: 'Billing Address Zip', field: 'billing.address.zip', filterable: true, sortable: true },
    ];

    this.gridOptions = {
      backendServiceApi: {
        service: new GraphqlService(),
        process: (query) => this.userService.getAll<User[]>(query),
        options: {
          datasetName: 'users',
          columnDefinitions: this.columnDefinitions,
          keepArgumentFieldDoubleQuotes: true // FALSE by default
        }
      }
    };
  }
}

GraphQL Query

// the orderBy/filterBy fields will keep the dot notation while nodes are exploded
{
  users(first: 20, offset: 0, filterBy: [{field: "billing.address.street", operator: EQ, value: "123 Queens Street"}]) {
    totalCount,
    nodes {
      name,
      company,
      billing {
        address {
          street,
          zip
        }
      }
    }
  }
}

From the previous example, you can see that the orderBy keeps the (.) dot notation, while the nodes is exploded as it should billing { street }}. So keep this in mind while building your backend GraphQL service.

Extra Query Arguments

If you want to pass extra arguments outside of the filterBy argument, that will be used for the life of the grid. You can do so by using the extraQueryArguments which accept an array of field/value. For example, let say you want to load a grid of items that belongs to a particular user (with a userId). You can pass the extraQueryArguments to the backendServiceApi options property, like so

Component

this.gridOptions = {
  backendServiceApi: {
    service: new GraphqlService(),
    process: (query) => this.userService.getAll<User[]>(query),
    options: {
      columnDefinitions: this.columnDefinitions,
      datasetName: 'users',
      extraQueryArguments: [{
        field: 'userId',
        value: 123
      }]
    }
  }
};

GraphQL Query

// the orderBy/filterBy fields will keep the dot notation while nodes are exploded
{
  users(first: 20, offset: 0, userId: 123) {
    totalCount,
    nodes {
      name,
      company,
      billing {
        address {
          street,
          zip
        }
      }
    }
  }
}

Override the filter query

Column filters may have a Custom Operator, that acts as a placeholder for you to define your own logic. To do so, the easiest way is to provide the filterQueryOverride callback in the GraphQL Options. This method will be called with BackendServiceFilterQueryOverrideArgs to let you decide dynamically on how the filter should be assembled.

E.g. you could listen for a specific column and the active OperatorType.custom in order to switch the filter to an SQL LIKE search in GraphQL:

Note technically speaking GraphQL isn't a database query language like SQL, it's an application query language. Depending on your configuration, your GraphQL Server might already support regex querying (e.g. Hasura _regex) or you could add your own implementation (e.g. see this SO: https://stackoverflow.com/a/37981802/1212166). Just make sure that whatever custom operator that you want to use, is already included in your GraphQL Schema.

backendServiceApi: {
  options: {
    filterQueryOverride: ({ fieldName, columnDef, columnFilterOperator, searchValues }) => {
      if (columnFilterOperator === OperatorType.custom && columnDef?.id === 'name') {
        // the `operator` is a string, make sure to implement this new operator in your GraphQL Schema
        return { field: fieldName, operator: 'Like', value: searchValues[0] };
      }
    },
  }
}

Last updated