MongoDB QueryBuilder

This library exists to ease the creation of MongoDB filter and FindOptions structs when using the MongoDB driver in combination with a JSONAPI query parser.

Installation

go get -u go.jtlabs.io/mongo

Usage

Example code below translated from examples/example.go - for more info, see the example file run running instructions.

package main

import (
	"context"
	"encoding/json"
	"fmt"
	"log"
	"net/http"
	"time"

	mongobuilder "go.jtlabs.io/mongo"
	queryoptions "go.jtlabs.io/query"
	"go.mongodb.org/mongo-driver/bson"
	"go.mongodb.org/mongo-driver/mongo"
	"go.mongodb.org/mongo-driver/mongo/options"
)

// schema for things collection (used by mongo query builder)
var thingsSchema = bson.M{
	"$jsonSchema": bson.M{
		"bsonType": "object",
		"required": []string{"thingID"},
		"properties": bson.M{
			"thingID": bson.M{
				"bsonType":    "string",
				"description": "primary identifier for the thing",
			},
			"created": bson.M{
				"bsonType":    "date",
				"description": "time at which the thing was created",
			},
			"name": bson.M{
				"bsonType":    "string",
				"description": "name of the thing",
			},
			"types": bson.M{
				"bsonType":    "array",
				"description": "type tags for the thing",
				"items": bson.M{
					"bsonType": "string",
				},
			},
		},
	},
}

// create a new MongoDB QueryBuilder (with strict validation set to true)
var builder = querybuilder.NewQueryBuilder("things", thingsSchema, true)

//pointer for the mongo collection to query from
var collection *mongo.Collection

func getThings(w http.ResponseWriter, r *http.Request) {
	opt, err := queryoptions.FromQuerystring(r.URL.RawQuery)
	if err != nil {
		fmt.Fprint(w, err)
		return
	}

	// build a bson.M filter for the Find based on queryoptions filters
	filter, err := builder.Filter(opt)
	if err != nil {
		// NOTE: will only error when strictValidation is true
		fmt.Fprint(w, err)
		return
	}

	// build options (pagination, sorting, field projection) based on queryoptions
	fo, err := builder.FindOptions(opt)
	if err != nil {
		// NOTE: will only error when strictValidation is true
		fmt.Fprint(w, err)
		return
	}

	// now use the filter and options in a Find call to the Mongo collection
	cur, err := collection.Find(context.TODO(), filter, fo)
	if err != nil {
		fmt.Fprint(w, err)
		return
	}

	defer cur.Close(context.TODO())

	data := []struct {
		ThingID string    `bson:"thingID"`
		Name    string    `bson:"name"`
		Created time.Time `bson:"created"`
		Types   []string  `bson:"types"`
	}{}
	if err = cur.All(context.TODO(), &data); err != nil {
		fmt.Fprint(w, err)
		return
	}

	re, _ := json.Marshal(data)
	fmt.Fprint(w, string(re))
}

func main() {
	// create a MongoDB client
	mc, err := mongo.NewClient(options.Client().ApplyURI("mongodb://localhost:27017"))
	if err != nil {
		log.Fatal(err)
	}

	// connect to MongoDB
	if err := mc.Connect(context.TODO()); err != nil {
		log.Fatal(err)
	}

	// create a collection with the schema
	colOpts := options.CreateCollection().SetValidator(thingsSchema)
	mc.Database("things-db").CreateCollection(context.TODO(), "things", colOpts)

	// set the collection pointer
	collection = mc.Database("things-db").Collection("things")

	http.HandleFunc("/v1/things", getThings)
	log.Fatal(http.ListenAndServe(":8080", nil))
}

QueryBuilder

A QueryBuilder struct can be created per MongoDB collection and will look to a JSONSchema defined (and often used as a validator) for the MongoDB collection to build queries and propery coerce types for parameters that are provided in a JSON API Query Options object as filters.

NewQueryBuilder

New QueryBuilder instances can be created using the NewQueryBuilder function:

jsonSchema := bson.D{ /* a JSON schema here... */ }
qb := querybuilder.NewQueryBuilder("collectionName", jsonSchema)

By default, the QueryBuilder does not perform strict schema validation when constructing filter instances and options for Find queries. Strict schema validation can be enabled which will result in an error when trying to build a filter referencing any fields that do not exist within the provided schema or when trying to sort or project based on fields that do not exist in the schema.

// With strict validation enabled
jsonSchema := bson.D{ /* a JSON schema here... */ }
qb := querybuilder.NewQueryBuilder("collectionName", jsonSchema, true)

Filter

The filter method returns a bson.M{} that can be used for excuting Find operations in Mongo.

func getAllThings(w http.ResponseWriter, r *http.Request) {
  // hydrate a QueryOptions index from the request
  opt, err := queryoptions.FromQuerystring(r.URL.RawQuery)
	if err != nil {
		fmt.Fprint(w, err)
		return
	}

  // a query filter in a bson.M based on QueryOptions filters
	f, _ := builder.Filter(opt)

	// options (pagination, sorting, field projection) based on QueryOptions
	fo, _ := builder.FindOptions(opt)

  // now use the filter and options in a Find call to the Mongo collection
	cur, err := collection.Find(context.TODO(), f, fo)
	if err != nil {
		fmt.Fprint(w, err)
		return
	}

  /* do cool stuff with the cursor... */
}

Filter

The QueryOptions (https://github.com/jtlabsio/query) package is great for parsing JSONAPI compliant filter, fields, sort and page details that are provided in the querystring of an API request. There are some nuances in the way in which filters are constructed based on the parameters.

// a query filter in a bson.M based on QueryOptions Filter values
f, err := builder.Filter(opt)
if err != nil {
  // this only occurs when strict schema validation is true
  // and a field is named in the querystring that doesn't actually
  // exist as defined in the schema... this is NOT the default
  // behavior
}

Query Operators

note: to illustrate the concept for example purposes, the querystring samples shown below are not URL encoded, but should be under normal circumstances...

string bsonType

For string bsonType fields in the schema, the following operators can be leveraged with specific querystring hints:

• begins with (i.e. { "name": { "regex": /^term/, "options": "i" } }): ?filter[name]=term*
• ends with (i.e. { "name": { "regex": /term$/, "options": "i" } }): ?ilter[name]=*term
• exact match (i.e. { "name": { "regex": /^term$/ } }): ?filter[name]="term"
• not equal (i.e. { "name": { "$ne": "term" } }): ?filter[name]=!=term
• in (i.e. { "name": { "$in": [ ... ] } }): ?filter[name]=term1,term2,term3,term4
• standard comparison (i.e. { "name": "term" }): ?filter[name]=term
• null is translated to null in the query (i.e. { 'name': null }): ?filter[name]=null

numeric bsonType

For numeric bsonType fields in the schema (int, long, decimal, and double), any values provided in the querystring that are parsed by QueryOptions are coerced to the appropriate type when constructing the filter. Additionally, the following operators can be used in combination with querystring hints:

• less than (i.e. { "age": { "$lt": 5 } }): ?filter[age]=<5
• less than equal (i.e. { "age": { "$lte": 5 } }): ?filter[age]=<=5
• greater than (i.e. { "age": { "$gt": 5 } }): ?filter[age]=>5
• greater than equal (i.e. { "age": { "$gte": 5 } }): ?filter[age]=>=5
• not equals (i.e. { "age": { "$ne": 5 } }): ?filter[age]=!=5
• in (i.e. { "age": { "$in": [1,2,3,4,5] } }): ?filter[age]=1,2,3,4,5
• standard comparison (i.e. { "age": 5 }): ?filter[age]=5

date bsonType

For date bsonType fields in the schema (date and timestamp), any values in the querystring are converted according to RFC3339 and used in the filter. The following operators can be used in combination with querystring hints:

• less than (i.e. { "someDate": { "$lt": new Date("2021-02-16T02:04:05.000Z") } }): ?filter[someDate]=<2021-02-16T02:04:05.000Z
• less than equal (i.e. { "someDate": { "$lte": new Date("2021-02-16T02:04:05.000Z") } }): ?filter[someDate]=<=2021-02-16T02:04:05.000Z
• greater than (i.e. { "someDate": { "$gt": new Date("2021-02-16T02:04:05.000Z") } }): ?filter[someDate]=>2021-02-16T02:04:05.000Z
• greater than equal (i.e. { "someDate": { "$gte": new Date("2021-02-16T02:04:05.000Z") } }): ?filter[someDate]=>=2021-02-16T02:04:05.000Z
• not equals (i.e. { "someDate": { "$ne": new Date("2021-02-16T02:04:05.000Z") } }): ?filter[someDate]=!=2021-02-16T02:04:05.000Z
• in (i.e. { "someDate": { "$in": [ ... ] } }): ?filter[someDate]=2021-02-16T00:00:00.000Z,2021-02-15T00:00:00.000Z
• standard comparison (i.e. { "someDate": new Date("2021-02-16T02:04:05.000Z") }): ?filter[someDate]=2021-02-16T02:04:05.000Z

FindOptions

Pagination, sorting and field projection are defined in options that are provided via QueryOptions can be extracted in used in MongoDB Find calls using the FindOptions method:

fo, err := builder.FindOptions(opt)
if err != nil {
  // this only occurs when strict schema validation is true
  // and a field is named in the querystring that doesn't actually
  // exist as defined in the schema... this is NOT the default
  // behavior
}

Projection

Projection is supported by specifying fields via the fields querystring parameter. By default, no projection is specified and all fields are returned. A - prefix can be used before a field name to exclude it from the results.

• ?fields=name,age,-someDate: includes the fields name, age, but not someDate
• ?fields=-_id,someID,name: excludes _id, but includes name and someID

Pagination

The QueryBuilder attempts to determine if page and size or offset and limit is used as a pagination strategy based on the QueryOptions values.

• ?page[limit]=100&page[offset]=0: sets skip to 0 and limit to 100
• ?page[size]=100&page[page]=1: sets skip to 100 and limit to 100

Sort

Sort is supported by specifying fields in the sort querystring parameter.

• ?sort=-someDate,name: sorts descending by someDate and ascending by name