This shard provides a basic ORM for using MongoDB wth the Crystal programming language. Mongo ORM is based on Granite ORM, and provides basic querying, associations, and model lifecycle capabilities. Mongo ORM is intended to be used with the Amber Framework, but can be used with vanilla crystal or any web framework.

Suggestions, feature requests, bug fixes, and pull requests are always welcome.

First you will need to install MongoDB (unless you are running a remote server), as well as the dependencies for Mongo.cr. On Arch Linux and Ubuntu, this can be done using your package manager as shown below. For other linux distrubutions, you may be able to use the script shown below.

Simply run:

$ sudo pacman -Syu libbson mongodb
$ sudo systemctl start mongodb

Simply run:

sudo apt update
sudo apt install libmongoc-dev libmongoc-1.0-0 libmongoclient-dev

Simply run ./install_linux_deps, the contents of which are shown below:

# install_linux_deps.sh
#!/bin/bash
mkdir -p lib || exit 1
cd lib || exit 1
wget https://github.com/mongodb/mongo-c-driver/releases/download/1.1.0/mongo-c-driver-1.1.0.tar.gz || exit 1
tar -zxvf mongo-c-driver-1.1.0.tar.gz && cd mongo-c-driver-1.1.0/ || exit 1
./configure --prefix=/usr --libdir=/usr/lib || exit 1
make -j4 || exit 1
sudo make install -j4 || exit 1

You can just run ./install_macos_deps, the contents of which are shown below:

#!/bin/bash
curl -LO https://github.com/mongodb/mongo-c-driver/releases/download/1.9.4/mongo-c-driver-1.9.4.tar.gz || exit 1
tar xzf mongo-c-driver-1.9.4.tar.gz || exit 1
cd mongo-c-driver-1.9.4 || exit 1
./configure || exit 1
make || exit 1
sudo make install || exit 1

Next, add the following to the shard.yml file in your project and run shards install:

# shard.yml
dependencies:
  mongo_orm:
    github: sam0x17/mongo_orm

By default (with zero configuration), Mongo ORM will attempt to connect to a database running at localhost:27017 which is the default MongoDB port, with the database name monogo_orm_db.

If the environment variable DATABASE_URL is present, Mongo ORM will connect using this variable instead. You can also specify the database name using the DATABASE_NAME environment variable. For example:

$ DATABASE_URL=mongodb://localhost:11771;DATABASE_NAME=my_db crystal app.cr

If the DATABASE_URL environment variable is not present, Mongo ORM will look for the file config/database.yml within your project directory. If the file exists, Mongo ORM will expect the following format (specify the keys database_url and database_name):

# config/database.yml
database_url: mongodb://localhost:11771
database_name: my_db

MongoDB is different from conventional relational database systems mainly because there is no set-in-stone schema, but instead a wilderness of BSON-based "documents" that may or may not roughly follow the same schema. Declaring static fields works much the same as it does in Granite ORM:

require "mongo_orm"

class User < Mongo::ORM::Document
  field name : String
  field age : Int32
  field deleted_at : Time
  field turned_on : Bool
  timestamps
end

This will declare a model called User with a string field called name, a 32-bit integer field called age, a Time field called deleted_at, a boolean field called turned_on, and the standard created_at and updated_at fields you will recognize from Rails that are created because we specified timestamps.

Note: all Time fields are presently locked into UTC because of some conversion bugs that arise when changing time zones and converting between BSON and crystal models.

To instantiate a User and save it to the database, you can do:

user = User.new
user.name = "Sam"
user.age = 248
user.turned_on = true
user.save!
puts user.inspect # print the created user
puts "id: #{user._id}" # print the ID of the created user

Note that the ID field is named _id, as in standard MongoDB.

You can also use the more compact create notation:

user = User.create name: "Sam", age: 248, turned_on: true

Currently, you can define has_many associations directly on a model, and they will behave roughly the same way they would in standard Rails. For example:

class Group < Mongo::ORM::Document
  field name : String
  has_many :users
end

This defines another model called Group. A group has a String field name, and has an ID-based collection of User documents called users which can be accessed via group.users where group is an instance of Group. To make a User a member of a Group, user.group_id can be set to the document ID of an already-created Group. Note that when you specify that model A has_many model B, the b.a_id field is also automatically created.

In addition to conventional table-style models/documents, Mongo ORM supports the ability to embed documents or collections of documents within documents, as per the BSON standard. This is sometimes a more convenient or more efficient alternative to spreading data out across multiple document collections (tables) and fully leverages the document-based nature of MongoDB. Note that you can also nest embedded documents. See the example below:

class Topic < Mongo::ORM::Document
  embeds top_comment : Comment # e.g. topic.top_comment.body
end

class Tag < Mongo::ORM::EmbeddedDocument
  field topic : String
end

class Comment < Mongo::ORM::EmbeddedDocument
  field body : String
  embeds_many :tags # e.g. comment.tags[0]
end

Mongo ORM also allows you to make use of document fields that are not specified explicitly in the model schema. In fact, it is possible to use Mongo ORM without specifying any model schema at all, however we have provided both options, as schemas provide sane defaults, type checking, and consistency, whereas extended fields (our name for fields not specified in a model schema) make it easy to do dynamic things that would be difficult or impossible in traditional relational databases, and require zero configuration.

For example, suppose you have an Admin collection in your database, and that some (but not all) Admin documents have a field called alias:

admin = Admin.find(4)
puts admin.alias

If the document does indeed have a field named alias, then this will print its value. If such a field is not defined, then nil (nothing) will be printed. This will also work on nested documents, for example blog.header.tag where blog is a Blog document and header is a Header embedded document.

If you know for a fact that alias should be defined on this particular document, you can use the following syntax to be more explicit:

admin = Admin.find(4)
puts admin.alias!

The ! syntax will cause an error to be thrown (undefined method) in the event that the alias field is not defined.