How to cross reference data in MongoDB - php

Question

I am trying to make my first application using PHP and MongoDB.
What I am trying to make is a listing of Fruit trees in a given area, and Also I want to have a list of fruit trees that could be grown in an area.
I come from MySQL background, so if in sql, I would have a trees table that had information about trees. Than I would have a table of actual trees, which would be left joined on a IDs with the trees table,a nd also have location and other information.
However in Mongo, I am not sure how this is done, or HOW it should be done.
Baiscally want I want is a list of trees, and than users can make a reference about their own tree.
Any help or direction on how this would be great.
$db->trees->fruittrees
$db->tress->userstrees
Chris

Answer 1

There are quite a few things missing from your question however:
What I am trying to make is a listing of Fruit trees in a given area, and Also I want to have a list of fruit trees that could be grown in an area.
Smells like a geospatial query: http://docs.mongodb.org/manual/core/geospatial-indexes/ .
So the user would have a location:
{
name: 'sammaye',
location: [107,206]
}
And each growing tree could take advantage of an array of areas:
{
name: 'apple tree'
locations_of_growth: [[74,59],[67,-45]]
}
And you would simply do a $near query on the user comparable to the distance of the Earth ( http://docs.mongodb.org/manual/core/geospatial-indexes/#distance-calculation ) to calculate what trees exist near that user.
The thing about geospatial queries is that they can be constrained with further information as well, so say you want to allow the user to compare the leaf colour of their apple trees in the area to find out how many apple trees in the area have green leafs you can simply add that as a conditiojn within the tree document and add it to your $near.
There is one downside to a geospatial query and it is that you can only have one index per collection ( http://docs.mongodb.org/manual/core/geospatial-indexes/#create-a-geospatial-index ) so this means that if you wish to also provide a list of trees that can be planted in that area then you might need two collections, one called other_trees and one called growable_species. When you want to allow the user to compare their tree to others in the area you query other_trees and when you wish to allow the user to view what trees can be grown in the area you can query growable_trees.
This is of course just one way of doing this.
Edit
I would not "recommend" using Doctrine, it is upto you and it depends on your scenario. Doctrine is a very heavy and surprisingly slow ORM and there are a lot of faster and more Lightweight ORMs for PHP out there: http://docs.mongodb.org/ecosystem/drivers/php-libraries/ if you want to a use an ORM that is, maybe you have a full stack framework or you want to go it alone with the driver.
Also MongoDB does not require JSON PHP (what-ever that is) handling knowledge. All results from MongoDB come into the driver as BSON and are then de-serialzed to a standard dict within PHP, namely an associative array. You do not require JSON PHP handling knowledge here.
Also unlike other answers, be very wary about "store all that information in a single JSON document", namely because MongoDB is BSON not JSON but also because embedding should be thought about very carefully and should be judged upon whether it fits your scenario or not. In fact you will find that embedding anything more than _ids in most cases to other rows can cause problems, but as I said it is scenario dependant.

Answer 2

MongoDB is a document-oriented database. This means you should design your model completely different as you are used in MySQL or another relational database. For example, if you want to store users information (including the trees that they have) you should store all that information in a single JSON document. For more reference please read this
I also recommend using Doctrine which lets you store objects into your MongoDB, so it will be far easier to model your database, you will only need to create an Object Oriented Model that represents your objects. Please refer to this page
For this you will need knowledge in JSON php handling and object oriented programming. One thing that always helps in the beginning of the transition from using MySql to a No-Sql database is to model your database and then denormalize the tables, that way you will know all the information you need to put in your objects/documents.
Hope this helps you

Answer 3

I've found quite often when modeling data in Mongo it depends on how you will use the data in the UI. Mongo isn't good at doing joins (as you know), but you can still do them. You can do them in a middle tier or in the UI.
So for your example, I would probably do the following:
A tree in the tree collection:
{
"_id" : 123,
"name" : "Orange",
"price" : 10,
"climate" : "Any"
}
User:
{
"user" : "Bob",
"selected_trees" : [ 123, 124, 156 ]
}
Then in your UI, you load Bob and make one call to the available tree collection to get the details of the trees for him.
Also a common practice is to actually include the full tree object with the user. You will recoil in disgust at this idea coming from a relational background, but it's very common with Mongo. You have copies of the same data all over. That is normal and expected.