Three-Way Relationship in Firebase - json

Question

I've been learning a lot about denormalised data over the past few months, but I wanted to know if the following is possible in a flattened architecture world. I know how to handle two-way relationships in Firebase, but what about a three-way relationship. Let me explain...
I have 5 items in my database, services, providers, serviceAtProvider, reviews and users. I want to be able to add providers to services and vice versa.
I'd also like there to be a specific page for a provider inside a service, and for there to be reviews linked to the provider at that specific service. The page url might look like this (site.com/serviceId/providerId). The rating is unique to that providerId inside of that serviceId – you can't rate serviceIds or providerIds separately.
I'm not sure how to go about creating such a complex relationship. How would I join the serviceId and providerId in that serviceAtProvider item?
This is what I've got so far:
"services": {
"service1": {
"name": "Hernia Repair",
"providers": {
"provider1": true,
"provider2": true
}
}
},
"providers": {
"provider1": { "name": "The Whittington Hospital" },
"provider2": { "name": "Homerton Hospital" }
},
"serviceAtProvider": {
"service1AtProvider1": { "rating": 4 },
"service1AtProvider2": { "rating": 3 }
},
"reviews": {
"service1AtProvider1": {
"review1": {
"body": "A review from user 1",
"user": "user1"
}
},
"service1AtProvider2": {
"review1": {
"body": "A review from user 2",
"user": "user2"
}
}
},
"users": {
"user1": { "name": "Ben Thomas" },
"user2": { "name": "Beatrix Potter" }
}
I don't know how to create that serviceAtProviderjoin, or how would I go about accessing the service1.name, provider1.name, service1AtProvider1.rating, reviews.service1AtProvider1 on one page. Can anyone explain how to do this?
Also, are there any best practices I should follow?
Any help is appreciated. Thanks in advance!
UPDATE
{
"availableServices": {
"service1": { "name": "Hernia Repair" },
"service2": { "name": "Chemotherapy" }
},
"services": {
"provider": {
"name": "The Whittington Hospital",
"service": {
"service1": {
"rating": 4,
"reviewId1": true
},
"service2": {
"rating": 3,
"reviewId2": true
},
}
}
},
"reviews": {
"reviewId1": {
"review1": {
"rating": 4,
"body": "A review from user 1",
"user": "user1"
}
}
},
"users": {
"user1": { "name": "Raphael Essoo-Snowdon" },
"user2": { "name": "Sharlyne Slassi" }
}
}

Answer 1

I would start by making the data structure a bit simpler and more direct. It's hard to determine the correct data structure for your needs without a detailed use case. I'll do my best to make some generic assumptions here. You'll have to adapt as necessary.
{
"service": {
"service1": { "name": "Foo Service" },
...
},
"provider": {
"provider1": { name: "Foo Place" },
...
},
"ratings": {
"service1": { // service id
"provider1": { // provider id
"average_rating": 4
},
...
},
...
},
"reviews": {
"service1": { // service id
"provider1": { // provider id
"user": "user1",
"rating": 4
},
...
},
...
},
"user": {
"user1": { "name": "Foo Bar" },
...
}
}
Now, to look up the providers who offer a given service, and grab their reviews, I would do the following:
var ref = new Firebase(...);
ref.child('ratings/service1').on('child_added', function(reviewSnap) {
console.log(
'Provider ' + reviewSnap.key(),
'Average rating ' + reviewSnap.val().average_rating
);
});
Joining in the names of the services and providers could be done in several ways. Here's a manual technique:
var ref = new Firebase(...);
ref.child('ratings/service1').on('child_added', accumulateReview);
function accumulateReview(reviewSnap) {
var reviewData = reviewSnap.val();
var reviewid = reviewSnap.key();
fetchService(reviewSnap.parent().key(), function(serviceSnap) {
loadRec('provider', reviewSnap.key(), function(providerSnap) {
console.log('Provider: ', providerSnap.key(), providerSnap.val().name);
console.log('Service: ', serviceSnap.key(), serviceSnap.val().name);
console.log('Average rating: ', reviewData.average_rating);
});
});
}
var serviceCache = {};
function fetchService(serviceid, done) {
// demonstrates creating a local cache for things that will be
// looked up frequently
if( !serviceCache.hasOwnProperty(serviceid) ) {
cacheService(serviceid, done);
}
else {
done(serviceCache[serviceid]);
}
}
function cacheService(serviceid, done) {
loadRec('service', function(ss) {
serviceCache[serviceid] = ss;
fetchService(serviceid, done);
});
}
function loadRec(type, key, done) {
ref.child(type).child(key).once('value', done);
}
I could also automate some of this process with Firebase.util's NormalizedCollection:
var ref = new Firebase(...);
var nc = Firebase.util.NormalizedCollection(
[ref.child('reviews/service1'), 'review'],
ref.child('provider'),
ref.child('user')
)
.select('review.rating', {key: 'provider.name', alias: 'providerName'}, {key: 'user.name', alias: 'userName'})
.ref();
nc.on('child_added', function(snap) {
var data = snap.val();
console.log('Provider', data.providerName);
console.log('User', data.userName);
console.log('Rating', data.rating);
});
Note that nothing here is set in stone. This is how I would approach it. There are probably dozens of structures at least as good or better.