Reputation: 203
I'm looking for some information on the best way to retrieve data from a local JSON file and handle the response. After browsing through Stack Overflow, I have some mixed thoughts as I've seen multiple ways of doing the same thing (although no explanation on why one may or may not be preferred).
Essentially, I have an Angular app that is utilising a factory to retrieve data from a JSON file; I'm then waiting for the response to resolve in my controller before using it in my html file, similar to the below:
Factory:
comparison.factory('Info', ['$http', function($http) {
var retrievalFile = 'retrievalFile.json';
return {
retrieveInfo: function() {
return $http.get(retrievalFile);
}
}
}]);
Controller:
comparison.controller('comparisonController', ['$scope', 'Info', function($scope, Info) {
Info.retrieveInfo().then(function(response) {
$scope.info = response.data;
});
}]);
My main point of contention is figuring out when it's best to wait for the response to resolve, or if it even matters. I'm toying with the idea of having the factory return the fulfilled promise, and wait for the controller to retrieve the data also. In my view, it's best to abstract all data retrieval out of the controller and into the factory, but I'm not sure if this extends to waiting for the actual data to be returned within the factory itself. With this in mind, I'm confused about whether to opt for option 1 or option 2 and would really appreciate some feedback from more experienced/qualified developers!
Factory:
comparison.factory('Info', ['$http', function($http) {
var retrievalFile = 'retrievalFile.json';
return {
retrieveInfo: function() {
return $http.get(retrievalFile).then(function(response) {
return response.data;
});
}
}
}]);
Controller:
comparison.controller('comparisonController', ['$scope', 'Info', function($scope, Info) {
Info.retrieveInfo().then(function(response) {
$scope.info = response;
});
}]);
Thank you for any input/suggestions in advance!
Upvotes: 11
Views: 1807
Reputation: 7256
It depends on what your controller is expecting and how you set up your application. Generally, I always go with the second option. Its because I usually have global error or success handlers in all api requests and I have a shared api service
. Something like below.
var app = angular.module('app', []);
app.service('ApiService', ['$http', function($http) {
var get = function(url, params) {
$http.get(url, { params: params })
.then(handleSuccess, handleError);
};
// handle your global errors here
// implementation will vary based upon how you handle error
var handleError = function(response) {
return $q.reject(response);
};
// handle your success here
// you can return response.data or response based upon what you want
var handleSuccess = function(response) {
return response.data;
};
}]);
app.service('InfoService', ['ApiService', function(ApiService) {
var retrieveInfo = function() {
return ApiService.get(retrievalFile);
/**
// or return custom object that your controller is expecting
return ApiService.get.then(function(data) {
return new Person(data);
});
**//
};
// I prefer returning public functions this way
// as I can just scroll down to the bottom of service
// to see all public functions at one place rather than
// to scroll through the large file
return { retrieveInfo: retrieveInfo };
}]);
app.controller('InfoController', ['InfoService', function(InfoService) {
InfoService.retrieveInfo().then(function(info) {
$scope.info = info;
});
}])
Or if you are using router you can resolve the data into the controller. Both ngRouter and uiRouter support resolves:
$stateProvider.state({
name: 'info',
url: '/info',
controller: 'InfoController',
template: 'some template',
resolve: {
// this injects a variable called info in your controller
// with a resolved promise that you return here
info: ['InfoService', function(InfoService) {
return InfoService.retrieveInfo();
}]
}
});
// and your controller will be like
// much cleaner right
app.controller('InfoController', ['info', function(info) {
$scope.info = info;
}]);
Upvotes: 4
Reputation: 916
Good question. A couple of points:
Models (M in MVC) are a data representation of your application and will house the data logic. In Angular case this would be a service or factory class as you rightfully pointed out. Why is that well for example:
2.1 AccountsController (might have multiple data models injected)
2.1.1 UserModel
2.1.2 AuthModel
2.1.3 SubscriptionModel
2.1.4 SettingsModel
There are numerous ways to approach the data model approach, but I would say your service class should be the data REST model i.e. getting, storing, caching, validating, etc. I've included a basic example, but suggest you investigate JavaScript OOP as that will help point you in the right direction as to how to build data models, collections, etc.
Below is an example of service class to manage your data.Note I have not tested this code but it should give you a start.
EXAMPLE:
(function () {
'use strict';
ArticleController.$inject = ['$scope', 'Article'];
function ArticleController($scope, Article) {
var vm = this,
getArticles = function () {
return Article.getArticles()
.then(function (result) {
if (result) {
return vm.articles = result;
}
});
};
vm.getArticles = getArticles;
vm.articles = {};
// OR replace vm.articles with $scope if you prefer e.g.
$scope.articles = {};
$scope.userNgClickToInit = function () {
vm.getArticles();
};
// OR an init on document ready
// BUT to honest I would put all init logic in service class so all in calling is init in ctrl and model does the rest
function initArticles() {
vm.getArticles();
// OR chain
vm.getArticles()
.then(getCategories); // doesn't here, just an example
}
initArticles();
}
ArticleModel.$inject = ['$scope', '$http', '$q'];
function ArticleModel($scope, $http, $q) {
var model = this,
URLS = {
FETCH: 'data/articles.json'
},
articles;
function extract(result) {
return result.data;
}
function cacheArticles(result) {
articles = extract(result);
return articles;
}
function findArticle(id) {
return _.find(articles, function (article) {
return article.id === parseInt(id, 10);
})
}
model.getArticles = function () {
return (articles) ? $q.when(articles) : $http.get(URLS.FETCH).then(cacheArticles);
};
model.getArticleById = function (id) {
var deferred = $q.defer();
if (articles) {
deferred.resolve(findArticle(id))
} else {
model.getBookmarks().then(function () {
deferred.resolve(findArticle(id))
})
}
return deferred.promise;
};
model.createArticle = function (article) {
article.id = articles.length;
articles.push(article);
};
model.updateArticle = function (bookmark) {
var index = _.findIndex(articles, function (a) {
return a.id == article.id
});
articles[index] = article;
};
model.deleteArticle = function (article) {
_.remove(articles, function (a) {
return a.id == article.id;
});
};
}
angular.module('app.article.model', [])
.controller('ArticleController', ArticleController)
.service('Article', ArticleModel);
})()
Upvotes: 1
Reputation: 59173
It's really just preference. I like to think of it in terms of API. What is the API you want to expose? Do you want your controller to receive the entire response or do you want your controller to just have the data the response wraps? If you're only ever going to use response.data
then option 2 works great as you never have to deal with anything but the data you're interested in.
A good example is the app we just wrote where I work. We have two apps: a back-end API and our front-end Angular application. We created an API wrapper service in the front-end application. In the service itself we place a .catch
for any of the API endpoints that have documented error codes (we used Swagger to document and define our API). In that .catch
we handle those error codes and return a proper error. When our controllers/directives consume the service they get back a much stricter set of data. If an error occurs then the UI is usually safe to just display the error message sent from the wrapper service and won't have to worry about looking at error codes.
Likewise for successful responses we do much of what you're doing in option 2. In many cases we refine the data down to what is minimally useful in the actual app. In this way we keep a lot of the data churning and formatting in the service and the rest of the app has a lot less to do. For instance, if we need to create an object based on that data we'll just do that in return the object to the promise chain so that controllers aren't doing that all over the place.
Upvotes: 2
Reputation: 5407
I would choose option two, as it your options are really mostly the same. But let see when we add a model structure like a Person
suppose.
comparison.factory('Info', ['$http', function($http) {
var retrievalFile = 'retrievalFile.json';
return {
retrieveInfo: function() {
return $http.get(retrievalFile).then(function(response) {
//we will return a Person...
var data = response.data;
return new Person(data.name, data.age, data.gender);
});
}
}
}]);
This is really simple, but if you have to map more complex data into object models (you retrieve a list of people with their own items... etc), that's when things get more complicated, you will probably want to add a service to handle the mapping between data and models. Well you have another service DataMapper
(example), if you choose your first option you will have to inject DataMapper
into your controller and you will have to make your request through your factory, and map the response with the injected service. And then you probably say, Should I have all this code here? ... Well probably no.
That is an hypothetical case, something that count a lot is how you feel structuring your code, won't architecture it in a way you won't understand. And at the end take a look at this: https://en.wikipedia.org/wiki/SOLID_(object-oriented_design) and research more information about this principles but focused to javascript.
Upvotes: 1