(working)johnpapa的Angular Style Guild正體中文版
Angular風格指南 by @john_papa
If you are looking for an opinionated style guide for syntax, conventions, and structuring Angular applications, then step right in. These styles are based on my development experience with Angular, presentations, Pluralsight training courses and working in teams.
如果你在尋找一份關於Angular app的語法、格式以及結構的風格指南的話,那就接著看下去吧。這份風格指南是基於我在Angular上的開發經驗、一些關於這方面的演講、Pluralsight training courses以及在團隊中工作的心得而寫成。
The purpose of this style guide is to provide guidance on building Angular applications by showing the conventions I use and, more importantly, why I choose them.
這份風格指南的目的是在你建立Angular app的過程中提供一些幫助;我會在裡面展示一些我常用的方法,當然更重要的是,為什麼我會這樣做。
If you like this guide, check out my Angular Patterns: Clean Code course at Pluralsight which is a companion to this guide.
如果你喜歡這份指南,可以參考我在Pluralsight上開的課程Angular Patterns: Clean Code,他會是這份指南的好夥伴。
Never work in a vacuum. I find that the Angular community is an incredible group who are passionate about sharing experiences. As such, a friend and Angular expert Todd Motto and I have collaborated on many styles and conventions. We agree on most, and some we diverge. I encourage you to check out Todd’s guidelines to get a sense for his approach and how it compares.
永遠不要閉門造車。我發現Angular社群是一個充滿熱情且樂於分享經驗的地方。因此,我和的朋友-Angular專家Todd Motto在許多風格和規範上合作過。大部份的時候我們的看法是相同的,不過在某些地方仍有所岐異。
我鼓勵你參考Todd’s guidelines並且感受兩者間的差異。
Many of my styles have been from the many pair programming sessions Ward Bell and I have had. My friend Ward has certainly helped influence the ultimate evolution of this guide.
我很多的風格來自於跟Ward Bell結對編程(pair programming)的過程中,這位好友對於這篇指南的幫助很大。
While this guide explains the what, why and how, I find it helpful to see them in practice. This guide is accompanied by a sample application that follows these styles and patterns. You can find the sample application (named modular) here in the modular
folder. Feel free to grab it, clone it, or fork it. Instructions on running it are in its readme.
雖然這份指南會解釋what、why以及how,不過我發現在實例中使用它們會更有幫助。這份指南伴隨著這個範例app(modular),
歡迎grab、clone或fork他。
Translations of this Angular style guide are maintained by the community and can be found here.
這份指南的翻譯版本是由社群來維護的,可以在這裡查看。
一個檔案裡面只定義一個Component
The following example defines the app
module and its dependencies, defines a controller, and defines a factory all in the same file.
下面這個範例在同一個檔案裡面定義了app
module以及他的依賴,另外又定義了一個controller以及一個factory。
/* 不建議 */
angular
.module('app', ['ngRoute'])
.controller('SomeController', SomeController)
.factory('someFactory', someFactory);
function SomeController() { }
function someFactory() { }
The same components are now separated into their own files.
現在我們把每個Componet分別寫在不同的檔案裡
/* 建議 */
// app.module.js
angular
.module('app', ['ngRoute']);
/* 建議 */
// someController.js
angular
.module('app')
.controller('SomeController', SomeController);
function SomeController() { }
/* 建議 */
// someFactory.js
angular
.module('app')
.factory('someFactory', someFactory);
function someFactory() { }
把Angular componets包裝到一個立即式(IIFE, Immediately Invoked Function Expression)裡面。
Why?: An IIFE removes variables from the global scope. This helps prevent variables and function declarations from living longer than expected in the global scope, which also helps avoid variable collisions.
為什麼?: IIFE會把來自全域的變數移除,這可以防止變數以及函式在全域中有比預期長的生命週期,也可以避免產生衝突。
Why?: When your code is minified and bundled into a single file for deployment to a production server, you could have collisions of variables and many global variables. An IIFE protects you against both of these by providing variable scope for each file.
為什麼?: 當你在壓縮或合併文件時,IIFE可以給每個文件一個單獨的function scope,避免變數產生衝突。
/* avoid */
// logger.js
angular
.module('app')
.factory('logger', logger);
// logger function is added as a global variable
function logger() { }
// storage.js
angular
.module('app')
.factory('storage', storage);
// storage function is added as a global variable
function storage() { }
/**
* recommended
*
* no globals are left behind
*/
// logger.js
(function() {
'use strict';
angular
.module('app')
.factory('logger', logger);
function logger() { }
})();
// storage.js
(function() {
'use strict';
angular
.module('app')
.factory('storage', storage);
function storage() { }
})();
Note: For brevity only, the rest of the examples in this guide may omit the IIFE syntax.
註:為了方便,下面的範例會省略IIFE
Note: IIFE’s prevent test code from reaching private members like regular expressions or helper functions which are often good to unit test directly on their own. However you can test these through accessible members or by exposing them through their own component. For example placing helper functions, regular expressions or constants in their own factory or constant.
為sub-modules命名時使用帶有separators的命名規範。(譯註:指的就是parent.sub這樣的方式)
Why?: Unique names help avoid module name collisions. Separators help define modules and their submodule hierarchy. For example app
may be your root module while app.dashboard
and app.users
may be modules that are used as dependencies of app
.
為什麼?:避免module的名稱衝突,而且separators可以幫助定義模組的層次。例如app
可能是你的root module,而app.dashboard
跟app.users
則是依賴於app
的sub module。
宣告module的時候不要使用含有setter語法的變數。(譯註:簡單來說就是不需要宣告一個變數給module)
Why?: With 1 component per file, there is rarely a need to introduce a variable for the module.
為什麼?:當每個檔案中只有一個component的時候,不太需要宣告一個變數給他。
/* avoid */
var app = angular.module('app', [
'ngAnimate',
'ngRoute',
'app.shared',
'app.dashboard'
]);
Instead use the simple setter syntax.
用簡單的setter語法來取代。
/* recommended */
angular
.module('app', [
'ngAnimate',
'ngRoute',
'app.shared',
'app.dashboard'
]);
當使用一個module的時候,儘量用鏈式語法(chaining),而不是使用變數。
Why?: This produces more readable code and avoids variable collisions or leaks.
為什麼?:這樣的寫法可讀性更高,而且可以避免衝突或者洩漏。
/* avoid */
var app = angular.module('app');
app.controller('SomeController', SomeController);
function SomeController() { }
/* recommended */
angular
.module('app')
.controller('SomeController', SomeController);
function SomeController() { }
應該只set一次,並且讓其他的實例取用
Why?: A module should only be created once, then retrieved from that point and after.
為什麼?:一個module應該只建立一次,在這之後的地方都應該取用他
angular.module('app', []);
to set a module.angular.module('app');
to get a module.在callback中用具名函式取代匿名函式
Why?: This produces more readable code, is much easier to debug, and reduces the amount of nested callback code.
為什麼?:這樣的寫法可讀性更高,更容易debug,而且可以避免太多層的callback code。
/* avoid */
angular
.module('app')
.controller('Dashboard', function() { })
.factory('logger', function() { });
/* recommended */
// dashboard.js
angular
.module('app')
.controller('Dashboard', Dashboard);
function Dashboard() { }
// logger.js
angular
.module('app')
.factory('logger', logger);
function logger() { }
controllerAs
syntax over the classic controller with $scope
syntax.用controllerAs
語法來取代一般的語法。
Why?: Controllers are constructed, “newed” up, and provide a single new instance, and the controllerAs
syntax is closer to that of a JavaScript constructor than the classic $scope syntax
.
為什麼?:Controller會被建構,new出來,並且提供一個新的實例;而且controllerAs
在語法上比較接近javascript的建構式。
Why?: It promotes the use of binding to a “dotted” object in the View (e.g. customer.name
instead of name
), which is more contextual, easier to read, and avoids any reference issues that may occur without “dotting”.
為什麼?:促使你在View裡面使用有.
的Object做binding(例如用customer.name
取代name
),這種寫法提高了可讀性、跟上下文的關聯性,並且避免在使用一般寫法時可能會發生的reference issue。
Why?: Helps avoid using $parent
calls in Views with nested controllers.
為什麼?:避免你在多層嵌套的controller裡面使用$parent
呼叫。
<!-- avoid -->
<div ng-controller="Customer">
{{ name }}
</div>
<!-- recommended -->
<div ng-controller="Customer as customer">
{{ customer.name }}
</div>
controllerAs
syntax over the classic controller with $scope
syntax.用controllerAs
語法來取代傳統的controller with $scope
語法。
The controllerAs
syntax uses this
inside controllers which gets bound to $scope
controllerAs
語法在controller中會用this
來取代$scope
Why?: controllerAs
is syntactic sugar over $scope
. You can still bind to the View and still access $scope
methods.
為什麼?:controllerAs
語法只是用來取代$scope
的語法糖,你仍然可以把資料綁定到view上,也可以使用$scope
的方法。
Why?: Helps avoid the temptation of using $scope
methods inside a controller when it may otherwise be better to avoid them or move them to a factory. Consider using $scope
in a factory, or if in a controller just when needed. For example when publishing and subscribing events using $emit
, $broadcast
, or $on
consider moving these uses to a factory and invoke from the controller.
為什麼?:這種語法可以幫助你抵擋在不正確時機使用$scope
的方法的誘惑,例如你有更好的做法,或是他可以搬到factory裡面時。你應該只在必要時才在controller裡使用$scope
,其他的時候應該在factory裡面使
用。例如當你要使用$emit
,$broadcast
, 以及$on
來收發事件時,應該把它們寫在factory裡,然後再從controller取用。
/* avoid */
function Customer($scope) {
$scope.name = {};
$scope.sendMessage = function() { };
}
/* recommended - but see next section */
function Customer() {
this.name = {};
this.sendMessage = function() { };
}
this
when using the controllerAs
syntax. Choose a consistent variable name such as vm
, which stands for ViewModel.當你在使用controllerAs
語法的時候,給this
一個capture variable,例如vm
(ViewModel)。
Why?: The this
keyword is contextual and when used within a function inside a controller may change its context. Capturing the context of this
avoids encountering this problem.
Why?:因為this會受到function scope的影響而指向不同的context,所以在一開始就給他一個capture variable來避免這些問題。
/* avoid */
function Customer() {
this.name = {};
this.sendMessage = function() { };
}
/* recommended */
function Customer() {
var vm = this;
vm.name = {};
vm.sendMessage = function() { };
}
Note: You can avoid any jshint warnings by placing the comment below above the line of code. However it is not needed when the function is named using UpperCasing, as this convention means it is a constructor function, which is what a controller is in Angular.
註:你可以用下面的方式來避免jshint拋出警告。不過這在函式用大寫開頭命名(通常是代表他是構造函式)的時候是不必要的。
/* jshint validthis: true */
var vm = this;
Note: When creating watches in a controller using controller as
, you can watch the vm.*
member using the following syntax. (Create watches with caution as they add more load to the digest cycle.)
註:當你用controllerAs
語法在controller中建立watches時,可以參考下面的方式。(建立watches的時候要謹慎,因為他會對digest cycle帶來負擔)
<input ng-model="vm.title"/>
function SomeController($scope, $log) {
var vm = this;
vm.title = 'Some Title';
$scope.$watch('vm.title', function(current, original) {
$log.info('vm.title was %s', original);
$log.info('vm.title is now %s', current);
});
}
Place bindable members at the top of the controller, alphabetized, and not spread through the controller code.
Why?: Placing bindable members at the top makes it easy to read and helps you instantly identify which members of the controller can be bound and used in the View.
Why?: Setting anonymous functions in-line can be easy, but when those functions are more than 1 line of code they can reduce the readability. Defining the functions below the bindable members (the functions will be hoisted) moves the implementation details down, keeps the bindable members up top, and makes it easier to read.
/* avoid */
function Sessions() {
var vm = this;
vm.gotoSession = function() {
/* ... */
};
vm.refresh = function() {
/* ... */
};
vm.search = function() {
/* ... */
};
vm.sessions = [];
vm.title = 'Sessions';
/* recommended */
function Sessions() {
var vm = this;
vm.gotoSession = gotoSession;
vm.refresh = refresh;
vm.search = search;
vm.sessions = [];
vm.title = 'Sessions';
////////////
function gotoSession() {
/* */
}
function refresh() {
/* */
}
function search() {
/* */
}
Note: If the function is a 1 liner consider keeping it right up top, as long as readability is not affected.
/* avoid */
function Sessions(data) {
var vm = this;
vm.gotoSession = gotoSession;
vm.refresh = function() {
/**
* lines
* of
* code
* affects
* readability
*/
};
vm.search = search;
vm.sessions = [];
vm.title = 'Sessions';
/* recommended */
function Sessions(dataservice) {
var vm = this;
vm.gotoSession = gotoSession;
vm.refresh = dataservice.refresh; // 1 liner is OK
vm.search = search;
vm.sessions = [];
vm.title = 'Sessions';
Use function declarations to hide implementation details. Keep your bindable members up top. When you need to bind a function in a controller, point it to a function declaration that appears later in the file. This is tied directly to the section Bindable Members Up Top. For more details see this post.
Why?: Placing bindable members at the top makes it easy to read and helps you instantly identify which members of the controller can be bound and used in the View. (Same as above.)
Why?: Placing the implementation details of a function later in the file moves that complexity out of view so you can see the important stuff up top.
Why?: Function declaration are hoisted so there are no concerns over using a function before it is defined (as there would be with function expressions).
Why?: You never have to worry with function declarations that moving var a
before var b
will break your code because a
depends on b
.
Why?: Order is critical with function expressions
/**
* avoid
* Using function expressions.
*/
function Avengers(dataservice, logger) {
var vm = this;
vm.avengers = [];
vm.title = 'Avengers';
var activate = function() {
return getAvengers().then(function() {
logger.info('Activated Avengers View');
});
}
var getAvengers = function() {
return dataservice.getAvengers().then(function(data) {
vm.avengers = data;
return vm.avengers;
});
}
vm.getAvengers = getAvengers;
activate();
}
Notice that the important stuff is scattered in the preceding example. In the example below, notice that the important stuff is up top. For example, the members bound to the controller such as vm.avengers
and vm.title
. The implementation details are down below. This is just easier to read.
/*
* recommend
* Using function declarations
* and bindable members up top.
*/
function Avengers(dataservice, logger) {
var vm = this;
vm.avengers = [];
vm.getAvengers = getAvengers;
vm.title = 'Avengers';
activate();
function activate() {
return getAvengers().then(function() {
logger.info('Activated Avengers View');
});
}
function getAvengers() {
return dataservice.getAvengers().then(function(data) {
vm.avengers = data;
return vm.avengers;
});
}
}
Defer logic in a controller by delegating to services and factories.
Why?: Logic may be reused by multiple controllers when placed within a service and exposed via a function.
Why?: Logic in a service can more easily be isolated in a unit test, while the calling logic in the controller can be easily mocked.
Why?: Removes dependencies and hides implementation details from the controller.
Why?: Keeps the controller slim, trim, and focused.
/* avoid */
function Order($http, $q, config, userInfo) {
var vm = this;
vm.checkCredit = checkCredit;
vm.isCreditOk;
vm.total = 0;
function checkCredit() {
var settings = {};
// Get the credit service base URL from config
// Set credit service required headers
// Prepare URL query string or data object with request data
// Add user-identifying info so service gets the right credit limit for this user.
// Use JSONP for this browser if it doesn't support CORS
return $http.get(settings)
.then(function(data) {
// Unpack JSON data in the response object
// to find maxRemainingAmount
vm.isCreditOk = vm.total <= maxRemainingAmount
})
.catch(function(error) {
// Interpret error
// Cope w/ timeout? retry? try alternate service?
// Re-reject with appropriate error for a user to see
});
};
}
/* recommended */
function Order(creditService) {
var vm = this;
vm.checkCredit = checkCredit;
vm.isCreditOk;
vm.total = 0;
function checkCredit() {
return creditService.isOrderTotalOk(vm.total)
.then(function(isOk) { vm.isCreditOk = isOk; })
.catch(showServiceError);
};
}
Define a controller for a view, and try not to reuse the controller for other views. Instead, move reusable logic to factories and keep the controller simple and focused on its view.
Why?: Reusing controllers with several views is brittle and good end to end (e2e) test coverage is required to ensure stability across large applications.
When a controller must be paired with a view and either component may be re-used by other controllers or views, define controllers along with their routes.
Note: If a View is loaded via another means besides a route, then use the ng-controller="Avengers as vm"
syntax.
Why?: Pairing the controller in the route allows different routes to invoke different pairs of controllers and views. When controllers are assigned in the view using ng-controller
, that view is always associated with the same controller.
/* avoid - when using with a route and dynamic pairing is desired */
// route-config.js
angular
.module('app')
.config(config);
function config($routeProvider) {
$routeProvider
.when('/avengers', {
templateUrl: 'avengers.html'
});
}
<!-- avengers.html -->
<div ng-controller="Avengers as vm">
</div>
/* recommended */
// route-config.js
angular
.module('app')
.config(config);
function config($routeProvider) {
$routeProvider
.when('/avengers', {
templateUrl: 'avengers.html',
controller: 'Avengers',
controllerAs: 'vm'
});
}
<!-- avengers.html -->
<div>
</div>
Services are instantiated with the new
keyword, use this
for public methods and variables. Since these are so similar to factories, use a factory instead for consistency.
Note: All Angular services are singletons. This means that there is only one instance of a given service per injector.
// service
angular
.module('app')
.service('logger', logger);
function logger() {
this.logError = function(msg) {
/* */
};
}
// factory
angular
.module('app')
.factory('logger', logger);
function logger() {
return {
logError: function(msg) {
/* */
}
};
}
Factories are singletons and return an object that contains the members of the service.
Expose the callable members of the service (its interface) at the top, using a technique derived from the Revealing Module Pattern.
Why?: Placing the callable members at the top makes it easy to read and helps you instantly identify which members of the service can be called and must be unit tested (and/or mocked).
Why?: This is especially helpful when the file gets longer as it helps avoid the need to scroll to see what is exposed.
Why?: Setting functions as you go can be easy, but when those functions are more than 1 line of code they can reduce the readability and cause more scrolling. Defining the callable interface via the returned service moves the implementation details down, keeps the callable interface up top, and makes it easier to read.
/* avoid */
function dataService() {
var someValue = '';
function save() {
/* */
};
function validate() {
/* */
};
return {
save: save,
someValue: someValue,
validate: validate
};
}
/* recommended */
function dataService() {
var someValue = '';
var service = {
save: save,
someValue: someValue,
validate: validate
};
return service;
////////////
function save() {
/* */
};
function validate() {
/* */
};
}
This way bindings are mirrored across the host object, primitive values cannot update alone using the revealing module pattern
Use function declarations to hide implementation details. Keep your accessible members of the factory up top. Point those to function declarations that appears later in the file. For more details see this post.
Why?: Placing accessible members at the top makes it easy to read and helps you instantly identify which functions of the factory you can access externally.
Why?: Placing the implementation details of a function later in the file moves that complexity out of view so you can see the important stuff up top.
Why?: Function declaration are hoisted so there are no concerns over using a function before it is defined (as there would be with function expressions).
Why?: You never have to worry with function declarations that moving var a
before var b
will break your code because a
depends on b
.
Why?: Order is critical with function expressions
/**
* avoid
* Using function expressions
*/
function dataservice($http, $location, $q, exception, logger) {
var isPrimed = false;
var primePromise;
var getAvengers = function() {
// implementation details go here
};
var getAvengerCount = function() {
// implementation details go here
};
var getAvengersCast = function() {
// implementation details go here
};
var prime = function() {
// implementation details go here
};
var ready = function(nextPromises) {
// implementation details go here
};
var service = {
getAvengersCast: getAvengersCast,
getAvengerCount: getAvengerCount,
getAvengers: getAvengers,
ready: ready
};
return service;
}
/**
* recommended
* Using function declarations
* and accessible members up top.
*/
function dataservice($http, $location, $q, exception, logger) {
var isPrimed = false;
var primePromise;
var service = {
getAvengersCast: getAvengersCast,
getAvengerCount: getAvengerCount,
getAvengers: getAvengers,
ready: ready
};
return service;
////////////
function getAvengers() {
// implementation details go here
}
function getAvengerCount() {
// implementation details go here
}
function getAvengersCast() {
// implementation details go here
}
function prime() {
// implementation details go here
}
function ready(nextPromises) {
// implementation details go here
}
}
Refactor logic for making data operations and interacting with data to a factory. Make data services responsible for XHR calls, local storage, stashing in memory, or any other data operations.
Why?: The controller’s responsibility is for the presentation and gathering of information for the view. It should not care how it gets the data, just that it knows who to ask for it. Separating the data services moves the logic on how to get it to the data service, and lets the controller be simpler and more focused on the view.
Why?: This makes it easier to test (mock or real) the data calls when testing a controller that uses a data service.
Why?: Data service implementation may have very specific code to handle the data repository. This may include headers, how to talk to the data, or other services such as $http. Separating the logic into a data service encapsulates this logic in a single place hiding the implementation from the outside consumers (perhaps a controller), also making it easier to change the implementation.
/* recommended */
// dataservice factory
angular
.module('app.core')
.factory('dataservice', dataservice);
dataservice.$inject = ['$http', 'logger'];
function dataservice($http, logger) {
return {
getAvengers: getAvengers
};
function getAvengers() {
return $http.get('/api/maa')
.then(getAvengersComplete)
.catch(getAvengersFailed);
function getAvengersComplete(response) {
return response.data.results;
}
function getAvengersFailed(error) {
logger.error('XHR Failed for getAvengers.' + error.data);
}
}
}
Note: The data service is called from consumers, such as a controller, hiding the implementation from the consumers, as shown below.
/* recommended */
// controller calling the dataservice factory
angular
.module('app.avengers')
.controller('Avengers', Avengers);
Avengers.$inject = ['dataservice', 'logger'];
function Avengers(dataservice, logger) {
var vm = this;
vm.avengers = [];
activate();
function activate() {
return getAvengers().then(function() {
logger.info('Activated Avengers View');
});
}
function getAvengers() {
return dataservice.getAvengers()
.then(function(data) {
vm.avengers = data;
return vm.avengers;
});
}
}
When calling a data service that returns a promise such as $http
, return a promise in your calling function too.
Why?: You can chain the promises together and take further action after the data call completes and resolves or rejects the promise.
/* recommended */
activate();
function activate() {
/**
* Step 1
* Ask the getAvengers function for the
* avenger data and wait for the promise
*/
return getAvengers().then(function() {
/**
* Step 4
* Perform an action on resolve of final promise
*/
logger.info('Activated Avengers View');
});
}
function getAvengers() {
/**
* Step 2
* Ask the data service for the data and wait
* for the promise
*/
return dataservice.getAvengers()
.then(function(data) {
/**
* Step 3
* set the data and resolve the promise
*/
vm.avengers = data;
return vm.avengers;
});
}
Create one directive per file. Name the file for the directive.
Why?: It is easy to mash all the directives in one file, but difficult to then break those out so some are shared across apps, some across modules, some just for one module.
Why?: One directive per file is easy to maintain.
Note: “Best Practice: Directives should clean up after themselves. You can use
element.on('$destroy', ...)
orscope.$on('$destroy', ...)
to run a clean-up function when the directive is removed” … from the Angular documentation
/* avoid */
/* directives.js */
angular
.module('app.widgets')
/* order directive that is specific to the order module */
.directive('orderCalendarRange', orderCalendarRange)
/* sales directive that can be used anywhere across the sales app */
.directive('salesCustomerInfo', salesCustomerInfo)
/* spinner directive that can be used anywhere across apps */
.directive('sharedSpinner', sharedSpinner);
function orderCalendarRange() {
/* implementation details */
}
function salesCustomerInfo() {
/* implementation details */
}
function sharedSpinner() {
/* implementation details */
}
/* recommended */
/* calendarRange.directive.js */
/**
* @desc order directive that is specific to the order module at a company named Acme
* @example <div acme-order-calendar-range></div>
*/
angular
.module('sales.order')
.directive('acmeOrderCalendarRange', orderCalendarRange);
function orderCalendarRange() {
/* implementation details */
}
/* recommended */
/* customerInfo.directive.js */
/**
* @desc sales directive that can be used anywhere across the sales app at a company named Acme
* @example <div acme-sales-customer-info></div>
*/
angular
.module('sales.widgets')
.directive('acmeSalesCustomerInfo', salesCustomerInfo);
function salesCustomerInfo() {
/* implementation details */
}
/* recommended */
/* spinner.directive.js */
/**
* @desc spinner directive that can be used anywhere across apps at a company named Acme
* @example <div acme-shared-spinner></div>
*/
angular
.module('shared.widgets')
.directive('acmeSharedSpinner', sharedSpinner);
function sharedSpinner() {
/* implementation details */
}
Note: There are many naming options for directives, especially since they can be used in narrow or wide scopes. Choose one that makes the directive and its file name distinct and clear. Some examples are below, but see the Naming section for more recommendations.
When manipulating the DOM directly, use a directive. If alternative ways can be used such as using CSS to set styles or the animation services, Angular templating, ngShow
or ngHide
, then use those instead. For example, if the directive simply hides and shows, use ngHide/ngShow.
Why?: DOM manipulation can be difficult to test, debug, and there are often better ways (e.g. CSS, animations, templates)
Provide a short, unique and descriptive directive prefix such as acmeSalesCustomerInfo
which would be declared in HTML as acme-sales-customer-info
.
Why?: The unique short prefix identifies the directive’s context and origin. For example a prefix of cc-
may indicate that the directive is part of a CodeCamper app while acme-
may indicate a directive for the Acme company.
Note: Avoid ng-
as these are reserved for Angular directives. Research widely used directives to avoid naming conflicts, such as ion-
for the Ionic Framework.
When creating a directive that makes sense as a stand-alone element, allow restrict E
(custom element) and optionally restrict A
(custom attribute). Generally, if it could be its own control, E
is appropriate. General guideline is allow EA
but lean towards implementing as an element when it’s stand-alone and as an attribute when it enhances its existing DOM element.
Why?: It makes sense.
Why?: While we can allow the directive to be used as a class, if the directive is truly acting as an element it makes more sense as an element or at least as an attribute.
Note: EA is the default for Angular 1.3 +
<!-- avoid -->
<div class="my-calendar-range"></div>
/* avoid */
angular
.module('app.widgets')
.directive('myCalendarRange', myCalendarRange);
function myCalendarRange() {
var directive = {
link: link,
templateUrl: '/template/is/located/here.html',
restrict: 'C'
};
return directive;
function link(scope, element, attrs) {
/* */
}
}
<!-- recommended -->
<my-calendar-range></my-calendar-range>
<div my-calendar-range></div>
/* recommended */
angular
.module('app.widgets')
.directive('myCalendarRange', myCalendarRange);
function myCalendarRange() {
var directive = {
link: link,
templateUrl: '/template/is/located/here.html',
restrict: 'EA'
};
return directive;
function link(scope, element, attrs) {
/* */
}
}
Use controller as
syntax with a directive to be consistent with using controller as
with view and controller pairings.
Why?: It makes sense and it’s not difficult.
Note: The directive below demonstrates some of the ways you can use scope inside of link and directive controllers, using controllerAs. I in-lined the template just to keep it all in one place.
Note: Regarding dependency injection, see Manually Identify Dependencies.
Note: Note that the directive’s controller is outside the directive’s closure. This style eliminates issues where the injection gets created as unreachable code after a return
.
<div my-example max="77"></div>
angular
.module('app')
.directive('myExample', myExample);
function myExample() {
var directive = {
restrict: 'EA',
templateUrl: 'app/feature/example.directive.html',
scope: {
max: '='
},
link: linkFunc,
controller: ExampleController,
controllerAs: 'vm',
bindToController: true // because the scope is isolated
};
return directive;
function linkFunc(scope, el, attr, ctrl) {
console.log('LINK: scope.min = %s *** should be undefined', scope.min);
console.log('LINK: scope.max = %s *** should be undefined', scope.max);
console.log('LINK: scope.vm.min = %s', scope.vm.min);
console.log('LINK: scope.vm.max = %s', scope.vm.max);
}
}
ExampleController.$inject = ['$scope'];
function ExampleController($scope) {
// Injecting $scope just for comparison
var vm = this;
vm.min = 3;
console.log('CTRL: $scope.vm.min = %s', $scope.vm.min);
console.log('CTRL: $scope.vm.max = %s', $scope.vm.max);
console.log('CTRL: vm.min = %s', vm.min);
console.log('CTRL: vm.max = %s', vm.max);
}
<!-- example.directive.html -->
<div>hello world</div>
<div>max={{vm.max}}<input ng-model="vm.max"/></div>
<div>min={{vm.min}}<input ng-model="vm.min"/></div>
Use bindToController = true
when using controller as
syntax with a directive when you want to bind the outer scope to the directive’s controller’s scope.
Why?: It makes it easy to bind outer scope to the directive’s controller scope.
Note: bindToController
was introduced in Angular 1.3.0.
<div my-example max="77"></div>
angular
.module('app')
.directive('myExample', myExample);
function myExample() {
var directive = {
restrict: 'EA',
templateUrl: 'app/feature/example.directive.html',
scope: {
max: '='
},
controller: ExampleController,
controllerAs: 'vm',
bindToController: true
};
return directive;
}
function ExampleController() {
var vm = this;
vm.min = 3;
console.log('CTRL: vm.min = %s', vm.min);
console.log('CTRL: vm.max = %s', vm.max);
}
<!-- example.directive.html -->
<div>hello world</div>
<div>max={{vm.max}}<input ng-model="vm.max"/></div>
<div>min={{vm.min}}<input ng-model="vm.min"/></div>
Resolve start-up logic for a controller in an activate
function.
Why?: Placing start-up logic in a consistent place in the controller makes it easier to locate, more consistent to test, and helps avoid spreading out the activation logic across the controller.
Why?: The controller activate
makes it convenient to re-use the logic for a refresh for the controller/View, keeps the logic together, gets the user to the View faster, makes animations easy on the ng-view
or ui-view
, and feels snappier to the user.
Note: If you need to conditionally cancel the route before you start use the controller, use a route resolve instead.
/* avoid */
function Avengers(dataservice) {
var vm = this;
vm.avengers = [];
vm.title = 'Avengers';
dataservice.getAvengers().then(function(data) {
vm.avengers = data;
return vm.avengers;
});
}
/* recommended */
function Avengers(dataservice) {
var vm = this;
vm.avengers = [];
vm.title = 'Avengers';
activate();
////////////
function activate() {
return dataservice.getAvengers().then(function(data) {
vm.avengers = data;
return vm.avengers;
});
}
}
When a controller depends on a promise to be resolved before the controller is activated, resolve those dependencies in the $routeProvider
before the controller logic is executed. If you need to conditionally cancel a route before the controller is activated, use a route resolver.
Use a route resolve when you want to decide to cancel the route before ever transitioning to the View.
Why?: A controller may require data before it loads. That data may come from a promise via a custom factory or $http. Using a route resolve allows the promise to resolve before the controller logic executes, so it might take action based on that data from the promise.
Why?: The code executes after the route and in the controller’s activate function. The View starts to load right away. Data binding kicks in when the activate promise resolves. A “busy” animation can be shown during the view transition (via ng-view
or ui-view
)
Note: The code executes before the route via a promise. Rejecting the promise cancels the route. Resolve makes the new view wait for the route to resolve. A “busy” animation can be shown before the resolve and through the view transition. If you want to get to the View faster and do not require a checkpoint to decide if you can get to the View, consider the controller activate
technique instead.
/* avoid */
angular
.module('app')
.controller('Avengers', Avengers);
function Avengers(movieService) {
var vm = this;
// unresolved
vm.movies;
// resolved asynchronously
movieService.getMovies().then(function(response) {
vm.movies = response.movies;
});
}
/* better */
// route-config.js
angular
.module('app')
.config(config);
function config($routeProvider) {
$routeProvider
.when('/avengers', {
templateUrl: 'avengers.html',
controller: 'Avengers',
controllerAs: 'vm',
resolve: {
moviesPrepService: function(movieService) {
return movieService.getMovies();
}
}
});
}
// avengers.js
angular
.module('app')
.controller('Avengers', Avengers);
Avengers.$inject = ['moviesPrepService'];
function Avengers(moviesPrepService) {
var vm = this;
vm.movies = moviesPrepService.movies;
}
Note: The example below shows the route resolve points to a named function, which is easier to debug and easier to handle dependency injection.
/* even better */
// route-config.js
angular
.module('app')
.config(config);
function config($routeProvider) {
$routeProvider
.when('/avengers', {
templateUrl: 'avengers.html',
controller: 'Avengers',
controllerAs: 'vm',
resolve: {
moviesPrepService: moviesPrepService
}
});
}
function moviePrepService(movieService) {
return movieService.getMovies();
}
// avengers.js
angular
.module('app')
.controller('Avengers', Avengers);
Avengers.$inject = ['moviesPrepService'];
function Avengers(moviesPrepService) {
var vm = this;
vm.movies = moviesPrepService.movies;
}
Note: The code example’s dependency on movieService
is not minification safe on its own. For details on how to make this code minification safe, see the sections on dependency injection and on minification and annotation.
Avoid using the shortcut syntax of declaring dependencies without using a minification-safe approach.
Why?: The parameters to the component (e.g. controller, factory, etc) will be converted to mangled variables. For example, common
and dataservice
may become a
or b
and not be found by Angular.
/* avoid - not minification-safe*/
angular
.module('app')
.controller('Dashboard', Dashboard);
function Dashboard(common, dataservice) {
}
This code may produce mangled variables when minified and thus cause runtime errors.
/* avoid - not minification-safe*/
angular.module('app').controller('Dashboard', d);function d(a, b) { }
Use $inject
to manually identify your dependencies for Angular components.
Why?: This technique mirrors the technique used by ng-annotate
, which I recommend for automating the creation of minification safe dependencies. If ng-annotate
detects injection has already been made, it will not duplicate it.
Why?: This safeguards your dependencies from being vulnerable to minification issues when parameters may be mangled. For example, common
and dataservice
may become a
or b
and not be found by Angular.
Why?: Avoid creating in-line dependencies as long lists can be difficult to read in the array. Also it can be confusing that the array is a series of strings while the last item is the component’s function.
/* avoid */
angular
.module('app')
.controller('Dashboard',
['$location', '$routeParams', 'common', 'dataservice',
function Dashboard($location, $routeParams, common, dataservice) {}
]);
/* avoid */
angular
.module('app')
.controller('Dashboard',
['$location', '$routeParams', 'common', 'dataservice', Dashboard]);
function Dashboard($location, $routeParams, common, dataservice) {
}
/* recommended */
angular
.module('app')
.controller('Dashboard', Dashboard);
Dashboard.$inject = ['$location', '$routeParams', 'common', 'dataservice'];
function Dashboard($location, $routeParams, common, dataservice) {
}
Note: When your function is below a return statement the $inject
may be unreachable (this may happen in a directive). You can solve this by moving the Controller outside of the directive.
/* avoid */
// inside a directive definition
function outer() {
var ddo = {
controller: DashboardPanelController,
controllerAs: 'vm'
};
return ddo;
DashboardPanelController.$inject = ['logger']; // Unreachable
function DashboardPanelController(logger) {
}
}
/* recommended */
// outside a directive definition
function outer() {
var ddo = {
controller: DashboardPanelController,
controllerAs: 'vm'
};
return ddo;
}
DashboardPanelController.$inject = ['logger'];
function DashboardPanelController(logger) {
}
Use $inject
to manually identify your route resolver dependencies for Angular components.
Why?: This technique breaks out the anonymous function for the route resolver, making it easier to read.
Why?: An $inject
statement can easily precede the resolver to handle making any dependencies minification safe.
/* recommended */
function config($routeProvider) {
$routeProvider
.when('/avengers', {
templateUrl: 'avengers.html',
controller: 'AvengersController',
controllerAs: 'vm',
resolve: {
moviesPrepService: moviePrepService
}
});
}
moviePrepService.$inject = ['movieService'];
function moviePrepService(movieService) {
return movieService.getMovies();
}
Use ng-annotate for Gulp or Grunt and comment functions that need automated dependency injection using /** @ngInject */
Why?: This safeguards your code from any dependencies that may not be using minification-safe practices.
Why?: ng-min
is deprecated
I prefer Gulp as I feel it is easier to write, to read, and to debug.
The following code is not using minification safe dependencies.
angular
.module('app')
.controller('Avengers', Avengers);
/* @ngInject */
function Avengers(storageService, avengerService) {
var vm = this;
vm.heroSearch = '';
vm.storeHero = storeHero;
function storeHero() {
var hero = avengerService.find(vm.heroSearch);
storageService.save(hero.name, hero);
}
}
When the above code is run through ng-annotate it will produce the following output with the $inject
annotation and become minification-safe.
angular
.module('app')
.controller('Avengers', Avengers);
/* @ngInject */
function Avengers(storageService, avengerService) {
var vm = this;
vm.heroSearch = '';
vm.storeHero = storeHero;
function storeHero() {
var hero = avengerService.find(vm.heroSearch);
storageService.save(hero.name, hero);
}
}
Avengers.$inject = ['storageService', 'avengerService'];
Note: If ng-annotate
detects injection has already been made (e.g. @ngInject
was detected), it will not duplicate the $inject
code.
Note: When using a route resolver you can prefix the resolver’s function with /* @ngInject */
and it will produce properly annotated code, keeping any injected dependencies minification safe.
Note: Starting from Angular 1.3 you can use the
ngApp
directive’sngStrictDi
parameter to detect any potentially missing magnification safe dependencies. When present the injector will be created in “strict-di” mode causing the application to fail to invoke functions which do not use explicit function annotation (these may not be minification safe). Debugging info will be logged to the console to help track down the offending code. I prefer to only useng-strict-di
for debugging purposes only.<body ng-app="APP" ng-strict-di>
Use gulp-ng-annotate or grunt-ng-annotate in an automated build task. Inject /* @ngInject */
prior to any function that has dependencies.
Why?: ng-annotate will catch most dependencies, but it sometimes requires hints using the /* @ngInject */
syntax.
The following code is an example of a gulp task using ngAnnotate
gulp.task('js', ['jshint'], function() {
var source = pkg.paths.js;
return gulp.src(source)
.pipe(sourcemaps.init())
.pipe(concat('all.min.js', {newLine: ';'}))
// Annotate before uglify so the code get's min'd properly.
.pipe(ngAnnotate({
// true helps add where @ngInject is not used. It infers.
// Doesn't work with resolve, so we must be explicit there
add: true
}))
.pipe(bytediff.start())
.pipe(uglify({mangle: true}))
.pipe(bytediff.stop())
.pipe(sourcemaps.write('./'))
.pipe(gulp.dest(pkg.paths.dev));
});
Use a decorator, at config time using the $provide
service, on the $exceptionHandler
service to perform custom actions when exceptions occur.
Why?: Provides a consistent way to handle uncaught Angular exceptions for development-time or run-time.
Note: Another option is to override the service instead of using a decorator. This is a fine option, but if you want to keep the default behavior and extend it a decorator is recommended.
/* recommended */
angular
.module('blocks.exception')
.config(exceptionConfig);
exceptionConfig.$inject = ['$provide'];
function exceptionConfig($provide) {
$provide.decorator('$exceptionHandler', extendExceptionHandler);
}
extendExceptionHandler.$inject = ['$delegate', 'toastr'];
function extendExceptionHandler($delegate, toastr) {
return function(exception, cause) {
$delegate(exception, cause);
var errorData = {
exception: exception,
cause: cause
};
/**
* Could add the error to a service's collection,
* add errors to $rootScope, log errors to remote web server,
* or log locally. Or throw hard. It is entirely up to you.
* throw exception;
*/
toastr.error(exception.msg, errorData);
};
}
Create a factory that exposes an interface to catch and gracefully handle exceptions.
Why?: Provides a consistent way to catch exceptions that may be thrown in your code (e.g. during XHR calls or promise failures).
Note: The exception catcher is good for catching and reacting to specific exceptions from calls that you know may throw one. For example, when making an XHR call to retrieve data from a remote web service and you want to catch any exceptions from that service and react uniquely.
/* recommended */
angular
.module('blocks.exception')
.factory('exception', exception);
exception.$inject = ['logger'];
function exception(logger) {
var service = {
catcher: catcher
};
return service;
function catcher(message) {
return function(reason) {
logger.error(message, reason);
};
}
}
Handle and log all routing errors using $routeChangeError
.
Why?: Provides a consistent way to handle all routing errors.
Why?: Potentially provides a better user experience if a routing error occurs and you route them to a friendly screen with more details or recovery options.
/* recommended */
var handlingRouteChangeError = false;
function handleRoutingErrors() {
/**
* Route cancellation:
* On routing error, go to the dashboard.
* Provide an exit clause if it tries to do it twice.
*/
$rootScope.$on('$routeChangeError',
function(event, current, previous, rejection) {
if (handlingRouteChangeError) { return; }
handlingRouteChangeError = true;
var destination = (current && (current.title ||
current.name || current.loadedTemplateUrl)) ||
'unknown target';
var msg = 'Error routing to ' + destination + '. ' +
(rejection.msg || '');
/**
* Optionally log using a custom service or $log.
* (Don't forget to inject custom service)
*/
logger.warning(msg, [current]);
/**
* On routing error, go to another route/state.
*/
$location.path('/');
}
);
}
Use consistent names for all components following a pattern that describes the component’s feature then (optionally) its type. My recommended pattern is feature.type.js
. There are 2 names for most assets:
avengers.controller.js
)AvengersController
)Why?: Naming conventions help provide a consistent way to find content at a glance. Consistency within the project is vital. Consistency with a team is important. Consistency across a company provides tremendous efficiency.
Why?: The naming conventions should simply help you find your code faster and make it easier to understand.
Use consistent names for all components following a pattern that describes the component’s feature then (optionally) its type. My recommended pattern is feature.type.js
.
Why?: Provides a consistent way to quickly identify components.
Why?: Provides pattern matching for any automated tasks.
/**
* common options
*/
// Controllers
avengers.js
avengers.controller.js
avengersController.js
// Services/Factories
logger.js
logger.service.js
loggerService.js
/**
* recommended
*/
// controllers
avengers.controller.js
avengers.controller.spec.js
// services/factories
logger.service.js
logger.service.spec.js
// constants
constants.js
// module definition
avengers.module.js
// routes
avengers.routes.js
avengers.routes.spec.js
// configuration
avengers.config.js
// directives
avenger-profile.directive.js
avenger-profile.directive.spec.js
Note: Another common convention is naming controller files without the word controller
in the file name such as avengers.js
instead of avengers.controller.js
. All other conventions still hold using a suffix of the type. Controllers are the most common type of component so this just saves typing and is still easily identifiable. I recommend you choose 1 convention and be consistent for your team. My preference is avengers.controller.js
.
/**
* recommended
*/
// Controllers
avengers.js
avengers.spec.js
Name test specifications similar to the component they test with a suffix of spec
.
Why?: Provides a consistent way to quickly identify components.
Why?: Provides pattern matching for karma or other test runners.
/**
* recommended
*/
avengers.controller.spec.js
logger.service.spec.js
avengers.routes.spec.js
avenger-profile.directive.spec.js
Use consistent names for all controllers named after their feature. Use UpperCamelCase for controllers, as they are constructors.
Why?: Provides a consistent way to quickly identify and reference controllers.
Why?: UpperCamelCase is conventional for identifying object that can be instantiated using a constructor.
/**
* recommended
*/
// avengers.controller.js
angular
.module
.controller('HeroAvengersController', HeroAvengersController);
function HeroAvengersController() { }
Append the controller name with the suffix Controller
.
Why?: The Controller
suffix is more commonly used and is more explicitly descriptive.
/**
* recommended
*/
// avengers.controller.js
angular
.module
.controller('AvengersController', AvengersController);
function AvengersController() { }
Use consistent names for all factories named after their feature. Use camel-casing for services and factories.
Why?: Provides a consistent way to quickly identify and reference factories.
/**
* recommended
*/
// logger.service.js
angular
.module
.factory('logger', logger);
function logger() { }
Use consistent names for all directives using camel-case. Use a short prefix to describe the area that the directives belong (some example are company prefix or project prefix).
Why?: Provides a consistent way to quickly identify and reference components.
/**
* recommended
*/
// avenger-profile.directive.js
angular
.module
.directive('xxAvengerProfile', xxAvengerProfile);
// usage is <xx-avenger-profile> </xx-avenger-profile>
function xxAvengerProfile() { }
When there are multiple modules, the main module file is named app.module.js
while other dependent modules are named after what they represent. For example, an admin module is named admin.module.js
. The respective registered module names would be app
and admin
.
Why?: Provides consistency for multiple module apps, and for expanding to large applications.
Why?: Provides easy way to use task automation to load all module definitions first, then all other angular files (for bundling).
Separate configuration for a module into its own file named after the module. A configuration file for the main app
module is named app.config.js
(or simply config.js
). A configuration for a module named admin.module.js
is named admin.config.js
.
Why?: Separates configuration from module definition, components, and active code.
Why?: Provides an identifiable place to set configuration for a module.
app.route.js
for the main module and admin.route.js
for the admin
module. Even in smaller apps I prefer this separation from the rest of the configuration.Structure your app such that you can L
ocate your code quickly, I
dentify the code at a glance, keep the F
lattest structure you can, and T
ry to stay DRY. The structure should follow these 4 basic guidelines.
Why LIFT?: Provides a consistent structure that scales well, is modular, and makes it easier to increase developer efficiency by finding code quickly. Another way to check your app structure is to ask yourself: How quickly can you open and work in all of the related files for a feature?
When I find my structure is not feeling comfortable, I go back and revisit these LIFT guidelines
L
ocating our code is easyI
dentify code at a glanceF
lat structure as long as we canT
ry to stay DRY (Don’t Repeat Yourself) or T-DRYMake locating your code intuitive, simple and fast.
Why?: I find this to be super important for a project. If the team cannot find the files they need to work on quickly, they will not be able to work as efficiently as possible, and the structure needs to change. You may not know the file name or where its related files are, so putting them in the most intuitive locations and near each other saves a ton of time. A descriptive folder structure can help with this.
/bower_components
/client
/app
/avengers
/blocks
/exception
/logger
/core
/dashboard
/data
/layout
/widgets
/content
index.html
.bower.json
When you look at a file you should instantly know what it contains and represents.
Why?: You spend less time hunting and pecking for code, and become more efficient. If this means you want longer file names, then so be it. Be descriptive with file names and keeping the contents of the file to exactly 1 component. Avoid files with multiple controllers, multiple services, or a mixture. There are deviations of the 1 per file rule when I have a set of very small features that are all related to each other, they are still easily identifiable.
Keep a flat folder structure as long as possible. When you get to 7+ files, begin considering separation.
Why?: Nobody wants to search 7 levels of folders to find a file. Think about menus on web sites … anything deeper than 2 should take serious consideration. In a folder structure there is no hard and fast number rule, but when a folder has 7-10 files, that may be time to create subfolders. Base it on your comfort level. Use a flatter structure until there is an obvious value (to help the rest of LIFT) in creating a new folder.
Be DRY, but don’t go nuts and sacrifice readability.
Why?: Being DRY is important, but not crucial if it sacrifices the others in LIFT, which is why I call it T-DRY. I don’t want to type session-view.html for a view because, well, it’s obviously a view. If it is not obvious or by convention, then I name it.
Have a near term view of implementation and a long term vision. In other words, start small and but keep in mind on where the app is heading down the road. All of the app’s code goes in a root folder named app
. All content is 1 feature per file. Each controller, service, module, view is in its own file. All 3rd party vendor scripts are stored in another root folder and not in the app
folder. I didn’t write them and I don’t want them cluttering my app (bower_components
, scripts
, lib
).
Note: Find more details and reasoning behind the structure at this original post on application structure.
Place components that define the overall layout of the application in a folder named layout
. These may include a shell view and controller may act as the container for the app, navigation, menus, content areas, and other regions.
Why?: Organizes all layout in a single place re-used throughout the application.
Create folders named for the feature they represent. When a folder grows to contain more than 7 files, start to consider creating a folder for them. Your threshold may be different, so adjust as needed.
Why?: A developer can locate the code, identify what each file represents at a glance, the structure is flat as can be, and there is no repetitive nor redundant names.
Why?: The LIFT guidelines are all covered.
Why?: Helps reduce the app from becoming cluttered through organizing the content and keeping them aligned with the LIFT guidelines.
Why?: When there are a lot of files (10+) locating them is easier with a consistent folder structures and more difficult in flat structures.
/**
* recommended
*/
app/
app.module.js
app.config.js
components/
calendar.directive.js
calendar.directive.html
user-profile.directive.js
user-profile.directive.html
layout/
shell.html
shell.controller.js
topnav.html
topnav.controller.js
people/
attendees.html
attendees.controller.js
people.routes.js
speakers.html
speakers.controller.js
speaker-detail.html
speaker-detail.controller.js
services/
data.service.js
localstorage.service.js
logger.service.js
spinner.service.js
sessions/
sessions.html
sessions.controller.js
sessions.routes.js
session-detail.html
session-detail.controller.js
Note: Do not use structuring using folders-by-type. This requires moving to multiple folders when working on a feature and gets unwieldy quickly as the app grows to 5, 10 or 25+ views and controllers (and other features), which makes it more difficult than folder-by-feature to locate files.
/*
* avoid
* Alternative folders-by-type.
* I recommend "folders-by-feature", instead.
*/
app/
app.module.js
app.config.js
app.routes.js
controllers/
attendees.js
session-detail.js
sessions.js
shell.js
speakers.js
speaker-detail.js
topnav.js
directives/
calendar.directive.js
calendar.directive.html
user-profile.directive.js
user-profile.directive.html
services/
dataservice.js
localstorage.js
logger.js
spinner.js
views/
attendees.html
session-detail.html
sessions.html
shell.html
speakers.html
speaker-detail.html
topnav.html
Create small modules that encapsulate one responsibility.
Why?: Modular applications make it easy to plug and go as they allow the development teams to build vertical slices of the applications and roll out incrementally. This means we can plug in new features as we develop them.
Create an application root module whose role is pull together all of the modules and features of your application. Name this for your application.
Why?: Angular encourages modularity and separation patterns. Creating an application root module whose role is to tie your other modules together provides a very straightforward way to add or remove modules from your application.
Only put logic for pulling together the app in the application module. Leave features in their own modules.
Why?: Adding additional roles to the application root to get remote data, display views, or other logic not related to pulling the app together muddies the app module and make both sets of features harder to reuse or turn off.
Why?: The app module becomes a manifest that describes which modules help define the application.
Create modules that represent feature areas, such as layout, reusable and shared services, dashboards, and app specific features (e.g. customers, admin, sales).
Why?: Self contained modules can be added to the application with little or no friction.
Why?: Sprints or iterations can focus on feature areas and turn them on at the end of the sprint or iteration.
Why?: Separating feature areas into modules makes it easier to test the modules in isolation and reuse code.
Create modules that represent reusable application blocks for common services such as exception handling, logging, diagnostics, security, and local data stashing.
Why?: These types of features are needed in many applications, so by keeping them separated in their own modules they can be application generic and be reused across applications.
The application root module depends on the app specific feature modules and any shared or reusable modules.
Why?: The main app module contains a quickly identifiable manifest of the application’s features.
Why?: Each feature area contains a manifest of what it depends on, so it can be pulled in as a dependency in other applications and still work.
Why?: Intra-App features such as shared data services become easy to locate and share from within app.core
(choose your favorite name for this module).
Note: This is a strategy for consistency. There are many good options here. Choose one that is consistent, follows Angular’s dependency rules, and is easy to maintain and scale.
My structures vary slightly between projects but they all follow these guidelines for structure and modularity. The implementation may vary depending on the features and the team. In other words, don’t get hung up on an exact like-for-like structure but do justify your structure using consistency, maintainability, and efficiency in mind.
In a small app, you can also consider putting all the shared dependencies in the app module where the feature modules have no direct dependencies. This makes it easier to maintain the smaller application, but makes it harder to reuse modules outside of this application.
Inject code into module configuration that must be configured before running the angular app. Ideal candidates include providers and constants.
Why?: This makes it easier to have a less places for configuration.
angular
.module('app')
.config(configure);
configure.$inject =
['routerHelperProvider', 'exceptionHandlerProvider', 'toastr'];
function configure (routerHelperProvider, exceptionHandlerProvider, toastr) {
exceptionHandlerProvider.configure(config.appErrorPrefix);
configureStateHelper();
toastr.options.timeOut = 4000;
toastr.options.positionClass = 'toast-bottom-right';
////////////////
function configureStateHelper() {
routerHelperProvider.configure({
docTitle: 'NG-Modular: '
});
}
}
Any code that needs to run when an application starts should be declared in a factory, exposed via a function, and injected into the run block.
Why?: Code directly in a run block can be difficult to test. Placing in a factory makes it easier to abstract and mock.
angular
.module('app')
.run(runBlock);
runBlock.$inject = ['authenticator', 'translator'];
function runBlock(authenticator, translator) {
authenticator.initialize();
translator.initialize();
}
Use $document
and $window
instead of document
and window
.
Why?: These services are wrapped by Angular and more easily testable than using document and window in tests. This helps you avoid having to mock document and window yourself.
Use $timeout
and $interval
instead of setTimeout
and setInterval
.
Why?: These services are wrapped by Angular and more easily testable and handle Angular’s digest cycle thus keeping data binding in sync.
Unit testing helps maintain clean code, as such I included some of my recommendations for unit testing foundations with links for more information.
Write a set of tests for every story. Start with an empty test and fill them in as you write the code for the story.
Why?: Writing the test descriptions helps clearly define what your story will do, will not do, and how you can measure success.
it('should have Avengers controller', function() {
// TODO
});
it('should find 1 Avenger when filtered by name', function() {
// TODO
});
it('should have 10 Avengers', function() {
// TODO (mock data?)
});
it('should return Avengers via XHR', function() {
// TODO ($httpBackend?)
});
// and so on
Use Jasmine or Mocha for unit testing.
Why?: Both Jasmine and Mocha are widely used in the Angular community. Both are stable, well maintained, and provide robust testing features.
Note: When using Mocha, also consider choosing an assert library such as Chai. I prefer Mocha.
Use Karma as a test runner.
Why?: Karma is easy to configure to run once or automatically when you change your code.
Why?: Karma hooks into your Continuous Integration process easily on its own or through Grunt or Gulp.
Why?: Some IDE’s are beginning to integrate with Karma, such as WebStorm and Visual Studio.
Why?: Karma works well with task automation leaders such as Grunt (with grunt-karma) and Gulp (with gulp-karma).
Use Sinon for stubbing and spying.
Why?: Sinon works well with both Jasmine and Mocha and extends the stubbing and spying features they offer.
Why?: Sinon makes it easier to toggle between Jasmine and Mocha, if you want to try both.
Why?: Sinon has descriptive messages when tests fail the assertions.
Use PhantomJS to run your tests on a server.
Why?: PhantomJS is a headless browser that helps run your tests without needing a “visual” browser. So you do not have to install Chrome, Safari, IE, or other browsers on your server.
Note: You should still test on all browsers in your environment, as appropriate for your target audience.
Run JSHint on your tests.
Why?: Tests are code. JSHint can help identify code quality issues that may cause the test to work improperly.
Relax the rules on your test code to allow for common globals such as describe
and expect
. Relax the rules for expressions, as Mocha uses these.
Why?: Your tests are code and require the same attention and code quality rules as all of your production code. However, global variables used by the testing framework, for example, can be relaxed by including this in your test specs.
/* jshint -W117, -W030 */
Or you can add the following to your JSHint Options file.
"jasmine": true,
"mocha": true,
Place unit test files (specs) side-by-side with your client code. Place specs that cover server integration or test multiple components in a separate tests
folder.
Why?: Unit tests have a direct correlation to a specific component and file in source code.
Why?: It is easier to keep them up to date since they are always in sight. When coding whether you do TDD or test during development or test after development, the specs are side-by-side and never out of sight nor mind, and thus more likely to be maintained which also helps maintain code coverage.
Why?: When you update source code it is easier to go update the tests at the same time.
Why?: Placing them side-by-side makes it easy to find them and easy to move them with the source code if you move the source.
Why?: Having the spec nearby makes it easier for the source code reader to learn how the component is supposed to be used and to discover its known limitations.
Why?: Separating specs so they are not in a distributed build is easy with grunt or gulp.
/src/client/app/customers/customer-detail.controller.js
/customer-detail.controller.spec.js
/customers.controller.spec.js
/customers.controller-detail.spec.js
/customers.module.js
/customers.route.js
/customers.route.spec.js
Use subtle animations with Angular to transition between states for views and primary visual elements. Include the ngAnimate module. The 3 keys are subtle, smooth, seamless.
Why?: Subtle animations can improve User Experience when used appropriately.
Why?: Subtle animations can improve perceived performance as views transition.
Use short durations for animations. I generally start with 300ms and adjust until appropriate.
Why?: Long animations can have the reverse affect on User Experience and perceived performance by giving the appearance of a slow application.
Use animate.css for conventional animations.
Why?: The animations that animate.css provides are fast, smooth, and easy to add to your application.
Why?: Provides consistency in your animations.
Why?: animate.css is widely used and tested.
Note: See this great post by Matias Niemel? on Angular animations
If planning to produce documentation, use jsDoc
syntax to document function names, description, params and returns. Use @namespace
and @memberOf
to match your app structure.
Why?: You can generate (and regenerate) documentation from your code, instead of writing it from scratch.
Why?: Provides consistency using a common industry tool.
/**
* Logger Factory
* @namespace Factories
*/
(function() {
angular
.module('app')
.factory('logger', logger);
/**
* @namespace Logger
* @desc Application wide logger
* @memberOf Factories
*/
function logger($log) {
var service = {
logError: logError
};
return service;
////////////
/**
* @name logError
* @desc Logs errors
* @param {String} msg Message to log
* @returns {String}
* @memberOf Factories.Logger
*/
function logError(msg) {
var loggedMsg = 'Error: ' + msg;
$log.error(loggedMsg);
return loggedMsg;
};
}
})();
Use JS Hint for linting your JavaScript and be sure to customize the JS Hint options file and include in source control. See the JS Hint docs for details on the options.
Why?: Provides a first alert prior to committing any code to source control.
Why?: Provides consistency across your team.
{
"bitwise": true,
"camelcase": true,
"curly": true,
"eqeqeq": true,
"es3": false,
"forin": true,
"freeze": true,
"immed": true,
"indent": 4,
"latedef": "nofunc",
"newcap": true,
"noarg": true,
"noempty": true,
"nonbsp": true,
"nonew": true,
"plusplus": false,
"quotmark": "single",
"undef": true,
"unused": false,
"strict": false,
"maxparams": 10,
"maxdepth": 5,
"maxstatements": 40,
"maxcomplexity": 8,
"maxlen": 120,
"asi": false,
"boss": false,
"debug": false,
"eqnull": true,
"esnext": false,
"evil": false,
"expr": false,
"funcscope": false,
"globalstrict": false,
"iterator": false,
"lastsemic": false,
"laxbreak": false,
"laxcomma": false,
"loopfunc": true,
"maxerr": false,
"moz": false,
"multistr": false,
"notypeof": false,
"proto": false,
"scripturl": false,
"shadow": false,
"sub": true,
"supernew": false,
"validthis": false,
"noyield": false,
"browser": true,
"node": true,
"globals": {
"angular": false,
"$": false
}
}
Use JSCS for checking your coding styles your JavaScript and be sure to customize the JSCS options file and include in source control. See the JSCS docs for details on the options.
Why?: Provides a first alert prior to committing any code to source control.
Why?: Provides consistency across your team.
{
"excludeFiles": ["node_modules/**", "bower_components/**"],
"requireCurlyBraces": [
"if",
"else",
"for",
"while",
"do",
"try",
"catch"
],
"requireOperatorBeforeLineBreak": true,
"requireCamelCaseOrUpperCaseIdentifiers": true,
"maximumLineLength": {
"value": 100,
"allowComments": true,
"allowRegex": true
},
"validateIndentation": 4,
"validateQuoteMarks": "'",
"disallowMultipleLineStrings": true,
"disallowMixedSpacesAndTabs": true,
"disallowTrailingWhitespace": true,
"disallowSpaceAfterPrefixUnaryOperators": true,
"disallowMultipleVarDecl": null,
"requireSpaceAfterKeywords": [
"if",
"else",
"for",
"while",
"do",
"switch",
"return",
"try",
"catch"
],
"requireSpaceBeforeBinaryOperators": [
"=", "+=", "-=", "*=", "/=", "%=", "<<=", ">>=", ">>>=",
"&=", "|=", "^=", "+=",
"+", "-", "*", "/", "%", "<<", ">>", ">>>", "&",
"|", "^", "&&", "||", "===", "==", ">=",
"<=", "<", ">", "!=", "!=="
],
"requireSpaceAfterBinaryOperators": true,
"requireSpacesInConditionalExpression": true,
"requireSpaceBeforeBlockStatements": true,
"requireLineFeedAtFileEnd": true,
"disallowSpacesInsideObjectBrackets": "all",
"disallowSpacesInsideArrayBrackets": "all",
"disallowSpacesInsideParentheses": true,
"validateJSDoc": {
"checkParamNames": true,
"requireParamTypes": true
},
"disallowMultipleLineBreaks": true,
"disallowCommaBeforeLineBreak": null,
"disallowDanglingUnderscores": null,
"disallowEmptyBlocks": null,
"disallowMultipleLineStrings": null,
"disallowTrailingComma": null,
"requireCommaBeforeLineBreak": null,
"requireDotNotation": null,
"requireMultipleVarDecl": null,
"requireParenthesesAroundIIFE": true
}
Create an Angular Constant for vendor libraries’ global variables.
Why?: Provides a way to inject vendor libraries that otherwise are globals. This improves code testability by allowing you to more easily know what the dependencies of your components are (avoids leaky abstractions). It also allows you to mock these dependencies, where it makes sense.
// constants.js
/* global toastr:false, moment:false */
(function() {
'use strict';
angular
.module('app.core')
.constant('toastr', toastr)
.constant('moment', moment);
})();
Use constants for values that do not change and do not come from another service. When constants are used only for a module that may be reused in multiple applications, place constants in a file per module named after the module. Until this is required, keep constants in the main module in a constants.js
file.
Why?: A value that may change, even infrequently, should be retrieved from a service so you do not have to change the source code. For example, a url for a data service could be placed in a constants but a better place would be to load it from a web service.
Why?: Constants can be injected into any angular component, including providers.
Why?: When an application is separated into modules that may be reused in other applications, each stand-alone module should be able to operate on its own including any dependent constants.
// Constants used by the entire app
angular
.module('app.core')
.constant('moment', moment);
// Constants used only by the sales module
angular
.module('app.sales')
.constant('events', {
ORDER_CREATED: 'event_order_created',
INVENTORY_DEPLETED: 'event_inventory_depleted'
});
Use file templates or snippets to help follow consistent styles and patterns. Here are templates and/or snippets for some of the web development editors and IDEs.
Angular snippets that follow these styles and guidelines.
TAB
ngcontroller // creates an Angular controller
ngdirective // creates an Angular directive
ngfactory // creates an Angular factory
ngmodule // creates an Angular module
Angular file templates that follow these styles and guidelines can be found at SideWaffle
Angular snippets and file templates that follow these styles and guidelines. You can import them into your WebStorm settings:
File
menuImport Settings
menu optionOK
TAB
:
ng-c // creates an Angular controller
ng-f // creates an Angular factory
ng-m // creates an Angular module
Angular snippets that follow these styles and guidelines.
apm install angularjs-styleguide-snippets
or
In a JavaScript file type these commands followed by a TAB
ngcontroller // creates an Angular controller
ngdirective // creates an Angular directive
ngfactory // creates an Angular factory
ngmodule // creates an Angular module
You can use the HotTowel yeoman generator to create an app that serves as a starting point for Angular that follows this style guide.
Install generator-hottowel
npm install -g generator-hottowel
Create a new folder and change directory to it
mkdir myapp
cd myapp
Run the generator
yo hottowel helloWorld
Client-side routing is important for creating a navigation flow between views and composing views that are made of many smaller templates and directives.
Use the AngularUI Router for client-side routing.
Why?: UI Router offers all the features of the Angular router plus a few additional ones including nested routes and states.
Why?: The syntax is quite similar to the Angular router and is easy to migrate to UI Router.
Define routes for views in the module where they exist. Each module should contain the routes for the views in the module.
Why?: Each module should be able to stand on its own.
Why?: When removing a module or adding a module, the app will only contain routes that point to existing views.
Why?: This makes it easy to enable or disable portions of an application without concern over orphaned routes.
Use Gulp or Grunt for creating automated tasks. Gulp leans to code over configuration while Grunt leans to configuration over code. I personally prefer Gulp as I feel it is easier to read and write, but both are excellent.
Learn more about gulp and patterns for task automation in my Gulp Pluralsight course
Use task automation to list module definition files *.module.js
before all other application JavaScript files.
Why?: Angular needs the module definitions to be registered before they are used.
Why?: Naming modules with a specific pattern such as *.module.js
makes it easy to grab them with a glob and list them first.
var clientApp = './src/client/app/';
// Always grab module files first
var files = [
clientApp + '**/*.module.js',
clientApp + '**/*.js'
];
Avoid using filters for scanning all properties of a complex object graph. Use filters for select properties.
Why?: Filters can easily be abused and negatively effect performance if not used wisely, for example when a filter hits a large and deep object graph.
For anything else, API reference, check the Angular documentation.
Open an issue first to discuss potential changes/additions. If you have questions with the guide, feel free to leave them as issues in the repository. If you find a typo, create a pull request. The idea is to keep the content up to date and use github’s native feature to help tell the story with issues and PR’s, which are all searchable via google. Why? Because odds are if you have a question, someone else does too! You can learn more here at about how to contribute.
By contributing to this repository you are agreeing to make your content available subject to the license of this repository.
1. Discuss the changes in a GitHub issue.
2. Open a Pull Request, reference the issue, and explain the change and why it adds value.
3. The Pull Request will be evaluated and either merged or declined.
tldr; Use this guide. Attributions are appreciated.
Copyright (c) 2014-2015 John Papa
Permission is hereby granted, free of charge, to any person obtaining
a copy of this software and associated documentation files (the
‘Software’), to deal in the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Software, and to
permit persons to whom the Software is furnished to do so, subject to
the following conditions:
The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED ‘AS IS’, WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.