A fake AP module to help develop and test Atlassian Connect applications.
Atlassian Connect apps often use the Atlassian Connect JavaScript API, also called AP, to overcome the limitations due to the app existing in an iframe from an Atlassian page.
AP is typically included by calling the following script:
<script src="https://connect-cdn.atl-paas.net/all.js"></script>
However this script only work when in an iframe from an Atlassian page. It means that when developing or testing, it is not possible to directly call the page if it is using AP.
This package provides a way to make a fake AP that can be used instead of the real one. It includes the most commonly used features of AP, including:
- Generating JWT tokens
- Dialogs
- Events
- Flags
- History
- Request
- User locale
Note: This package should never be used on a production environment.
Using npm:
npm install --save-dev @smartbear/fake-ap
Using yarn:
yarn add -D @smartbear/fake-ap
Simply create a Fake AP instance an make it available globally:
import FakeAP from '@smartbear/fake-ap'
window.AP = new FakeAP()
To display modal dialogs and flags (using AP.dialog.create
and AP.flag.create
), Fake AP provides two React components that you should mount or render.
These components are React portals, which means you can safely insert them anywhere in your React component tree as they will be rendered in another element outside.
For instance, given the following HTML:
<body>
<div id="root" />
</body>
You can mount a React component with Fake AP like this:
import React from 'react'
import ReactDOM from 'react-dom/client'
import { APDialogs, APFlags } from '@smartbear/fake-ap'
const root = ReactDOM.createRoot(document.getElementById('root'))
root.render(
<div>
<APDialogs />
<APFlags />
<div>Some content</div>
</div>
)
This will render the following document:
<body>
<div id="root">
Some content
</div>
<div id="ap_dialogs" />
<div id="ap_flags" />
</body>
The ap_dialogs
and ap_flags
element will contain the working Fake AP components.
The fake AP creation should be done in a script included instead of the real one. For instance, in for the project fake AP was originally created for, the script is available as a pack which is included when a flag is set. Here is an example using Rails and Webpacker:
<% if ENV['USE_FAKE_AP'] %>
<%= javascript_pack_tag 'fake_ap' %>
<% else %>
<script src="https://connect-cdn.atl-paas.net/all.js"></script>
<% end %>
While most features work with no configuration, some methods require configuration to tell our fake AP which values to deal with.
Configuration can be done when creating the fake AP, or at any time later using the special AP.configure
method:
const AP = new FakeAP({
locale: 'en_US'
})
AP.configure({
locale: 'fr_FR'
})
Here is a list of all available configuration (refer to their own section for details):
Configuration | Default value | Description |
---|---|---|
clientKey |
null |
The client key for AP.context.getToken |
sharedSecret |
null |
The shared secret for AP.context.getToken |
userId |
null |
The user ID for AP.context.getToken |
context |
null |
The context for AP.context.getToken and AP.context.getContext |
dialogUrls |
{} |
URLs to call when using AP.dialog.create |
locale |
en_US |
The user locale for AP.user.getLocale |
requestAdapter |
RequestAdapter |
The request adapter for AP.request |
notImplementedAction |
() => {} |
The method called when using a method that is not implemented |
missingConfigurationAction |
throw new Error() |
The method called when a configuration is missing |
Note: when using AP.configure
, all previous configuration is kept, only conflicting configuration is replaced. All new configuration is added.
To use AP.context.getToken
, which creates a valid JWT token as Atlassian would do, it is required to provide the tenant client key and shared secret, as well as a user ID:
AP.configure({
clientKey: 'key',
sharedSecret: 'secret',
userId: 'user'
})
You can also configure a context that will be added to the payload (see AP.context.getContext
). By default that context is an empty object.
To use AP.context.getContext
, you can provide a context object to the configuration:
AP.configure({
context: {
jira: {
project: {
id: '10000'
}
}
}
})
By default the context is an empty object. The same context will also be added to the payload of AP.context.getToken
.
To use dialogs (AP.dialog.create
), you need to provide the dialog keys and URLs as they are describe in the descriptor:
AP.configure({
dialogUrls: {
'custom-dialog': 'https://localhost:3000/dialog'
}
})
AP.dialog.create({
key: 'custom-dialog'
})
You can specify the user locale (defaults to en_US
):
AP.configure({
locale: 'fr_FR'
})
AP.request
is the most complex to implement. Because of same-origin policy, it is not possible to implement a native request method using a shared secret configuration. Since several solutions are possible to work around this limitation, adapters have been designed to provide a request method.
If you do not specify any adapter, the default behavior when using AP.request
is to do nothing.
It will just call the notImplementedAction
method if provided:
AP.configure({
notImplementedAction: console.log
})
// This will call console.log('AP.request', 'path', { method: 'POST' })
AP.request('path', { method: 'POST' })
The backend adapter is a way to make actual requests to the Jira API using your own backend. It requires some work from the backend though:
- Implement a Jira client to interact with the Jira API
- Provide an API endpoint for the fake AP to call
This API endpoint should not require any authentication method, as the adapter does not provide any. For obvious reasons, ensure that this API endpoint is never available on production.
Backend adapter should be configured like this:
import FakeAP, { BackendRequestAdapter } from '@smartbear/fake-ap'
const backendRequestAdapter = new BackendRequestAdapter('/path/to/fake_ap/api')
const AP = new FakeAP({
requestAdapter: backendRequestAdapter
})
Then a POST request to the backend API will be made using the AP.request
options.
For instance, consider following request:
AP.request(
'/rest/api/3/search',
{
data: { some: 'data' }
}
)
This will make a POST request to your backend with a request body containing:
method = GET
path = '/rest/api/3/search'
data = { some: 'data' }
To make this adapter as generic as possible, there is no additional information sent to the backend. If you need to know which tenant is this request for, you need to specify it with the configuration URL. For instance:
AP.configure({
requestAdapter: new BackendRequestAdapter('/path/to/fake_ap/api/tenants/2')
})
It is possible to create a custom adapter by extending from the default adapter:
import { RequestAdapter } from '@smartbear/fake-ap'
class CustomAdapter extends RequestAdapter {
async request() {
return {
body: '{}'
}
}
}
Here are the specifications for a custom adapter:
- Extend from
RequestAdapter
- Implement a
request
method that is async (or returns a promise) - If the request is a success,
request
should return an object with abody
property that contained the JSON response as a string - If the request is a failure,
request
should throw the response body with a specific object:// Assuming the response body was '{}' and status was 400 { err: '{}', xhr: { responseText: '{}', status: 400, statusText: 'Bad Request' } }
It is possible to configure the behavior of AP methods that are not implemented in fake AP:
AP.configure({
notImplementedAction: console.log
})
When called, any method that is not implemented will call the notImplementedAction
method with the method name and all the arguments.
Using the above configuration, calling AP.context.getContext('a', 'b')
will call console.log('AP.context.getContext', 'a', 'b')
.
It is possible to have more advanced behaviors:
// This will forward all arguments to console.log, but will silence AP.resize calls
AP.configure({
notImplementedAction: (method, ...args) => {
if (method !== 'AP.resize') {
console.log(method, ...args)
}
}
})
When a method requires some configuration that is not provided (for instance AP.context.getToken
), Fake AP will throw an error by default. It is possible to change this behavior:
AP.configure({
clientKey: 'key',
userId: 'user',
missingConfigurationAction: console.log
})
// This will call console.log('AP.context.getToken', 'sharedSecret', callback)
const callback = () => {}
await AP.context.getToken(callback)
AP.context
:getToken
getContext
AP.dialog
:create
close
getCustomData
AP.event
:on
once
off
emit
AP.flag.create
AP.history
:getState
popState
pushState
replaceState
AP.request
AP.user.getLocale
Note: AP.dialog.create
does not handle every options. It handles the key
option properly, but will show a dialog as if the options were:
{
width: '100%',
height: '100%',
chrome: false
}
Fake AP is still missing a lot of methods from the actual AP:
AP.cookie
AP.dialog
:getButton
disableCloseOnSubmit
createButton
isCloseOnEscape
AP.event
: all any and public eventsAP.history
:back
forward
go
AP.host
AP.iframe
:AP.resize
andAP.sizeToParent
AP.inlineDialog
AP.jira
AP.navigator
AP.user
:getCurrentUser
getTimeZone
Some methods like AP.resize
do not really make sense in a development or testing environment, so they may not be implemented before a long time.
Once Fake AP is created, it is possible to add any custom implementation that is specific to your application.
The example below is a custom implementation of AP.navigator.go
. It checks that you are on a specific page (/normal_page
) and are trying to navigate to an issue. If it is the case it will check the correct issue ID using AP.request
, then redirect to the correct page using that issue ID.
AP.navigator.go = async (target, context) => {
if (target === 'issue' && window.location.pathname === '/normal_page') {
const response = await AP.request(`/rest/api/3/issue/${context.issueKey}`)
const issueId = JSON.parse(response.body).id
window.location = `/issue_page?issueId=${issueId}`
}
}