View on GitHub

faast-path-to-beautiful-apps

Adobe I/O Lab - 2019

Chapter 4. Call Sensei API from Runtime

Learning Objectives

Write an action that makes an API call to Adobe Content Sensei Services

Lab Tasks

Add Sensei as a Service
Create Runtime action to make an API call to Sensei Service
- Create an action that gets an Access Token for Sensei Service
- Make a call to Sensei Service

1. Add Adobe Content Sensei Services as a Service

GOAL: We will add Adobe Content Sensei Services as a service in your integration

Navigate to Adobe I/O Console at https://console.adobe.io in your browser
Find the Integration that you created during Chapter 1 and open it
Select the Services tab in your Integration
Under Available Services, select Content AI (Beta) from the Adobe Sensei category.
Click Add Service to complete
You should now see 2 services under Configured Services, Adobe Content AI (Beta) and I/O Management API

2. Create Runtime action to make an API call to Sensei

GOAL: We will start with an action that gets an access token for Sensei, then add an API call to the action so that your action can get tags for an image using Sensei Content Service.

Set up an action that gets an Access Token for Sensei Content AI

Any API that accesses a service or content on behalf of an end user authenticates using the OAuth and JSON Web Token standards. For service-to-service integrations, you will also need a JSON Web Token (JWT) that encapsulates your client credentials and authenticates the identity of your integration. You exchange the JWT for the OAuth token that authorizes access.

For more details, please refer to Service Account Integration documentation

The JWT Workflow contains 6 steps

1- Create a Public Key Certificate
2- Subscribe to a Service or Event Provider
3- Configure a Service Account Integration
4- Secure your Client Credentials
5- Create your JSON Web Token (JWT)
6- Exchange your JWT for an Access Token

Let’s set up an action that helps you create and store an Access Token in Runtime

In the resources you have downloaded, navigate to adobeio-cna-lib-auth-ims-master folder
```
 $ cd ~/Desktop/adobeio-cna-lib-auth-ims-master/
```
For Git users, you can also get the latest copy off our GitHub repo.
We’ll create a .env file to configure this library to your integration. Run the following commands
```
 $ touch .env
 $ open .env
```

Copy the following block in as place holder:

 OW_NAMESPACE="<change-me>"
 # Optional parameters (should use 'npm run configure' if used)
 IMS_AUTH_TYPE=<code/jwt>

You can find the first two values using your Adobe I/O CLI
```
 $ aio runtime property get
```
Update OW_NAMESPACE in the .env file to match to your whisk namespace.
Save your .env file.
Now, we’ll also populate the jwt.json file.
```
 $ open jwt.json
```
Most values can be found in the Overview tab in Console. The metascope can be seen in your JWT tab, it should look something like this
```
 "meta_scopes": ["https://ims-na1.adobelogin.com/s/ent_adobeio_sdk", "https://ims-na1.adobelogin.com/s/ent_sensei_image_sdk"]
```
Next, run the following command,
```
 awk -v ORS='\\n' '1' ~/Desktop/private.key | pbcopy
```
The private key has now been copied to your clipboard. Paste it into jwt_secret field. Starting from -----BEGIN PRIVATE KEY----- to -----END PRIVATE KEY-----.
Save your jwt.json file.

Your AUTH library is now set up. Let’s deploy it into your namespace by running the following command.

 $ npm install
 $ npm run deploy-jwt

Your should get something that looks like

 endpoints (web actions): https://https://adobeioruntime.net/api/v1/web/<YOUR_NAMESPACE>/jwtauthp/jwtauthenticate

Run the action to get an Access Token

You now have 2 new packages for Auth in your namespace. You can look them up by

 $ aio runtime package list

You should see

 /NAMESPACE/jwtauthp-shared                      private 
 /NAMESPACE/jwtauthp                             private 

Now, let’s invoke the Auth aciton to get a token
```
 $ aio runtime action invoke jwtauthp/jwtauthenticate --blocking
```
You can see that this is actually a sequence, and has invoked multiple actions.
You can now look up your access token at
```
 $ aio runtime action invoke jwtauthp-shared/jwtauth --result
```
You should see your access token returned.
``` { “accessToken”: “xxxxxxxxxxxxxxxxxxx”, “accessTokenExpiry”: xxxxxxxxxxxxxxxxxxx } }

3. Invoke Sensei API using this Access Token

There’s a sample action on your Desktop sensei.js. Let’s create a new action with it.
```
 $ aio runtime action update sensei sensei.js
```

Now let’s try to invoke this action manually by passing in some parameters. Build the following command in your text editor

 aio runtime action invoke sensei --param accessToken <insert accessToken from above> --param ent_api_key <insert API key from Console Integration> --param image_url https://www.jcpportraits.com/sites/jcpportraits.com/files/portrait/1712/7_182-456_FamilyGallery6.jpg --param results 10 --param confidence 0.5 --result

When ready, paste this command into your Terminal to invoke the action, you should get something that looks like

 {
 "tags": [
 {
   "confidence": 0.9170970402413429,
   "tag": "family"
 },
 {
   "confidence": 0.8392684959100764,
   "tag": "child"
 },
 {
   "confidence": 0.8171461274057861,
   "tag": "happy"
 }....

This is a list of tags for the image and the confidence score for each tag

Open a browser window and go to https://www.jcpportraits.com/sites/jcpportraits.com/files/portrait/1712/7_182-456_FamilyGallery6.jpg Does the tags look accurate?

4. Create Runtime action to make an API call to Sensei Content AI

Challenge: Using the sample call above, can you figure out how to use the access token you just create to make an API call to Sensei Content API? (Hint: default parameter and sequences are your best friends.)

If you got it right, the result should look something similar to your last call.

Navigate

Next:
Lesson 5 - Use Project Starter to build an app with UI

Return Home: Workbook Index