Call SAP Conversational AI API Using OAuth

1. In the file explorer, create a new folder for the project and call it chatbot-api.

2. Open VS Code.

   Make sure you have installed the Microsoft extension for Python, as well as all the prerequisite packages.

3. Go to File > Add Folder to Workspace, and select the project folder.

4. Inside the folder, create a folder called static. Download the SAP Conversational AI icon and place it in the folder.

Your project should look like this:

1. In your project, create a file called configproperties.py.

2. Add properties for the credentials you will need:

   Python Copy

   clientid = "<your client ID>"
   secret = "<your secret>"
   requesttoken = "<your request token>"
   oauthURL = "<URL for getting OAuth token>"
   requestURL = "<URL for calling request API>"

3. Retrieve your credentials from your chatbot and place as values in the configproperties.py file.

   • For the client ID, secret and OAuth URL, go to Settings > Tokens, and click Generate under Runtime APIs.

     

     Select Client Credentials, and click Next. After a few seconds, the credentials are created and you can click Close to close the dialog.

     

     Copy the client ID and secret.

     

   • For the request token and request API URL, go to Settings > Versions and open the version of your chatbot that you want to call.

     

Now your configproperties.py file should have your credentials.

1. In your project, create a file called chatbot-api.py.

2. In the file add the following skeleton code:

   Python Copy

   from flask import Flask, request, jsonify
   import os
   import json
   import requests
   from flask_caching import Cache
   from datetime import datetime
   import configproperties
   
   app = Flask(__name__)
   cf_port = os.getenv("PORT")
   
   #Other code will go here
   
   if __name__ == '__main__':
   	if cf_port is None:
   		app.run(host='0.0.0.0', port=5000, debug=True)
   	else:
   		app.run(host='0.0.0.0', port=int(cf_port), debug=True)

The above is standard code, and will create the localhost server on port 5000. Also, debug=True will allow you make changes to the web server without having to manually restart the server.

You will have to have installed the Flask, flask_caching, and requests packages.

All subsequent code goes where it is marked: #Other code will go here

Add the following to enable the caching of the OAuth token:

config = {
    "DEBUG": True,          # some Flask specific configs
    "CACHE_TYPE": "SimpleCache",  # Flask-Caching related configs
    "CACHE_DEFAULT_TIMEOUT": 300
}
app.config.from_mapping(config)
cache = Cache(app)

For more information on configuring and using caching, see Flask-Caching.

Add the following function to retrieve the OAuth token:

@cache.cached(timeout=43200, key_prefix='token')
def get_token():
    print("GETTING TOKEN")
    result = requests.post(configproperties.oauthURL, data={ 'grant_type':'client_credentials' }, auth=(configproperties.clientid, configproperties.secret))
    token = json.loads(result.content)
    return token["access_token"]

Notice the following:

• You set a print statement so you will be able to see whether this function is executed based on the cache (more on this later).

• In addition to sending the credentials as headers, you also send a the grant_type key-value pair as form-encoded data.

• In the @cache decorator for this function, you set the caching to be valid for 12 hours, which is the validity for the token. You also set the cache field name to token.

Ideally, the caching timeout would not be hard-coded, but would be retrieved from within the JSON when you get the OAuth token, in the expires_in field.

Add the following function to create an endpoint, and when called, to call the SAP Conversational AI API and return the intent that was detected.

# Route for calling SAP Conversational AI API
@app.route('/api')
def api():
    # Get text to send to NLP (from URL parameter)
    text = request.args.get("text")

    # Make Post request, with text and credentials
    result = requests.post(configproperties.requestURL, json={"text" : text}, headers={ "Authorization" : "Bearer " + get_token(), "X-Token" : "Token " + configproperties.requesttoken })
    resultsjson = json.loads(result.content)

    # Return web page, including the intent
    if resultsjson["results"]["intents"]:
        return f'<h1>SAP Conversational AI - API</h1><body>The intent was <b>{resultsjson["results"]["intents"][0]["slug"]}</b><br><img src="static/283370-pictogram-purple.svg" width=260px></body>'
    else:
        return f'<h1>SAP Conversational AI - API</h1><body>No intents.<br><img src="static/283370-pictogram-purple.svg" width=260px></body>'

Notice that you must send the OAuth token as bearer authorization token, and the X-Token header with your request token.

The format of the response for the Runtime request and dialog APIs are described under Runtime API.

1. Start the server by clicking the Run in Terminal icon.

   

2. Call the server with the URL:

   URL Copy

   http://localhost:5000/api?text=hi

   The web page shows the intent that was detected, greetings.

   

3. Call the server with the URL:

   URL Copy

   http://localhost:5000/api?text=bye

   The web page shows the intent that was detected, now goodbye.

   

4. Call the server with the URL:

   URL Copy

   http://localhost:5000/api?text=abcdef

   The web page now says that no intent was detected.

   

If you look at the terminal, you will see that we retrieved the token in the first call but then used the cached token in the next 2 calls.