Skip to main content

Lab 1: Provision an AI Resource

Building your first AI service

Provision in Azure​

  1. Open up the Azure Portal
  2. Search for Azure AI Services
  3. Create a multi-service account
  • subscription
  • choose your RG or create a new one
  • region
  • pricing tier is Standard S0

image.png

  1. Wait for this to complete.
    image.png

Connecting to the AI Service​

You connect to a multi-service account using Keys and an Endpoint. These can be found on the Keys and Endpoints tab.

image.png

  1. Take note of the:
  • key
  • endpoint
  • region (some services need this as well)

  1. Inside of the lab folder, navigate to the .env file The path for me was C:\coding\openai\mslearn-ai-services\Labfiles\01-use-azure-ai-services\Python\rest-client\.env
    image.png

  2. Enter in your key and endpoint here and save. Normally you will want to extract these from a Key Vault, but we will learn about that in a later lab.

  3. Save the .env file

Using the service using a Rest API​

  1. Open the rest-client.py file and read through it.
rest-client.py
from dotenv import load_dotenv
import os
import http.client, base64, json, urllib
from urllib import request, parse, error

def main():
    global ai_endpoint
    global ai_key

    try:
        # Get Configuration Settings
        load_dotenv()
        ai_endpoint = os.getenv('AI_SERVICE_ENDPOINT')
        ai_key = os.getenv('AI_SERVICE_KEY')

        # Get user input (until they enter "quit")
        userText =''
        while userText.lower() != 'quit':
            userText = input('Enter some text ("quit" to stop)\n')
            if userText.lower() != 'quit':
                GetLanguage(userText)


    except Exception as ex:
        print(ex)

def GetLanguage(text):
    try:
        # Construct the JSON request body (a collection of documents, each with an ID and text)
        jsonBody = {
            "documents":[
                {"id": 1,
                 "text": text}
            ]
        }

        # Let's take a look at the JSON we'll send to the service
        print(json.dumps(jsonBody, indent=2))

        # Make an HTTP request to the REST interface
        uri = ai_endpoint.rstrip('/').replace('https://', '')
        conn = http.client.HTTPSConnection(uri)

        # Add the authentication key to the request header
        headers = {
            'Content-Type': 'application/json',
            'Ocp-Apim-Subscription-Key': ai_key
        }

        # Use the Text Analytics language API
        conn.request("POST", "/text/analytics/v3.1/languages?", str(jsonBody).encode('utf-8'), headers)

        # Send the request
        response = conn.getresponse()
        data = response.read().decode("UTF-8")

        # If the call was successful, get the response
        if response.status == 200:

            # Display the JSON response in full (just so we can see it)
            results = json.loads(data)
            print(json.dumps(results, indent=2))

            # Extract the detected language name for each document
            for document in results["documents"]:
                print("\nLanguage:", document["detectedLanguage"]["name"])

        else:
            # Something went wrong, write the whole response
            print(data)

        conn.close()


    except Exception as ex:
        print(ex)


if __name__ == "__main__":
    main()

  1. Run this file by running python rest-client.py

  2. Enter in some text and it will predict what language it is.

image.png

  1. Type quit to exit as the instructions show.

Using the service using an SDK​

An SDK will be more powerful than using the REST APIs