Skip to content

Latest commit

 

History

History
216 lines (158 loc) · 9.36 KB

README-LUIS.md

File metadata and controls

216 lines (158 loc) · 9.36 KB

Setting up LUIS via CLI:

This README contains information on how to create and deploy a LUIS application. When the bot is ready to be deployed to production, we recommend creating a LUIS Endpoint Resource for usage with your LUIS App.

For instructions on how to create a LUIS Application via the LUIS portal, see these Quickstart steps:

  1. Quickstart: Create a new app in the LUIS portal
  2. Quickstart: Deploy an app in the LUIS portal

Table of Contents:


Install Azure CLI >=2.0.61:

Visit the following page to find the correct installer for your OS:

Install LUIS CLI >=2.4.0:

Open a CLI of your choice and type the following:

npm i -g luis-apis@^2.4.0

LUIS portal account:

You should already have a LUIS account with either https://luis.ai, https://eu.luis.ai, or https://au.luis.ai. To determine where to create a LUIS account, consider where you will deploy your LUIS applications, and then place them in the corresponding region.

After you've created your account, you need your Authoring Key and a LUIS application ID.


1. Import the local LUIS application to luis.ai

luis import application --region "LuisAppAuthoringRegion" --authoringKey "LuisAuthoringKey" --appName "FlightBooking" --in "./cognitiveModels/FlightBooking.json"

Outputs the following JSON:

{
    "id": "########-####-####-####-############",
    "name": "FlightBooking",
    "description": "A LUIS model that uses intent and entities.",
    "culture": "en-us",
    "usageScenario": "",
    "domain": "",
    "versionsCount": 1,
    "createdDateTime": "2019-03-29T18:32:02Z",
    "endpoints": {},
    "endpointHitsCount": 0,
    "activeVersion": "0.1",
    "ownerEmail": "bot@contoso.com",
    "tokenizerVersion": "1.0.0"
}

For the next step, you'll need the "id" value for --appId and the "activeVersion" value for --versionId.

2. Train the LUIS Application

luis train version --region "LuisAppAuthoringRegion" --authoringKey "LuisAuthoringKey" --appId "LuisAppId" --versionId "LuisAppversion" --wait

3. Publish the LUIS Application

luis publish version --region "LuisAppAuthoringRegion" --authoringKey "LuisAuthoringKey" --appId "LuisAppId" --versionId "LuisAppversion" --publishRegion "LuisAppPublishRegion"

--region corresponds to the region you author your application in. The regions available for this are "westus", "westeurope" and "australiaeast".
These regions correspond to the three available portals, https://luis.ai, https://eu.luis.ai, or https://au.luis.ai.
--publishRegion corresponds to the region of the endpoint you're publishing to, (e.g. "westus", "southeastasia", "westeurope", "brazilsouth").
See the reference docs for a list of available publish/endpoint regions.

Outputs the following:

 {
    "versionId": "0.1",
    "isStaging": false,
    "endpointUrl": "https://westus.api.cognitive.microsoft.com/luis/v2.0/apps/########-####-####-####-############",
    "region": "westus",
    "assignedEndpointKey": null,
    "endpointRegion": "westus",
    "failedRegions": "",
    "publishedDateTime": "2019-03-29T18:40:32Z",
    "directVersionPublish": false
}

To see how to create an LUIS Cognitive Service Resource in Azure, please see the next README. This Resource should be used when you want to move your bot to production. The instructions will show you how to create and pair the resource with a LUIS Application.


1. Create a new LUIS Cognitive Services resource on Azure via Azure CLI

Note:
If you don't have a Resource Group in your Azure subscription, you can create one through the Azure portal or through using:

az group create --subscription "AzureSubscriptionGuid" --location "westus" --name "ResourceGroupName"

To see a list of valid locations, use az account list-locations

# Use Azure CLI to create the LUIS Key resource on Azure
az cognitiveservices account create --kind "luis" --name "NewLuisResourceName" --sku "S0" --location "westus" --subscription "AzureSubscriptionGuid" -g "ResourceGroupName"

The command will output a response similar to the JSON below:

{
  "endpoint": "https://westus.api.cognitive.microsoft.com/luis/v2.0",
  "etag": "\"########-####-####-####-############\"",
  "id": "/subscriptions/########-####-####-####-############/resourceGroups/ResourceGroupName/providers/Microsoft.CognitiveServices/accounts/NewLuisResourceName",
  "internalId": "################################",
  "kind": "luis",
  "location": "westus",
  "name": "NewLuisResourceName",
  "provisioningState": "Succeeded",
  "resourceGroup": "ResourceGroupName",
  "sku": {
    "name": "S0",
    "tier": null
  },
  "tags": null,
  "type": "Microsoft.CognitiveServices/accounts"
}

Take the output from the previous command and create a JSON file in the following format:

{
    "azureSubscriptionId": "00000000-0000-0000-0000-000000000000",
    "resourceGroup": "ResourceGroupName",
    "accountName": "NewLuisResourceName"
}

2. Retrieve ARM access token via Azure CLI

az account get-access-token --subscription "AzureSubscriptionGuid"

This will return an object that looks like this:

{
  "accessToken": "eyJ0eXAiOiJKVtokentokentokentokentokeng1dCI6Ik4tbEMwbi05REFMcXdodUhZbkhRNjNHZUNYYyIsItokenI6Ik4tbEMwbi05REFMcXdodUhZbkhRNjNHZUNYYyJ9.eyJhdWQiOiJodHRwczovL21hbmFnZW1lbnQuY29yZS53aW5kb3dzLm5ldC8iLCJpc3MiOiJodHRwczovL3N0cy53aW5kb3dzLm5ldC83MmY5ODhiZi04NmYxLTQxYWYtOTFhYi0yZDdjZDAxMWRiNDcvIiwiaWF0IjoxNTUzODc3MTUwLCJuYmYiOjE1NTM4NzcxNTAsImV4cCI6MTU1Mzg4MTA1MCwiX2NsYWltX25hbWVzIjp7Imdyb3VwcyI6InNyYzEifSwiX2NsYWltX3NvdXJjZXMiOnsic3JjMSI6eyJlbmRwb2ludCI6Imh0dHBzOi8vZ3JhcGgud2luZG93cy5uZXQvNzJmOTg4YmYtODZmMS00MWFmLTkxYWItMmQ3Y2QwMTFkYjQ3L3VzZXJzL2ZmZTQyM2RkLWJhM2YtNDg0Ny04NjgyLWExNTI5MDA4MjM4Ny9nZXRNZW1iZXJPYmplY3RzIn19LCJhY3IiOiIxIiwiYWlvIjoiQVZRQXEvOEtBQUFBeGVUc201NDlhVHg4RE1mMFlRVnhGZmxxOE9RSC9PODR3QktuSmRqV1FqTkkwbmxLYzB0bHJEZzMyMFZ5bWZGaVVBSFBvNUFFUTNHL0FZNDRjdk01T3M0SEt0OVJkcE5JZW9WU0dzd0kvSkk9IiwiYW1yIjpbIndpYSIsIm1mYSJdLCJhcHBpZCI6IjA0YjA3Nzk1LThkZGItNDYxYS1iYmVlLTAyZjllMWJmN2I0NiIsImFwcGlkYWNyIjoiMCIsImRldmljZWlkIjoiNDhmNDVjNjEtMTg3Zi00MjUxLTlmZWItMTllZGFkZmMwMmE3IiwiZmFtaWx5X25hbWUiOiJHdW0iLCJnaXZlbl9uYW1lIjoiU3RldmVuIiwiaXBhZGRyIjoiMTY3LjIyMC4yLjU1IiwibmFtZSI6IlN0ZXZlbiBHdW0iLCJvaWQiOiJmZmU0MjNkZC1iYTNmLTQ4NDctODY4Mi1hMTUyOTAwODIzODciLCJvbnByZW1fc2lkIjoiUy0xLTUtMjEtMjEyNzUyMTE4NC0xNjA0MDEyOTIwLTE4ODc5Mjc1MjctMjYwOTgyODUiLCJwdWlkIjoiMTAwMzdGRkVBMDQ4NjlBNyIsInJoIjoiSSIsInNjcCI6InVzZXJfaW1wZXJzb25hdGlvbiIsInN1YiI6Ik1rMGRNMWszN0U5ckJyMjhieUhZYjZLSU85LXVFQVVkZFVhNWpkSUd1Nk0iLCJ0aWQiOiI3MmY5ODhiZi04NmYxLTQxYWYtOTFhYi0yZDdjZDAxMWRiNDciLCJ1bmlxdWVfbmFtZSI6InN0Z3VtQG1pY3Jvc29mdC5jb20iLCJ1cG4iOiJzdGd1bUBtaWNyb3NvZnQuY29tIiwidXRpIjoiT2w2NGN0TXY4RVNEQzZZQWRqRUFtokenInZlciI6IjEuMCJ9.kFAsEilE0mlS1pcpqxf4rEnRKeYsehyk-gz-zJHUrE__oad3QjgDSBDPrR_ikLdweynxbj86pgG4QFaHURNCeE6SzrbaIrNKw-n9jrEtokenlosOxg_0l2g1LeEUOi5Q4gQREAU_zvSbl-RY6sAadpOgNHtGvz3Rc6FZRITfkckSLmsKAOFoh-aWC6tFKG8P52rtB0qVVRz9tovBeNqkMYL49s9ypduygbXNVwSQhm5JszeWDgrFuVFHBUP_iENCQYGQpEZf_KvjmX1Ur1F9Eh9nb4yI2gFlKncKNsQl-tokenK7-tokentokentokentokentokentokenatoken",
  "expiresOn": "2200-12-31 23:59:59.999999",
  "subscription": "AzureSubscriptionGuid",
  "tenant": "tenant-guid",
  "tokenType": "Bearer"
}

The value needed for the next step is the "accessToken".

3. Use luis add appazureaccount to pair your LUIS resource with a LUIS Application

luis add appazureaccount --in "path/to/created/requestBody.json" --appId "LuisAppId" --authoringKey "LuisAuthoringKey" --armToken "accessToken"

If successful, it should yield a response like this:

{
  "code": "Success",
  "message": "Operation Successful"
}

4. See the LUIS Cognitive Services' keys

az cognitiveservices account keys list --name "NewLuisResourceName" --subscription "AzureSubscriptionGuid" -g "ResourceGroupName"

This will return an object that looks like this:

{
  "key1": "9a69####dc8f####8eb4####399f####",
  "key2": "####f99e####4b1a####fb3b####6b9f"
}