glennwedin / serverless-functions-dialogflow

How to host Dialogflow fulfillment on different cloud providers

Actions-on-Google on different cloud function providers

All the functions in this project has been set up using each cloud providers default cli. An alternative to doing this is using something like the Serverless framework. In this experiment it made the most sense to do it the best documented and tested way possible.

Content:

1. AWS Lambda
2. Azure Functions
3. Google Cloud Functions

AWS Lambda

Requirements:

• Docker
• AWS cli
• AWS SAM cli

Run locally

Run these commands

    cd aws-dialogflow/aws-dialogflow
    npm install
    cd ..
    sam local start-api

Test using postman: http://127.0.0.1:3000/aws-dialogflow Post body:

  "responseId": "e9453cdc-32ab-4569-b8ef-3ddb3bcb3376",
  "queryResult": {
    "queryText": "hi",
    "action": "input.welcome",
    "parameters": {},
    "allRequiredParamsPresent": true,
    "fulfillmentMessages": [
      {
        "text": {
          "text": [
            ""
          ]
        }
      }
    ],
    "intent": {
      "name": "sdfsdf",
      "displayName": "Welcome"
    },
    "intentDetectionConfidence": 1,
    "languageCode": "en"
  },
  "originalDetectIntentRequest": {
    "payload": {}
  },
  "session": "dsfsfdfs"
}"

Deploy

Create a storage bucket on AWS S3 for your function

sam package --template-file template.yaml --s3-bucket aws-dialogflowbucket --output-template-file packaged.yaml

Deploy the packaged function

sam deploy --template-file ./packaged.yaml --stack-name aws-dialogflow --capabilities CAPABILITY_IAM

Azure functions

Requirements:

• Azure cli
• Azure core tools

Run locally

Run this command:

cd azure-dialogflow
npm install
func host start

Test using postman: http://localhost:7071/api/azure-dialogflow Post body:

  "responseId": "e9453cdc-32ab-4569-b8ef-3ddb3bcb3376",
  "queryResult": {
    "queryText": "hi",
    "action": "input.welcome",
    "parameters": {},
    "allRequiredParamsPresent": true,
    "fulfillmentMessages": [
      {
        "text": {
          "text": [
            ""
          ]
        }
      }
    ],
    "intent": {
      "name": "sdfsdf",
      "displayName": "Welcome"
    },
    "intentDetectionConfidence": 1,
    "languageCode": "en"
  },
  "originalDetectIntentRequest": {
    "payload": {}
  },
  "session": "dsfsfdfs"
}"

Deploy

Create resource group

az group create --name azure-dialogflow --location westeurope

Create a storage for the function

az storage account create --name azuredialogflow --location westeurope --resource-group azure-dialogflow --sku Standard_LRS

Create a functionapp on Azure

az functionapp create --resource-group azure-dialogflow --consumption-plan-location westeurope --name azure-dialogflow --storage-account azuredialogflow --runtime node

Publish the function

func azure functionapp publish azure-dialogflow

Google Cloud Functions

Requirements:

• Google Cloud Cli
• Google Cloud function emulator

Run locally

Set a projectID for your function project and start the emulator

functions config set projectId YOUR_PROJECT_ID
functions start

Deploy function locally

functions deploy googleDialogflow --trigger-http

Test it with postman: http://localhost:8010/YOUR_PROJECT_ID/us-central1/googleDialogflow

Post body:

  "responseId": "e9453cdc-32ab-4569-b8ef-3ddb3bcb3376",
  "queryResult": {
    "queryText": "hi",
    "action": "input.welcome",
    "parameters": {},
    "allRequiredParamsPresent": true,
    "fulfillmentMessages": [
      {
        "text": {
          "text": [
            ""
          ]
        }
      }
    ],
    "intent": {
      "name": "sdfsdf",
      "displayName": "Welcome"
    },
    "intentDetectionConfidence": 1,
    "languageCode": "en"
  },
  "originalDetectIntentRequest": {
    "payload": {}
  },
  "session": "dsfsfdfs"
}"

Deploy

gcloud functions deploy --runtime=nodejs8 --trigger-http