Dialogflow - Node SDK does not preserve contexts within session - node.js

I'm using the dialogflow Node SDK to send textRequests and eventRequests to dialog flow.
The fulfillment webhook shows that the context is not preserved though the sessionId is the same.
Working with the same dialogflow agent from actions-on-google assistant, the context is preserved.
so the only difference is that i'm using the Node SDK to send the text.
this.app = apiai(CLIENT_ACCESS_TOKEN);
this.options = {
sessionId: 'abc',
originalRequest: {
data: {
user: 'temp_user'
},
conversation: {
"conversationId": "123456789"
}
}
};
const request = this.app.textRequest('This is captured by INTENT_1 that triggers
a webhook that sets context to MY_CONTEXT', options);
const request = this.app.textRequest('This should be captured by INTENT_2
that has an input context of MY_CONTEXT', options);
the second request does not trigger INTENT_2, but the default fallback intent, unless I remove the input context from INTENT_2 in dialogflow and then it's triggered

Might you be using resetContexts by mistake? That would explain it.

Related

How to update existing Intent in Dialogflow(V2) using nodejs SDK?

I am using Dialogflow nodejs SDK(V2) to Integrate Dialogflow in my nodjs application for that I am using dialogflow npm node library. I'm able to create the Intent and get the list of Intents and I'm able to query as well. But I can't find any method for updating the existing Intent and getting Intent details based on the Intent ID.
can you please help me or guide me how to resolve this issue?
Thanks.
To update an intent, first, you need to get the intent details. If you have the intent name or ID, then you can simply make a request to list intent API and find the intent details with matching intent name.
Once you have the intent details you wish to update( here referred as existingIntent), you can use the below code to update it.
async function updateIntent(newTrainingPhrases) {
// Imports the Dialogflow library
const dialogflow = require("dialogflow");
// Instantiates clients
const intentsClient = new dialogflow.IntentsClient();
const intent = existingIntent; //intent that needs to be updated
const trainingPhrases = [];
let previousTrainingPhrases =
existingIntent.trainingPhrases.length > 0
? existingIntent.trainingPhrases
: [];
previousTrainingPhrases.forEach(textdata => {
newTrainingPhrases.push(textdata.parts[0].text);
});
newTrainingPhrases.forEach(phrase => {
const part = {
text: phrase
};
// Here we create a new training phrase for each provided part.
const trainingPhrase = {
type: "EXAMPLE",
parts: [part]
};
trainingPhrases.push(trainingPhrase);
});
intent.trainingPhrases = trainingPhrases;
const updateIntentRequest = {
intent,
languageCode: "en-US"
};
// Send the request for update the intent.
const result = await intentsClient.updateIntent(updateIntentRequest);
return result;
}

Knowledge works in "try it", but agent does not recognize intent and retrieve $Knowledge.Answer

I know this is in Beta, but I've set up a Knowledge base for my agent and the intent doesn't seem to get recognized.
When setting up the Knowledge base, the "try it" test works and retrieves successfully, but when trying the same request from the simple chat bot, the intent is not recognized. What else is necessary to hook the Knowledge feature to the agent?
What is the medium of that simple chat bot you are using?Is it android/web?
Assuming are using dialogflow v2 node.js library,We have to pass the full path of the knowledgeBase in the queryParams inside the detectIntent function request object.Then only dialogflow will look into knowledgeBase for matching the user input to the Knowledge base Intents.
Sample of request object-
// const projectId = 'ID of GCP project associated with your Dialogflow agent';
// const sessionId = `user specific ID of session, e.g. 12345`;
const sessionPath = sessionClient.sessionPath(projectId, sessionId);
let request = {
session: sessionPath,
queryInput: {
text: {
text: 'hi,how are you?',
languageCode: 'en-US',
},
},
queryParams: {
knowledgeBaseNames:['projects/stockmarket-XXXX/knowledgeBases/XXXXXXXXXXXXXXx'] //Paste your knowledge base path,Check this out from the diagnostic info
}
};
checkout https://github.com/googleapis/nodejs-dialogflow/blob/master/samples/detect.v2beta1.js#L438
Let me know if you have any questions :)

Clearing incoming agent contexts in node sdk is not working

I have a dialogflow agent with a 'test_contexts' intent, with webhook enabled.
When matching this intent, I would like to control the incoming/outgoing contexts in the conversation, using fullfillments.
I cannot remove a context from the incoming request from dialogflow.
For example in the code below, if the incoming request contains context 'loggedin', it will not be removed.
Fullfillments code
'use strict';
const { WebhookClient } = require('dialogflow-fulfillment');
const { Card, Suggestion } = require('dialogflow-fulfillment');
exports.MyAgent = (request, response) => {
// Create a webhookclient class
const agent = new WebhookClient({ request, response });
// Create function to test context control
const test_context = function(agent){
// fails to remove incoming context, set previously in the conversation
agent.clearContext('loggedin');
// fails to remove/modify incoming context, set previously in the conversation
agent.setContext({ name: 'loggedin', lifespan: 0 });
// successfully removes contexts set at the current point in the conversation, however, fails to remove any incoming contexts (as the name suggests)
agent.clearOutgoingContexts();
}
// Define which functions are called for which intents
let intentMap = new Map();
intentMap.set('test_contexts', test_context);
agent.handleRequest(intentMap);
}
How can I remove a context incoming from Dialogflow?!
agent.setContext({ name: 'loggedin', lifespan: -1 }); successfully removed the context.

Dialogflow assistant app exiting after successful fulfillment

I have a dialogflow assistant app with 3 intents. The first intent asks the user for location and name details from google. I am using a webhook for the fulfillment of this intent. I am able to extract the user information name and location, but after it is showing output from webhook, it is exiting from flow. But it is supposed to pass the location parameters to next intent and stay on the flow. Can anybody help me how to stop assistant from exiting?
Here is the webhook code
'use strict';
const functions = require('firebase-functions');
const DialogflowApp = require('actions-on-google').DialogflowApp;
exports.dialogflowFirebaseFulfillment = functions.https.onRequest((request, response) => {
const requestPermission = (app) => {
app.askForPermissions('To report ', [app.SupportedPermissions.NAME, app.SupportedPermissions.DEVICE_PRECISE_LOCATION]);
};
const userInfo = (app) => {
if (app.isPermissionGranted()) {
const address = app.getDeviceLocation().coordinates;
const name = app.getUserName().givenName;
if (name) {
app.tell(`You are name ${name}`);
}
else {
// Note: Currently, precise locaton only returns lat/lng coordinates on phones and lat/lng coordinates
// and a geocoded address on voice-activated speakers.
// Coarse location only works on voice-activated speakers.
app.tell('Sorry, I could not figure out where you are.Plaese try again');
}
} else {
app.tell('Sorry, I could not figure out where you are.Please try again');
}
};
const app = new DialogflowApp({request, response});
const actions = new Map();
actions.set('request_permission', requestPermission);
actions.set('user_info', userInfo);
app.handleRequest(actions);
});
The problem is that you are calling app.tell() in your code which is a signal to the Assistant to send the message and then end the conversation.
If you want to send the message and then leave the microphone open for the user to reply, you should use app.ask() instead. It takes the same parameters - the only difference is that it expects the user to reply.
So that portion of your code might look something like
if (name) {
app.ask(`You are name ${name}. What would you like to do now?`);
}
(You should make sure that the prompt for the user is one that they will expect to reply. The review process will reject your Action if you reply and it isn't obvious that the user is supposed to reply to you.)

GoogleActions Account not linked yet error

I'm trying to implement oauth2 authentication on my nodejs Google Assistant app developed using (DialogFlow or API.ai and google actions).
So I followed this answer. But I'm always getting "It looks like your test oauth account is not linked yet. " error. When I tried to open the url shown on the debug tab, it shows 500 broken url error.
Dialogflow fullfillment
index.js
'use strict';
const functions = require('firebase-functions'); // Cloud Functions for Firebase library
const DialogflowApp = require('actions-on-google').DialogflowApp; // Google Assistant helper library
const googleAssistantRequest = 'google'; // Constant to identify Google Assistant requests
exports.dialogflowFirebaseFulfillment = functions.https.onRequest((request, response) => {
console.log('Request headers: ' + JSON.stringify(request.headers));
console.log('Request body: ' + JSON.stringify(request.body));
// An action is a string used to identify what needs to be done in fulfillment
let action = request.body.result.action; // https://dialogflow.com/docs/actions-and-parameters
// Parameters are any entites that Dialogflow has extracted from the request.
const parameters = request.body.result.parameters; // https://dialogflow.com/docs/actions-and-parameters
// Contexts are objects used to track and store conversation state
const inputContexts = request.body.result.contexts; // https://dialogflow.com/docs/contexts
// Get the request source (Google Assistant, Slack, API, etc) and initialize DialogflowApp
const requestSource = (request.body.originalRequest) ? request.body.originalRequest.source : undefined;
const app = new DialogflowApp({request: request, response: response});
// Create handlers for Dialogflow actions as well as a 'default' handler
const actionHandlers = {
// The default welcome intent has been matched, welcome the user (https://dialogflow.com/docs/events#default_welcome_intent)
'input.welcome': () => {
// Use the Actions on Google lib to respond to Google requests; for other requests use JSON
//+app.getUser().authToken
if (requestSource === googleAssistantRequest) {
sendGoogleResponse('Hello, Welcome to my Dialogflow agent!'); // Send simple response to user
} else {
sendResponse('Hello, Welcome to my Dialogflow agent!'); // Send simple response to user
}
},
// The default fallback intent has been matched, try to recover (https://dialogflow.com/docs/intents#fallback_intents)
'input.unknown': () => {
// Use the Actions on Google lib to respond to Google requests; for other requests use JSON
if (requestSource === googleAssistantRequest) {
sendGoogleResponse('I\'m having trouble, can you try that again?'); // Send simple response to user
} else {
sendResponse('I\'m having trouble, can you try that again?'); // Send simple response to user
}
},
// Default handler for unknown or undefined actions
'default': () => {
// Use the Actions on Google lib to respond to Google requests; for other requests use JSON
if (requestSource === googleAssistantRequest) {
let responseToUser = {
//googleRichResponse: googleRichResponse, // Optional, uncomment to enable
//googleOutputContexts: ['weather', 2, { ['city']: 'rome' }], // Optional, uncomment to enable
speech: 'This message is from Dialogflow\'s Cloud Functions for Firebase editor!', // spoken response
displayText: 'This is from Dialogflow\'s Cloud Functions for Firebase editor! :-)' // displayed response
};
sendGoogleResponse(responseToUser);
} else {
let responseToUser = {
//richResponses: richResponses, // Optional, uncomment to enable
//outputContexts: [{'name': 'weather', 'lifespan': 2, 'parameters': {'city': 'Rome'}}], // Optional, uncomment to enable
speech: 'This message is from Dialogflow\'s Cloud Functions for Firebase editor!', // spoken response
displayText: 'This is from Dialogflow\'s Cloud Functions for Firebase editor! :-)' // displayed response
};
sendResponse(responseToUser);
}
}
};
// If undefined or unknown action use the default handler
if (!actionHandlers[action]) {
action = 'default';
}
// Run the proper handler function to handle the request from Dialogflow
actionHandlers[action]();
// Function to send correctly formatted Google Assistant responses to Dialogflow which are then sent to the user
function sendGoogleResponse (responseToUser) {
if (typeof responseToUser === 'string') {
app.ask(responseToUser); // Google Assistant response
} else {
// If speech or displayText is defined use it to respond
let googleResponse = app.buildRichResponse().addSimpleResponse({
speech: responseToUser.speech || responseToUser.displayText,
displayText: responseToUser.displayText || responseToUser.speech
});
// Optional: Overwrite previous response with rich response
if (responseToUser.googleRichResponse) {
googleResponse = responseToUser.googleRichResponse;
}
// Optional: add contexts (https://dialogflow.com/docs/contexts)
if (responseToUser.googleOutputContexts) {
app.setContext(...responseToUser.googleOutputContexts);
}
app.ask(googleResponse); // Send response to Dialogflow and Google Assistant
}
}
// Function to send correctly formatted responses to Dialogflow which are then sent to the user
function sendResponse (responseToUser) {
// if the response is a string send it as a response to the user
if (typeof responseToUser === 'string') {
let responseJson = {};
responseJson.speech = responseToUser; // spoken response
responseJson.displayText = responseToUser; // displayed response
response.json(responseJson); // Send response to Dialogflow
} else {
// If the response to the user includes rich responses or contexts send them to Dialogflow
let responseJson = {};
// If speech or displayText is defined, use it to respond (if one isn't defined use the other's value)
responseJson.speech = responseToUser.speech || responseToUser.displayText;
responseJson.displayText = responseToUser.displayText || responseToUser.speech;
// Optional: add rich messages for integrations (https://dialogflow.com/docs/rich-messages)
responseJson.data = responseToUser.richResponses;
// Optional: add contexts (https://dialogflow.com/docs/contexts)
responseJson.contextOut = responseToUser.outputContexts;
response.json(responseJson); // Send response to Dialogflow
}
}
});
// Construct rich response for Google Assistant
const app = new DialogflowApp();
const googleRichResponse = app.buildRichResponse()
.addSimpleResponse('This is the first simple response for Google Assistant')
.addSuggestions(
['Suggestion Chip', 'Another Suggestion Chip'])
// Create a basic card and add it to the rich response
.addBasicCard(app.buildBasicCard(`This is a basic card. Text in a
basic card can include "quotes" and most other unicode characters
including emoji 📱. Basic cards also support some markdown
formatting like *emphasis* or _italics_, **strong** or __bold__,
and ***bold itallic*** or ___strong emphasis___ as well as other things
like line \nbreaks`) // Note the two spaces before '\n' required for a
// line break to be rendered in the card
.setSubtitle('This is a subtitle')
.setTitle('Title: this is a title')
.addButton('This is a button', 'https://assistant.google.com/')
.setImage('https://developers.google.com/actions/images/badges/XPM_BADGING_GoogleAssistant_VER.png',
'Image alternate text'))
.addSimpleResponse({ speech: 'This is another simple response',
displayText: 'This is the another simple response 💁' });
// Rich responses for both Slack and Facebook
const richResponses = {
'slack': {
'text': 'This is a text response for Slack.',
'attachments': [
{
'title': 'Title: this is a title',
'title_link': 'https://assistant.google.com/',
'text': 'This is an attachment. Text in attachments can include \'quotes\' and most other unicode characters including emoji 📱. Attachments also upport line\nbreaks.',
'image_url': 'https://developers.google.com/actions/images/badges/XPM_BADGING_GoogleAssistant_VER.png',
'fallback': 'This is a fallback.'
}
]
},
'facebook': {
'attachment': {
'type': 'template',
'payload': {
'template_type': 'generic',
'elements': [
{
'title': 'Title: this is a title',
'image_url': 'https://developers.google.com/actions/images/badges/XPM_BADGING_GoogleAssistant_VER.png',
'subtitle': 'This is a subtitle',
'default_action': {
'type': 'web_url',
'url': 'https://assistant.google.com/'
},
'buttons': [
{
'type': 'web_url',
'url': 'https://assistant.google.com/',
'title': 'This is a button'
}
]
}
]
}
}
}
};
Actually I deployed the code exists in the dialog flow inline editor. But don't know how to implement an oauth endpoint, whether it should be a separate cloud function or it has to be included within the existsing one. And also I am so confused with how oauth authorization code flow will actually work.. Let's assume we are on the Assistant app, once the user say "talk to foo app", does it automatically opens a web browser for oauth code exchange process?
The answer you referenced had an update posted on October 25th indicating they had taken action to prevent you from entering in a google.com endpoint as your auth provider for Account Linking. It seems possible that they may have taken other actions to prevent using Google's auth servers in this way.
If you're using your own auth server, the error 500 would indicate an error on your oauth server, and you should check your oauth server for errors.
Update to answer some of your other questions.
But don't know how to implement an oauth endpoint
Google provides guidance (but not code) on what you need to do for a minimal OAuth service, either using the Implicit Flow or the Authorization Code Flow, and how to test it.
whether it should be a separate cloud function or it has to be included within the existing one
It should be separate - it is even arguable that it must be separate. In both the Implicit Flow and the Authorization Code Flow, you need to provide a URL endpoint where users will be redirected to log into your service. For the Authorization Code Flow, you'll also need an additional webhook that the Assistant will use to exchange tokens.
The function behind these needs to be very very different than what you're doing for the Dialogflow webhook. While someone could probably make a single function that handles all of the different tasks - there is no need to. You'll be providing the OAuth URLs separately.
However, your Dialogflow webhook does have some relationship with your OAuth server. In particular, the tokens that the OAuth server hands to the Assistant will be handed back to the Dialogflow webhook, so Dialogflow needs some way to get the user's information based on that token. There are many ways to do this, but to list just a few:
The token could be a JWT and contain the user information as claims in the body. The Dialogflow webhook should use the public key to verify the token is valid and needs to know the format of the claims.
The OAuth server and the Dialogflow webhook could use a shared account database, and the OAuth server store the token as a key to the user account and delete expired keys. The Dialogflow webhook could then use the token it gets as a key to look up the user.
The OAuth server might have a(nother) webhook where Dialogflow could request user information, passing the key as an Authorization header and getting a reply. (This is what Google does, for example.)
The exact solutions depends on your needs and what resources you have available to you.
And also I am so confused with how oauth authorization code flow will actually work.. Let's assume we are on the Assistant app, once the user say "talk to foo app", does it automatically opens a web browser for oauth code exchange process?
Broadly speaking - yes. The details vary (and can change), but don't get too fixated on the details.
If you're using the Assistant on a speaker, you'll be prompted to open the Home app which should be showing a card saying what Action wants permission. Clicking on the card will open a browser or webview to the Actions website to begin the flow.
If you're using the Assistant on a mobile device, it prompts you directly and then opens a browser or webview to the Actions website to begin the flow.
The auth flow basically involves:
Having the user authenticate themselves, if necessary.
Having the user authorize the Assistant to access your resources on the user's behalf.
It then redirects to Google's servers with a one-time code.
Google's servers then take the code... and close the window. That's the extent of what the user's see.
Behind the scenes, Google takes this code and, since you're using the Authorization Code Flow, exchanges it for an auth token and a refresh token at the token exchange URL.
Then, whenever the user uses your Action, it will send an auth token along with the rest of the request to your server.
Plz suggest the necessary package for OAuth2 configuration
That I can't do. For starters - it completely depends on your other resources and requirements. (And this is why StackOverflow doesn't like people asking for suggestions like this.)
There are packages out there (you can search for them) that let you setup an OAuth2 server. I'm sure someone out there provides OAuth-as-a-service, although I don't know any offhand. Finally, as noted above, you can write a minimal OAuth2 server using the guidance from Google.
Trying to create a proxy for Google's OAuth is... probably possible... not as easy as it first seems... likely not as secure as anyone would be happy with... and possibly (but not necessarily, IANAL) a violation of Google's Terms of Service.
can't we store the user's email address by this approach?
Well, you can store whatever you want in the user's account. But this is the user's account for your Action.
You can, for example, access Google APIs on behalf of your user to get their email address or whatever else they have authorized you to do with Google. The user account that you have will likely store the OAuth tokens that you use to access Google's server. But you should logically think of that as separate from the code that the Assistant uses to access your server.
My implementation of a minimal oauth2 server(works for the implicit flow but doesn't store the user session).
taken from https://developers.google.com/identity/protocols/OAuth2UserAgent.
function oauth2SignIn() {
// Google's OAuth 2.0 endpoint for requesting an access token
var oauth2Endpoint = 'https://accounts.google.com/o/oauth2/v2/auth';
// Create element to open OAuth 2.0 endpoint in new window.
var form = document.createElement('form');
form.setAttribute('method', 'GET'); // Send as a GET request.
form.setAttribute('action', oauth2Endpoint);
//Get the state and redirect_uri parameters from the request
var searchParams = new URLSearchParams(window.location.search);
var state = searchParams.get("state");
var redirect_uri = searchParams.get("redirect_uri");
//var client_id = searchParams.get("client_id");
// Parameters to pass to OAuth 2.0 endpoint.
var params = {
'client_id': YOUR_CLIENT_ID,
'redirect_uri': redirect_uri,
'scope': 'email',
'state': state,
'response_type': 'token',
'include_granted_scopes': 'true'
};
// Add form parameters as hidden input values.
for (var p in params) {
var input = document.createElement('input');
input.setAttribute('type', 'hidden');
input.setAttribute('name', p);
input.setAttribute('value', params[p]);
form.appendChild(input);
}
// Add form to page and submit it to open the OAuth 2.0 endpoint.
document.body.appendChild(form);
form.submit();
}
This implementation isn't very secure but it's the only code I've gotten to work as OAuth server for the Assistant.
I am able to make it work after a long time. We have to enable the webhook first and we can see how to enable the webhook in the dialog flow fulfillment docs If we are going to use Google Assistant, then we have to enable the Google Assistant Integration in the integrations first. Then follow the steps mentioned below for the Account Linking in actions on google:-
Go to google cloud console -> APIsand Services -> Credentials -> OAuth 2.0 client IDs -> Web client -> Note the client ID, client secret from there -> Download JSON - from json note down the project id, auth_uri, token_uri -> Authorised Redirect URIs -> White list our app's URL -> in this URL fixed part is https://oauth-redirect.googleusercontent.com/r/ and append the project id in the URL -> Save the changes
Actions on Google -> Account linking setup 1. Grant type = Authorisation code 2. Client info 1. Fill up client id,client secrtet, auth_uri, token_uri 2. Enter the auth uri as https://www.googleapis.com/auth and token_uri as https://www.googleapis.com/token 3. Save and run 4. It will show an error while running on the google assistant, but dont worry 5. Come back to the account linking section in the assistant settings and enter auth_uri as https://accounts.google.com/o/oauth2/auth and token_uri as https://accounts.google.com/o/oauth2/token 6. Put the scopes as https://www.googleapis.com/auth/userinfo.profile and https://www.googleapis.com/auth/userinfo.email and weare good to go. 7. Save the changes.
In the hosting server(heroku)logs, we can see the access token value and through access token, we can get the details regarding the email address.
Append the access token to this link "https://www.googleapis.com/oauth2/v1/userinfo?access_token=" and we can get the required details in the resulting json page.
`accessToken = req.get("originalRequest").get("data").get("user").get("accessToken")
r = requests.get(link)
print("Email Id= " + r.json()["email"])
print("Name= " + r.json()["name"])`

Resources