I want to send a request to this Amazon Alexa API.
That page contains the last 50 activities I made with my Amazon Echo. The page returns JSON. Before you can request that page, you need to authorize your account, so the proper cookies are set in your browser.
If I do something simple as:
const rp = require("request-promise");
const options = {
method: "GET",
uri: "https://alexa.amazon.com/api/activities?startTime=&size=50&offset=-1",
json: true
};
rp(options).then(function(data) {
console.log(data);
}).catch(function(err) {
console.log(err);
});
I can send a GET request to that URL. This works fine, except Amazon has no idea it's me who's sending the request, because I haven't authorized my NodeJS application.
I've successfully copied ~10 cookies from my regular browser into an incognito tab and authorized that way, so I know copying the cookies will work. After adding them all using tough-cookie, it didn't work, unfortunately. I still got redirected to the signin page (according to the error response).
How do I authorize for this API, so I can send my requests?
I have been looking for a solution for this too. The best idea I have is to use account linking, but I haven't try it yet. Looks like ASK-CLI has interface for this also, but I can't figure it out how to use it (what is that URL?). For linking account to 3rd party server is not easy, but link it back to Amazon for the json API should not be that complicated.
Related
✨ Hello everyone!✨
General Problem:
I have a web app that has about 50 images that shouldn't be able to be accessed before the user logs into the site. This should be a simple answer I suspect, there are plenty of sites that also require this basic protection. Maybe I do not know the right words to google here, but I am having a bit of trouble. Any help is appreciated.
App details:
My web app is built in typescript react, with a node.js/express/mongoDB backend. Fairly typical stuff.
What I have tried:
My best thought so far was to upload them into the public folder on the backend server hosted on heroku. Then I protected the images with authenication middlewear to any url that had "/images/" as a part of it. This works, partially. I am able to see the images when I call the api from postman with the authenication header. But I cannot figure out a way to display that image in my react web app. Here is the basic call I used.
fetch(url,
{
headers: {
Authorization:token,
},
}
);
and then the actual response is just an empty object when I try to copy it
{}
but I also get this when I console log the pure response, some kind of readable stream:
from following related question
I came up with the following: (which is normally wrapped in a asyc function)
const image = await fetch(url,{headers:{ Authorization:token}});
const theBlob = await image.blob();
console.log(URL.createObjectURL(theBlob));
which gives me the link: http://localhost:3000/b299feb8-6ee2-433d-bf05-05bce01516b3 which only displays a blank page.
Any help is very much appreciated! Thanks! 😄
After lots of work trying to understand whats going on, here is my own answer:
const image = await axios(url, { responseType: "blob", headers: {Authorization: token }});
const srcForImage = URL.createObjectURL(image.data)
Why it makes sense now
So I did not understand the innerworkings of what was going on. Please correct me, but the following is my understanding:
So the image was being sent in binary. What I had to do to fix that was to set the reponseType in axios as "blob", which then sent a blob, which I believe means its base 64 encoded instead. Then the function URL.createObjectURL does some magic, and must save it to the browser as part of the page. Then we can just use that as the image url. When you visit it yourself, you must type the 'blob:' part of the url it give you too, otherwise its blank, or stick it in <img src={srcForImage}/> and it works great. I bet it would've worked in the original fetch example in the question, I just never put the url in a tag or included 'blob:' as part of the URL.
That's correct, you send the auth token and the backend uses that to auth the user (check that he exists in the DB, that he has the correct Role and check the jwt too)
The server only responds with the images if the above is true
If your server is responding with an empty object then the problem is the backend not the frontend, console.log what you're sending to the frontend
This is an Express.js server. I'm trying to authenticate my Instagram API.
const express = require('express');
const path = require('path');
const bodyParser = require('body-parser');
const axios = require('axios');
const ejs = require('ejs');
var app = express();
// bodyparser middleware setup
app.use(bodyParser.urlencoded({ extended: false }));
app.use(bodyParser.json());
var instagramClientId = '123123';
app.get('/instagram', (req, res) => {
axios({
method: 'post',
url: `https://api.instagram.com/oauth/authorize/?client_id=${instagramClientId}&redirect_uri=REDIRECT-URI&response_type=code`,
}).then((response) => {
res.send(response.data);
console.log(response.data);
}).catch((e) => {
console.log(e);
});
});
// port set-up
const port = process.env.PORT || 3000;
app.listen(port, () => {
console.log(`app fired up on port ${port}`);
});
This is the error I got. Looks like the html file was sent just fine, but the css and js weren't executed. You can see the error messages were all about style and js not being excuted.
Is this because I need to enable some options in my res.send() code?
I'm going to answer two questions here, the first is what I think you're actually having problems with and the second is what would technically be the answer to this question.
What I think your actual problem is: A misunderstanding of OAuth2.0
Looking at your code, it looks like you're trying to get a user to authenticate with Instagram. This is great, OAuth2.0 is fantastic and a good way to authenticate at the moment, but you've got the wrong end of the stick with how to implement it. OAuth2.0 is about redirects, not proxying HTML to the user. In this case it looks like you're using axios to make a server side call to the instagram OAuth endpoint and then sending the user that HTML. What you should actually be doing is redirecting the user to the Instagram URL you've built.
A high level of the "dance" you go through is the following.
The user requests to login with instagram by pressing a button on your website.
You send the user to an instagram URL, that URL contains your applications token plus an "approved" redirect url. Once the user has logged in with Instagram, Instagram will redirect the user to your approved redirect url.
The users browser has now been redirected to a second endpoint on your server, this endpoint recieves a one-time token from Instagram. You take that token on your server side and use axios (or similar) to make a server side request to fetch some user information such as their profile. Once you have that data, you can then create a user in your the database if needed and issue a new session token to them. Along with the profile call on this, you'll also get a token given directly to you (different from the one the users browser gave you) which will allow you to make requests to the Instagram API for the privileges you requested from the user originally.
This means you have 2 endpoints on your service, the "hello, I'd like to log in with instagram, please redirect me to the instagram login page" and then "hello, instagram said I'm all good and gave me this token to prove it, you can now check with them directly" (this is the callback endpoint).
You can manage this whole process manually which is great for understanding OAuth, or you can use something like Passport.js to abstract this for you. This lets you inject your own logic in a few places and handles a lot of the back and forth dance for you. In this instance, I'd probably suggest handling it yourself to learn how it all works.
Ultimately, you are not sending the user any HTML via res.send or anything similar. Instead your first endpoint simply uses a res.redirect(instagramUrl). You also thus do not make any HTTP requests during this portion, you do that on the "callback" after they've entered their username and password with Instagram.
Technically the correct answer to this question: proxy the JS and CSS calls, but this is really bad for security!
You're sending some HTML from a 3rd party in this case. So you will need to also allow the user access to the HTML and CSS. Security wise, this is quite iffy and you should really consider not doing this as it's bad practice. All of the JS and CSS links in the page are most likely relative links, meaning they're asking you for some JS and CSS which you are not hosting. Your best bet is to find these exact paths (ie: /js/app.min.js) and to then proxy these requests. So you'll create a new endpoint on your service which will make a request to instagrams /js/app.min.js and then send that back down with res.send.
Again, please do not do this, pretending to be another service is a really bad idea. If you need something from instagram, use OAuth2.0 to authenticate the user and then make requests using their tokens and the official instagram API.
I am developing a website in Node.JS / Express, and have until now used JWTs to store authentication data.
To summarize, I have an AngularJS app which calls a REST service on Node, and for authentication the user POSTs username and data to the server and gets a JWT containing the username and expiration, which then gets sent along with every request which requires authentication.
Now, though, I need to be able to connect to a third party service (which is a REST service that runs on Delphi) which uses sessions to save authentication data instead.
Using Fiddler, I can see that the service, when used in another implementation (third party) responds with JSON authentication data. In all subsequent requests, the client sends a cookie called dsessionid, although it doesn't appear in the "response cookies" in the previous response.
My question is: how do I take that dsessionid cookie in my NodeJS app and put it in my JWT so I can send it to my actual client?
I managed to solve my problem by getting the session ID in the "Pragma" header of the response (I don't know if the header is something "default" for Delphi applications or if it was specific to this application, I had to look through Fiddler entries to find out what header had the data) and then using a custom cookie jar with the request module to put the session ID in the cookies of subsequent requests.
const request = require('request-promise-native')
return request.get({
url: 'http://example.com:20202/auth',
resolveWithFullResponse: true
})
.then(res => {
return res.headers['pragma']
})
.then(sid => {
let j = request.jar()
j.setCookie(request.cookie('sessionid=' + sid), 'http://example.com')
return request.get({
url:'http://example.com:20202/fetchData',
resolveWithFullResponse: true,
jar: j
})
Hope this helps somebody.
there is a website that works with virtual items for an online game. I made a chrome extension that automates some actions on that website. Since I'd like to make this run on my raspberryPi (and chromium with the extension seems to be too slow and unefficient) I am trying to move this into node.js.
The login for the website works with Steam OpenID. It allows you to select items from a list, click a few buttons, then it sends you a tradeoffer on steam.
My extension works with the website while I was logged in there. It receives their database with jQuery getJSON, loops through the array, pushes some values into an array and then sends a post request telling the website which items I want and which items I am offering.
Here is how I am sending the request from chrome:
function withdrawXHR(botId, playerItems, botItems) {
$.ajax({
url: websiteURL,
type: 'post',
data: {
"steamid": botId,
"peopleItems": playerItems,
"botItems": botItems
},
success: function (data) {
console.error('>> Done: ' + data)
console.log("")
},
error: function(XMLHttpRequest, textStatus, errorThrown) {
console.error('>> Error: ' + errorThrown)
console.log("")
}
});
}
I can do everything in node so far like receiving their database, working through it, filter out the values I need, but I can't manage to send a working request. The problem is probably the login / how the website knows who I am.
I used wrapAPI (a chrome extension) to catch the request that is being sent when manually working with the website. Here is what it looks like:
So these are the things I am wondering about:
How would I send this request from node?
How does the website know who I am? They obviously know, because they are sending me an offer, but I can't see any "personal" data in that request.
Would I need to log into Steam OpenId from Node in some way? Is that possible?
What is a CF-RAY? (See the end of the captured request).
I am quite new to JS and requests in general and even "newer" to Node.js. I don't fully understand how the background of sending requests works. I would just need some tips, ideas on how to achieve my goal here.
Any help is greatly appreciated! Thank you! :)
You cannot use XMLHttpRequest for resources across domains. ( incidentally, unless you are using an extension)
I would look into grabbing express.js, and something called CORS. CORS permits cross-domain requests.
Here: http://enable-cors.org/server_expressjs.html
And here is some information on XHR requests in browser extensions: https://developer.chrome.com/extensions
I have a node.js server which authenticates using google-passport-oauth2. My server-side code looks like that from the documentation:
app.get('/auth/google',
passport.authenticate('google', { scope:
[ 'https://www.googleapis.com/auth/plus.login',
, 'https://www.googleapis.com/auth/plus.profile.emails.read' ] }
));
app.get( '/auth/google/callback',
passport.authenticate( 'google', {
successRedirect: '/auth/google/success',
failureRedirect: '/auth/google/failure'
}));
I figure that /auth/google redirects to google's login, and when permissions are recieved, the google profile and token are sent to the callback /auth/google/callback.
Now I am making an android app which wants to authenticate with this API. I'm using the directions for integrating Google Sign-In to do the authentication on google's end. Now my android app has the profile and token and wants to verify it with my server.
I've tried doing this with passport-google-token and passport-google-id-token (not sure the difference...), but it didn't work for whatever reason. Now I'm looking at other possibilities, like a Google Client API library for node.js, but it seems bulky. Then there's the tokeninfo endpoint, which involves an extra request and more latency. Or maybe I should look at express-jwt?
And suddenly, I wonder... couldn't I just pass the token from my android app to the server at auth/google/callback? That would make things a little simpler. I think this must be a pipe dream, because I haven't found any information about doing it. But if it's possible, how should I format the token/profile data in the request so the passport.authenticate() method recognizes it? (JSON, form data, headers)
If this can't be done, I'm taking suggestions for well-documented token verification libraries for node...
I still don't know about reusing the google-passport-oauth2 route, but I did figure out how to validate Google's idToken using passport-google-id-token.
The documentation says:
The post request to this route should include a JSON object with the
key id_token set to the one the client received from Google (e.g.
after successful Google+ sign-in).
But it only works if it's sent as a query string (GET or POST works).
https://localhost:8888/auth/googletoken?id_token=xxxxxxxxxx
I have a feeling this is not the most secure method, but I'll have to deal with that later.
EDIT: It turns out, the token is useless without the client ID (in your app), so it's OK to send it by querystring.
EDIT 2: One of the google-id-token devs has reminded me that the JSON will only be received if body-parser has been installed.