net:ERR_CONNECTION_RESET from a specific website in electron app - electron

I Have an Electron app with a Webview. When i set the webview src to the website https://www.zap.co.il
i'm receiving the error: net:ERR_CONNECTION_RESET. The weird things are:
It's happening only for that website, i'm not receiving this error from any other website.
I'm able to reach the website from any other browser on my computer (Chrome, Chromium).
This is happening in both Windows and Mac platforms.
A colleague of mine have the same App installed on her computer, and she is experiencing the same behaviour from a different computer (and network)
There is no antivirus Installed that may intercept and no proxy configured. I have tried to reset the network, clear cache and cookies, change the MTU and basically any solution i found on the internet, and since it's happening for my colleague too, i suppose it has nothing to do with a network issues.
And since i can reach the website from other browsers, I think the issue is probably relates to Electron.
I'm Using the versions:
electron: 8.2.1
electron-builder: 22.4.1

The problem was that electron add's few properties to the user agent string. and this specific website is not accepting custom properties in the user agent (probably for security reasons). I fixed it by removing the additional properties from the webview's user agent:
webview.addEventListener('dom-ready', () => {
const uaArr = webview.getUserAgent().split(" ");
const newUaArr = uaArr.filter((uar => !uar.startsWith('Electron')));
webview.setUserAgent(newUaArr.join(" "));
});

Related

Does electron's registerHttpProtocol work in development?

I'm trying to register a custom protocol with electron. I want it to be a redirect location that a website can use to provide an api key (like myprotocol://example/payload=api-key). I have been using electron's registerHttpProtocol and also tried electron's interceptHttpProtocol.
But, when the website tries to redirect to my protocol my electron app doesn't do anything. The website goes to myprotocol://example/payload=api-key, and registers a "page doesn't exist error"--while nothing happens in my app.
This is in a development environment. I've seen some discussion about custom protocols that assume a production environment.
Can you register a custom protocol with electron in development?
Why am I not able to intercept the website's going to the protocol I've set out?
Here's my code:
main.js:
app.whenReady().then(() => {
protocol.registerHttpProtocol('examplep', (request, callback) => {
console.log("examplep", request);
callback('it-worked');
}, (error) => {
if (error) console.error('Failed to register protocol = ' + error)
})
protocol.interceptHttpProtocol("examplep", function (request, callback) { //I've tried both registerHttp... and interceptHttp... methods, so including both here; though I think in practice only one should be required
console.log('intercepted!' + request)
callback(request);
});
})
redirect url provided to website:
'http://examplep'
And I've whitelisted this url on the website itself.
I've also tried related methods registerStringProtocol, interceptStringProtocol, registerFileProtocol, and interceptFileProtocol, without success.
What am I missing?
Sounds like you need to support deep linking fora desktop app, which is done via a Custom URI Scheme and is registered with setAsDefaultProtocolClient.
When your Electron app starts up write this code to register the scheme, on the main side of your app:
const customScheme = 'x-mycompany-myapp';
app.setAsDefaultProtocolClient(customScheme);
The custom scheme can be tested from the command line like this, depending whether you are running macOS or Windows:
open x-mycompany-myapp:/some/location
start x-mycompany-myapp:/some/location
A web client will just invoke a URL as in this Javascript code of mine;
The notification will be received within the main side of your app and on Windows will attempt to create a new instance of the app, in which case you need to detect this condition, process the notification then cancel the new app instance.
On MacOS it will be received within the open-url event, so you register it like this:
app.on('open-url', this._onOpenUrl);
Once the main side of the Electron app has the notification, it needs to get the URL information and forward it to the renderer process. You can use ipcMain events for this.
Finally the code for receiving the notification in running instances and starting the app from a deep link are different.
EXAMPLE APP
Since the code is a little tricky, here is some example code that may be useful, to give you something to compare against. If it helps you can also run the app by following the instructions in the blog post:
Code
Blog Post
My use case is around receiving OAuth responses after signing in from the system browser. Hopefully you can borrow some ideas from it related to deep linking though.
INFO.PLIST
My understand is that in a development environment (on macOS) deep links work when the app is running, but if you stop the app and attempt a deep link it will not start the app.
You can only resolve this for a packaged app, which requires an info.plist. In my code sample the info.plist is generated from build protocol entries in the package.json file.
My code sample is packaged in a basic way by the Electron Packager, so when I run npm run pack, the app is built to a dist folder. I can then run the packaged version of the app and it gets registered with the system - as can be seen in the Default Apps tool. See screenshots in the blog post.
SECRETS
Secrets for a desktop app should be stored using operating system secure storage. There are screenshots of credential storage in the blog post.
On Electron, have a look at the keytar component - and this wrapper class of mine. I am storing tokens (strings) so you should be able to adapt the code for your API keys.

Previewing rails server on AWS Cloud 9: VFS Connection does not exist [Firefox only]

Update: This is specific to Firefox. This does not occur using Chrome. I will leave the text below as is, but now the main question is, how and what in Firefox is causing this. It occurs regardless of cookie settings, such as allowing third-party cookies and trackers.
Preface: I've searched and read dozens of pages trying to resolve this issue I've gone through all the troubleshooting steps in the AWS documentation, eg,
https://docs.aws.amazon.com/cloud9/latest/user-guide/app-preview.html#app-preview-share
and this thread:
https://forums.aws.amazon.com/message.jspa?messageID=849200 as well as resources on stackoverflow
and the c9 forums
https://community.c9.io/t/what-is-vfs-connection-does-not-exist-on-aws-c9/22697/4)
I am trying to preview a running rails app from the Cloud 9 IDE. When I run the app using rails server, there are no error messages indicating anything wrong. The output shows Started GET '/' for [my home IP] ...... 200 Completed OK.
Console Output:
Trying to preview the running app only produces a "OOPS VFS connection does not exist" screen. The preview url is always https://[user_generated_sequence].vfs.cloud9.[region].amazonaws.com/ All TCP traffic is allowed as per the rules set up by following the troubleshooting guides.
OOPS error message page:
The same issue occurs whether using puma or WEBrick. With WEBrick, a popup for the preview link appears, but leads to the same error message page. With puma, the running app is listening on tcp://[localhost/0.0.0.0]:[8080 or 3000]. With WEBrick, it listens instead on http://localhost:8080.
I have followed the instructions exactly in the rails tutorial to set up a simple app. Everything in the code itself should be fine. How can I get this to work correctly? I'm very confused and about to give up on rails.
I recently had the same issues after updating Firefox because cross-site cookies are required to preview a running rails application. If, like me, you have disabled that in Firefox as part of strict Enhanced Tracking Protection, they will need to be enabled at least for this specific site.
Make sure you are accessing the preview from the same browser:
This URL works only when the IDE for the environment is open and the
application is running in the same web browser
https://docs.aws.amazon.com/cloud9/latest/user-guide/app-preview.html#app-preview-preview-app
Something that worked for me, do the following:
run:
rails s
the in the AWS EDE click on 'Preview' and 'Select Running Application'
Once it will open click on the button/link and it should open in another window with the running application.
Note: Doing this manually by copying the URL link and pasting in another tab did not work for me.
Try disable Ad Blockers and privacy Extention. It worked for me.
Basically you need to allow third party cookies
in chrome it's
Settings >> Security >> Cookies and web data >> Choose second option(block cookies only in incognito)
That works for me
On cloud9, click the shield logo in the address bar and switch "Enhanced Tracking Protection" to be off. Then refresh the page and the preview.
I was struggling with this as well. Finally, I changed the cookie setting from all third-party cookies to cookies from unvisited sites (See image)
Cookie Settings for Firefox

Can't connect to local web server in my cordova iOS app

When switching from UIWebView to WKWebView in a Cordova iOS app there are common known problems with same origin policy when loading a web app from the local file system.
The typical fix is to use a local web server (listening on localhost) which serves your HTML,CSS,JS.
This is why I tried to switch to this cordova plugin:
https://github.com/apache/cordova-plugins/tree/wkwebview-engine-localhost
Unfortunately the result is a blank screen. In the web debug console I called
console.log(location.href)
which returns about:blank. Then I tryed navigating to the local web server with this JS call from the console:
location.href = 'http://localhost:8080';
But the web view stays at about:blank. I also tried with different ports and of course I checked my config.xml (e.g. <allow-navigation href="*" />)
I assumed that there is a problem with with the plugin so I switched to Ionic Webview. Interestingly it has the exact same behavior.
In general I am able to navigate to other sites, but not to localhost:[port], so I guess that those plugins are not able to start the web server at all. However, I couldn't find anyone else with this problem.
What could be the reason for this problem? Is there a way to debug this? Unfortunately my web console doesn't report any problems.
The problem here was not to start the web server but connecting to it. The communication has been prevented by VPN (MobileIron) which tries to tunnel those connections. Unfortunately, at the moment there is no solution for WkWebView at all.

Facebook iOS Mobile Web JS SDK Error

We already have Facebook working on our main desktop app without an issue. When I added the same setup code to our mobile site I received an error from the JavaScript SDK:
Received message of type object from https://s-static.ak.facebook.com, expected a string
Then, I changed the setup code to an exact replica from here: https://developers.facebook.com/docs/guides/mobile/web/ and received the exact same message.
This error only shows when the user-agent is iOS or Android in Chrome Developer Tools or in the iOS Simulator. When it's running from the desktop, we aren't seeing this error.
Any idea what's going on here?
Thanks :)
With all due respect, I believe that bokonic's answer is missing the point. The immediate reason of the message you're receiving is most likely javascript postMessage method. It is used to communicate your website with an object returned by window.open or an iframe's embedded website.
window.parent.postMessage(message, targetOrigin);
You might want to have a look at this article to see how it works.
According to MDN reference first argument of the postMessage method can be of any type. Still, Prior to Gecko 6.0 (Firefox 6.0 / Thunderbird 6.0 / SeaMonkey 2.3), the message parameter must be a string.
I tested the method on Chrome 28.0.1500.72 (latest version when writing this answer) and it works fine when passing an object but still throws the warning message to the console. I believe that facebook feels uncomfortable with passing some rich data as a string so they pass it as an object and simply doesn't care about the warning. Or, since there is a workaround (e.g. JSON.stringify), they have a bug they don't know about.
I had the same error. This isn't necessarily the reason you're getting it, but I ended up looking at the packets being received by my iphone simulator, and here's the error from s-statick.ak:
[truncated] <span>Given URL is not allowed by the Application configuration.: One or more of the given URLs is not allowed by the App's settings. It must match the Website URL or Canvas URL, or the domain must be a subdomain of one o
For me, the error is that on my device(s), I'm accessing the site through an ip address (192.168.xx etc) and on local development I have hosts routing localhost to a url that is in my facebook app config... so the IP isn't configured for the app.
Facebook won't let me use an IP address, so it seems like the only option is a test deploy or a DNS server. Hope this helps you out.

SUP101: blackberry works on simulator but not device

I'm having a problem with getting the tutorial to work on the device
it works fine in the simulator, but when I try to run it on the device it doesn't work.
I put some logs and it seems to come from the login function
===JASONROGERS292C 2480 supAdmin s3pAdmin
[0.0] TAC:TAI.handleRegRsp : ALREADY_REGISTERED
[0.0] TAC:TAI.handleRegRsp : (ALREADY_REGISTERED), registered entry found:{{APN=,id=2,state=CLOSED,waf=WAF[0.0] _3GPP,keepAlive=300,IP=0.0.0.0,dnsSrv=0.0.0.0}}
[0.0] [TMM.TunnelAllocator]:runNotReadyListeners: registered listeners found
[0.0] TAC:TAI.handleListenerDeRegRsp: not existent tunnel in TAC, tid=2
[0.0] SIM cache not loaded
has anybody got an idea what I forgot to do when running the code on a device?
Cheers
Jason
edit:
a couple of links towards what I'm doing (I can post all the code people want but its the standard generated code form SUP, which mean a lot of code lol)
The tutorial
http://infocenter.sybase.com/help/topic/com.sybase.infocenter.dc01214.0200/doc/html/title.html
the zip to the already done tutorial :
http://www.sdn.sap.com/irj/bpx/index?rid=/webcontent/uuid/40ea4956-b95c-2e10-11b3-e68c73b2280e
Solution:
I was missing a couple of things:
1) Specify that the app should use Wifi
SUP101DB.getSynchronizationProfile().setString("transport", "WIFI");
2) Check that the DNS: in my case the server was local to my windows (through parallels) so I had to specify the ip and not the namespace
getSynchronizationProfile().setServerName("10.50.30.108");//"JASONSERVER");
I downloaded the SUP 101 - BlackBerry Project and checked the sources.
It does not contain networking code in the form a usual BB app would do.
However I see the SUP101.SUP101DB class has the following method:
private static String getSyncTransportSuffix() {
com.sybase.persistence.ConnectionProfile syncProfile =
getSynchronizationProfile();
return com.sybase.afx.util.NetworkUtil.getTransportUrlSuffix(syncProfile);
}
So the code uses com.sybase.afx.util.NetworkUtil to get url suffix. By appending such suffix to a url we say to the underlaying BB API what network transport to use. Unfortunatelly sources do not contain com.sybase.afx.util.NetworkUtil to look how exactly it works.
I should say that even with native Java BB development detecting a proper network transport may turn to be a very tricky part. OS 5+ provides a new networking API that is much better, but for the older OS it is hard. This SUP project is built with BlackBerry JRE 4.6.1, so it can not use new OS 5+ networking API. Who knows maybe SUP simply fails to detect proper network transport?
Any way, just a few ideas to try:
check for APN settings on the device, sometimes they have to be populated in order to use Direct TCP transport (it depends on wireless provider requirements). APN name/pass are wireless provider specific. Usually googling allows to find proper APN settings for a provider.
if your device has WiFi try enabling it. SUP should be smart enough to prefer WiFi if available. In this case I assume networking should have no issues.

Resources