I've been working on an app that bridges http calls over an established tunnel between two JID's. I have my tunneling setup correctly, but I cannot figure out how I'm supposed to use the HttpServer to serve requests. Using talk_base::HttpListenServer I can accept connections:
server.Listen(talk_base::SocketAddress("127.0.0.1", 59000));
But even after attaching a signal:
server.SignalHttpRequest.connect(this, &TunnelManager::HandleTouchRequest);
I have no idea how I'm supposed to respond to the requests. At this point, the requests just hang after connecting with no apparent signals being called. I was wondering if there are any better resources, possibly code examples, for me to look at to get a working http server?
So it looks like there might be a problem with using the PhysicalSocketServer on OSX instead of the platform specific wrapper. Setting up the correct socket server before anything else fixed the problem. Apparently my usage above is correct.
talk_base::Thread::Current()->set_socketserver(new talk_base::MacCFSocketServer());
This appears to have fixed problems with the HttpClient as well.
Related
I have been trying to connect two machines: both Virtual Machines, one is Xubuntu and the other is Ubuntu. I'm also very new to OpenDDS, but the best way -or so it seems- to do it is to use the .ini files.
However, when I try to connect, I seem to fail in changing the Discovery Server, since the default is localhost:12345. Can somebody help me with that so I can configure the file properly?
I have tried using the dds_udp_conf.ini and the tcp one, but it doesn't seem to work.
Also, I tried using unicast, but failed.
the ini file:
[common]
DCPSDebugLevel=0
DCPSInfoRepo=corbaloc::localhost::12345/DCPSInfoRepo
DCPSGlobalTransportConfig=config1
[config/config1]
transports=udp1
[transport/udp1]
transport_type=udp
And I use the syntax:
./publisher -DCPSConfigFile conf.ini
Well, the publisher and subscriber are supposed to connect, but the publisher sends some error messages and in the other VM nothing happens.
I seem to fail because I cant change the configuration in the localhost for discovery.
When I try to run the server with a different parameter than localhost:12345 it always sends error messages too.
It's unclear to me where you're running the InfoRepo if both the publisher and subscriber are told the InfoRepo is running at localhost. Regardless I would recommend using the RTPS discovery and transport instead. It's easy to set up because the participants can find each other through the network's multicast without InfoRepo. This config is the simplest way to use RTPS with OpenDDS:
[common]
DCPSDefaultDiscovery=DEFAULT_RTPS
DCPSGlobalTransportConfig=$file
[transport/the_rtps_transport]
transport_type=rtps_udp
Just give this to both the programs and they should find each other. If not that would mean there's probably something's wrong with how the networking is set up on your VMs.
As I understand it, it is not simply possible to implement a rest webservice on a device which is behind NAT. So i was searching for some solutions for this.
Is it possible to use long polling in order to implement the webservice? this way, the local device will make a call to the remote client (which is exactly what i want), the client has to keep the connection open (keep alive?) until the client want to call a webservice method. It can do so, because the connection is still open. After the call the client will immediately send another poll to the client ... etc..
Is it possible to implement it this way?
Another solutions on which i came across:
ReverseHTTP - I don't know very much about this, but it sounds like i can implement the webservice with this. right?
There are several other solutions, like TURN or STUN but they seem to be very complicated.
Do you have any suggestions?
I am using c++/linux on my network devices.
EDIT: Port Forwarding is not an option.
You've got a lot of different concepts here in this question. You can certainly implement a RESTful service behind a firewall/NAT... you just need to configure your firewall/NAT to forward connections to your service. There are issues of firewall/NAT devices timing out connections... here again, you can configure your device to not do that, or you can update your communication mechanism with some kind of "keep-alive". "long polling" is somewhat unrelated, and is used as a way of getting an "interactive like response" from a server... basically the server sits on a poll request from a client until it has something to respond with, or the request times out and the client makes another one. STUN and TURN are more voice/video communications-related technologies. I suggest starting with simply having your firewall/NAT device forward web-based requests to your web server.
You don't say what transfer protocol you are using, I'm assuming HTTP.
HTTP uses TCP/IP, so your device NAT needs to redirect the connection request to your server.
There's others ways, like if you have more than one internet IP address, so the requests could be directed to the server too, but thats more complicated than port forward so I think its not what you have.
So basically you need to configure the port forward. Take it like a PABX, calls from the exterior lines needs to know a ramal to reach a phone, thats a distant mean to think of it.
And as said, the suggestions you said, are not intended for that, is mainly for client connection, what for many NAT is not necessary, as the NAT is prepared for doing that.
I'm brand new to c++ and know next to nothing about web protocols or websockets, so this may seem ridiculous.
I make websites that are 100% ajax and want to incorporate websockets. Fastcgi++ is everything I could hope for for the ajax demands, but it doesn't have websockets, and I chose websocket++ over libwebsockets since websocket++ is more or less a simple #include, so I assumed that I could incorporate it into fastcgi++.
I think I've figured out fastcgi++, and it looks like most of the action happens in Fastcgipp::Request then Fastcgipp::Http::Sessions for session data http://www.nongnu.org/fastcgipp/doc/2.1/a00005.html; however, I think I have to do the same thing with websocket++'s server::handler for handling the websocket https://github.com/zaphoyd/websocketpp/wiki/Creating-Applications-using-WebSocket--, and now I'm lost at sea.
Enter my complete inexperience with c++: I think I have to use virtual inheritance, but I have no idea. Also, if I could even properly "subclass" both, how do I make sure that they don't run over each other?
Please show me a basic example of how websocket++ can use fastcgi++'s session management.
A WebSocket connection cannot be processed by an HTTP request/response workflow. In order to use something like fastcgi++ with both regular HTTP requests and with WebSocket requests it would need to have some way of recognizing a WebSocket handshake and piping that off to another handler instead of processing it as HTTP. I don't see an obvious pass through mode of that sort in its documentation, but I could be missing something.
If such a feature exists, WebSocket++ can be used in stream mode where it disables all of its network elements and just processes streams of bytes piped in from another networking library.
Some alternatives:
WebSocket++ supports HTTP pass through. This is essentially the opposite of what is described above. WebSocket++ would be used as the networking layer. It would process incoming WebSocket connections and would pass off HTTP requests to some other subsystem.
WebSocket++ and fastcgi++ could be run on different ports or different hostnames. This could be done in the same program or separate programs. With client side requests directed to the appropriate host/port.
Disclaimer: I am the author of WebSocket++
Is it possible to create an HTTP tunnel in Delphi or C++?
My application connects to several HTTP servers that do not belong to the company I work for. Because of that, our users need to open their firewall ports to allow those connections. I thought about creating a tunnel at my company and redirecting HTTP requests made by my application through this tunnel. This way, my clients will only need to open one port and the tunnel will handle all requests. All requests are made with POST or GET using indy components.
EDIT: I can't use an HTTP proxy. Some of my users have already got their own HTTP proxy and it is going to be impossible to connect to two different proxy servers at the same time.
Here is a free component is kind of old but it works you can get yourself inspired from there
TGpHTTPProxy
Or you can try this samples
https://sites.google.com/site/delphibasics/home/delphibasicssnippets/examplesocks4proxybyaphex
https://sites.google.com/site/delphibasics/home/delphibasicssnippets/multi-threadedhttpproxyserver
As Warren P. and Rob Kennedy suggest, you really just need a proxy server. Don't write a tunnel yourself, it's a huge overkill and it's far from easy (writing a robust socket application is more time consuming than it first appears to be).
If you want something dead simple look for datapipe.c or netcat (nc) unix command. SSH can create tunnels too (look in OpenSSH and PuTTy docs).
Here is a free open source HTTP-Tunnel and UDP-Tunnel: http://barbatunnel.codeplex.com/
My objective is to have an event that is triggered when a website is accessed.
Now, maybe through the window title, or the text in the window. Or maybe even reading a URL.
As of now I am aware of FindWindow (class,title);
However all attempts to put this line of code into a loop and it's exit condition being when the window appears have been fruitless.
Any assistance would be very helpful.
That's not possible. At least if I understood you correctly.
You want to register a callback when ANY software on your machine accesses a specific website?
Just imagine a browser uses SSL, there is no way to detect this by listening to the traffic or something similar.
However, if you want to be notified about all connections to a specific IP, then you could use sniffing mechanisms of your kernel or even redirect all traffic to this IP to a proxy you have set up with iptables or similar.
Windows has a sniffing library called WinPCap, on linux you could use tcpdump.
Though, more information about your problem would be nice.
Looking for window titles can be a bit problematic. I don't know how much control you have over the desktop, but you might consider building an addon for Firefox (or the equivalent in IE) to look for this particular site.
https://developer.mozilla.org/En/Extensions/Firefox
You might also consider building a simple local proxy server (depending on what you are doing) that looks for this site and performs some action. You would have to make sure all the browsers on the machine point to this local proxy to get it working correctly. See the link below for some discussion on a custom proxy server:
How to create a simple proxy in C#?