-
-
Save SeanCline/6005128 to your computer and use it in GitHub Desktop.
#define _CRT_SECURE_NO_DEPRECATE | |
#include <cpprest/http_listener.h> | |
#include <iostream> | |
#include <iomanip> | |
#include <sstream> | |
#include <thread> | |
#include <chrono> | |
#include <ctime> | |
// cpprest provides macros for all streams but std::clog in basic_types.h | |
#ifdef _UTF16_STRINGS | |
// On Windows, all strings are wide | |
#define uclog std::wclog | |
#else | |
// On POSIX platforms, all strings are narrow | |
#define uclog std::clog | |
#endif // endif _UTF16_STRINGS | |
using namespace std; | |
using namespace web::http::experimental::listener; | |
using namespace web::http; | |
using namespace web; | |
void respond(const http_request& request, const status_code& status, const json::value& response) { | |
json::value resp; | |
resp[U("status")] = json::value::number(status); | |
resp[U("response")] = response; | |
// Pack in the current time for debugging purposes. | |
time_t now = time(0); | |
utility::stringstream_t ss; | |
ss << put_time(localtime(&now), L"%Y-%m-%dT%H:%S:%MZ"); | |
resp[U("server_time")] = json::value::string(ss.str()); | |
request.reply(status, resp); | |
} | |
int main() | |
{ | |
// Synchronously bind the listener to all nics. | |
uclog << U("Starting listener.") << endl; | |
http_listener listener(U("http://*/json")); | |
listener.open().wait(); | |
// Handle incoming requests. | |
uclog << U("Setting up JSON listener.") << endl; | |
listener.support(methods::GET, [] (http_request req) { | |
auto http_get_vars = uri::split_query(req.request_uri().query()); | |
auto found_name = http_get_vars.find(U("request")); | |
if (found_name == end(http_get_vars)) { | |
auto err = U("Request received with get var \"request\" omitted from query."); | |
uclog << err << endl; | |
respond(req, status_codes::BadRequest, json::value::string(err)); | |
return; | |
} | |
auto request_name = found_name->second; | |
uclog << U("Received request: ") << request_name << endl; | |
respond(req, status_codes::OK, json::value::string(U("Request received for: ") + request_name)); | |
}); | |
// Wait while the listener does the heavy lifting. | |
// TODO: Provide a way to safely terminate this loop. | |
uclog << U("Waiting for incoming connection...") << endl; | |
while(true) { | |
this_thread::sleep_for(chrono::milliseconds(2000)); | |
} | |
// Nothing left to do but commit suicide. | |
uclog << U("Terminating JSON listener.") << endl; | |
listener.close(); | |
return 0; | |
} |
Does it works for IPv6 for me "listener.open().wait();" is throwing below error.
terminate called after throwing an instance of 'boost::wrapexceptboost::system::system_error'
what(): resolve: Service not found
Using address as "http://[::1]:8585"
@mpanwar-github Hmm, it seems like a couple people have reported similar behaviour.
- https://stackoverflow.com/questions/45879633/how-to-listen-on-an-ipv6-address-using-cpprestsdk-http-listener
- https://user-images.githubusercontent.com/3418587/167885315-1de2d03b-d860-44e6-9f29-f5068ae434ec.png
On my Windows box I also get an error when listening on an IPv6 address. When I listen on a wildcard address like http://*/json
it listens on both IPv4 and IPv6, so that might be a workaround for you.
@SeanCline Thanks for response, I have already tried with "*" that shows port listening on IPv6. I didn't find anything if this works for IPv6 at all.
Even i tried with interface as wel this was throwing below error.
Starting listener at url.http://[fe80::a3fe:944b:534f:7ec%enp44s0f1]:8585
terminate called after throwing an instance of 'web::uri_exception'
what(): Invalid hexadecimal digit
further for me with hostname it is working not with IP "::1" that conclude that IPv6 support is there but format is different. Though i haven't done trying with other server IP with IPv6.
@samalitu That
U()
is specific to the C++ REST SDK.First, some history...
On Windows, it's common to use wide strings. This is because Windows NT settled on 16-bit (wide) characters before UTF-8 gained traction. The C++ REST SDK builds on several operating systems, including Windows and Linux. The developers decided to keep using wide strings when building for Windows to so they don't need to do a conversion from UTF-8 for every system call. But on Linux-like OS's they can use UTF-8 (narrow) strings everywhere.
So,
U()
is just a macro. On Windows it expands to something like:#define U(x) L x
but on Linux it's just#define U(x) x
. This puts anL
on the front of string literals when compiled on Windows, but leaves it alone with compiling on Linux.This design is starting to feel more dated as Microsoft slowly starts recommending using UTF-8 in more places.