GitHub - Edgio/is2: embedded RESTy http(s) server library from Edgio

Event driven http(s) API server library -in C++

Background

Abstract

is2 is a reasonably performant event driven http(s) API server library written in C++ with support for url routing, tls (with openssl), serving static files, proxying, and subrequests.

Supports

threading
r3 like url routing with support for slugs
URL handler registration inspired by flask blueprints
tls with openssl
handler support for files, proxies, and subrequests.

Architecture Overview

is2 starts with a server object.
listener objects -listening on a particular port with a given scheme (TCP/TLS) can be registered with a server object.
handler objects can be routed to the listener objects at specified urls.

A basic example of instantation/registration:

  // create a TCP listener on port 12345
  ns_is2::lsnr *lsnr =
    new ns_is2::lsnr(12345, ns_is2::SCHEME_TCP);
  
  // create request handler --based on ns_is2::default_rqst_h
  ns_is2::rqst_h *rqst_h = new request_handler();

  // specify route to request handler on the listener
  lsnr->add_route("/path1/path2", rqst_h);  

  // create a server object
  ns_is2::srvr *srvr = new ns_is2::srvr();

  // register listener with server object
  srvr->register_lsnr(lsnr);

The Reactor

is2 has an event system similar to libevent or libuv. The reactor is a std::priority_queue of timer events with inverted ordering, for dequeue'ing the next nearest timeout. The "next nearest timeout" is used for the subsequent call to wait for events (epoll_wait, select, kqueue).

Reactor loop psuedo code:

  while true:

    // dequeue / handle timer events until timer > now
    do 
      last_timer = timer_queue.dequeue
    while(last_timer < now)

    // wait for an event -up to the next timer event
    events = wait_event(last_timer)

    // handle events if any
    for each event
      handle (readable/writeable/etc)

Install

Build requirements

Packages

openssl
pthread

Building example

g++ ./test.cc -lis2 -lpthread -o test

Usage

Basic example

Code

#include <is2/srvr/srvr.h>
#include <is2/srvr/lsnr.h>
#include <is2/srvr/default_rqst_h.h>
#include <is2/srvr/api_resp.h>

// define handler
class base_handler: public ns_is2::default_rqst_h
{
public:
        // GET
        ns_is2::h_resp_t do_get(ns_is2::session &a_session,
                                ns_is2::rqst &a_rqst,
                                const ns_is2::url_pmap_t &a_url_pmap)
        {
                // send json response
                return send_json_resp(a_session,
                                      true,
                                      ns_is2::HTTP_STATUS_OK,
                                      "{\"msg\": \"Hello World\"}");
        }
};

int main(void)
{
        // create server
        ns_is2::srvr *l_srvr = new ns_is2::srvr();

        // set server name -for server response
        l_srvr->set_server_name("hello world server");

        // create listener
        ns_is2::lsnr *l_lsnr = new ns_is2::lsnr(12345, ns_is2::SCHEME_TCP);

        // create handler
        ns_is2::rqst_h *l_rqst_h = new base_handler();

        // add route to listener
        l_lsnr->add_route("/hello", l_rqst_h);

        // register listener with server
        l_srvr->register_lsnr(l_lsnr);

        // num_threads == 0 means run single thread in foreground
        l_srvr->set_num_threads(0);
        l_srvr->run();
        
        // cleanup
        if(l_srvr) {delete l_srvr; l_srvr = NULL;}
        if(l_rqst_h) {delete l_rqst_h; l_rqst_h = NULL;}
        return 0;
}

Running

Testing

>curl 'http://127.0.0.1:12345/hello' -v
*   Trying 127.0.0.1...
* Connected to 127.0.0.1 (127.0.0.1) port 12345 (#0)
> GET /hello HTTP/1.1
> Host: 127.0.0.1:12345
> User-Agent: curl/7.47.0
> Accept: */*
> 
< HTTP/1.1 200 OK
< Connection: keep-alive
< Content-Length: 22
< Content-type: application/json
< Date: Thu, 15 Feb 2018 00:42:40 GMT
< Server: hello world server
< 
* Connection #0 to host 127.0.0.1 left intact
{"msg": "Hello World"}

Performance with hurl

>hurl 'http://127.0.0.1:12345/hello' -t1 -p100 -l5
Running 1 threads 100 parallel connections per thread with infinite requests per connection
+-----------/-----------+-----------+-----------+--------------+-----------+-------------+-----------+
| Completed / Requested |    IdlKil |    Errors | kBytes Recvd |   Elapsed |       Req/s |      MB/s |
+-----------/-----------+-----------+-----------+--------------+-----------+-------------+-----------+
|     58456 /     58556 |         0 |         0 |     11064.94 |     1.00s |   62255.49s |    10.81s |
|     89685 /     89785 |         0 |         0 |     11100.93 |     1.50s |   62458.00s |    10.84s |
|    120803 /    120903 |         0 |         0 |     11061.48 |     2.00s |   62236.00s |    10.80s |
|    151836 /    151936 |         0 |         0 |     11031.26 |     2.50s |   62066.00s |    10.77s |
|    182631 /    182731 |         0 |         0 |     10946.66 |     3.00s |   61590.00s |    10.69s |
|    213663 /    213763 |         0 |         0 |     11030.91 |     3.50s |   62064.00s |    10.77s |
|    244744 /    244844 |         0 |         0 |     11048.32 |     4.00s |   62162.00s |    10.79s |
|    275814 /    275914 |         0 |         0 |     11044.41 |     4.50s |   62140.00s |    10.79s |
|    306890 /    306990 |         0 |         0 |     11046.55 |     5.00s |   62152.00s |    10.79s |
| RESULTS:             ALL
| fetches:             306890
| max parallel:        100
| bytes:               7.949221e+07
| seconds:             5.001000
| mean bytes/conn:     259.025090
| fetches/sec:         61365.726855
| bytes/sec:           1.589526e+07
| HTTP response codes: 
| 200 -- 306890

Contribute

We welcome issues, questions and pull requests.

License

This project is licensed under the terms of the Apache 2.0 open source license. Please refer to the LICENSE-2.0.txt file for the full terms.

Table of Contents