• 分享一个嵌入式httpdserver开发库


    http://sourceforge.net/projects/boahttpd/

        一个C接口的开发库,适用于 windows/linux/或其它嵌入式平台,支持CGI扩展,支持多线程。採用面向对象开发,一个应用里能够同一时候开多个http server,仅仅要port不同就互不影响。主要应用场景应该是嵌入式应用(所谓boa-server的概念),在须要做一个基本web的设备管理使用。在样例中就是展示,从页面信息的提交、到处理、到结果的返回,里面结合jquery/bootstrap网页技术,能够作出一个精致的设备管理控制页面。 

    - Web file downloading 网页文件浏览
    - CGI support , CGI支持
    - Multi-thread, Multi-instance  多线程、多实例
    - Cross-Platform (Windows, Linux, other Embedded Linux) 跨平台支持
    - Filter & Authentication support  过滤器 / 鉴权password认证
    - Cookie support  , Cookie支持

         这个开发库体积较小,已经编译好了windows, linux(fedora14,x86)版本号的库。使用起来就是一个头文件、一个库文件。比方说以下几句话就完毕了一个简单的httpserver:

    #include "../boahttpd.h"
    int main()
    {
        boa_server_t server;
        if(boa_server_create(&server, NULL,
              "wwwroot", 
               8888, "0.0.0.0") < 0)
        {
           printf("failed to create server!
    ");
           return 0;
        } 
    
        getchar();
        boa_server_destroy(&server);
        return 0;
    }

               事实上就是一个简单的create操作,传入主文件夹文件夹的位置,port号就能够了。在当前文件夹下创建一个wwwroot, 在wwwroot放一个index.html,执行程序试试吧。在浏览器里输入http://127.0.0.1:8888试着訪问一下。
          上面是完毕一个普通的httpserver,能够浏览下载网页了。可是若要支持cgi的处理,则须要注冊一些回调函数。每一个cgi服务有一个唯一的名称,并相应一个回调函数。当然,能够用一个回调函数来处理全部的cgi服务。示范代码例如以下:

    static int cgi_handler(void* context, void* userdata)
    {
        … handle the cgi request …
        return 0;
    }
    int main()
    {
        boa_cgi_t cgi ;
        boa_cgi_create(&cgi);
        boa_cgi_register(&cgi, "test", &cgi_handler, NULL);
     
        boa_server_t server;
        if(boa_server_create(&server,&cgi,
                wwwroot", 8888, "0.0.0.0") < 0)
        {
           printf("failed to create server!
    ");
           return 0;
        }
         getchar();
    
        boa_cgi_destroy(&cgi);
        boa_server_destroy(&server);
        return 0;
    }
    

             上面是先定义若干个handler(即callback),然后把这些handler都加入�一个boa_cgi_t对象里,然后把这个boa_cgi_t对象交给server就可以。在执行的时候,server在接收到cgi请求后,会依据cgi服务的名称、找到对应的handler来处理。
              比方,输入cgi请求: http://127.0.0.1:8888/cgi/test?name=someone
              server收到此cgi请求后,检查到服务名称为test,于是检查boa_cgi_t里有没有注冊了名称为"test"的handler,假设有的话就调用之。在handler里,能够获取cgi请求的相关信息,比如參数"name=someone"、以及HTTP头部的其它field字段。假设是个POST请求,能够获得正文。

    static int cgi_handler(void* context, void* userdata)
    {
        const char* service = boa_cgi_name(context);
        const char* method = boa_cgi_method(context);
        const char* argument = boa_cgi_args(context);
        const char* content = boa_cgi_content(context);
        const char* anyfield = boa_cgi_field(context,”Content-Length”);
          int sock = boa_cgi_socket(context);
    }
    

           另外提供了2个函数,用来对此cgi请求做出回复。

         boa_cgi_send_error ( context, status, reason); // 错误请求,如404, "Not Found"
         boa_cgi_send_text(context, "正文text/json/html什么都能够", CONTENT_TYPE_TXT); // 200 OK + 正文
            

    须要铭记的是,此server是多线程的,每一个cgi请求都新开了一个线程处理,因此这个handler回调函数事实上是在各个线程里面被调用的。
            注意这个userdata的使用,事实上是面向对象的一个设计方法。比方,你想封装一个httpserver对象: 

    ////////////////  .h /////////////////////
    class MyHttpd
    {
    public:
    	int Open(int port, const char* ip);
    	void Close();
    	int HandleCgiRequest(void* context);
    private:    
    	boa_cgi_t    m_cgi;
    	boa_server_t  m_server;
    }
    
    ///////////////////////////// *.cpp /////////////////////////////
    static int cgi_handler(void* context, void* userdata)
    {
    	MyHttpd*  server = (MyHttpd*) userdata;                   
    	server->HandleCgiRequest(context); 
    	return 0;
    }
    int MyHttpd::Open(int port, const char* ip)
    {
    	boa_cgi_create(&cgi);
    	boa_cgi_register(&cgi, "test", &cgi_handler, this);  //这里将userdata指定为this
    	boa_server_create(&server,&cgi,  wwwroot", 8888, "0.0.0.0") ;   
    	return 0;
    }
    int int MyHttpd::HandleCgiRequest(void* context)
    { 
    	const char* service = boa_cgi_name(context);
    	...
    }
    
    
    

                 通过上面的转换,把一个C风格的回调转成了C++风格的回调,这样你能够有一个自己的server类了。

            

         

    boahttpd Library Development Manual

     

    The library, boahttpd, is designed to create a cgi-featured http server (normally called as boa-server)easy and fast. That means you can easily add CGI support to your server application. It support file downloading by default, like any other web servers.

    Its written in C++ inside, but provides a C-Style API which you will find it to be Object-Oriented essentially. Multiple platforms are supported with a uqique set of interface, including Windows, Linux, or Embedded systems. It will take a few minutes to get to know how to use it.

    Features:
    - Web file downloading
    - CGI support
    - Multi-thread, Multi-instance
    - Cross-Platform (Windows, Linux, other Embedded Linux)
    - Filter & Authentication support
    - Cookie support
    2. Getting Started

    First you have to decide which platform your application is running on. For windows you have to take  ‘ boahttpd.lib/.dll’  into your project, plus the header file ‘boahttpd.h’.  For linux(x86),  a release built under Fedora14 is provided , which will work correctly under  those systems similar  (I mean the kernel version ) to Fedora 14.

    For embedded systems, such as linux (arm), your have to provide the specific toolchains and ask for a specific build for your target platform.

    2.1 A ‘helloworld’ Application
    A most simple usage of the library would be like this: (see sample code `ex1`)

    #include <stdio.h>
    #include <stdlib.h>

    /* include boahttpd library */
    #include "../boahttpd.h"
    #pragma comment(lib, "../boahttpd.lib")

    /* start a simple web server */
    int main()
    {
     boa_server_t server;
     if(boa_server_create(&server, NULL,
    "wwwroot",
    8888, "0.0.0.0",
    NULL, NULL) < 0)
     {
      printf("failed to create server! ");
      return 0;
     }

     getchar();
     boa_server_destroy(&server);
     return 0;
    }

    The code above  sets up a http server running on port 8888, and the root directory of web pages is ‘wwwroot’ folder which should be located on the current directory.  So to make things right, your have to put some files in your ‘wwwroot’ folder, for example, a ‘index.html’ as the default home page.

    Now run the program,  then try the following url in your web browser:
         http://127.0.0.1:8888
    or
         http://127.0.0.1:8888/image/abc.jpg
    if there is a picture file located in that sub-directory: wwwroot/image/abc.jpg.


    2.2 Easy CGI Application
    Now we get to the key feature of the library: CGI support.  It also easy to know and to use, see  the  following snippet : (see sample code `ex2`)

    … … …
    static int cgi_handler(void* context, void* userdata)
    {
     … handle the cgi request …
     return 0;
    }

    int main()
    {
     boa_cgi_t cgi ;
     boa_cgi_create(&cgi);
     boa_cgi_register(&cgi, "test", &cgi_handler, NULL);

     boa_server_t server;
     if(boa_server_create(&server, &cgi,
    "wwwroot", 8888, "0.0.0.0", NULL, NULL) < 0)
     {
      printf("failed to create server! ");
      return 0;
     }

     getchar();

     boa_cgi_destroy(&cgi);
     boa_server_destroy(&server);
     return 0;
    }


    By register a callback function , cgi_handler, you could have the chance to handle the CGI request. A number of such callbacks can be register to the single boa_cgi_t object .

    You can register a default handler for all request , whose name is `*`:
     boa_cgi_register(&m_cgi, "*", &cgi_handler, this); 

    (Note: each CGI request is called in a NEW thread, thus the callback handler is called in that NEW thread)

    Take the follow request as an example.  When you type
       http://127.0.0.1:8888/cgi/mytest?name=shaofa&height=175
    in your browser, your httpd server will got a CGI request with:

    Name Value Meaning
    method GET The HTTP method, ‘GET’ or ‘POST’
    service mytest The name of the CGI service
    argument name=shaofa&height=175 The argument passed with the URL
    content  Vaild if the method is ‘POST’
    HTTP FIELD  Any other HTTP head field
    Socket   The socket handle

       The following snippet shows how to retrieve  those values:

    static int cgi_handler(void* context, void* userdata)
    {
     const char* service = boa_cgi_name(context);
     const char* method = boa_cgi_method(context);
     const char* argument = boa_cgi_args(context);
     const char* content = boa_cgi_content(context);

     const char* anyfield = boa_cgi_field(context,”Content-Length”);
          int sock = boa_cgi_socket(context);
    }

    If you want to reply an error message , just call:
         boa_cgi_send_error ( context, status, reason);
    If you want to reply an 200 OK message with a text content, just call:
         boa_cgi_send_text(context, “……”, CONTENT_TYPE_TXT);

    Keep in mind that the framework has already created a thread for each CGI request,  you need not to create a thread yourself.
    2.3 Advanced CGI Application
    Now we are talking about how to use the API in C++ manner. Support we want to create a http server object,  the sample code would be like below. (see sample code `complex`)

    ////////////////  .h /////////////////////  
    class MyHttpd 

    public: 
        int Open(int port, const char* ip); 
        void Close(); 
        int HandleCgiRequest(void* context); 
    private:     
        boa_cgi_t    m_cgi; 
        boa_server_t  m_server; 

     
    ///////////////////////////// *.cpp /////////////////////////////  
    static int cgi_handler(void* context, void* userdata) 

        MyHttpd*  server = (MyHttpd*) userdata;                    
        server->HandleCgiRequest(context);  
        return 0; 

    int MyHttpd::Open(int port, const char* ip) 

    boa_cgi_create(&cgi); 
    // set userdata to `this`  
    boa_cgi_register(&cgi, "test", &cgi_handler, this); 
    boa_server_create(&server,&cgi,  "wwwroot",
    8888, "0.0.0.0", NULL, NULL) ;    
        return 0; 

    int int MyHttpd::HandleCgiRequest(void* context) 
    {  
        const char* service = boa_cgi_name(context); 
        ... 

    3. Form Request
    This section is about how to deal with FORM request. For clear demonstration, a demo page is built like this:

    When user clicked OK button after filling the Name and URL fields,  a POST request is triggered (here jquery is used).

     function onAddStream()
     {
      var request = $.ajax({
        type: "POST",
        url: "cgi/addStream",
        data: {
          name: $("#strName").val() ,
        url: $("#strUrl").val()
       }
      });
      
      request.done(function(reply) {
          var errorCode = reply.errorCode;
       if(errorCode == 0)
       {
       }
       
      });
     }

    The .ajax call send a POST request the boahttpd server, with a content string whose format is like “field1=value1&field2=value2&…”.  So in the cgi handler you have to parse the content and extract the fields. The sample code parses the field, and make a reply text with JSON format.

     const char* service = boa_cgi_name(context); 
     const char* content = boa_cgi_content(context);

     if(strcmp(service, "addStream") == 0)
     {
      string name, url;

      // parse request:
      Property props(content, "&", "=");
      props.Get("name", name);
      props.Get("url", url);

      // make JSON reply
      json::Object obj;
      obj["errorCode"] = 0;
      obj["name"] = name;
      obj["url"] = url;

      // reply
      std::string replytext = json::Serialize(obj);
       boa_cgi_send_text(context, replytext.c_str(),
    "application/json");
     } 

    4. Filter & Authentication
    The  filter  mechanism  is designed to support Session / Authentication , which is used to prevent illegal user access. After filter handler is setup for a boa_server_t,  it will be called on each request before other handler (cgi handler or other inner handler).
    4.1 Filter Handler
    Firstly, define a handler,
    static int filter_handler(void* context, void* userdata) 

     MyHttpd*  server = (MyHttpd*) userdata;                    
     server->OnFilter(context);  
     return 0; 

    Then register it to the server,
     boa_server_create(&m_server ,
      &m_cgi, 
      "wwwroot",
      8888, "0.0.0.0",
      filter_handler, this) ;

    You have full privileges to decide if the request should be rejected (returns non-zero) or continue (returns zero). For example,  you define all files located in /images/ are always accessible.
    int MyHttpd::OnFilter(void* context)
    {
     const char* filepath = boa_cgi_path(context);
     const char* uri = boa_cgi_uri(context);

     // Allow: ALL files under /images/ can be accessed
      if(strcmp(filepath, "/images/")==0)
       return 0;

    4.2 Authentication
    Here I will give a sample to illustrate how to setup a simple authentication.  Session based authentication is also be feasible with the sample approach.

    Setup A /cgi/login Service
    User opens a browser, comes to the login page, fills in the password, and presses submit. The service gets the requests and invokes the handler, which will in turn check the password, and, if correct, set a cookie.

       if(strcmp(m_passwd, passwd.c_str()) == 0)
    {
     boa_cgi_set_cookie(context, "Authentication", passwd.c_str());
     boa_cgi_send_redirect(context, "/index.html");
    }
           ( Alternatively, you could returns a Session ID rather than the password)

    Test In The Filter
       Henceforth, the browser will send a Cookie field containing the password in its HTTP request. In the filter handler, you check this Cookie field, and test the password.
     string cookie = boa_cgi_field(context, "Cookie");
     printf("--------- cookie: %s ", cookie.c_str());

     // GET authentication
     Property props(cookie.c_str(), "; ", "=");  
     string auth;
     props.Get("Authentication", auth);
     if(auth.compare(m_passwd) != 0)
     {
      // redirect to login page
      boa_cgi_send_redirect(context, "/login.html");
      return -1;
     }
    If no appropriate information (correct password or session id) is found in the cookie, the request will be redirected to the login page.

     

     

                                                             

     

  • 相关阅读:
    页面跳转
    vue项目流程
    前端框架发展史
    webpack构建react项目和发布流程
    React工作原理
    React项目中的registerServiceWorker作用?
    学习react总结
    浏览器的渲染:过程与原理
    浮动相关
    块级元素与内嵌元素
  • 原文地址:https://www.cnblogs.com/hrhguanli/p/3927581.html
Copyright © 2020-2023  润新知