【Unix/Linux编程实践】DIY简单web服务器

一.服务器的设计问题：DIY or 代理

1.DIY——服务器接收请求，自己处理工作——用于快速简单的任务；

2.代理——服务器接收请求，然后创建一个新进程来处理工作——用于慢速的更加复杂的任务。

二.web服务器功能

常用的3种用户操作：

1.列举目录信息；

2.cat文件；

3.运行程序

请求处理： 目录：显示目录列表
文件：显示内容
.cgi文件：运行
不存在：错误消息

三.设计web服务器

1.建立服务器

2.接收请求

3.读取请求

4.处理请求

5.发送应答

四.HTTP基本结构

1.客户发送请求

GET filename HTTP/version
可选参数
空行

2.服务器发送应答

HTTP/version status-code status-message
附加信息
空行
内容

五.编写Web服务器

为简单起见，我们的Web服务器做以下约定：

1.服务器只支持GET命令，只接收请求航，跳过其余参数，然后处理请求和发送应答，另外，我们忽略出错检查。

2.服务器为每一个请求创建一个新的进程来处理，子进程将请求分割成命令和参数。如果命令不是GET，服务器应答HTTP返回码表示未实现的命令。如果命令是GET，服务器将期望得到目录名，一个以.cgi结尾的可执行程序或文件名。如果没有该目录或指定的文件，服务器报错。如果存在，服务器决定执行的操作：ls，exec或cat。

代码：

/*
 *  socklib.c
 *
 *  This file contains functions used lots when writing internet
 *  client/server programs.  The two main functions here are:
 *
 *  make_server_socket( portnum )   returns a server socket
 *                  or -1 if error
 *      make_server_socket_q(portnum,backlog)
 *
 *  connect_to_server(char *hostname, int portnum)
 *                  returns a connected socket
 *                  or -1 if error
 */ 

#include    <stdio.h>
#include    <unistd.h>
#include    <sys/types.h>
#include    <sys/socket.h>
#include    <netinet/in.h>
#include    <netdb.h>
#include    <time.h>
#include    <strings.h>

#define   HOSTLEN  256
#define   BACKLOG  1

int make_server_socket_q(int , int );

int make_server_socket(int portnum)
{
    return make_server_socket_q(portnum, BACKLOG);
}
int make_server_socket_q(int portnum, int backlog)
{
    struct  sockaddr_in   saddr;   /* build our address here */
    struct  hostent     *hp;   /* this is part of our    */
    char    hostname[HOSTLEN];     /* address            */
    int sock_id;           /* the socket             */

    sock_id = socket(PF_INET, SOCK_STREAM, 0);  /* get a socket */
    if ( sock_id == -1 ) 
        return -1;

    /** build address and bind it to socket **/

    bzero((void *)&saddr, sizeof(saddr));   /* clear out struct     */
    gethostname(hostname, HOSTLEN);         /* where am I ?         */
    hp = gethostbyname(hostname);           /* get info about host  */
                                            /* fill in host part    */
    bcopy( (void *)hp->h_addr, (void *)&saddr.sin_addr, hp->h_length);
    saddr.sin_port = htons(portnum);        /* fill in socket port  */
    saddr.sin_family = AF_INET ;            /* fill in addr family  */
    if ( bind(sock_id, (struct sockaddr *)&saddr, sizeof(saddr)) != 0 )
           return -1;

    /** arrange for incoming calls **/

    if ( listen(sock_id, backlog) != 0 ) 
        return -1;
    return sock_id;
}

int connect_to_server(char *host, int portnum)
{
    int sock;
    struct sockaddr_in  servadd;        /* the number to call */
    struct hostent      *hp;            /* used to get number */

    /** Step 1: Get a socket **/

    sock = socket( AF_INET, SOCK_STREAM, 0 );    /* get a line   */
    if ( sock == -1 ) 
        return -1;

    /** Step 2: connect to server **/

    bzero( &servadd, sizeof(servadd) );     /* zero the address     */
    hp = gethostbyname( host );             /* lookup host's ip #   */
    if (hp == NULL) 
        return -1;
    bcopy(hp->h_addr, (struct sockaddr *)&servadd.sin_addr, hp->h_length);
    servadd.sin_port = htons(portnum);      /* fill in port number  */
    servadd.sin_family = AF_INET ;          /* fill in socket type  */

    if ( connect(sock,(struct sockaddr *)&servadd, sizeof(servadd)) !=0)
           return -1;

    return sock;
}

/* webserv.c - a minimal web server (version 0.2)
 *      usage: ws portnumber
 *   features: supports the GET command only
 *             runs in the current directory
 *             forks a new child to handle each request
 *             has MAJOR security holes, for demo purposes only
 *             has many other weaknesses, but is a good start
 *      build: cc webserv.c socklib.c -o webserv
 */
#include    <stdio.h>
#include    <sys/types.h>
#include    <sys/stat.h>
#include    <string.h>

main(int ac, char *av[])
{
    int     sock, fd;
    FILE    *fpin;
    char    request[BUFSIZ];

    if ( ac == 1 ){
        fprintf(stderr,"usage: ws portnum\n");
        exit(1);
    }
    sock = make_server_socket( atoi(av[1]) );
    if ( sock == -1 ) exit(2);

    /* main loop here */

    while(1){
        /* take a call and buffer it */
        fd = accept( sock, NULL, NULL );
        fpin = fdopen(fd, "r" );

        /* read request */
        fgets(request,BUFSIZ,fpin);
        printf("got a call: request = %s", request);
        read_til_crnl(fpin);

        /* do what client asks */
        process_rq(request, fd);
        fclose(fpin);
    }
}

/* ------------------------------------------------------ *
   read_til_crnl(FILE *)
   skip over all request info until a CRNL is seen
   ------------------------------------------------------ */

read_til_crnl(FILE *fp)
{
    char    buf[BUFSIZ];
    while( fgets(buf,BUFSIZ,fp) != NULL && strcmp(buf,"\r\n") != 0 )
        ;
}

/* ------------------------------------------------------ *
   process_rq( char *rq, int fd )
   do what the request asks for and write reply to fd 
   handles request in a new process
   rq is HTTP command:  GET /foo/bar.html HTTP/1.0
   ------------------------------------------------------ */

process_rq( char *rq, int fd )
{
    char    cmd[BUFSIZ], arg[BUFSIZ];

    /* create a new process and return if not the child */
    if ( fork() != 0 )
    {
        while (waitpid(-1, NULL, 0) > 0);   /*wait for all children*/
        return;
    }

    /*avoid zombie*/
    if ( fork() != 0)
        exit(0);

    strcpy(arg, "./");      /* precede args with ./ */
    if ( sscanf(rq, "%s%s", cmd, arg+2) != 2 )
        return;

    if ( strcmp(cmd,"GET") != 0 )
        cannot_do(fd);
    else if ( not_exist( arg ) )
        do_404(arg, fd );
    else if ( isadir( arg ) )
        do_ls( arg, fd );
    else if ( ends_in_cgi( arg ) )
        do_exec( arg, fd );
    else
        do_cat( arg, fd );
}

/* ------------------------------------------------------ *
   the reply header thing: all functions need one
   if content_type is NULL then don't send content type
   ------------------------------------------------------ */

header( FILE *fp, char *content_type )
{
    fprintf(fp, "HTTP/1.0 200 OK\r\n");
    if ( content_type )
        fprintf(fp, "Content-type: %s\r\n", content_type );
}

/* ------------------------------------------------------ *
   simple functions first:
        cannot_do(fd)       unimplemented HTTP command
    and do_404(item,fd)     no such object
   ------------------------------------------------------ */

cannot_do(int fd)
{
    FILE    *fp = fdopen(fd,"w");

    fprintf(fp, "HTTP/1.0 501 Not Implemented\r\n");
    fprintf(fp, "Content-type: text/plain\r\n");
    fprintf(fp, "\r\n");

    fprintf(fp, "That command is not yet implemented\r\n");
    fclose(fp);
}

do_404(char *item, int fd)
{
    FILE    *fp = fdopen(fd,"w");

    fprintf(fp, "HTTP/1.0 404 Not Found\r\n");
    fprintf(fp, "Content-type: text/plain\r\n");
    fprintf(fp, "\r\n");

    fprintf(fp, "The item you requested: %s\r\nis not found\r\n", 
            item);
    fclose(fp);
}

/* ------------------------------------------------------ *
   the directory listing section
   isadir() uses stat, not_exist() uses stat
   do_ls runs ls. It should not
   ------------------------------------------------------ */

isadir(char *f)
{
    struct stat info;
    return ( stat(f, &info) != -1 && S_ISDIR(info.st_mode) );
}

not_exist(char *f)
{
    struct stat info;
    return( stat(f,&info) == -1 );
}

do_ls(char *dir, int fd)
{
    FILE    *fp ;

    fp = fdopen(fd,"w");
    header(fp, "text/plain");
    fprintf(fp,"\r\n");
    fflush(fp);

    dup2(fd,1);
    dup2(fd,2);
    close(fd);
    execlp("ls","ls","-l",dir,NULL);
    perror(dir);
    exit(1);
}

/* ------------------------------------------------------ *
   the cgi stuff.  function to check extension and
   one to run the program.
   ------------------------------------------------------ */

char * file_type(char *f)
/* returns 'extension' of file */
{
    char    *cp;
    if ( (cp = strrchr(f, '.' )) != NULL )
        return cp+1;
    return "";
}

ends_in_cgi(char *f)
{
    return ( strcmp( file_type(f), "cgi" ) == 0 );
}

do_exec( char *prog, int fd )
{
    FILE    *fp ;

    fp = fdopen(fd,"w");
    header(fp, NULL);
    fflush(fp);
    dup2(fd, 1);
    dup2(fd, 2);
    close(fd);
    execl(prog,prog,NULL);
    perror(prog);
}
/* ------------------------------------------------------ *
   do_cat(filename,fd)
   sends back contents after a header
   ------------------------------------------------------ */

do_cat(char *f, int fd)
{
    char    *extension = file_type(f);
    char    *content = "text/plain";
    FILE    *fpsock, *fpfile;
    int c;

    if ( strcmp(extension,"html") == 0 )
        content = "text/html";
    else if ( strcmp(extension, "gif") == 0 )
        content = "image/gif";
    else if ( strcmp(extension, "jpg") == 0 )
        content = "image/jpeg";
    else if ( strcmp(extension, "jpeg") == 0 )
        content = "image/jpeg";

    fpsock = fdopen(fd, "w");
    fpfile = fopen( f , "r");
    if ( fpsock != NULL && fpfile != NULL )
    {
        header( fpsock, content );
        fprintf(fpsock, "\r\n");
        while( (c = getc(fpfile) ) != EOF )
            putc(c, fpsock);
        fclose(fpfile);
        fclose(fpsock);
    }
    exit(0);
}

六.运行web服务器

编译运行：

$ cc webserv.c socklib.c -o webserv
$ ./webserv 12345

访问文件：

我们将html文件放到webserv同一目录下，并且用http://yourhostname:12345/filename.html来打开它。

执行程序：

我们创建以下脚本（命名为hello.cgi，权限为755）：

#！/bin/sh
#hello.cgi
printf "Content-type:text/plain\n\nhello\n";

打开http://yourhostname:12345/hello.cgi即可执行它。

查看目录：

新建一个目录hello，访问http://yourhostname:12345/hello即可。

七.改进建议

1.缓存溢出保护；

2.CGI程序需要设计一些环境变量；

3.HTTP头部可以包含更多的信息；

……