介紹一個(gè)C語言實(shí)現(xiàn)的http下載器

????做OTA升級功能時(shí)，能直接拿到的往往只是升級包的鏈接，需要我們自己去下載，這時(shí)候就需要用到http下載器，下文介紹一個(gè)C語言實(shí)現(xiàn)的http下載器。

功能

1、支持chunked方式傳輸?shù)南螺d 2、被重定向時(shí)能下載重定向頁面 3、要實(shí)現(xiàn)的接口為int http_download(char *url, char *save_path)

思路

1、解析輸入的URL，分離出主機(jī)，端口號，文件路徑的信息 2、解析主機(jī)的DNS 3、填充http請求的頭部，給服務(wù)器發(fā)包 4、解析收到的http頭，提取狀態(tài)碼，Content-length, Transfer-Encoding等字段信息 （1）如果是普通的頭則進(jìn)行接下來的正常收包流程 （2）如果狀態(tài)碼為302，則從頭里提取出重定向地址，用新的地址重新開始下載動作 （3）如果傳送方式是chunked的，則進(jìn)行分段讀取數(shù)據(jù)并拼接 （4）如果是404或其他狀態(tài)碼則打印錯(cuò)誤信息

缺陷

太多錯(cuò)誤處理，讓代碼看起來不太舒服

其他

1、如何移植到?jīng)]有文件系統(tǒng)的系統(tǒng)中？ 修改sava_data接口里面的保存就好了 2、如何提高下載速度？

增大讀寫buffer緩沖區(qū)

改為多線程，使用Range字段分段讀取，最后再拼在一起

代碼

?

/************************************************************
Copyright (C), 2016, Leon, All Rights Reserved.
FileName: download.c
coding: UTF-8
Description: 實(shí)現(xiàn)簡單的http下載功能
Author: Leon
Version: 1.0
Date: 2016-12-2 1032
Function:


History:
         
 Leon


************************************************************/


#include 
#include 
#include 
#include 
#include 
#include 
#include 
#include 
#include 
#include 


#define HOST_NAME_LEN   256
#define URI_MAX_LEN     2048
#define RECV_BUF        8192
#define RCV_SND_TIMEOUT (10*1000)   //收發(fā)數(shù)據(jù)超時(shí)時(shí)間(ms)


typedef struct {
    int sock;                       //與服務(wù)器通信的socket
    FILE *in;                       //sock描述符轉(zhuǎn)為文件指針，方便讀寫
    char host_name[HOST_NAME_LEN];  //主機(jī)名
    int port;                       //主機(jī)端口號
    char uri[URI_MAX_LEN];          //資源路徑
    char buffer[RECV_BUF];          //讀寫緩沖
    int status_code;                //http狀態(tài)碼
    int chunked_flag;               //chunked傳輸?shù)臉?biāo)志位
    int len;                        //Content-length里的長度
    char location[URI_MAX_LEN];     //重定向地址
    char *save_path;                //保存內(nèi)容的路徑指針
    FILE *save_file;                //保存內(nèi)容的文件指針
    int recv_data_len;              //收到數(shù)據(jù)的總長度
    time_t start_recv_time;         //開始接受數(shù)據(jù)的時(shí)間
    time_t end_recv_time;           //結(jié)束接受數(shù)據(jù)的時(shí)間
} http_t;


/* 打印宏 */
#define MSG_DEBUG   0x01
#define MSG_INFO    0x02
#define MSG_ERROR   0x04


static int print_level = /*MSG_DEBUG |*/ MSG_INFO | MSG_ERROR;


#define lprintf(level, format, argv...) do{     
    if(level & print_level)     
        printf("[%s][%s(%d)]:"format, #level, __FUNCTION__, __LINE__, ##argv);  
}while(0)


#define MIN(x, y) ((x) > (y) ? (y) : (x))


#define HTTP_OK         200
#define HTTP_REDIRECT   302
#define HTTP_NOT_FOUND  404


/* 不區(qū)分大小寫的strstr */
char *strncasestr(char *str, char *sub)
{
    if(!str || !sub)
        return NULL;


    int len = strlen(sub);
    if (len == 0)
    {
        return NULL;
    }


    while (*str)
    {
        if (strncasecmp(str, sub, len) == 0)
        {
            return str;
        }
        ++str;
    }
    return NULL;
}


/* 解析URL, 成功返回0，失敗返回-1 */
/* http://127.0.0.1:8080/testfile */
int parser_URL(char *url, http_t *info)
{
    char *tmp = url, *start = NULL, *end = NULL;
    int len = 0;


    /* 跳過http:// */
    if(strncasestr(tmp, "http://"))
    {   
        tmp += strlen("http://");
    }
    start = tmp;
    if(!(tmp = strchr(start, '/')))
    {
        lprintf(MSG_ERROR, "url invaild
");
        return -1;      
    }
    end = tmp;


    /*解析端口號和主機(jī)*/
    info->port = 80;   //先附默認(rèn)值80


    len = MIN(end - start, HOST_NAME_LEN - 1);
    strncpy(info->host_name, start, len);
    info->host_name[len] = '?';


    if((tmp = strchr(start, ':')) && tmp < end)
    {
        info->port = atoi(tmp + 1);
        if(info->port <= 0 || info->port >= 65535)
        {
            lprintf(MSG_ERROR, "url port invaild
");
            return -1;
        }
        /* 覆蓋之前的賦值 */
        len = MIN(tmp - start, HOST_NAME_LEN - 1);
        strncpy(info->host_name, start, len);
        info->host_name[len] = '?';
    }


    /* 復(fù)制uri */
    start = end;
    strncpy(info->uri, start, URI_MAX_LEN - 1);


    lprintf(MSG_INFO, "parse url ok
host:%s, port:%d, uri:%s
", 
        info->host_name, info->port, info->uri);
    return 0;
}


/* dns解析,返回解析到的第一個(gè)地址，失敗返回-1，成功則返回相應(yīng)地址 */
unsigned long dns(char* host_name)
{


    struct hostent* host;
    struct in_addr addr;
    char **pp;


    host = gethostbyname(host_name);
    if (host == NULL)
    {
        lprintf(MSG_ERROR, "gethostbyname %s failed
", host_name);
        return -1;
    }


    pp = host->h_addr_list;


    if (*pp!=NULL)
    {
        addr.s_addr = *((unsigned int *)*pp);
        lprintf(MSG_INFO, "%s address is %s
", host_name, inet_ntoa(addr));
        pp++;
        return addr.s_addr;
    }


    return -1;
}


/* 設(shè)置發(fā)送接收超時(shí) */
int set_socket_option(int sock)
{
    struct timeval timeout;


    timeout.tv_sec = RCV_SND_TIMEOUT/1000;
    timeout.tv_usec = RCV_SND_TIMEOUT%1000*1000;
    lprintf(MSG_DEBUG, "%ds %dus
", (int)timeout.tv_sec, (int)timeout.tv_usec);
    //設(shè)置socket為非阻塞
    // fcntl(sock ,F_SETFL, O_NONBLOCK); //以非阻塞的方式，connect需要重新處理


    // 設(shè)置發(fā)送超時(shí)
    if(-1 == setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (char *)&timeout, 
            sizeof(struct timeval)))
    {
        lprintf(MSG_ERROR, "setsockopt error: %m
");
        return -1;
    }


    // 設(shè)置接送超時(shí)
    if(-1 == setsockopt(sock, SOL_SOCKET, SO_RCVTIMEO, (char *)&timeout, 
            sizeof(struct timeval)))
    {
        lprintf(MSG_ERROR, "setsockopt error: %m
");
        return -1;
    }


    return 0;
}


/* 連接到服務(wù)器 */
int connect_server(http_t *info)
{
    int sockfd;
    struct sockaddr_in server;
    unsigned long addr = 0;
    unsigned short port = info->port;


    sockfd = socket(AF_INET, SOCK_STREAM, 0);
    if (-1 == sockfd)
    {
        lprintf(MSG_ERROR, "socket create failed
");
        goto failed;
    }


    if(-1 == set_socket_option(sockfd))
    {
        goto failed;
    }


    if ((addr = dns(info->host_name)) == -1)
    {
        lprintf(MSG_ERROR, "Get Dns Failed
");
        goto failed;
    }
    memset(&server, 0, sizeof(server));
    server.sin_family = AF_INET; 
    server.sin_port = htons(port); 
    server.sin_addr.s_addr = addr;


    if (-1 == connect(sockfd, (struct sockaddr *)&server, sizeof(struct sockaddr)))
    {
        lprintf(MSG_ERROR, "connect failed: %m
");
        goto failed;
    }


    info->sock = sockfd;
    return 0;


failed:
    if(sockfd != -1)
        close(sockfd);
    return -1;
}


/* 發(fā)送http請求 */
int send_request(http_t *info)
{
    int len;


    memset(info->buffer, 0x0, RECV_BUF);
    snprintf(info->buffer, RECV_BUF - 1, "GET %s HTTP/1.1
"
        "Accept: */*
"
        "User-Agent: Mozilla/5.0 (compatible; MSIE 5.01; Windows NT 5.0)
"
        "Host: %s
"
        "Connection: Close

", info->uri, info->host_name);


    lprintf(MSG_DEBUG, "request:
%s
", info->buffer);
    return send(info->sock, info->buffer, strlen(info->buffer), 0);
}


/* 解析http頭 */
int parse_http_header(http_t *info)
{
    char *p = NULL;


    // 解析第一行
    fgets(info->buffer, RECV_BUF, info->in);
    p = strchr(info->buffer, ' ');
    //簡單檢查http頭第一行是否合法
    if(!p || !strcasestr(info->buffer, "HTTP"))
    {
        lprintf(MSG_ERROR, "bad http head
");
        return -1;
    }
    info->status_code = atoi(p + 1);   
    lprintf(MSG_DEBUG, "http status code: %d
", info->status_code);


    // 循環(huán)讀取解析http頭
    while(fgets(info->buffer, RECV_BUF, info->in))
    {
        // 判斷頭部是否讀完
        if(!strcmp(info->buffer, "
"))
        {
            return 0;   /* 頭解析正常 */
        }
        lprintf(MSG_DEBUG, "%s", info->buffer);
        // 解析長度 Content-length: 554
        if(p = strncasestr(info->buffer, "Content-length"))
        {
            p = strchr(p, ':');
            p += 2;     // 跳過冒號和后面的空格
            info->len = atoi(p);
            lprintf(MSG_INFO, "Content-length: %d
", info->len);
        }
        else if(p = strncasestr(info->buffer, "Transfer-Encoding"))
        {
            if(strncasestr(info->buffer, "chunked"))
            {
                info->chunked_flag = 1;
            }
            else
            {
                /* 不支持其他編碼的傳送方式 */
                lprintf(MSG_ERROR, "Not support %s", info->buffer);
                return -1;
            }
            lprintf(MSG_INFO, "%s", info->buffer);
        }
        else if(p = strncasestr(info->buffer, "Location"))
        {
            p = strchr(p, ':');
            p += 2;     // 跳過冒號和后面的空格
            strncpy(info->location, p, URI_MAX_LEN - 1);
            lprintf(MSG_INFO, "Location: %s
", info->location);
        }
    }
    lprintf(MSG_ERROR, "bad http head
");
    return -1;  /* 頭解析出錯(cuò) */
}


/* 保存服務(wù)器響應(yīng)的內(nèi)容 */
int save_data(http_t *info, const char *buf, int len)
{
    int total_len = len;
    int write_len = 0;


    // 文件沒有打開則先打開
    if(!info->save_file)
    {
        info->save_file = fopen(info->save_path, "w");
        if(!info->save_file)
        {
            lprintf(MSG_ERROR, "fopen %s error: %m
", info->save_path);
            return -1;
        }
    }


    while(total_len)
    {
        write_len = fwrite(buf, sizeof(char), len, info->save_file);
        if(write_len < len && errno != EINTR)
        {
            lprintf(MSG_ERROR, "fwrite error: %m
");
            return -1;
        }
        total_len -= write_len;
    }
}


/* 讀數(shù)據(jù) */
int read_data(http_t *info, int len)
{
    int total_len = len;
    int read_len = 0;
    int rtn_len = 0;


    while(total_len)
    {
        read_len = MIN(total_len, RECV_BUF);
        // lprintf(MSG_DEBUG, "need read len: %d
", read_len);
        rtn_len = fread(info->buffer, sizeof(char), read_len, info->in);
        if(rtn_len < read_len)
        {
            if(ferror(info->in))
            {
                if(errno == EINTR) /* 信號中斷了讀操作 */
                {
                    ;   /* 不做處理繼續(xù)往下走 */
                }
                else if(errno == EAGAIN || errno == EWOULDBLOCK) /* 超時(shí) */
                {
                    lprintf(MSG_ERROR, "socket recvice timeout: %dms
", RCV_SND_TIMEOUT);
                    total_len -= rtn_len;
                    lprintf(MSG_DEBUG, "read len: %d
", rtn_len);
                    break;
                }
                else    /* 其他錯(cuò)誤 */
                {
                    lprintf(MSG_ERROR, "fread error: %m
");
                    break;
                }
            }
            else    /* 讀到文件尾 */
            {
                lprintf(MSG_ERROR, "socket closed by peer
");
                total_len -= rtn_len;
                lprintf(MSG_DEBUG, "read len: %d
", rtn_len);
                break;
            }
        }


        // lprintf(MSG_DEBUG, " %s
", info->buffer);
        total_len -= rtn_len;
        lprintf(MSG_DEBUG, "read len: %d
", rtn_len);
        if(-1 == save_data(info, info->buffer, rtn_len))
        {
            return -1;
        }
        info->recv_data_len += rtn_len;
    }
    if(total_len != 0)
    {
        lprintf(MSG_ERROR, "we need to read %d bytes, but read %d bytes now
", 
            len, len - total_len);
        return -1;
    }
}


/* 接收服務(wù)器發(fā)回的chunked數(shù)據(jù) */
int recv_chunked_response(http_t *info)
{
    long part_len;


    //有chunked，content length就沒有了
    do{
        // 獲取這一個(gè)部分的長度
        fgets(info->buffer, RECV_BUF, info->in);
        part_len = strtol(info->buffer, NULL, 16);
        lprintf(MSG_DEBUG, "part len: %ld
", part_len);
        if(-1 == read_data(info, part_len))
            return -1;


        //讀走后面的
兩個(gè)字符
        if(2 != fread(info->buffer, sizeof(char), 2, info->in))
        {
            lprintf(MSG_ERROR, "fread \r\n error : %m
");
            return -1;
        }
    }while(part_len);
    return 0;
}


/* 計(jì)算平均下載速度，單位byte/s */
float calc_download_speed(http_t *info)
{
    int diff_time = 0; 
    float speed = 0.0;


    diff_time = info->end_recv_time - info->start_recv_time;
    /* 最小間隔1s，避免計(jì)算浮點(diǎn)數(shù)結(jié)果為inf */
    if(0 == diff_time)
        diff_time = 1;
    speed = (float)info->recv_data_len / diff_time;


    return  speed;
}


/* 接收服務(wù)器的響應(yīng)數(shù)據(jù) */
int recv_response(http_t *info)
{
    int len = 0, total_len = info->len;


    if(info->chunked_flag)
        return recv_chunked_response(info);


    if(-1 == read_data(info, total_len))
        return -1;


    return 0;
}


/* 清理操作 */
void clean_up(http_t *info)
{
    if(info->in)
        fclose(info->in);
    if(-1 != info->sock)
        close(info->sock);
    if(info->save_file)
        fclose(info->save_file);
    if(info)
        free(info);
}


/* 下載主函數(shù) */
int http_download(char *url, char *save_path)
{
    http_t *info = NULL;
    char tmp[URI_MAX_LEN] = {0};


    if(!url || !save_path)
        return -1;


    //初始化結(jié)構(gòu)體
    info = malloc(sizeof(http_t));
    if(!info)
    {
        lprintf(MSG_ERROR, "malloc failed
");
        return -1;
    }
    memset(info, 0x0, sizeof(http_t));
    info->sock = -1;
    info->save_path = save_path;


    // 解析url
    if(-1 == parser_URL(url, info))
        goto failed;


    // 連接到server
    if(-1 == connect_server(info))
        goto failed;


    // 發(fā)送http請求報(bào)文
    if(-1 == send_request(info))
        goto failed;


    // 接收響應(yīng)的頭信息
    info->in = fdopen(info->sock, "r");
    if(!info->in)
    {
        lprintf(MSG_ERROR, "fdopen error
");
        goto failed;
    }


    // 解析頭部
    if(-1 == parse_http_header(info))
        goto failed;


    switch(info->status_code)
    {
        case HTTP_OK:
            // 接收數(shù)據(jù)
            lprintf(MSG_DEBUG, "recv data now
");
            info->start_recv_time = time(0);
            if(-1 == recv_response(info))
                goto failed;


            info->end_recv_time = time(0);
            lprintf(MSG_INFO, "recv %d bytes
", info->recv_data_len);
            lprintf(MSG_INFO, "Average download speed: %.2fKB/s
", 
                    calc_download_speed(info)/1000);
            break;
        case HTTP_REDIRECT:
            // 重啟本函數(shù)
            lprintf(MSG_INFO, "redirect: %s
", info->location);
            strncpy(tmp, info->location, URI_MAX_LEN - 1);
            clean_up(info);
            return http_download(tmp, save_path);


        case HTTP_NOT_FOUND:
            // 退出
            lprintf(MSG_ERROR, "Page not found
");
            goto failed;
            break;


        default:
            lprintf(MSG_INFO, "Not supported http code %d
", info->status_code);
            goto failed;
    }


    clean_up(info);
    return 0;
failed:
    clean_up(info);
    return -1;
}


/****************************************************************************
測試用例:
(1)chunked接收測試
./a.out "http://www.httpwatch.com/httpgallery/chunked/chunkedimage.aspx" test.aspx
(2)重定向測試
./a.out "192.168.10.1/main.html" test.txt
(3)錯(cuò)誤輸入測試
./a.out "32131233" test.txt
(4)根目錄輸入測試
./a.out "www.baidu.com/" test.txt
(5)端口號訪問測試
./a.out "192.168.0.200:8000/FS_AC6V1.0BR_V15.03.4.12_multi_TD01.bin" test.txt
****************************************************************************/


int main(int argc, char *argv[])
{
    if(argc < 3)
        return -1;


    http_download(argv[1], argv[2]);
    return 0;
}

　　審核編輯：湯梓紅