晓宇IT

晓宇IT

关注
发信
关注(28)粉丝(399)

C语言 HTTP 下载文件

       C语言 HTTP协议下载文件，实现断点续传，socket通讯，目前只支持ip和port方式连接，有兴趣的读者可完善域名方式。
      代码分为 http.c： 实现http协议下载文件 ，socket.c： 封装linux socket函数，移植时只需修改socket.c中的函数即可。
     希望对大家有帮助，本人亲测可用！
http.c

点击(此处)折叠或打开

  1. //http.c
  2. //作者：王振
  3. #include <stdio.h>
  4. #include <stdlib.h>
  5. #include <errno.h>
  6. #include <string.h>

  8. #include <sys/types.h>
  9. #include <sys/stat.h>
  10. #include <fcntl.h>

  12. #include "socket.h"
  13. #include "http.h"


  16. #define MAX_RECV_SIZE    1440//硬件单包最大的接收字节数
  17. char g_host[URL_LEN];
  18. char g_ip[URL_LEN+1];//ip/域名
  19. char g_port[5+1];

  21. char g_buf_send[4*1024];//发送数据暂存区
  22. char g_buf_recv[10*1024];//接收数据暂存区

  24. BreakPoint_ST g_break_point;

  26. /*
  27. 功能:判断断点有效性,现在校验url是否一致
  28. 参数:
  29. 返回:
  30. >0---------有效,已下载文件大小
  31. -1----------无效
  32. */
  33. int Get_Breakpoint_Available(BreakPoint_ST *breakpoint,char *url,char *file_crc)
  34. {
  35.     
  36.     //判断断点是否有效,后续加入文件校验码
  37.     if((memcmp(breakpoint->url,url,strlen(url))== 0)&&(breakpoint->recv_size== MAX_RECV_SIZE))
  38.         return breakpoint->download_size;
  39.     else
  40.     {
  41.         return -1;
  42.     }
  43.     
  44. }


  47. /*
  48. 功能:判断要下载文件是否存在断点
  49. 参数:
  50. filename---要下载的文件名
  51. file_crc----服务器返回下载文件的校验码
  52. 返回:
  53. 0---------无断点
  54. >0--------有断点,已下载文件大小
  55. */
  56. int Get_Breakpoint(char *url,char *filename,char *file_crc)
  57. {
  58.     char filename_bp[64];
  59.     int fd = -1;
  60.     int ret;
  61.     BreakPoint_ST break_point;
  62.     
  63.     //断点文件名 filename+bp
  64.     sprintf(filename_bp,"%s.bp",filename);

  66.     //检测是否存在filename断点文件
  67.     fd = open(filename_bp,O_RDONLY,S_IRUSR|S_IWUSR);
  68.     if(fd == -1)
  69.     {    
  70.         #ifdef DEBUG_HTTP
  71.         printf("no exsit %s\n",filename_bp);
  72.         #endif
  73.         return 0;
  74.     }

  76.     //存在断点
  77.     ret = read(fd,&break_point,sizeof(break_point));
  78.     if(ret != sizeof(break_point))
  79.     {
  80.         perror("ERR:Get_Breakpoint read");
  81.         exit(-1);
  82.     }

  84.     close(fd);

  86.     //判断断点是否有效
  87.     ret = Get_Breakpoint_Available(&break_point,url,file_crc);
  88.     if(ret > 0)
  89.         return ret;
  90.     else
  91.     {
  92.         
  93.         printf("%s not available\n",filename_bp);
  94.         remove(filename);
  95.         remove(filename_bp);
  96.         return 0;
  97.         
  98.     }
  99. }

  101. /*
  102. 功能:保存断点信息,文件名filename.bp
  103. 参数:
  104. filename---要下载的文件名
  105. file_crc----服务器返回下载文件的校验码
  106. 返回:
  107. 0---------成功
  108. >0--------有断点,已下载文件大小
  109. */
  110. int Save_Breakpoint(char *url,char *filename,int download_size,char *file_crc)
  111. {
  112.     int fd;
  113.     BreakPoint_ST breakpoint;
  114.     char filename_bp[128];//断点信息文件名，包含路径

  116.     sprintf(filename_bp,"%s.bp",filename);
  117.     /* 创建目的文件 */
  118.     if((fd=open(filename_bp,O_WRONLY|O_CREAT,S_IRUSR|S_IWUSR))==-1)
  119.     {
  120.         fprintf(stderr,"Open %s Error：%s\n",filename_bp,strerror(errno));
  121.         exit(1);
  122.     }
  123.     memset(&breakpoint,0x0,sizeof(breakpoint));
  124.     strcpy(breakpoint.url,url);
  125.     //strcpy(breakpoint.crc,file_crc);
  126.     strcpy(breakpoint.filename,filename);
  127.     breakpoint.download_size = download_size;
  128.     breakpoint.recv_size= MAX_RECV_SIZE;
  129.     
  130.     //xu tioa zheng wei fen ci xie ru
  131.     if(write(fd,&breakpoint,sizeof(breakpoint)) != sizeof(breakpoint))
  132.     {
  133.         perror("ERR:Save_Breakpoint");
  134.         exit(1);
  135.     }

  137.     close(fd);

  139.     return 0;



  143. }

  145. /*
  146. 功能:保存文件,追加写
  147. 参数:
  148. 返回:
  149. 0---------成功
  150. */

  152. int Save_File(char *filebuf,int filelength,char *filename)
  153. {
  154.     int fd;
  155.     /* 创建目的文件追加写 */
  156.     if((fd=open(filename,O_WRONLY|O_CREAT|O_APPEND,S_IRUSR|S_IWUSR))==-1)
  157.     {
  158.         fprintf(stderr,"Open %s Error：%s\n",filename,strerror(errno));
  159.         exit(1);
  160.     }
  161.     //xu tioa zheng wei fen ci xie ru
  162.     if(write(fd,filebuf,filelength) != filelength)
  163.     {
  164.         perror("ERR:Save_File");
  165.         exit(1);
  166.     }

  168.     close(fd);

  170.     return 0;


  173. }

  175.  
  176. int HTTP_GetResponseCode(void)
  177. {
  178.  
  179.  
  180. }

  182.  /*
  183. 功能:读取http返回的协议实体主体长度
  184. 参数:
  185. revbuf--------接收到的返回值
  186. 返回值:
  187. >=0---------内容(实体主体)的长度
  188. -1-----------数据返回错误
  189. */
  190. int HTTP_GetRecvLength(char *revbuf)
  191. {
  192.     char *p1 = NULL;
  193.     int HTTP_Body = 0;//内容体长度
  194.     int HTTP_Head = 0;//HTTP 协议头长度


  197.     HTTP_Body = HTTP_GetContentLength(revbuf);
  198.     if(HTTP_Body == -1)
  199.         return -1;

  201.     p1=strstr(revbuf,"\r\n\r\n");
  202.     if(p1==NULL)
  203.         return -1;
  204.     else
  205.     {
  206.         HTTP_Head = p1- revbuf +4;// 4是\r\n\r\n的长度
  207.         return HTTP_Body+HTTP_Head;
  208.     }


  211. }


  214. /*
  215. 功能:读取http返回的Content-Length长度
  216. 参数:
  217. revbuf--------接收到的数据
  218. 返回值:
  219. >=0---------Content-Length长度
  220. -1-----------数据返回错误
  221. */
  222. int HTTP_GetContentLength(char *revbuf)
  223. {
  224.     char *p1 = NULL, *p2 = NULL;
  225.     int HTTP_Body = 0;//内容体长度

  227.     p1 = strstr(revbuf,"Content-Length");
  228.     if(p1 == NULL)
  229.         return -1;
  230.     else
  231.     {
  232.         p2 = p1+strlen("Content-Length")+ 2;
  233.         HTTP_Body = atoi(p2);
  234.         return HTTP_Body;
  235.     }

  237. }

  239.  /*
  240.  功能:
  241.  参数:
  242.  sockfd--------接收到的返回值
  243.  返回值:
  244.  >0---------接收到长度
  245.  -1----------失败
  246.  =0---------服务端断开连接
  247.  注:内部接收缓冲10k
  248.  */

  250. int HTTP_Recv(int sockfd,char *buf_recv)
  251. {
  252.     int ret;
  253.     int recvlen=0;
  254.     int downloadlen = 0;
  255.     //int contentlen=0;
  256.     char buf_recv_tmp[10*1024+1];
  257.     
  258.     memset(buf_recv_tmp,0x0,sizeof(buf_recv_tmp));
  259.     while(1)
  260.     {
  261.         ret = Recv(sockfd,buf_recv_tmp+recvlen,sizeof(buf_recv_tmp)-1,0);

  263.         if(ret <= 0)//下载失败
  264.         {
  265.             perror("ERR:recv fail");
  266.             return ret;
  267.         }
  268.     
  269.     
  270.         if(recvlen == 0)
  271.         {
  272.             #ifdef DEBUG_HTTP_RECV
  273.             printf("recv len = %d\n", ret);
  274.              printf("recv = %s\n", buf_recv_tmp);
  275.             #endif
  276.             //获取需要下载长度;
  277.             downloadlen = HTTP_GetRecvLength(buf_recv_tmp);


  280.             #ifdef DEBUG_HTTP_RECV
  281.             printf("downloadlen = %d\n",downloadlen);
  282.             #endif
  283.         }

  285.         recvlen += ret;
  286.         #ifdef DEBUG_HTTP_RECV
  287.         printf("total recvlen = %d\n",recvlen);
  288.         #endif

  290.         if(downloadlen == recvlen)//下载完成
  291.             break;


  294.     }
  295.     memcpy(buf_recv,buf_recv_tmp,downloadlen);
  296.     return recvlen;

  298. }

  300. /*
  301. 功能:获取下载url中的文件名,最后一个/后的字符
  302. 参数:
  303. 返回值:
  304. 0-----------成功
  305. -1----------失败
  306. 注:内部接收缓冲10k
  307. */

  309. int HTTP_GetFileName(char *url,char *filename)
  310. {
  311.     //提取url中最后一个/后的内容
  312.     int len;
  313.     int i;

  315.     len = strlen(url);
  316.     for(i=len-1;i>0;i--)
  317.     {
  318.         if(url[i] == '/')
  319.             break;
  320.     }
  321.     if(i == 0)//下载地址错误
  322.     {
  323.         printf("url not contain '/'\n");
  324.         return -1;
  325.     }
  326.     else
  327.     {
  328.     
  329.         strcpy(filename,url+i+1);
  330.         #ifdef DEBUG_HTTP
  331.         printf("filename=%s\n",filename);
  332.         #endif
  333.         return 0;
  334.     }
  335. }

  337. /*
  338. 功能:获取下载url中的路径,第一个/后的字符
  339. 参数:
  340. 返回值:
  341. 0-----------成功
  342. -1----------失败
  343. 注:url ex "http://host:port/path"
  344. */
  345. int HTTP_GetPath(char *url,char *path)
  346. {
  347.     char *p;

  349.     p = strstr(url,"http://");
  350.     if(p == NULL)
  351.     {
  352.         p = strchr(url,'/');
  353.         if(p == NULL)
  354.             return -1;
  355.         else
  356.         {
  357.             strcpy(path,p);
  358.             return 0;
  359.         }
  360.     }
  361.     else
  362.     {
  363.         p = strchr(url+strlen("http://"),'/');
  364.         if(p == NULL)
  365.             return -1;
  366.         else
  367.         {
  368.             strcpy(path,p);
  369.             return 0;
  370.         }
  371.     }

  373. }
  374. /*
  375. 功能:获取下载url中的ip和port,ip支持域名,端口默认为80
  376. 参数:
  377. 返回值:
  378. 1-----------域名式
  379. 2-----------ip port式
  380. -1----------失败
  381. 注:url ex "http://host:port/path"
  382. */

  384. int HTTP_Get_IP_PORT(char *url,char *ip,char *port)
  385. {
  386.     char *p = NULL;
  387.     int offset = 0;
  388.     char DOMAIN_NAME[128];

  390.     p = strstr(url,"http://");
  391.     if(p == NULL)
  392.     {
  393.         offset = 0;
  394.     }
  395.     else
  396.     {
  397.         offset = strlen("http://");
  398.     }

  400.     p = strchr(url+offset,'/');
  401.     if(p == NULL)
  402.     {
  403.         printf("url:%s format error\n",url);
  404.         return -1;
  405.         
  406.     }
  407.     else
  408.     {

  410.         memset(DOMAIN_NAME,0x0,sizeof(DOMAIN_NAME));
  411.         memcpy(DOMAIN_NAME,url+offset,(p-url-offset));
  412.         p = strchr(DOMAIN_NAME,':');
  413.         if(p == NULL)
  414.         {
  415.             strcpy(ip,DOMAIN_NAME);
  416.             strcpy(port,"80");
  417.             //printf("ip %p,port %p\n",ip,port);
  418.             
  419.             #ifdef DEBUG_HTTP
  420.             printf("ip=%s,port=%s\n",ip,port);//debug info
  421.             #endif
  422.             return 1;

  424.         }
  425.         else
  426.         {    
  427.             *p = '\0';

  429.             strcpy(ip,DOMAIN_NAME);
  430.             strcpy(port,p+1);
  431.             
  432.             #ifdef DEBUG_HTTP
  433.             printf("ip=%s,port=%s\n",ip,port);//debug info
  434.             #endif
  435.             return 2;

  437.         }


  440.         return 0;
  441.     }
  442.     
  443. }
  444. void Package_Url_Get_File(char *path, char *range)
  445. {
  446.     char buf[64];
  447.     memset(g_buf_send,0x0,sizeof(g_buf_send));        
  448.     sprintf(g_buf_send, "GET %s",path);

  450.     
  451.     //HTTP/1.1\r\n 前面需要一个空格
  452.     strcat(g_buf_send," HTTP/1.1\r\n");
  453.     strcat(g_buf_send, "Host: ");
  454.     strcat(g_buf_send, g_host);
  455.     //strcat(g_buf_send, ":");
  456.     //strcat(g_buf_send, PORT);
  457.     
  458.     sprintf(buf, "\r\nRange: bytes=%s",range);
  459.     strcat(g_buf_send,buf);
  460.     strcat(g_buf_send, "\r\nKeep-Alive: 200");
  461.     strcat(g_buf_send,"\r\nConnection: Keep-Alive\r\n\r\n");
  462.     

  464. }

  466. int Package_Url_Get_FileSize(char *url)
  467. {
  468.     
  469.     memset(g_buf_send,0x0,sizeof(g_buf_send));        
  470.     sprintf(g_buf_send, "HEAD %s",url);

  472.         //HTTP/1.1\r\n 前面需要一个空格
  473.     strcat(g_buf_send," HTTP/1.1\r\n");
  474.     strcat(g_buf_send, "Host: ");
  475.     strcat(g_buf_send, g_host);
  476.     //strcat(g_buf_send, ":");
  477.     //strcat(g_buf_send, PORT);
  478.     strcat(g_buf_send,"\r\nConnection: Keep-Alive\r\n\r\n");

  480.     return 0;
  481. }


  484. int HTTP_GetFileSize(int sockfd,char *path)
  485. {
  486.     int ret = -1;
  487.     char buf_recv_tmp[10*1024+1];

  489.     Package_Url_Get_FileSize(path);
  490. #ifdef DEBUG_HTTP
  491.     printf("send = %s \n",g_buf_send);
  492. #endif

  494.     Send(sockfd, g_buf_send, strlen(g_buf_send), 0);

  496.     memset(buf_recv_tmp,0x0,sizeof(buf_recv_tmp));                                                
  497.     ret = Recv(sockfd,buf_recv_tmp,sizeof(buf_recv_tmp)-1,0);
  498. #ifdef DEBUG_HTTP
  499.     printf("recv len = %d\n", ret);
  500.     printf("recv = %s\n", buf_recv_tmp);
  501. #endif
  502.     if(ret <= 0)
  503.     {
  504.         perror("ERR:recv fail GetFileSize()");
  505.         return -1;

  507.     }
  508.     ret = HTTP_GetContentLength(buf_recv_tmp);
  509.     if(ret <= 0)
  510.         return -1;
  511.     else
  512.         return ret;


  515. }




  520. /*
  521. 功能:分段下载文件
  522. 参数:
  523. 返回值:
  524. >0----------已下载文件大小(不包含上次下载)
  525. -1----------失败
  526. */
  527. int HTTP_GetFile(int sockfd,char *path,int filelength,int download_size,char *filebuf)
  528. {
  529.     int count;
  530.     char range[32];
  531.     int i;
  532.     int j = 0;//成功下载次数
  533.     int ret = -1;
  534.     char *p = NULL;
  535.     int download_index;//下载开始索引

  537.     count = (filelength%MAX_RECV_SIZE)?(filelength/MAX_RECV_SIZE +1):(filelength/MAX_RECV_SIZE);

  539.     download_index = download_size/MAX_RECV_SIZE;

  541.     for(i=download_index;i<count;i++)
  542.     {
  543.         //if(i == 20)//测试断点
  544.             //break;


  547.         if((i == (count-1))&&(filelength%MAX_RECV_SIZE))
  548.             sprintf(range,"%d-%d",i*MAX_RECV_SIZE,filelength-1);
  549.         else
  550.             sprintf(range,"%d-%d",i*MAX_RECV_SIZE,(i+1)*MAX_RECV_SIZE-1);


  553.         Package_Url_Get_File(path,range);
  554.         #ifdef DEBUG_HTTP
  555.          printf("send = %s \n",g_buf_send);
  556.         #endif
  557.          Send(sockfd, g_buf_send, strlen(g_buf_send), 0);

  559.         /*需改为提取http 返回协议头和协议体总长,然后定长接收*/
  560.         memset(g_buf_recv,0x0,sizeof(g_buf_recv));                                             
  561.         ret = HTTP_Recv(sockfd,g_buf_recv);
  562.         if(ret < 0)
  563.             break;
  564.         if(ret == 0 )//服务端断开连接
  565.         {
  566.             sockfd = Socket_Connect(g_ip,g_port);
  567.              i--;
  568.             continue;
  569.         }
  570.         /*提取协议体数据,保存在filebuf中*/
  571.         p = strstr(g_buf_recv,"\r\n\r\n");
  572.         if(p == NULL)//jia ru duan dian baocun
  573.         {
  574.             printf("ERR:g_buf_recv not contain end flag\n");
  575.             break;
  576.         }
  577.          else
  578.          {
  579.              memcpy(filebuf+j*MAX_RECV_SIZE,p+4,MAX_RECV_SIZE);
  580.              j++;

  582.          }
  583.     }
  584.     if(i == count)
  585.         return (filelength-download_size);
  586.     else
  587.         return (i*MAX_RECV_SIZE-download_size);
  588. }

  590. /*
  591. 功能:HTTP下载文件
  592. 参数:
  593. 返回值:
  594. 0----------下载完成
  595. -1---------失败
  596. -2---------部分下载完成
  597. 注:保存文件到bin所在目录
  598. */
  599. int HTTP_DownloadFile(char *url,char *save_path)
  600. {
  601.     int ret;
  602.     int sockfd;
  603.     int filesize;
  604.     int download_size;
  605.     char filename[FILENAME_LEN+1];
  606.     char filename_bp[FILENAME_LEN+3+1];
  607.     char *filebuf;
  608.     char save_file_path[FILENAME_LEN+1];//保存下载文件的路径+文件名

  610.     char path[PATH_LEN+1];//url中的path

  612.     //提取ip和port或url(url 暂不实现,需要gethostbyname linux)
  613.     ret = HTTP_Get_IP_PORT(url,g_ip,g_port);
  614.     if(ret == -1)
  615.         return -1;
  616.     else
  617.     {
  618.         sprintf(g_host,"%s:%s",g_ip,g_port);
  619.     }
  620.     //提取下载文件名
  621.     ret = HTTP_GetFileName(url,filename);
  622.     if(ret == -1)
  623.         return -1;

  625.     ret = HTTP_GetPath(url,path);
  626.     if(ret == -1)
  627.         return -1;
  628.     //sleep(3);//debug info
  629.     //建立连接
  630.     sockfd = Socket_Connect(g_ip,g_port);

  632.     //获取下载文件总大小
  633.     filesize = HTTP_GetFileSize(sockfd,path);
  634.     if(filesize == -1)
  635.         return -1;
  636.     //#ifdef DEBUG_HTTP
  637.     printf("http need download size %d\n",filesize);
  638.     //#endif
  639.     //malloc分配存储文件空间
  640.     filebuf = (char *)malloc(filesize);
  641.     if(filebuf == NULL)
  642.     {
  643.         perror("malloc filebuf fail");
  644.         return -1;
  645.     }
  646.     else
  647.         memset(filebuf,0x0,filesize);

  649.     download_size = Get_Breakpoint(url,filename,NULL);
  650.     #ifdef DEBUG_HTTP
  651.     printf("breakpoint download_size=%d\n",download_size);//debug info
  652.     sleep(3);//debug info
  653.     #endif
  654.     //分段下载文件
  655.     ret = HTTP_GetFile(sockfd,path,filesize,download_size,filebuf);
  656.     Close(sockfd);
  657.     if(ret < 0)
  658.     {
  659.         free(filebuf);
  660.         return -1;
  661.     }
  662.     else
  663.     {

  665.         sprintf(save_file_path,"%s%s",save_path,filename);
  666.         
  667.         #ifdef DEBUG_HTTP
  668.         printf("save_path=%s\n",save_path);
  669.         printf("filename=%s\n",filename);
  670.         printf("save_file_path=%s\n",save_file_path);
  671.         printf("download_size = %d\n",ret);
  672.         #endif
  673.         Save_File(filebuf,ret,save_file_path);
  674.         free(filebuf);
  675.         if((ret+download_size) == filesize)//全部下载完成
  676.         {
  677.             sprintf(filename_bp,"%s.bp",filename);
  678.             remove(filename_bp);
  679.             
  680.      printf("download success\n");
  681.             return 0;
  682.         }
  683.         else//部分下载完成
  684.         {
  685.             printf("part download success\n");
  686.             //保存断点信息
  687.             Save_Breakpoint(url,save_file_path,ret+download_size,NULL);
  688.             return -2;
  689.         }
  690.     }

  692. }
socket.c 

点击(此处)折叠或打开

  1. //socket.c
  2. //作者：王振

  4. #include <stdio.h>
  5. #include <stdlib.h>
  6. #include <errno.h>
  7. #include <string.h>
  8. #include <sys/types.h>
  9. #include <netinet/in.h>
  10. #include <sys/socket.h>
  11. #include <sys/wait.h>

  13. #include "socket.h"

  15. int Connect(int fd, struct sockaddr *addr, socklen_t len)
  16. {
  17.     int result;
  18.     if ((result = connect(fd, addr, len)) == -1) {
  19.         perror("connect");
  20.         exit(EXIT_FAILURE);
  21.     }
  22.     return 0;
  23. }

  25. int Socket_Connect(char *ip,char *port)
  26. {
  27.     struct sockaddr_in addr;

  29.     int sockfd;
  30.     int len;

  32.     
  33.     addr.sin_family = AF_INET;
  34.     addr.sin_addr.s_addr = inet_addr(ip);//127.0.0.1为本机ip
  35.     addr.sin_port = htons(atoi(port));
  36.     len = sizeof(addr);
  37.         
  38.     sockfd = socket(AF_INET, SOCK_STREAM, 0);
  39.     Connect(sockfd, (struct sockaddr*)&addr, len);
  40.     return sockfd;

  42. }


  45. /*
  46. 功能:向socketfd发送数据,内部实现了循环发送len长度
  47. 参数:
  48. sockfd 是代表你与远程程序连接的套接字描述符。
  49. msg 是一个指针，指向你想发送的信息的地址。
  50. len 是你想发送信息的长度。
  51. flags 发送标记。一般都设为0
  52. 返回:
  53. 0-------- 成功
  54. 退出---失败
  55. 修改:
  56. 备注:
  57. 王振
  58. */
  59. int Send(int sockfd, char *sendbuf, int len, int flags)
  60. {
  61.     int sendlen = 0;
  62.     int ret = -1;

  64.     while(sendlen < len)
  65.     {
  66.         ret = send(sockfd, sendbuf+sendlen, len-sendlen, flags);
  67.         if(-1 == ret)
  68.         {
  69.             perror("send");
  70.             exit(EXIT_FAILURE);
  71.         }
  72.         else
  73.             sendlen += ret;

  75.     }

  77.     return 0;

  79. }

  81. int Close(int sockfd)
  82. {
  83.     return close(sockfd);
  84.     

  86. }
  87. int Recv(int sockfd, char *recvbuf, int len, int flags)
  88. {
  89.     int recv_len;

  91.     if ((recv_len = recv(sockfd, recvbuf, len, flags)) < 0)
  92.     {
  93.         perror("recv error");
  94.         exit(EXIT_FAILURE);
  95.     }

  97.     return recv_len;
  98. }

08-30 22:34
Powered by LMLPHP ©2024 晓宇IT 0.053294