我有一个用C语言编写的简单代理服务器程序。我遇到了一些数据不一致的问题。当我请求网站时，我将结果数据保存在服务器端的文件中，然后将其发送到客户端并将其保存在客户端。客户端和服务器上的结果将具有不同的大小，并且看起来某些HTML正在被复制。通常，保存在服务器上的文件将小于客户端上保存的文件，尽管两个文件仍然大于实际网页（即，如果我右键单击页面并“另存为”，则生成的页面小于我的代码返回的那些）。我已经尝试了各种方法来解决这个问题，似乎没有任何工作。结果甚至在同一网站的尝试之间似乎有所不同。例如，我可以两次请求同一个网站，但文件大小与两次尝试都不同。在非常罕见的情况下，特别是在小型网站上，客户端和服务器程序都会返回正确的网页，并且这两个文件都具有适当的大小。
注意：我知道代码仍然很混乱。在我继续前进之前，我更担心解决这个问题。我将在纠正此问题后解决问题（例如检查套接字是否无法打开），因此请仅解决我已概述的问题。
Server.c
#include <sys/types.h>
#include <sys/socket.h>
#include <netdb.h>
#include <stdio.h>
#include <string.h>
#include <arpa/inet.h>
#include <regex.h>
#include <time.h>
/*Code used to resolve URL into IP address adapted from following URL:
http://www.binarytides.com/hostname-to-ip-address-c-sockets-linux/
*/
//This code has been adapted from the server code provided in class
int main(int argc, char** argv)
{
char str[655360];
char recvline[655360];
char parsedRecv[655360];
char domain[1025];
char directory[1025];
char absoluteURL[1025];
char temp[1025];
int listen_fd, conn_fd, n, tempCount;
struct sockaddr_in servaddr;
int bytesRead;
int stringCounter;
int port;
FILE *fp;
//Variables used for second socket and resolving of host
char ip[100];
int sockfd, secondSocketCount;
int len = sizeof(struct sockaddr);
struct sockaddr_in secondServaddr;
struct addrinfo *servinfo, *p;
struct sockaddr_in *h;
int rv;
char simpleMatch[10];
int flag = 0;
//End
//Used for HTTP GET request
char request[2049];
listen_fd = socket(AF_INET, SOCK_STREAM, 0);
bzero(&servaddr, sizeof(servaddr));
servaddr.sin_family = AF_INET;
servaddr.sin_addr.s_addr = htons(INADDR_ANY);
if(argc < 2)
{
printf("Error! Enter a port number to run this server on.\n\tEx: ./server 22000\n\r\0");
return 0;
}
else
{
port = atoi(argv[1]);
}
servaddr.sin_port = htons(port);
printf("\n");
printf("Awaiting connections...\n");
bind(listen_fd, (struct sockaddr*)&servaddr, sizeof(servaddr));
listen(listen_fd, 10);
//Once the server is listening, enter an infinite loop to keep listening
while(1)
{
conn_fd = accept(listen_fd, (struct sockaddr*) NULL, NULL);
bytesRead = read(conn_fd, recvline, sizeof(recvline));
if(bytesRead > 0) //data read
{
recvline[bytesRead] = '\0';
bzero(absoluteURL, 1025);
strcpy(absoluteURL, recvline);
//Extract host and page from recvline
//For loop used to check if URL begins with HTTP or HTTPS
for(stringCounter = 0; stringCounter < 5; stringCounter++)
{
simpleMatch[stringCounter] = tolower(recvline[stringCounter]);
}
simpleMatch[strlen(simpleMatch)] = '\0';
if(strcmp("http:", simpleMatch) == 0)
{
for(stringCounter = 7, tempCount = 0; stringCounter < strlen(recvline); stringCounter++, tempCount++)
{
temp[tempCount] = recvline[stringCounter];
}
temp[strlen(temp)] = '\0';
strcpy(recvline, temp);
}
else if(strcmp("https", simpleMatch) == 0)
{
for(stringCounter = 8, tempCount = 0; stringCounter < strlen(recvline); stringCounter++, tempCount++)
{
temp[tempCount] = recvline[stringCounter];
}
temp[strlen(temp)] = '\0';
strcpy(recvline, temp);
}
//printf("\n\nAfter stripping HTTP, we are left with: %s\n\n", recvline);
//Now that HTTP:// or HTTPS:// has been stripped, can parse for domain
for(stringCounter = 0, tempCount = 0; stringCounter < strlen(recvline); stringCounter++)
{
//moving domain into the domain string
if(flag == 0)
{
if(recvline[stringCounter] != '/')
{
domain[stringCounter] = recvline[stringCounter];
}
else
{
domain[stringCounter + 1] = '\0';
//directory[tempCount] = recvline[stringCounter];
flag = 1;
tempCount++;
}
}
else
{
directory[tempCount] = recvline[stringCounter];
tempCount++;
}
}
//printf("\n\nDirectory is: %s\n\n", directory);
//reset flag and append '\0' to directory and domain
flag = 0;
if(tempCount < 1025)
{
directory[tempCount] = '\0';
}
//directory[strlen(directory)] = '\0';
//domain[strlen(domain)] = '\0';
//Done extracting
//Resolve hostname to IP
if((rv = getaddrinfo(domain, NULL, NULL, &servinfo)) != 0)
{
printf("Error: an IP address cannot be resolved for %s\n", domain);
return 0;
//fprintf(stderr, "getaddrinfo: %s\n", gai_strerror(rv));
}
else
{
for(p = servinfo; p != NULL; p = p->ai_next)
{
h = (struct sockaddr_in *) p->ai_addr;
strcpy(ip, inet_ntoa(h->sin_addr));
}
freeaddrinfo (servinfo);
printf("%s resolved to: %s\n", domain, ip);
}
//End Resolve
//Now that the IP is resolved, open a socket and connect to the IP
//Open socket
sockfd = socket(AF_INET, SOCK_STREAM, 0);
bzero(&secondServaddr, sizeof(secondServaddr));
secondServaddr.sin_family = AF_INET;
secondServaddr.sin_port = htons(80);
inet_pton(AF_INET, ip, &(secondServaddr.sin_addr)); //IP taken from earlier resolution
connect(sockfd, (struct sockaddr*) &secondServaddr, sizeof(secondServaddr));
//socket is open, can create and send request, finally
bzero(request, 2049);
//sprintf(request, "GET %s HTTP/1.1\r\nHost: %s\r\n\r\n", directory, domain);
//sprintf(request, "GET %s HTTP/1.1\r\n\r\n", absoluteURL);
//sprintf(request, "GET %s HTTP/1.1\r\nHost: %s\r\n\r\n", absoluteURL, domain);
sprintf(request, "GET /%s HTTP/1.1\r\nHost: %s\r\n\r\n", directory, domain);
write(sockfd, request, strlen(request));
printf("\tAttempting to retrieve data: this may be slow.\n");
bzero(recvline, 655360);
bzero(parsedRecv, 655360);
//Old method used to retrieve data
//This was changed when I began to run into issues
/*while(1)
{
secondSocketCount = read(sockfd, parsedRecv, sizeof(parsedRecv));
if(secondSocketCount == -1)
{
printf("Error receiving data: server terminating.\n");
return 0;
}
else if(secondSocketCount == 0)
{
//no more data
break;
}
strcat(recvline, parsedRecv);
}*/
//This while loop is used to read in data (the response from the server)
bzero(str, 655360);
while(secondSocketCount = read(sockfd, recvline, sizeof(recvline)) > 0)
{
strcat(str, recvline);
}
//bzero(parsedRecv, 655360);
//recvline[strlen(recvline)] = '\0';
printf("\tData retrieved from main server.\n");
//This for loop finds the end of the HTTP header and copies everything after into parsedRecv
for(stringCounter = 0, tempCount = 0; stringCounter < strlen(str); stringCounter++)
{
//lazy if statement to find two \r\n in a row to mark the end of the header
if(str[stringCounter] == '\r' && str[stringCounter + 1] == '\n' && str[stringCounter + 2] == '\r' && str[stringCounter + 3] == '\n' && flag == 0)
{
flag = 1;
stringCounter += 3;
}
if(flag == 1)
{
parsedRecv[tempCount] = str[stringCounter];
tempCount++;
}
}
flag = 0;
parsedRecv[strlen(parsedRecv)] = '\0';
fp = fopen("ReturnedPageServer.html", "w");
if(fp != NULL)
{
fprintf(fp, "%s", parsedRecv);
//fprintf(fp, "%s", recvline);
}
fclose(fp);
printf("\tData saved to ReturnedPageServer.html\n");
}
//strcpy(str, "This is a test of the Hello World Broadcast System.\n");
bzero(str, 655360);
strcpy(str, parsedRecv);
write(conn_fd, str, strlen(str));
close(conn_fd);
printf("\tData sent to client.\n");
printf("Awaiting further connections...\n");
//strcpy(directory, "");
//strcpy(domain, "");
//strcpy(recvline, "");
bzero(directory, 1025);
bzero(domain, 1025);
bzero(temp, 1025);
bzero(recvline, 655360);
}
return 0;
}
Client.c
#include <sys/types.h>
#include <sys/socket.h>
#include <netdb.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
//Code adapted from client code provided in class
int main(int argc, char** argv)
{
int sockfd, n, port;
int len = sizeof(struct sockaddr);
char sendline[10000];
char recvline[655360];
struct sockaddr_in servaddr;
FILE *fp;
sockfd = socket(AF_INET, SOCK_STREAM, 0);
bzero(&servaddr, sizeof(servaddr));
servaddr.sin_family = AF_INET;
if(argc < 2)
{
printf("Error! Enter the port number for the server.\n\tEx: ./client 22000\n\r\0");
return 0;
}
else
{
port = atoi(argv[1]);
}
servaddr.sin_port = htons(port);
inet_pton(AF_INET, "129.120.151.94", &(servaddr.sin_addr)); //CSE01 IP
connect(sockfd, (struct sockaddr*) &servaddr, sizeof(servaddr));
printf("url: ");
scanf("%s", sendline);
//strcpy(sendline, "The server should display this text.\n\0");
//printf("\nLength of string: %d\n", strlen(sendline));
//printf("\t%s\n", sendline);
write(sockfd, sendline, strlen(sendline));
fp = fopen("ReturnedPageClient.html", "w");
bzero(recvline, 655360);
while(n = read(sockfd, recvline, sizeof(recvline)) > 0)
{
//printf("%s", recvline);
if(fp != NULL)
{
fprintf(fp, "%s", recvline);
}
else
{
printf("\tError saving file: client terminating.\n");
fclose(fp);
return 0;
}
}
fclose(fp);
printf("\tResponse received from proxy server.\n\tFile saved as \"ReturnedPageClient.html\"\n");
close(sockfd);
return 0;
}
答案 0 :(得分：2)
一个问题 - 不确定它的'问题' - 是否期望TCP是面向消息的。即你发送一个500字节的消息，并期望接收一个500字节的块。这不是TCP的工作原理。 TCP是面向流的，500字节发送可以接收为服务器上的250个2字节读取，或1 500字节或100,50,100,2,2,2,2,2,240。您必须在服务器上循环直到你收到所有'消息'。这将导致不一致的行为，特别是在本地工作但不在“真正的”网络上工作
这反过来又提出了一个问题，即你如何知道自己收到了一条完整的“信息”。您需要一些允许消息框架的更高级别协议（比如发送固定大小的长度，然后是正文）
答案 1 :(得分：0)
要使用套接字对客户端和服务器之间的通信进行排序，您需要某种协议，以便连接中的每一方都可以确定消息何时完成。通过TCP套接字传输的消息可以以与发送方写入的大小不同的块接收。
您不会在程序中考虑到这一点，您认为成功的read请求会返回发送方write请求所写的数据。这是不正确的：您应该继续从套接字读取并将读取的块存储到请求缓冲区中，直到有完整请求或套接字关闭。
一个简单的协议是一次写一行。使用\n个字符对请求和答案进行排序。这或多或少是HTTP，SMTP和POP协议的作用。
代码很多问题：
它有语法错误阻止编译。
您将非常大的缓冲区定义为自动变量，超过2MB，您可能会在某些系统上出现堆栈溢出。
由于您不传输'\0'终结符，因此读入缓冲区的数据应为空终止。您似乎意识到了这个问题，但您的方法parsedRecv[strlen(parsedRecv)] = '\0';没有做任何有用的事情。根据定义，strlen(parsedRecv)是'\0'字节的偏移量。如果字符串未正确地终止，则strlen将扫描超出缓冲区的末尾并调用未定义的行为。成功'\0'后，您应该手动设置recvline[n] = '\0';字节与read。
您尝试将字符串片段与strcmp匹配：这不是最有效的，因为您需要先将片段复制到单独的缓冲区并将其终止，以便strcmp可以比较那是一个完整的字符串。请使用memcmp代替指定的片段长度。