用 C 语言编写一个网络蜘蛛，55个错误！
用C-Free 编译，出现55个错误。还算少！求完善！/************关于本文档********************************************
*filename: 用 C 语言编写一个网络蜘蛛来搜索网上出现的电子邮件地址
*purpose: 一个邮址搜索程序的雏形
*wrote by: zhoulifa(zhoulifa@) 周立发(http://zhoulifa.)
Linux爱好者 Linux知识传播者 SOHO族 开发者 最擅长C语言
*date time:2006-08-31 21:00:00
*Note: 任何人可以任意复制代码并运用这些文档，当然包括你的商业用途
* 但请遵循GPL
*Hope:希望越来越多的人贡献自己的力量，为科学技术发展出力
*********************************************************************/
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <sys/mman.h>
#include <unistd.h>
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <netdb.h>
#include <errno.h>
#include <locale.h>
#define USERAGENT "Wget/1.10.2"
#define ACCEPT "*/*"
#define ACCEPTLANGUAGE "zh-cn,zh;q=0.5"
#define ACCEPTENCODING "gzip,deflate"
#define ACCEPTCHARSET "gb2312,utf-8;q=0.7,*;q=0.7"
#define KEEPALIVE "300"
#define CONNECTION "keep-alive"
#define CONTENTTYPE "application/x-www-form-urlencoded"
#define MAXFILENAME 14
#define DEBUG 1
typedef struct webnode {
char * host; /* 网页所在的主机 */
int port; /* 网络服务器所使用的端口 */
char * dir; /* 网页所在的目录 */
char * page; /* 网页文件名 */
char * file; /* 本地保存的文件名 */
char IsHandled; /* 是否处理过 */
struct webnode * brother; /* 兄弟节点链表指针 */
struct webnode * child; /* 子节点链表指针 */
} WEBNODE;
struct sockaddr_in server_addr;
int sockfd = 0, dsend = 0, totalsend = 0, nbytes = 0, reqn = 0, i = 0, j = 0, ret = 0;
struct hostent *host;
char request[409600] = "", buffer[1024] = "", httpheader[1024] = "";
int FileNumber = 0;
char e[2] = "@/";
WEBNODE * NodeHeader, * NodeTail, * NodeCurr;
char * mapped_mem;
int GetHost(char * , char ** , char ** , int * , char ** ); /**/
void AnalyzePage(WEBNODE *); /**/
void AddInitNode(char *, char *, int, char * ); /**/
void HandleInitNode(WEBNODE *); /**/
void DisplayNode(WEBNODE *); /**/
void HandOneNode(WEBNODE *); /**/
void DoneWithList(int); /**/
void DoOnce(); /**/
void ConnectWeb(void); /**/
void SendRequest(void); /**/
void ReceiveResponse(void); /**/
void GetEmail(char * ); /**/
void GetLink(char * ); /**/
void GetBeforePos(char * , char ** ); /**/
void GetAfterPos(char * , char ** ); /**/
void AddChildNode(WEBNODE * , char * ); /**/
void GetAfterPosWithSlash(char * , char ** ); /**/
void GetMemory(char ** , int ); /**/
int IsExistWeb(WEBNODE * , char * , char * , int , char * ); /**/
void Rstrchr(char * , int , char ** ); /**/
int GetLocalAgent(char * UserAgent, char * Accept, char * AcceptLanguage, char * AcceptEncoding, char * AcceptCharset, char * KeepAlive, char * Connection, char * ContentType); /**/
/**************************************************************
功能：设置 HTTP 协议头内容的一些固定值
***************************************************************/
int GetLocalAgent(char * UserAgent, char * Accept, char * AcceptLanguage, char * AcceptEncoding, char * AcceptCharset, char * KeepAlive, char * Connection, char * ContentType)
{
memcpy(UserAgent, USERAGENT, strlen(USERAGENT));
memcpy(Accept, ACCEPT, strlen(ACCEPT));
memcpy(AcceptLanguage, ACCEPTLANGUAGE, strlen(ACCEPTLANGUAGE));
memcpy(AcceptEncoding, ACCEPTENCODING, strlen(ACCEPTENCODING));
memcpy(AcceptCharset, ACCEPTCHARSET, strlen(ACCEPTCHARSET));
memcpy(KeepAlive, KEEPALIVE, strlen(KEEPALIVE));
memcpy(Connection, CONNECTION, strlen(CONNECTION));
memcpy(ContentType, CONTENTTYPE, strlen(CONTENTTYPE));
return 0;
}
/**************************************************************
功能：在字符串 s 里搜索 x 字符，并设置指针 d 指向该位置
***************************************************************/
void Rstrchr(char * s, int x, char ** d)
{
int len = strlen(s) - 1;
while(len >= 0) {
if(x == s[len]) {(*d) = s + len; return;}
len–;
}
(*d) = 0;
}
/**************************************************************
功能：连接一个网站服务器
***************************************************************/
void ConnectWeb(void) { /* connect to web server */
/* create a socket descriptor */
if((sockfd=socket(PF_INET,SOCK_STREAM,0))==-1)
{
fprintf(stderr,"\tSocket Error:%s\a\n",strerror(errno));
exit(1);
}
/* bind address */
bzero(&server_addr, sizeof(server_addr));
server_addr.sin_family = AF_INET;
server_addr.sin_port = htons(NodeCurr->port);
server_addr.sin_addr = *((struct in_addr *)host->h_addr);
/* connect to the server */
if(connect(sockfd, (struct sockaddr *)(&server_addr), sizeof(struct sockaddr)) == -1)
{
fprintf(stderr, "\tConnect Error:%s\a\n", strerror(errno));
exit(1);
}
}
/**************************************************************
功能：向网站发送 HTTP 请求
***************************************************************/
void SendRequest(void) { /* send my http-request to web server */
dsend = 0;totalsend = 0;
nbytes=strlen(request);
while(totalsend < nbytes) {
dsend = write(sockfd, request + totalsend, nbytes - totalsend);
if(dsend==-1) {fprintf(stderr, "\tsend error!%s\n", strerror(errno));exit(0);}
totalsend+=dsend;
fprintf(stdout, "\n\tRequest.%d %d bytes send OK!\n", reqn, totalsend);
}
}
/**************************************************************
功能：接收网站的 HTTP 返回
***************************************************************/
void ReceiveResponse(void) { /* get response from web server */
fd_set writefds;
struct timeval tival;
int retry = 0;
FILE * localfp = NULL;
i=0; j = 0;
__ReCeive:
FD_ZERO(&writefds);
tival.tv_sec = 10;
tival.tv_usec = 0;
if(sockfd > 0) FD_SET(sockfd, &writefds);
else {fprintf(stderr, "\n\tError, socket is negative!\n"); exit(0);}
ret = select(sockfd + 1, &writefds, NULL, NULL, &tival);
if(ret ==0 ) {
if(retry++ < 10) goto __ReCeive;
}
if(ret <= 0) {fprintf(stderr, "\n\tError while receiving!\n"); exit(0);}
if(FD_ISSET(sockfd, &writefds)) {
memset(buffer, 0, 1024);
memset(httpheader, 0, 1024);
if((localfp = fopen(NodeCurr->file, "w")) == NULL) {if(DEBUG) fprintf(stderr, "create file ‘%s’ error\n", NodeCurr->file); return;}
/* receive data from web server */
while((nbytes=read(sockfd,buffer,1))==1)
{
if(i < 4) { /* 获取 HTTP 消息头 */
if(buffer[0] == ‘\r’ || buffer[0] == ‘\n’) i++;
else i = 0;
memcpy(httpheader + j, buffer, 1); j++;
}
else { /* 获取 HTTP 消息体 */
fprintf(localfp, "%c", buffer[0]); /* print content on the screen */
//fprintf(stdout, "%c", buffer[0]); /* print content on the screen */
i++;
}
}
fclose(localfp);
}
}
/**************************************************************
功能：执行一次 HTTP 请求
***************************************************************/
void DoOnce() { /* send and receive */
ConnectWeb(); /* connect to the web server */
/* send a request */
SendRequest();
/* receive a response message from web server */
ReceiveResponse();
close(sockfd); /* because HTTP protocol do something one connection, so I can close it after receiving */
}
/**************************************************************
功能：执行 HTTP 请求
***************************************************************/
void DoneWithList(int flag) {
if(flag) fprintf(stdout, "\tRequest.%d is:\n%s", ++reqn, request);
DoOnce();
if(flag) fprintf(stdout, "\n\tThe following is the response header:\n%s", httpheader);
}
/**************************************************************
功能：从字符串 src 中分析出网站地址和端口，并得到文件和目录
***************************************************************/
int GetHost(char * src, char ** web, char ** file, int * port, char ** dir) {
char * pA, * pB, * pC;
int len;
*port = 0;
if(!(*src)) return -1;
pA = src;
if(!strncmp(pA, "http://", strlen("http://"))) pA = src+strlen("http://");
/* else if(!strncmp(pA, "https://", strlen("https://"))) pA = src+strlen("https://"); */
else return 1;
pB = strchr(pA, ‘/’);
if(pB) {
len = strlen(pA) - strlen(pB);
GetMemory(web, len);
memcpy((*web), pA, len);
if(*(pB+1)) {
Rstrchr(pB + 1, ‘/’, &pC);
if(pC) len = strlen(pB + 1) - strlen(pC);
else len = 0;
if(len > 0) {
GetMemory(dir, len);
memcpy((*dir), pB + 1, len);
if(pC + 1) {
len = strlen(pC + 1);
GetMemory(file, len);
memcpy((*file), pC + 1, len);
}
else {
len = 1;
GetMemory(file, len);
memcpy((*file), e, len);
}
}
else {
len = 1;
GetMemory(dir, len);
memcpy((*dir), e + 1, len);
len = strlen(pB + 1);
GetMemory(file, len);
memcpy((*file), pB + 1, len);
}
}
else {
len = 1;
GetMemory(dir, len);
memcpy((*dir), e + 1, len);
len = 1;
GetMemory(file, len);
memcpy((*file), e, len);
}
}
else {
len = strlen(pA);
GetMemory(web, len);
memcpy((*web), pA, strlen(pA));
len = 1;
GetMemory(dir, len);
memcpy((*dir), e + 1, len);
len = 1;
GetMemory(file, len);
memcpy((*file), e, len);
}
pA = strchr((*web), ‘:’);
if(pA) *port = atoi(pA + 1);
else *port = 80;
return 0;
}
/*********************************************************************
*filename: mailaddrsearch.c
*purpose: 用 C 语言编写一个网络蜘蛛来搜索网上出现的电子邮件地址
*tidied by: zhoulifa(zhoulifa@) 周立发(http://zhoulifa.)
Linux爱好者 Linux知识传播者 SOHO族 开发者 最擅长C语言
*date time:2006-08-31 21:00:00
*Note: 任何人可以任意复制代码并运用这些文档，当然包括你的商业用途
* 但请遵循GPL
*Thanks to: www. 广东省 Linux 公共服务技术支持中心
*********************************************************************/
int main(int argc, char ** argv)
{
int WebPort;
char * WebHost = 0, * PageAddress = 0, * WebDir = 0;
if(argc < 2) {if(DEBUG) fprintf(stdout, "Command error, you should input like this:\n\t%s WebPageAddress1 WebPageAddress2 WebPageAddress3 …", argv[0]); exit(0);}
NodeHeader = NodeTail = NodeCurr = 0;
//setlocale(LC_ALL, "zh_CN.gb2312");
for(i = 1; i < argc; i++) {
ret = GetHost(argv, &WebHost, &PageAddress, &WebPort, &WebDir); /* Get web page info */
if(ret) {if(DEBUG) fprintf(stdout, "GetHost error from ‘%s’\n", argv); exit(0);}
AddInitNode(WebHost, PageAddress, WebPort, WebDir); /* add this page to chain */
}
free(WebHost); free(PageAddress);free(WebDir);
if(DEBUG) {
fprintf(stdout, "\nDisplay.%5d:", FileNumber);
DisplayNode(NodeHeader); /* display every node */
}
HandleInitNode(NodeHeader); /* handle every page */
return 0;
}
