encode.cpp

html文件编码转换与解析

/* encode.cpp -- code convert function
 * Designed by arthur, 2009-4-24
 * */

#include <iconv.h>
#include <string.h>
#include <errno.h>
#include <iostream>
#include <stdlib.h>
#include <fstream>

int bufferleft;

char inputbuffer[10240000];
char outputbuffer[10740000];

using namespace std;

char *code_knowned[] = { /*define you own codes here*/
"utf-8", "gbk", "gb2312", "gb18030" };

int code_convert(char *from_code, char *to_code, char *from, char *to) /*code from one to another*/
{
	static iconv_t cd;
	static char *tptr, *fptr;
	static size_t ileft, oleft, ret;

	tptr = fptr =NULL;

	if (from == NULL || to == NULL)
		return -1;

	cd = iconv_open((const char *)to_code, (const char *)from_code);
	if (cd == (iconv_t)-1) {
		return -1;
	}

	ileft = strlen(from);
	fptr = from;
	for (;;) {
		tptr = to;
		oleft = 10740000;

		ret = iconv(cd, &fptr, &ileft, &tptr, &oleft);
		if (ret != (size_t)-1) {
			break;
		}

		if (errno == EINVAL) {
			(void) memmove(from, fptr, ileft);
			(void) iconv(cd, &fptr, &ileft, &tptr, &oleft);
			(void) iconv_close(cd);
			bufferleft = oleft;
			return 1;
		} else if (errno == E2BIG) {
			continue;
		} else if (errno == EILSEQ) {
			(void) iconv(cd, &fptr, &ileft, &tptr, &oleft);
			(void) iconv_close(cd);
			bufferleft = oleft;
			return 1;
		} else if (errno == EBADF) {
			return -1;
		} else {
			return -1;
		}
	}

	(void) iconv(cd, &fptr, &ileft, &tptr, &oleft);
	bufferleft = oleft;
	cout<< bufferleft <<endl;
	//	cout<< oleft<<endl;
	(void) iconv_close(cd);
	return 0;
}

/*convert any code to the code you want*/
int code_convert_to(char *to_code, char *from, char *to) {
	static int i, ret, code_knowned_num;

	code_knowned_num = sizeof(code_knowned) / sizeof(code_knowned[0]);
	for (i = 0; i < code_knowned_num; i++) {
		ret = code_convert(code_knowned[i], to_code, from, to);
		if (ret == 0)
			return 0;
		else if (ret == 1)
			continue;
		else {
			return -1;
		}
	}
	return 0;
}

/*convert any code to unicode*/
char * c2utf8(char *from, char *to) {
	static int ret;

	ret = code_convert_to("utf-8", from, to);

	if (ret == -1) {
		return NULL;
	} else {
		return to;
	}
}

int main() {
	fstream fin;
	fstream fout;
	char infile[100];
	char outfile[100];
	cout<<"Please enter the input file name:"<<endl;
	cin >> infile;
	fin.open(infile, ios::in);
	cout<<"Please enter the output file name:"<<endl;
	cin >> outfile;
	fout.open(outfile, ios::out|ios::trunc);
	bzero(inputbuffer, sizeof(inputbuffer));
	fin.read(inputbuffer, sizeof(inputbuffer));
	c2utf8(inputbuffer, outputbuffer);
	fout.write(outputbuffer, sizeof(outputbuffer) - bufferleft);
	fin.close();
	fout.close();
	return 0;
}