pdf

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
 
#define ENDS_WITH(ch0, ch1) ((unsigned char)line[len - 3] == (ch1) && (unsigned char)line[len - 4] == (ch0))
 
/*
 * for gbk only.
 */
int main()
{
	FILE* fp = fopen("/mnt/hgfs/share/aaa1.txt", "r");
	FILE* fpOut = fopen("/mnt/hgfs/share/aaa2.txt", "w+");
	if (NULL == fp)
		return -1;
 
	char line[1024 * 10] = {0};
	while (NULL != fgets(line, sizeof(line) - 1, fp))
	{
		int len = strlen(line);
		if (0) // (len > 10)
		{
			printf("[%s]\n-1:%x\n-2:%x\n-3:%x\n-4:%x\n-5:%x\n-6:%x\n", 
					line,
					(unsigned int)(unsigned char)line[len - 1],
					(unsigned int)(unsigned char)line[len - 2],
					(unsigned int)(unsigned char)line[len - 3],
					(unsigned int)(unsigned char)line[len - 4],
					(unsigned int)(unsigned char)line[len - 5],
					(unsigned int)(unsigned char)line[len - 6]);
		}
		if (len >= 4)
		{
			do {
				if ((unsigned char)line[len - 1] != 0xa
						|| (unsigned char)line[len - 2] != 0xd)
					break;
				if (ENDS_WITH(0x80, 0x82))
				{
					// period
					break;
				}
 
				if (ENDS_WITH(0x80, 0x9d))
				{
					// double quotation mark
					break;
				}
				line[len - 2] = 0;
			} while (0);
		}
		fputs(line, fpOut);
	}
 
	fclose(fpOut);
	fclose(fp);
	return 1;
}