proxy2ch/stringEncodingConverter.c

#include <string.h>
#include <stdlib.h>
#if defined(__APPLE__)
#include <CoreFoundation/CoreFoundation.h>
#elif defined(_WIN32)
#include <windows.h>
#else
#include <iconv.h>
#include <errno.h>
#endif
#include "stringEncodingConverter.h"

char *convertShiftJISToUTF8(const char *in, size_t inLength)
{
#if defined(__APPLE__)
	CFStringRef inStr = CFStringCreateWithBytes(
		NULL,
		(const UInt8 *)in,
		inLength,
		kCFStringEncodingDOSJapanese,
		false
	);
	if (!inStr) return NULL;
	CFIndex outLength;
	CFRange range = CFRangeMake(0, CFStringGetLength(inStr));
	CFStringGetBytes(inStr, range, kCFStringEncodingUTF8, '?', false, NULL, 0, &outLength);
	char *out = calloc(outLength+1, 1);
	CFStringGetBytes(inStr, range, kCFStringEncodingUTF8, '?', false, (UInt8 *)out, outLength, NULL);
	CFRelease(inStr);
	return out;
#elif defined(_WIN32)
	int wideLength = MultiByteToWideChar(932, 0, in, inLength, NULL, 0);
	if (wideLength == 0) return NULL;
	LPWSTR inWideStr = (LPWSTR)malloc(wideLength * sizeof(WCHAR));
	wideLength = MultiByteToWideChar(932, 0, in, inLength, inWideStr, wideLength);
	int outLength = WideCharToMultiByte(CP_UTF8, 0, inWideStr, wideLength, NULL, 0, NULL, NULL);
	char *out = calloc(outLength+1, 1);
	WideCharToMultiByte(CP_UTF8, 0, inWideStr, wideLength, out, outLength, NULL, NULL);
	free(inWideStr);
	return out;
#else
	size_t inBytesLeft, outBytesLeft;
	inBytesLeft = inLength;
	outBytesLeft = inBytesLeft*3+1;
	char *out = (char *)malloc(outBytesLeft);
	char *out_p = out;
	iconv_t cd = iconv_open("UTF-8", "CP932");
	while (1) {
		int ret = iconv(cd, (char **)&in, &inBytesLeft, &out_p, &outBytesLeft);
		if (ret != -1) {
			break;
		}
		else if (errno == EILSEQ) {
			in += 2;
			inBytesLeft -= 2;
			*out_p++ = '?';
			outBytesLeft--;
		}
		else break;
	}
	iconv_close(cd);
	if (!out_p) {
		free(out);
		return NULL;
	}
	*out_p = 0;
	/* U+301C, U+2016, U+2212 should be converted to U+FF5E, U+2225, U+FF0D */
	out_p = out;
	while ((out_p = strstr(out_p, "\xe3\x80\x9c"))) {
		*out_p = 0xef;
		*(out_p+1) = 0xbd;
		*(out_p+2) = 0x9e;
	}
	out_p = out;
	while ((out_p = strstr(out_p, "\xe2\x80\x96"))) {
		*out_p = 0xe2;
		*(out_p+1) = 0x88;
		*(out_p+2) = 0xa5;
	}
	out_p = out;
	while ((out_p = strstr(out_p, "\xe2\x88\x92"))) {
		*out_p = 0xef;
		*(out_p+1) = 0xbc;
		*(out_p+2) = 0x8d;
	}
	return out;
#endif
}