proxy2ch/stringEncodingConverter.c

#include <string.h>
#include <stdlib.h>
#include <stdio.h>
#if defined(__APPLE__)
#include <CoreFoundation/CoreFoundation.h>
#elif defined(_WIN32)
#include <windows.h>
#else
#include <iconv.h>
#include <errno.h>
#include <stdint.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
}

char *convertUTF8ToShiftJIS(const char *in, size_t inLength)
{
#if defined(__APPLE__)
	CFStringRef inStr = CFStringCreateWithBytes(
		NULL,
		(const UInt8 *)in,
		inLength,
		kCFStringEncodingUTF8,
		false
	);
	if (!inStr) return NULL;
	CFIndex outLength;
	CFRange range = CFRangeMake(0, CFStringGetLength(inStr));
	CFStringGetBytes(inStr, range, kCFStringEncodingDOSJapanese, '?', false, NULL, 0, &outLength);
	char *out = calloc(outLength+1, 1);
	CFStringGetBytes(inStr, range, kCFStringEncodingDOSJapanese, '?', false, (UInt8 *)out, outLength, NULL);
	CFRelease(inStr);
	return out;
#elif defined(_WIN32)
	int wideLength = MultiByteToWideChar(CP_UTF8, 0, in, inLength, NULL, 0);
	if (wideLength == 0) return NULL;
	LPWSTR inWideStr = (LPWSTR)malloc(wideLength * sizeof(WCHAR));
	wideLength = MultiByteToWideChar(CP_UTF8, 0, in, inLength, inWideStr, wideLength);
	int outLength = WideCharToMultiByte(932, 0, inWideStr, wideLength, NULL, 0, NULL, NULL);
	char *out = calloc(outLength+1, 1);
	WideCharToMultiByte(932, 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("CP932", "UTF-8");
	while (1) {
		int ret = iconv(cd, (char **)&in, &inBytesLeft, &out_p, &outBytesLeft);
		if (ret != -1) {
			break;
		}
		else if (errno == EILSEQ) {
			in++;
			inBytesLeft--;
			*out_p++ = '?';
			outBytesLeft--;
		}
		else break;
	}
	iconv_close(cd);
	if (!out_p) {
		free(out);
		return NULL;
	}
	*out_p = 0;
	return out;
#endif
}

char *convertUTF8ToShiftJISWithNCR(const char *in, size_t inLength)
{
#if defined(__APPLE__)
	CFStringRef inStr = CFStringCreateWithBytes(
		NULL,
		(const UInt8 *)in,
		inLength,
		kCFStringEncodingUTF8,
		false
	);
	if (!inStr) return NULL;
	char *out = NULL;
	size_t bytesWritten = 0;
	CFIndex start = 0;
	CFIndex inputChars = CFStringGetLength(inStr);
	while (start < inputChars) {
		CFIndex outLength;
		CFRange range = CFRangeMake(start, inputChars - start);
		CFStringGetBytes(inStr, range, kCFStringEncodingDOSJapanese, 0, false, NULL, 0, &outLength);
		if (outLength) {
			out = realloc(out, bytesWritten+outLength+1);
			CFIndex converted = CFStringGetBytes(inStr, range, kCFStringEncodingDOSJapanese, 0, false, (UInt8 *)(out+bytesWritten), outLength, &outLength);
			if (converted) {
				start += converted;
				bytesWritten += outLength;
			}
			if (start >= inputChars) break;
		}
		unsigned int c1 = CFStringGetCharacterAtIndex(inStr, start++);
		if (c1 >= 0xd800 && c1 <= 0xdbff) {
			unsigned int c2 = CFStringGetCharacterAtIndex(inStr, start++);
			c1 = c1 - 0xd800;
			c2 = c2 - 0xdc00;
			c1 = 0x10000 + (c1 << 10) | c2;
		}
		out = realloc(out, bytesWritten+10+1);
		snprintf(out+bytesWritten, 11, "&#%u;", c1);
		bytesWritten += strlen(out+bytesWritten);
	}
	if (out) out[bytesWritten] = 0;
	CFRelease(inStr);
	return out;
#elif defined(_WIN32)
	int wideLength = MultiByteToWideChar(CP_UTF8, 0, in, inLength, NULL, 0);
	if (wideLength == 0) return NULL;
	LPWSTR inWideStr = (LPWSTR)malloc(wideLength * sizeof(WCHAR));
	wideLength = MultiByteToWideChar(CP_UTF8, 0, in, inLength, inWideStr, wideLength);
	BOOL isLossy;
	char *out = NULL;
	int outLength = WideCharToMultiByte(932, 0, inWideStr, wideLength, NULL, 0, NULL, &isLossy);
	if (!isLossy) {
		out = calloc(outLength+1, 1);
		WideCharToMultiByte(932, 0, inWideStr, wideLength, out, outLength, NULL, NULL);
	} else {
		int bytesWritten = 0;
		int start = 0;
		while (start < wideLength) {
			int lengthToConvert;
			for (lengthToConvert = 1; start+lengthToConvert < wideLength; lengthToConvert++) {
				WideCharToMultiByte(932, WC_NO_BEST_FIT_CHARS, inWideStr+start, lengthToConvert, NULL, 0, NULL, &isLossy);
				if (isLossy) {
					lengthToConvert--;
					break;
				}
			}
			if (lengthToConvert) {
				int outLength = WideCharToMultiByte(932, WC_NO_BEST_FIT_CHARS, inWideStr+start, lengthToConvert, NULL, 0, NULL, NULL);
				out = realloc(out, bytesWritten+outLength+1);
				bytesWritten += WideCharToMultiByte(932, WC_NO_BEST_FIT_CHARS, inWideStr+start, lengthToConvert, out+bytesWritten, outLength, NULL, NULL);
				start += lengthToConvert;
			}
			if (start >= wideLength) break;
			unsigned int c1 = inWideStr[start++];
			if (c1 >= 0xd800 && c1 <= 0xdbff) {
				unsigned int c2 = inWideStr[start++];
				c1 = c1 - 0xd800;
				c2 = c2 - 0xdc00;
				c1 = 0x10000 + (c1 << 10) | c2;
			}
			if (c1 == 0x301c) { /* wave dash */
				out = realloc(out, bytesWritten+3);
				strcpy(out+bytesWritten, "\x81\x60");
				bytesWritten += 2;
			} else {
				out = realloc(out, bytesWritten+10+1);
				snprintf(out+bytesWritten, 11, "&#%u;", c1);
				bytesWritten += strlen(out+bytesWritten);
			}
		}
		if (out) out[bytesWritten] = 0;
	}
	free(inWideStr);
	return out;
#else
	size_t inBytesLeft, outBytesLeft;
	inBytesLeft = inLength;
	outBytesLeft = inBytesLeft*4;
	uint32_t *utf32 = (uint32_t *)malloc(outBytesLeft);
	uint32_t *utf32_p = utf32;
	const char *codeset = "UCS-4-INTERNAL";
	iconv_t cd = iconv_open(codeset, "UTF-8");
	if (cd == (iconv_t)-1) {
		int x = 1;
		if (*(char *)&x) {
			codeset = "UCS-4LE";
		} else {
			codeset = "UCS-4BE";
		}
		cd = iconv_open(codeset, "UTF-8");
	}
	while (1) {
		int ret = iconv(cd, (char **)&in, &inBytesLeft, (char **)&utf32_p, &outBytesLeft);
		if (ret != -1) {
			break;
		}
		else if (errno == EILSEQ) {
			in++;
			inBytesLeft--;
		}
		else break;
	}
	iconv_close(cd);
	if (!utf32_p) {
		free(utf32);
		return NULL;
	}
	size_t numUnicodeWords = utf32_p - utf32;
	inBytesLeft = numUnicodeWords * 4;
	outBytesLeft = inBytesLeft*3 + 1; /* 4-byte utf-32 char may turn into 10-byte "&#xxxxxxx;" sequence - 3x multiplier is safe */
	char *out = (char *)malloc(outBytesLeft);
	char *out_p = out;
	int numInvalidConversions = 0;
	utf32_p = utf32;
	cd = iconv_open("CP932", codeset);
	while (1) {
		int ret = iconv(cd, (char **)&utf32_p, &inBytesLeft, &out_p, &outBytesLeft);
		if (ret != -1) {
			if (ret > 0) numInvalidConversions = ret;
			break;
		}
		else if (errno == EILSEQ) {
			if (*utf32_p == 0x301c) { /* wave dash */
				memcpy(out_p, "\x81\x60", 2);
				outBytesLeft -= 2;
				out_p += 2;
			} else {
				snprintf(out_p, outBytesLeft, "&#%u;", *utf32_p);
				outBytesLeft -= strlen(out_p);
				out_p += strlen(out_p);
			}
			utf32_p++;
			inBytesLeft -= 4;
		}
		else break;
	}
	if (numInvalidConversions) {
		/* fallback to per character conversion mode - FreeBSD/NetBSD iconv? */
		outBytesLeft = numUnicodeWords * 4 * 3 + 1;
		out_p = out;
		utf32_p = utf32;
		for (size_t wordsLeft = numUnicodeWords; wordsLeft > 0; wordsLeft--) {
			inBytesLeft = 4;
			int ret = iconv(cd, (char **)&utf32_p, &inBytesLeft, &out_p, &outBytesLeft);
			if (ret != -1 || (ret == -1 && errno == EILSEQ)) {
				if (ret != 0) {
					if (ret == -1) {
						utf32_p++;
						inBytesLeft -= 4;
					} else {
						outBytesLeft += 2;
						out_p -= 2;
					}
					if (*(utf32_p-1) == 0x301c) { /* wave dash */
						memcpy(out_p, "\x81\x60", 2);
						outBytesLeft -= 2;
						out_p += 2;
					} else {
						snprintf(out_p, outBytesLeft, "&#%u;", *(utf32_p-1));
						outBytesLeft -= strlen(out_p);
						out_p += strlen(out_p);
					}
				}
				continue;
			}
			else break;
		}
	}
	iconv_close(cd);
	free(utf32);
	if (!out_p) {
		free(out);
		return NULL;
	}
	*out_p = 0;
	return out;
#endif
}