From 3b73a03497e4fd67459960318308c9265bcd7805 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Vicent=20Mart=C3=AD?= <tanoku@gmail.com>
Date: Wed, 25 Apr 2012 16:26:12 -0700
Subject: UTF-8 changes yo

---
 src/win32/utf-conv.c | 92 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 92 insertions(+)

diff --git a/src/win32/utf-conv.c b/src/win32/utf-conv.c
index 0a705c0a..4b95001d 100644
--- a/src/win32/utf-conv.c
+++ b/src/win32/utf-conv.c
@@ -29,6 +29,98 @@ void gitwin_set_utf8(void)
 	_active_codepage = CP_UTF8;
 }
 
+#define U16_LEAD(c) (wchar_t)(((c)>>10)+0xd7c0)
+#define U16_TRAIL(c) (wchar_t)(((c)&0x3ff)|0xdc00)
+
+void git__utf8_to_16(wchar_t *dest, const char *src, size_t srcLength)
+{
+	wchar_t *pDest = dest;
+	uint32_t ch;
+	const uint8_t* pSrc = (uint8_t*) src;
+	const uint8_t *pSrcLimit = pSrc + srcLength;
+
+	assert(dest && src && srcLength > 0);
+
+	if ((pSrcLimit - pSrc) >= 4) {
+		pSrcLimit -= 3; /* temporarily reduce pSrcLimit */
+
+		/* in this loop, we can always access at least 4 bytes, up to pSrc+3 */
+		do {
+			ch = *pSrc++;
+			if(ch < 0xc0) {
+				/*
+				 * ASCII, or a trail byte in lead position which is treated like
+				 * a single-byte sequence for better character boundary
+				 * resynchronization after illegal sequences.
+				 */
+				*pDest++=(wchar_t)ch;
+			} else if(ch < 0xe0) { /* U+0080..U+07FF */
+				/* 0x3080 = (0xc0 << 6) + 0x80 */
+				*pDest++ = (wchar_t)((ch << 6) + *pSrc++ - 0x3080);
+			} else if(ch < 0xf0) { /* U+0800..U+FFFF */
+				/* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */
+				/* 0x2080 = (0x80 << 6) + 0x80 */
+				ch = (ch << 12) + (*pSrc++ << 6);
+				*pDest++ = (wchar_t)(ch + *pSrc++ - 0x2080);
+			} else /* f0..f4 */ { /* U+10000..U+10FFFF */
+				/* 0x3c82080 = (0xf0 << 18) + (0x80 << 12) + (0x80 << 6) + 0x80 */
+				ch = (ch << 18) + (*pSrc++ << 12);
+				ch += *pSrc++ << 6;
+				ch += *pSrc++ - 0x3c82080;
+				*(pDest++) = U16_LEAD(ch);
+				*(pDest++) = U16_TRAIL(ch);
+			}
+		} while(pSrc < pSrcLimit);
+
+		pSrcLimit += 3; /* restore original pSrcLimit */
+	}
+
+	while(pSrc < pSrcLimit) {
+		ch = *pSrc++;
+		if(ch < 0xc0) {
+			/*
+			 * ASCII, or a trail byte in lead position which is treated like
+			 * a single-byte sequence for better character boundary
+			 * resynchronization after illegal sequences.
+			 */
+			*pDest++=(wchar_t)ch;
+			continue;
+		} else if(ch < 0xe0) { /* U+0080..U+07FF */
+			if(pSrc < pSrcLimit) {
+				/* 0x3080 = (0xc0 << 6) + 0x80 */
+				*pDest++ = (wchar_t)((ch << 6) + *pSrc++ - 0x3080);
+				continue;
+			}
+		} else if(ch < 0xf0) { /* U+0800..U+FFFF */
+			if((pSrcLimit - pSrc) >= 2) {
+				/* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */
+				/* 0x2080 = (0x80 << 6) + 0x80 */
+				ch = (ch << 12) + (*pSrc++ << 6);
+				*pDest++ = (wchar_t)(ch + *pSrc++ - 0x2080);
+				pSrc += 3;
+				continue;
+			}
+		} else /* f0..f4 */ { /* U+10000..U+10FFFF */
+			if((pSrcLimit - pSrc) >= 3) {
+				/* 0x3c82080 = (0xf0 << 18) + (0x80 << 12) + (0x80 << 6) + 0x80 */
+				ch = (ch << 18) + (*pSrc++ << 12);
+				ch += *pSrc++ << 6;
+				ch += *pSrc++ - 0x3c82080;
+				*(pDest++) = U16_LEAD(ch);
+				*(pDest++) = U16_TRAIL(ch);
+				pSrc += 4;
+				continue;
+			}
+		}
+
+		/* truncated character at the end */
+		*pDest++ = 0xfffd;
+		break;
+	}
+
+	*pDest++ = 0x0;
+}
+
 wchar_t* gitwin_to_utf16(const char* str)
 {
 	wchar_t* ret;
-- 
cgit v1.2.1