More clear code to read from unicode console in Windows.

Avoid big stack/static buffers.

Author: hvlad

src/isql/isql.epp

@@ -304,16 +304,15 @@ static inline int fb_isdigit(const char c)
 	return isdigit((int)(UCHAR)c);
 }
 
+typedef Array<char> CharBuffer;
+GlobalPtr<CharBuffer> charBuffer;
 
 #ifdef WIN_NT
-// This function is highly based on code written by https://github.com/xenu
-// He permitted our usage here: https://github.com/Perl/perl5/pull/18702#issuecomment-1156050577
-static int win32ReadConsole(FILE* file, char* buffer, size_t bufferSize)
+
+GlobalPtr<Array<WCHAR>> wideBuffer;
+
+static int win32ReadConsole(FILE* file, CharBuffer& mbBuffer)
 {
-	// This function is a workaround for a bug in Windows:
-	// https://github.com/microsoft/terminal/issues/4551
-	// tl;dr: ReadFile() and ReadConsoleA() return garbage when reading
-	// non-ASCII characters from the console with the 65001 codepage.
 	auto handle = (HANDLE) _get_osfhandle(fileno(file));
 
 	if (handle == INVALID_HANDLE_VALUE)
@@ -322,107 +321,49 @@ static int win32ReadConsole(FILE* file, char* buffer, size_t bufferSize)
 		return -1;
 	}
 
-	DWORD mode;
-	if (!GetConsoleMode(handle, &mode))
+	DWORD oldMode, newMode;
+	if (GetConsoleMode(handle, &oldMode))
+	{
+		newMode = oldMode | (ENABLE_PROCESSED_INPUT | ENABLE_LINE_INPUT | ENABLE_ECHO_INPUT);
+		if (newMode != oldMode)
+			SetConsoleMode(handle, oldMode);
+	}
+	else
 	{
 		fb_assert(false);
-		return -1;
+		newMode = oldMode = 0;
 	}
 
-	const size_t MAX_LINE_LENGTH = 32765;
-	static WCHAR wideBuf[MAX_LINE_LENGTH +1];
+	Cleanup	consoleMode([&] {
+		if (oldMode != newMode)
+			SetConsoleMode(handle, oldMode);
+	});
 
-	WCHAR* pWide;
-	DWORD charsRead = 0;
-	size_t leftToRead = bufferSize;
+	const size_t MAX_LINE_LENGTH = MAX_USHORT;
+	WCHAR* wideBuf = wideBuffer->getBuffer(MAX_LINE_LENGTH, false);
 
-	while (leftToRead)
+	DWORD charsRead;
+	if (!ReadConsoleW(handle, wideBuf, MAX_LINE_LENGTH, &charsRead, NULL))
 	{
-		// The purpose of convertedBuf is to preserve partial UTF-8 (or of any
-		// other multibyte encoding) code points between read() calls. Since
-		// there's only one console, the buffer is global. It's needed because
-		// ReadConsoleW() returns a string of UTF-16 code units and its result,
-		// after conversion to the current console codepage, may not fit in the
-		// return buffer.
-		//
-		// The buffer's size is 8 because it will contain at most two UTF-8 code
-		// points.
-		static char convertedBuf[8];
-		static size_t convertedBufLen = 0;
-
-		if (convertedBufLen)
-		{
-			bool newline = false;
-			const size_t toWrite = MIN(convertedBufLen, leftToRead);
-
-			// Don't read anything if the *first* character is ^Z and
-			// ENABLE_PROCESSED_INPUT is enabled. On some versions of Windows,
-			// ReadFile() ignores ENABLE_PROCESSED_INPUT, but apparently it's a
-			// bug: https://github.com/microsoft/terminal/issues/4958
-			if (leftToRead == bufferSize && (mode & ENABLE_PROCESSED_INPUT) && convertedBuf[0] == 0x1A)
-				break;
-
-			// Are we returning a newline?
-			if (memchr(convertedBuf, '\n', toWrite) != 0)
-				newline = true;
-
-			memcpy(buffer, convertedBuf, toWrite);
-			buffer += toWrite;
-
-			// If there's anything left in convertedBuf, move it to the beginning of the buffer.
-			convertedBufLen -= toWrite;
-
-			if (convertedBufLen)
-				memmove(convertedBuf, convertedBuf + toWrite, convertedBufLen);
-
-			leftToRead -= toWrite;
-
-			// With ENABLE_LINE_INPUT enabled, we stop reading after the first
-			// newline, otherwise we stop reading after the first character.
-			if (!leftToRead || newline || (mode & ENABLE_LINE_INPUT) == 0)
-				break;
-		}
-
-		if (!charsRead)
-		{
-			if (!ReadConsoleW(handle, wideBuf, MAX_LINE_LENGTH, &charsRead, NULL))
-				return -1;
-
-			if (!charsRead)
-				break;
-
-			pWide = wideBuf;
-		}
-
-		DWORD wideBufLen = 1;
-		charsRead--;
-
-		if (IS_HIGH_SURROGATE(*pWide))
-		{
-			// High surrogate, read one more code unit.
-			if (charsRead)
-				charsRead--;
-			else
-			{
-				DWORD read2;
-				if (!ReadConsoleW(handle, pWide + 1, 1, &read2, NULL))
-					return -1;
-			}
+		fb_assert(false);
+		return -1;
+	}
 
-			++wideBufLen;
-		}
+	if (!charsRead)
+		return 0;
 
-		convertedBufLen = WideCharToMultiByte(GetConsoleCP(), 0, pWide, wideBufLen,
-			convertedBuf, sizeof(convertedBuf), NULL, NULL);
+	int mbLength = WideCharToMultiByte(GetConsoleCP(), 0, wideBuf, charsRead,
+		NULL, 0, NULL, NULL);
 
-		if (!convertedBufLen)
-			return -1;
+	if (!mbLength)
+		return -1;
 
-		pWide += wideBufLen;
-	}
+	mbLength = WideCharToMultiByte(GetConsoleCP(), 0, wideBuf, charsRead,
+		mbBuffer.getBuffer(mbLength, false), mbLength, NULL, NULL);
 
-	return bufferSize - leftToRead;
+	return mbLength;
 }
+
 #endif
 
 
@@ -1108,16 +1049,20 @@ static void readNextInputLine(const char* prompt)
 	do
 	{
 		// Read the line
-		char buffer[MAX_USHORT];
+		char* buffer;
 		int lineSize;
 
 #ifdef WIN_NT
 		if (!Input_file && isatty(fileno(Filelist->Ifp().indev_fpointer)))
-			lineSize = win32ReadConsole(Filelist->Ifp().indev_fpointer, buffer, sizeof(buffer));
+		{
+			lineSize = win32ReadConsole(Filelist->Ifp().indev_fpointer, charBuffer);
+			buffer = charBuffer->begin();
+		}
 		else
 #endif
 		{
-			if (fgets(buffer, sizeof(buffer), Filelist->Ifp().indev_fpointer) != NULL)
+			buffer = charBuffer->getBuffer(MAX_USHORT);
+			if (fgets(buffer, charBuffer->getCapacity(), Filelist->Ifp().indev_fpointer) != NULL)
 				lineSize = strlen(buffer);
 			else
 				lineSize = -1;