windows_code_page = UnicodeEncoding.UNICODE_CODE_PAGE;
}
+ #region GetByteCount()
+
// Internal version of "GetByteCount" which can handle a rolling
// state between multiple calls to this method.
- private static int InternalGetByteCount (char[] chars, int index, int count, uint leftOver, bool flush)
+ private static int InternalGetByteCount (char[] chars, int index, int count, ref char leftOver, bool flush)
{
// Validate the parameters.
if (chars == null) {
throw new ArgumentOutOfRangeException ("count", _("ArgRange_Array"));
}
+ if (index == chars.Length) {
+ if (flush && leftOver != '\0') {
+ // Flush the left-over surrogate pair start.
+ leftOver = '\0';
+ return 3;
+ }
+ return 0;
+ }
+
+ unsafe {
+ fixed (char* cptr = chars) {
+ return InternalGetByteCount (cptr + index, count, ref leftOver, flush);
+ }
+ }
+ }
+
+
+ private unsafe static int InternalGetByteCount (char* chars, int count, ref char leftOver, bool flush)
+ {
+ int index = 0;
+
// Determine the lengths of all characters.
char ch;
int length = 0;
- uint pair = leftOver;
+ char pair = leftOver;
while (count > 0) {
ch = chars[index];
if (pair == 0) {
length += 2;
} else if (ch >= '\uD800' && ch <= '\uDBFF') {
// This is the start of a surrogate pair.
- pair = (uint)ch;
+ pair = ch;
} else {
length += 3;
}
} else if (ch >= '\uDC00' && ch <= '\uDFFF') {
- // We have a surrogate pair.
- length += 4;
- pair = 0;
+ if (pair != 0) {
+ // We have a surrogate pair.
+ length += 4;
+ pair = '\0';
+ } else {
+ // We have a surrogate tail without
+ // leading surrogate. In NET_2_0 it
+ // uses fallback. In NET_1_1 we output
+ // wrong surrogate.
+ length += 3;
+ pair = '\0';
+ }
} else {
// We have a surrogate start followed by a
// regular character. Technically, this is
// We write out the surrogate start and then
// re-visit the current character again.
length += 3;
- pair = 0;
+ pair = '\0';
continue;
}
++index;
--count;
}
- if (flush && pair != 0) {
- // Flush the left-over surrogate pair start.
- length += 3;
+ if (flush) {
+ if (pair != '\0')
+ // Flush the left-over surrogate pair start.
+ length += 3;
+ leftOver = '\0';
}
+ else
+ leftOver = pair;
// Return the final length to the caller.
return length;
// Get the number of bytes needed to encode a character buffer.
public override int GetByteCount (char[] chars, int index, int count)
{
- return InternalGetByteCount (chars, index, count, 0, true);
+ char dummy = '\0';
+ return InternalGetByteCount (chars, index, count, ref dummy, true);
}
// Convenience wrappers for "GetByteCount".
throw new ArgumentNullException ("s");
}
- // Determine the lengths of all characters.
- char ch;
- int index = 0;
- int count = s.Length;
- int length = 0;
- uint pair;
- while (count > 0) {
- ch = s[index++];
- if (ch < '\u0080') {
- ++length;
- } else if (ch < '\u0800') {
- length += 2;
- } else if (ch >= '\uD800' && ch <= '\uDBFF' && count > 1) {
- // This may be the start of a surrogate pair.
- pair = (uint)(s[index]);
- if (pair >= (uint)0xDC00 && pair <= (uint)0xDFFF) {
- length += 4;
- ++index;
- --count;
- } else {
- length += 3;
- }
- } else {
- length += 3;
+ unsafe {
+ fixed (char* cptr = s) {
+ char dummy = '\0';
+ return InternalGetByteCount (cptr, s.Length, ref dummy, true);
}
- --count;
}
-
- // Return the final length to the caller.
- return length;
}
+ #endregion
+
+ #region GetBytes()
+
// Internal version of "GetBytes" which can handle a rolling
// state between multiple calls to this method.
private static int InternalGetBytes (char[] chars, int charIndex,
int charCount, byte[] bytes,
- int byteIndex, ref uint leftOver,
+ int byteIndex, ref char leftOver,
bool flush)
{
// Validate the parameters.
throw new ArgumentOutOfRangeException ("byteIndex", _("ArgRange_Array"));
}
+ if (charIndex == chars.Length) {
+ if (flush && leftOver != 0) {
+ // Flush the left-over surrogate pair start.
+ bytes [byteIndex++] = 0xEF;
+ bytes [byteIndex++] = 0xBB;
+ bytes [byteIndex++] = 0xBF;
+ leftOver = '\0';
+ return 3;
+ }
+ return 0;
+ }
+
+ unsafe {
+ fixed (char* cptr = chars) {
+ fixed (byte *bptr = bytes) {
+ return InternalGetBytes (
+ cptr + charIndex, charCount,
+ bptr + byteIndex, bytes.Length - byteIndex,
+ ref leftOver, flush);
+ }
+ }
+ }
+ }
+
+ private unsafe static int InternalGetBytes (char* chars, int charCount,
+ byte* bytes, int byteCount,
+ ref char leftOver, bool flush)
+ {
+ int charIndex = 0;
+ int byteIndex = 0;
+
+ // Convert the characters into bytes.
// Convert the characters into bytes.
char ch;
- int length = bytes.Length;
- uint pair;
- uint left = leftOver;
+ int length = byteCount;
+ char pair = leftOver;
int posn = byteIndex;
+ int code = 0;
+
while (charCount > 0) {
// Fetch the next UTF-16 character pair value.
- ch = chars[charIndex++];
- --charCount;
- if (left == 0) {
- if (ch >= '\uD800' && ch <= '\uDBFF') {
- // This is the start of a surrogate pair.
- left = (uint)ch;
+ ch = chars [charIndex];
+ if (pair == '\0') {
+ if (ch < '\uD800' || ch >= '\uE000')
+ code = ch;
+ else if (ch < '\uDC00') {
+ // surrogate start
+ pair = ch;
+ ++charIndex;
+ --charCount;
+ continue;
+ } else { // ch <= '\uDFFF'
+ // We have a surrogate tail without leading
+ // surrogate. In NET_2_0 it uses fallback.
+ // In NET_1_1 we output wrong surrogate.
+ if ((posn + 3) > length) {
+ throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
+ }
+ bytes [posn++] = (byte) (0xE0 | (ch >> 12));
+ bytes [posn++] = (byte) (0x80 | ((ch >> 6) & 0x3F));
+ bytes [posn++] = (byte) (0x80 | (ch & 0x3F));
+ ++charIndex;
+ --charCount;
continue;
- } else {
- // This is a regular character.
- pair = (uint)ch;
}
- } else if (ch >= '\uDC00' && ch <= '\uDFFF') {
- // We have a surrogate pair.
- pair = ((left - (uint)0xD800) << 10) +
- (((uint)ch) - (uint)0xDC00) +
- (uint)0x10000;
- left = 0;
} else {
- // We have a surrogate start followed by a
- // regular character. Technically, this is
- // invalid, but we have to do something.
- // We write out the surrogate start and then
- // re-visit the current character again.
- pair = (uint)left;
- left = 0;
- --charIndex;
- ++charCount;
+ if ('\uDC00' <= ch && ch <= '\uDFFF')
+ code = 0x10000 + (int) ch - 0xDC00 +
+ (((int) pair - 0xD800) << 10);
+ else {
+ // We have a surrogate start followed by a
+ // regular character. Technically, this is
+ // invalid, but we have to do something.
+ // We write out the surrogate start and then
+ // re-visit the current character again.
+ if ((posn + 3) > length) {
+ throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
+ }
+ bytes [posn++] = (byte) (0xE0 | (pair >> 12));
+ bytes [posn++] = (byte) (0x80 | ((pair >> 6) & 0x3F));
+ bytes [posn++] = (byte) (0x80 | (pair & 0x3F));
+ pair = '\0';
+ continue;
+ }
+ pair = '\0';
}
+ ++charIndex;
+ --charCount;
// Encode the character pair value.
- if (pair < (uint)0x0080) {
- if (posn >= length) {
+ if (code < 0x0080) {
+ if (posn >= length)
throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
- }
- bytes[posn++] = (byte)pair;
- } else if (pair < (uint)0x0800) {
- if ((posn + 2) > length) {
+ bytes [posn++] = (byte)code;
+ } else if (code < 0x0800) {
+ if ((posn + 2) > length)
throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
- }
- bytes[posn++] = (byte)(0xC0 | (pair >> 6));
- bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
- } else if (pair < (uint)0x10000) {
- if ((posn + 3) > length) {
+ bytes [posn++] = (byte) (0xC0 | (code >> 6));
+ bytes [posn++] = (byte) (0x80 | (code & 0x3F));
+ } else if (code < 0x10000) {
+ if ((posn + 3) > length)
throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
- }
- bytes[posn++] = (byte)(0xE0 | (pair >> 12));
- bytes[posn++] = (byte)(0x80 | ((pair >> 6) & 0x3F));
- bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
+ bytes [posn++] = (byte) (0xE0 | (code >> 12));
+ bytes [posn++] = (byte) (0x80 | ((code >> 6) & 0x3F));
+ bytes [posn++] = (byte) (0x80 | (code & 0x3F));
} else {
- if ((posn + 4) > length) {
+ if ((posn + 4) > length)
throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
- }
- bytes[posn++] = (byte)(0xF0 | (pair >> 18));
- bytes[posn++] = (byte)(0x80 | ((pair >> 12) & 0x3F));
- bytes[posn++] = (byte)(0x80 | ((pair >> 6) & 0x3F));
- bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
+ bytes [posn++] = (byte) (0xF0 | (code >> 18));
+ bytes [posn++] = (byte) (0x80 | ((code >> 12) & 0x3F));
+ bytes [posn++] = (byte) (0x80 | ((code >> 6) & 0x3F));
+ bytes [posn++] = (byte) (0x80 | (code & 0x3F));
}
}
- if (flush && left != 0) {
- // Flush the left-over surrogate pair start.
- if ((posn + 3) > length) {
- throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
+
+ if (flush) {
+ if (pair != '\0') {
+ // Flush the left-over incomplete surrogate.
+ if ((posn + 3) > length) {
+ throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
+ }
+ bytes [posn++] = (byte) (0xE0 | (pair >> 12));
+ bytes [posn++] = (byte) (0x80 | ((pair >> 6) & 0x3F));
+ bytes [posn++] = (byte) (0x80 | (pair & 0x3F));
}
- bytes[posn++] = (byte)(0xE0 | (left >> 12));
- bytes[posn++] = (byte)(0x80 | ((left >> 6) & 0x3F));
- bytes[posn++] = (byte)(0x80 | (left & 0x3F));
- left = 0;
+ leftOver = '\0';
}
- leftOver = left;
// Return the final count to the caller.
return posn - byteIndex;
}
+ private unsafe int Fallback (byte* bytes, int byteCount, char lead, char tail)
+ {
+ throw new NotImplementedException ();
+ }
+
// Get the bytes that result from encoding a character buffer.
public override int GetBytes (char[] chars, int charIndex, int charCount,
byte[] bytes, int byteIndex)
{
- uint leftOver = 0;
+ char leftOver = '\0';
return InternalGetBytes (chars, charIndex, charCount, bytes, byteIndex, ref leftOver, true);
}
throw new ArgumentOutOfRangeException ("byteIndex", _("ArgRange_Array"));
}
- // Convert the characters into bytes.
- char ch;
- int length = bytes.Length;
- uint pair;
- int posn = byteIndex;
- while (charCount > 0) {
- // Fetch the next UTF-16 character pair value.
- ch = s[charIndex++];
- if (ch >= '\uD800' && ch <= '\uDBFF' && charCount > 1) {
- // This may be the start of a surrogate pair.
- pair = (uint)(s[charIndex]);
- if (pair >= (uint)0xDC00 && pair <= (uint)0xDFFF) {
- pair = (pair - (uint)0xDC00) +
- ((((uint)ch) - (uint)0xD800) << 10) +
- (uint)0x10000;
- ++charIndex;
- --charCount;
- } else {
- pair = (uint)ch;
- }
- } else {
- pair = (uint)ch;
- }
- --charCount;
+ if (charIndex == s.Length)
+ return 0;
- // Encode the character pair value.
- if (pair < (uint)0x0080) {
- if (posn >= length) {
- throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
- }
- bytes[posn++] = (byte)pair;
- } else if (pair < (uint)0x0800) {
- if ((posn + 2) > length) {
- throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
- }
- bytes[posn++] = (byte)(0xC0 | (pair >> 6));
- bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
- } else if (pair < (uint)0x10000) {
- if ((posn + 3) > length) {
- throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
- }
- bytes[posn++] = (byte)(0xE0 | (pair >> 12));
- bytes[posn++] = (byte)(0x80 | ((pair >> 6) & 0x3F));
- bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
- } else {
- if ((posn + 4) > length) {
- throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
+ unsafe {
+ fixed (char* cptr = s) {
+ fixed (byte *bptr = bytes) {
+ char dummy = '\0';
+ return InternalGetBytes (
+ cptr + charIndex, charCount,
+ bptr + byteIndex, bytes.Length - byteIndex,
+ ref dummy, true);
}
- bytes[posn++] = (byte)(0xF0 | (pair >> 18));
- bytes[posn++] = (byte)(0x80 | ((pair >> 12) & 0x3F));
- bytes[posn++] = (byte)(0x80 | ((pair >> 6) & 0x3F));
- bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
}
}
-
- // Return the final count to the caller.
- return posn - byteIndex;
}
+ #endregion
+
// Internal version of "GetCharCount" which can handle a rolling
// state between multiple calls to this method.
#if NET_2_0
throw new ArgumentOutOfRangeException ("count", _("ArgRange_Array"));
}
+ int length = 0;
+
+ if (leftOverCount == 0) {
+ int end = index + count;
+ for (; index < end; index++, count--) {
+ if (bytes [index] < 0x80)
+ length++;
+ else
+ break;
+ }
+ }
+
// Determine the number of characters that we have.
uint ch;
- int length = 0;
uint leftBits = leftOverBits;
uint leftSoFar = (leftOverCount & (uint)0x0F);
uint leftSize = ((leftOverCount >> 4) & (uint)0x0F);
if (charIndex == chars.Length)
return 0;
+ int posn = charIndex;
+
+ if (leftOverCount == 0) {
+ int end = byteIndex + byteCount;
+ for (; byteIndex < end; posn++, byteIndex++, byteCount--) {
+ if (bytes [byteIndex] < 0x80)
+ chars [posn] = (char) bytes [byteIndex];
+ else
+ break;
+ }
+ }
+
// Convert the bytes into the output buffer.
uint ch;
int length = chars.Length;
- int posn = charIndex;
uint leftBits = leftOverBits;
uint leftSoFar = (leftOverCount & (uint)0x0F);
uint leftSize = ((leftOverCount >> 4) & (uint)0x0F);
#else
if (throwOnInvalid)
throw new ArgumentException (_("Overlong"), leftBits.ToString ());
+#endif
+ }
+ else if ((leftBits & 0xF800) == 0xD800) {
+ // UTF-8 doesn't use surrogate characters
+#if NET_2_0
+ Fallback (provider, ref fallbackBuffer, bytes, byteIndex, chars, ref posn);
+#else
+ if (throwOnInvalid)
+ throw new ArgumentException (_("Arg_InvalidUTF8"), "bytes");
#endif
}
else {
private class UTF8Encoder : Encoder
{
private bool emitIdentifier;
- private uint leftOver;
+ private char leftOverForCount;
+ private char leftOverForConv;
// Constructor.
public UTF8Encoder (bool emitIdentifier)
{
this.emitIdentifier = emitIdentifier;
- leftOver = 0;
+ leftOverForCount = '\0';
+ leftOverForConv = '\0';
}
// Override inherited methods.
public override int GetByteCount (char[] chars, int index,
int count, bool flush)
{
- return InternalGetByteCount (chars, index, count, leftOver, flush);
+ return InternalGetByteCount (chars, index, count, ref leftOverForCount, flush);
}
public override int GetBytes (char[] chars, int charIndex,
- int charCount, byte[] bytes, int byteCount, bool flush)
+ int charCount, byte[] bytes, int byteIndex, bool flush)
{
int result;
- result = InternalGetBytes (chars, charIndex, charCount, bytes, byteCount, ref leftOver, flush);
+ result = InternalGetBytes (chars, charIndex, charCount, bytes, byteIndex, ref leftOverForConv, flush);
emitIdentifier = false;
return result;
}
+#if NET_2_0
+ public unsafe override int GetByteCount (char* chars, int count, bool flush)
+ {
+ return InternalGetByteCount (chars, count, ref leftOverForCount, flush);
+ }
+
+ public unsafe override int GetBytes (char* chars, int charCount,
+ byte* bytes, int byteCount, bool flush)
+ {
+ int result;
+ result = InternalGetBytes (chars, charCount, bytes, byteCount, ref leftOverForConv, flush);
+ emitIdentifier = false;
+ return result;
+ }
+#endif
+
} // class UTF8Encoder
}; // class UTF8Encoding