public void GetIndexOfFirstCharacterToEncodeUtf8_AllDataValid()
{
// Loop from 96 elements all the way down to 0 elements, which tests that we're
// not overrunning our read buffers.
var span = _boundedBytes.Span;
_boundedBytes.MakeWriteable();
span.Fill((byte)_allowedChar); // make buffer all-valid
_boundedBytes.MakeReadonly();
for (int i = 96; i >= 0; i--)
{
Assert.Equal(-1, _encoder.FindFirstCharacterToEncodeUtf8(span.Slice(span.Length - i)));
}
// Also check from the beginning of the buffer just in case there's some alignment weirdness
// in the SIMD-optimized code that causes us to read past where we should.
_boundedBytes.MakeWriteable();
for (int i = 96; i >= 0; i--)
{
span[i] = (byte)_disallowedChar; // make this char invalid
Assert.Equal(-1, _encoder.FindFirstCharacterToEncodeUtf8(span.Slice(0, i)));
}
}
public void IsSuffix(CompareInfo compareInfo, string source, string value, CompareOptions options, bool expected, int expectedMatchLength)
{
if (options == CompareOptions.None)
{
Assert.Equal(expected, compareInfo.IsSuffix(source, value));
}
Assert.Equal(expected, compareInfo.IsSuffix(source, value, options));
if ((compareInfo == s_invariantCompare) && ((options == CompareOptions.None) || (options == CompareOptions.IgnoreCase)))
{
StringComparison stringComparison = (options == CompareOptions.IgnoreCase) ? StringComparison.InvariantCultureIgnoreCase : StringComparison.InvariantCulture;
Assert.Equal(expected, source.EndsWith(value, stringComparison));
Assert.Equal(expected, source.AsSpan().EndsWith(value.AsSpan(), stringComparison));
}
// Now test the span version - use BoundedMemory to detect buffer overruns
using BoundedMemory <char> sourceBoundedMemory = BoundedMemory.AllocateFromExistingData <char>(source);
sourceBoundedMemory.MakeReadonly();
using BoundedMemory <char> valueBoundedMemory = BoundedMemory.AllocateFromExistingData <char>(value);
valueBoundedMemory.MakeReadonly();
Assert.Equal(expected, compareInfo.IsSuffix(sourceBoundedMemory.Span, valueBoundedMemory.Span, options));
Assert.Equal(expected, compareInfo.IsSuffix(sourceBoundedMemory.Span, valueBoundedMemory.Span, options, out int actualMatchLength));
Assert.Equal(expectedMatchLength, actualMatchLength);
}
static void RunSpanCompareTest(CompareInfo compareInfo, ReadOnlySpan <char> string1, ReadOnlySpan <char> string2, CompareOptions options, int expected)
{
using BoundedMemory <char> string1BoundedMemory = BoundedMemory.AllocateFromExistingData(string1);
string1BoundedMemory.MakeReadonly();
using BoundedMemory <char> string2BoundedMemory = BoundedMemory.AllocateFromExistingData(string2);
string2BoundedMemory.MakeReadonly();
Assert.Equal(expected, Math.Sign(compareInfo.Compare(string1, string2, options)));
Assert.Equal(-expected, Math.Sign(compareInfo.Compare(string2, string1, options)));
}
public static void WidenLatin1ToUtf16()
{
using BoundedMemory <byte> latin1Mem = BoundedMemory.Allocate <byte>(128);
using BoundedMemory <char> utf16Mem = BoundedMemory.Allocate <char>(128);
// Fill the source with [deterministic] pseudo-random bytes, then make readonly.
new Random(0x12345).NextBytes(latin1Mem.Span);
latin1Mem.MakeReadonly();
// We'll write to the UTF-16 span.
// We test with a variety of span lengths to test alignment and fallthrough code paths.
ReadOnlySpan <byte> latin1Span = latin1Mem.Span;
Span <char> utf16Span = utf16Mem.Span;
for (int i = 0; i < latin1Span.Length; i++)
{
utf16Span.Clear(); // remove any data from previous iteration
// First, transcode the data from Latin-1 to UTF-16.
CallWidenLatin1ToUtf16(latin1Span.Slice(i), utf16Span.Slice(i));
// Then, validate that the data was transcoded properly.
for (int j = i; j < 128; j++)
{
Assert.Equal((ushort)latin1Span[i], (ushort)utf16Span[i]);
}
}
// Now run the test with a bunch of sliding 48-byte windows.
// This tests that we handle correctly the scenario where neither the beginning nor the
// end of the buffer is properly vector-aligned.
const int WindowSize = 48;
for (int i = 0; i < latin1Span.Length - WindowSize; i++)
{
utf16Span.Clear(); // remove any data from previous iteration
// First, transcode the data from Latin-1 to UTF-16.
CallWidenLatin1ToUtf16(latin1Span.Slice(i, WindowSize), utf16Span.Slice(i, WindowSize));
// Then, validate that the data was transcoded properly.
for (int j = 0; j < WindowSize; j++)
{
Assert.Equal((ushort)latin1Span[i + j], (ushort)utf16Span[i + j]);
}
}
}
static void RunSpanIndexOfTest(CompareInfo compareInfo, ReadOnlySpan <char> source, ReadOnlySpan <char> value, CompareOptions options, int expected)
{
using BoundedMemory <char> sourceBoundedMemory = BoundedMemory.AllocateFromExistingData(source);
sourceBoundedMemory.MakeReadonly();
using BoundedMemory <char> valueBoundedMemory = BoundedMemory.AllocateFromExistingData(value);
valueBoundedMemory.MakeReadonly();
Assert.Equal(expected, compareInfo.IndexOf(sourceBoundedMemory.Span, valueBoundedMemory.Span, options));
if (TryCreateRuneFrom(value, out Rune rune))
{
Assert.Equal(expected, compareInfo.IndexOf(sourceBoundedMemory.Span, rune, options)); // try the Rune-based version
}
}
private static unsafe void GetIndexOfFirstInvalidUtf8Sequence_Test_Core(byte[] input, int expectedRetVal, int expectedRuneCount, int expectedSurrogatePairCount)
{
// Arrange
using BoundedMemory <byte> boundedMemory = BoundedMemory.AllocateFromExistingData(input);
boundedMemory.MakeReadonly();
// Act
int actualRetVal;
int actualSurrogatePairCount;
int actualRuneCount;
fixed(byte *pInputBuffer = &MemoryMarshal.GetReference(boundedMemory.Span))
{
byte *pFirstInvalidByte = _getPointerToFirstInvalidByteFn.Value(pInputBuffer, input.Length, out int utf16CodeUnitCountAdjustment, out int scalarCountAdjustment);
long ptrDiff = pFirstInvalidByte - pInputBuffer;
Assert.True((ulong)ptrDiff <= (uint)input.Length, "ptrDiff was outside expected range.");
Assert.True(utf16CodeUnitCountAdjustment <= 0, "UTF-16 code unit count adjustment must be 0 or negative.");
Assert.True(scalarCountAdjustment <= 0, "Scalar count adjustment must be 0 or negative.");
actualRetVal = (ptrDiff == input.Length) ? -1 : (int)ptrDiff;
// The last two 'out' parameters are:
// a) The number to be added to the "bytes processed" return value to come up with the total UTF-16 code unit count, and
// b) The number to be added to the "total UTF-16 code unit count" value to come up with the total scalar count.
int totalUtf16CodeUnitCount = (int)ptrDiff + utf16CodeUnitCountAdjustment;
actualRuneCount = totalUtf16CodeUnitCount + scalarCountAdjustment;
// Surrogate pair count is number of UTF-16 code units less the number of scalars.
actualSurrogatePairCount = totalUtf16CodeUnitCount - actualRuneCount;
}
// Assert
Assert.Equal(expectedRetVal, actualRetVal);
Assert.Equal(expectedRuneCount, actualRuneCount);
Assert.Equal(expectedSurrogatePairCount, actualSurrogatePairCount);
}
unsafe static void RunSpanSortKeyTest(CompareInfo compareInfo, ReadOnlySpan <char> source, CompareOptions options, byte[] expectedSortKey)
{
using BoundedMemory <char> sourceBoundedMemory = BoundedMemory.AllocateFromExistingData(source);
sourceBoundedMemory.MakeReadonly();
Assert.Equal(expectedSortKey.Length, compareInfo.GetSortKeyLength(sourceBoundedMemory.Span, options));
using BoundedMemory <byte> sortKeyBoundedMemory = BoundedMemory.Allocate <byte>(expectedSortKey.Length);
// First try with a destination which is too small - should result in an error
Assert.Throws <ArgumentException>("destination", () => compareInfo.GetSortKey(sourceBoundedMemory.Span, sortKeyBoundedMemory.Span.Slice(1), options));
// Next, try with a destination which is perfectly sized - should succeed
Span <byte> sortKeyBoundedSpan = sortKeyBoundedMemory.Span;
sortKeyBoundedSpan.Clear();
Assert.Equal(expectedSortKey.Length, compareInfo.GetSortKey(sourceBoundedMemory.Span, sortKeyBoundedSpan, options));
Assert.Equal(expectedSortKey, sortKeyBoundedSpan[0..expectedSortKey.Length].ToArray());
private static void ToBytes_Test_Core(ReadOnlySpan <char> utf16Input, int destinationSize, bool replaceInvalidSequences, bool isFinalChunk, OperationStatus expectedOperationStatus, int expectedNumCharsRead, ReadOnlySpan <byte> expectedUtf8Transcoding)
{
// Arrange
using (BoundedMemory <char> boundedSource = BoundedMemory.AllocateFromExistingData(utf16Input))
using (BoundedMemory <byte> boundedDestination = BoundedMemory.Allocate <byte>(destinationSize))
{
boundedSource.MakeReadonly();
// Act
OperationStatus actualOperationStatus = Utf8.FromUtf16(boundedSource.Span, boundedDestination.Span, out int actualNumCharsRead, out int actualNumBytesWritten, replaceInvalidSequences, isFinalChunk);
// Assert
Assert.Equal(expectedOperationStatus, actualOperationStatus);
Assert.Equal(expectedNumCharsRead, actualNumCharsRead);
Assert.Equal(expectedUtf8Transcoding.Length, actualNumBytesWritten);
Assert.Equal(expectedUtf8Transcoding.ToArray(), boundedDestination.Span.Slice(0, actualNumBytesWritten).ToArray());
}
}
private static unsafe void GetIndexOfFirstInvalidUtf16Sequence_Test_Core(char[] input, int expectedRetVal, int expectedRuneCount, long expectedUtf8ByteCount)
{
// Arrange
using BoundedMemory <char> boundedMemory = BoundedMemory.AllocateFromExistingData(input);
boundedMemory.MakeReadonly();
// Act
int actualRetVal;
long actualUtf8CodeUnitCount;
int actualRuneCount;
fixed(char *pInputBuffer = &MemoryMarshal.GetReference(boundedMemory.Span))
{
char *pFirstInvalidChar = _getPointerToFirstInvalidCharFn.Value(pInputBuffer, input.Length, out long utf8CodeUnitCountAdjustment, out int scalarCountAdjustment);
long ptrDiff = pFirstInvalidChar - pInputBuffer;
Assert.True((ulong)ptrDiff <= (uint)input.Length, "ptrDiff was outside expected range.");
Assert.True(utf8CodeUnitCountAdjustment >= 0, "UTF-16 code unit count adjustment must be non-negative.");
Assert.True(scalarCountAdjustment <= 0, "Scalar count adjustment must be 0 or negative.");
actualRetVal = (ptrDiff == input.Length) ? -1 : (int)ptrDiff;
// The last two 'out' parameters are:
// a) The number to be added to the "chars processed" return value to come up with the total UTF-8 code unit count, and
// b) The number to be added to the "total UTF-16 code unit count" value to come up with the total scalar count.
actualUtf8CodeUnitCount = ptrDiff + utf8CodeUnitCountAdjustment;
actualRuneCount = (int)ptrDiff + scalarCountAdjustment;
}
// Assert
Assert.Equal(expectedRetVal, actualRetVal);
Assert.Equal(expectedRuneCount, actualRuneCount);
Assert.Equal(actualUtf8CodeUnitCount, expectedUtf8ByteCount);
}
public static void NarrowUtf16ToLatin1_AllLatin1Input()
{
using BoundedMemory <char> utf16Mem = BoundedMemory.Allocate <char>(128);
using BoundedMemory <byte> latin1Mem = BoundedMemory.Allocate <byte>(128);
// Fill the source with [deterministic] pseudo-random chars U+0000..U+00FF, then make readonly.
Random rnd = new Random(0x54321);
Span <char> utf16Span = utf16Mem.Span;
for (int i = 0; i < utf16Span.Length; i++)
{
utf16Span[i] = (char)(byte)rnd.Next();
}
utf16Mem.MakeReadonly();
// We'll write to the Latin-1 span.
// We test with a variety of span lengths to test alignment and fallthrough code paths.
Span <byte> latin1Span = latin1Mem.Span;
for (int i = 0; i < utf16Span.Length; i++)
{
latin1Span.Clear(); // remove any data from previous iteration
// First, validate that the workhorse saw the incoming data as all-Latin-1.
Assert.Equal(128 - i, CallNarrowUtf16ToLatin1(utf16Span.Slice(i), latin1Span.Slice(i)));
// Then, validate that the data was transcoded properly.
for (int j = i; j < 128; j++)
{
Assert.Equal((ushort)utf16Span[i], (ushort)latin1Span[i]);
}
}
}
public static void NarrowUtf16ToAscii_AllAsciiInput()
{
using BoundedMemory <char> utf16Mem = BoundedMemory.Allocate <char>(128);
using BoundedMemory <byte> asciiMem = BoundedMemory.Allocate <byte>(128);
// Fill source with 00 .. 7F.
Span <char> utf16Span = utf16Mem.Span;
for (int i = 0; i < utf16Span.Length; i++)
{
utf16Span[i] = (char)i;
}
utf16Mem.MakeReadonly();
// We'll write to the ASCII span.
// We test with a variety of span lengths to test alignment and fallthrough code paths.
Span <byte> asciiSpan = asciiMem.Span;
for (int i = 0; i < utf16Span.Length; i++)
{
asciiSpan.Clear(); // remove any data from previous iteration
// First, validate that the workhorse saw the incoming data as all-ASCII.
Assert.Equal(128 - i, CallNarrowUtf16ToAscii(utf16Span.Slice(i), asciiSpan.Slice(i)));
// Then, validate that the data was transcoded properly.
for (int j = i; j < 128; j++)
{
Assert.Equal((ushort)utf16Span[i], (ushort)asciiSpan[i]);
}
}
}
