public void TestUtf32BE()
{
Encoding penc = new PythonSurrogatePassEncoding(new UTF32Encoding(bigEndian: true, byteOrderMark: false));
// lone high surrogate
Assert.AreEqual("\ud810", penc.GetChars("\x00\x00\xd8\x10".AsBytes()));
// lone low surrogate
Assert.AreEqual("\udc0a", penc.GetChars("\x00\x00\xdc\n".AsBytes()));
// invalid surrogate pair (low, high)
Assert.AreEqual("\ude51\uda2f", penc.GetChars("\x00\x00\xdeQ\x00\x00\xda/".AsBytes()));
}
public void TestUtf16BE()
{
Encoding penc = new PythonSurrogatePassEncoding(Encoding.BigEndianUnicode);
// lone high surrogate
Assert.AreEqual("\ud810", penc.GetChars("\xd8\x10".AsBytes()));
// lone low surrogate
Assert.AreEqual("\udc0a", penc.GetChars("\xdc\n".AsBytes()));
// invalid surrogate pair (low, high)
Assert.AreEqual("\ude51\uda2f", penc.GetChars("\xdeQ\xda/".AsBytes()));
}
public void TestUtf16LE()
{
Encoding penc = new PythonSurrogatePassEncoding(Encoding.Unicode);
// lone high surrogate
Assert.AreEqual("\ud810", penc.GetChars("\x10\xd8".AsBytes()));
// lone low surrogate
Assert.AreEqual("\udc0a", penc.GetChars("\n\xdc".AsBytes()));
// invalid surrogate pair (low, high)
Assert.AreEqual("\ude51\uda2f", penc.GetChars("Q\xde/\xda".AsBytes()));
}
public void TestUtf7()
{
// "surrogatepass" is not supported for UTF-7 per se,
// but UTF-7 is supposed to decode any surogate characters from its ASCII mangled form
// without requiring any fallback support
Encoding penc = new PythonSurrogatePassEncoding(new UTF7Encoding(allowOptionals: true));
// lone high surrogate
Assert.AreEqual("abc\ud810xyz", penc.GetChars("abc+2BA-xyz".AsBytes()));
// lone low surrogate
Assert.AreEqual("abc\udc0axyz", penc.GetChars("abc+3Ao-xyz".AsBytes()));
// invalid surrogate pair (low, high)
Assert.AreEqual("abc\ude51\uda2fxyz", penc.GetChars("abc+3lHaLw-xyz".AsBytes()));
}
public void TestAscii()
{
// 'surrogatepass' is supported only for UTF-8, UTF-16LE, UTF-16BE, UTF-32LE, and UTF-32BE
// nevertheless, it can be used with other encodings as long as there are no encoding errors
Encoding penc = new PythonSurrogatePassEncoding(Encoding.ASCII);
// clean ASCII
Assert.AreEqual("abc", penc.GetChars("abc".AsBytes()));
// Attempting to decode surrogates from ASCII will throw an exception.
// Note that this is CPython 3.5 behaviour, CPython 3.4 will will blindly extract UTF-8 encoded surrogates from ASCII.
// lone high surrogate in UTF-8
Assert.Throws <DecoderFallbackException>(() => penc.GetChars("\xed\xa0\x90".AsBytes()));
// lone low surrogate in UTF-8
Assert.Throws <DecoderFallbackException>(() => penc.GetChars("\xed\xb0\x8a".AsBytes()));
// invalid surrogate pair (low, high) in UTF-8
Assert.Throws <DecoderFallbackException>(() => penc.GetChars("\xed\xb9\x91\xed\xa8\xaf".AsBytes()));
}
public void TestUtf8()
{
Encoding penc = new PythonSurrogatePassEncoding(Encoding.UTF8);
// lone high surrogate
Assert.AreEqual("abc\ud810xyz", penc.GetChars("abc\xed\xa0\x90xyz".AsBytes()));
// lone low surrogate
Assert.AreEqual("abc\udc0axyz", penc.GetChars("abc\xed\xb0\x8axyz".AsBytes()));
// invalid surrogate pair (low, high)
Assert.AreEqual("abc\ude51\uda2fxyz", penc.GetChars("abc\xed\xb9\x91\xed\xa8\xafxyz".AsBytes()));
// valid surrogate pair (high, low)
Assert.AreEqual("abc\uda2f\ude51xyz", penc.GetChars("abc\xed\xa8\xaf\xed\xb9\x91xyz".AsBytes()));
var chars = new char[9];
// broken lone high surrogate
var bytes = "abc\xed-\xa0\x90xyz".AsBytes();
Assert.That(() => penc.GetChars(bytes),
Throws.TypeOf <DecoderFallbackException>()
.With.Property("Index").EqualTo(3)
.And.Property("BytesUnknown").One.EqualTo(0xed));
var dec = penc.GetDecoder();
Assert.That(dec.GetCharCount(bytes, 0, 4, flush: false), Is.EqualTo(3));
Assert.That(dec.GetChars(bytes, 0, 4, chars, 0, flush: false), Is.EqualTo(3));
Assert.That(() => dec.GetCharCount(bytes, 4, 4, flush: false),
Throws.TypeOf <DecoderFallbackException>()
.With.Property("Index").EqualTo(-1)
.And.Property("BytesUnknown").One.EqualTo(0xed));
// broken in a different way
bytes = "abc\xed\xa0-\x90xyz".AsBytes();
Assert.That(() => penc.GetChars(bytes),
Throws.TypeOf <DecoderFallbackException>()
.With.Property("Index").EqualTo(3)
.And.Property("BytesUnknown").One.EqualTo(0xed));
dec.Reset();
Assert.That(dec.GetCharCount(bytes, 0, 4, flush: false), Is.EqualTo(3));
Assert.That(dec.GetChars(bytes, 0, 4, chars, 0, flush: false), Is.EqualTo(3));
Assert.That(() => dec.GetCharCount(bytes, 4, 4, flush: false),
Throws.TypeOf <DecoderFallbackException>()
.With.Property("Index").EqualTo(-1)
.And.Property("BytesUnknown").One.EqualTo(0xed));
dec.Reset();
Assert.That(dec.GetCharCount(bytes, 0, 5, flush: false), Is.EqualTo(3));
Assert.That(dec.GetChars(bytes, 0, 5, chars, 0, flush: false), Is.EqualTo(3));
Assert.That(() => dec.GetCharCount(bytes, 5, 3, flush: false),
Throws.TypeOf <DecoderFallbackException>()
.With.Property("Index").EqualTo(-2)
.And.Property("BytesUnknown").One.EqualTo(0xed));
// unfinished surrogate sequence in the middle
bytes = "abc\xed\xa0xyz".AsBytes();
Assert.That(() => penc.GetChars(bytes),
Throws.TypeOf <DecoderFallbackException>()
.With.Property("Index").EqualTo(3)
.And.Property("BytesUnknown").One.EqualTo(0xed));
dec.Reset();
Assert.That(dec.GetCharCount(bytes, 0, 5, flush: false), Is.EqualTo(3));
Assert.That(dec.GetChars(bytes, 0, 5, chars, 0, flush: false), Is.EqualTo(3));
Assert.That(() => dec.GetCharCount(bytes, 5, 2, flush: false),
Throws.TypeOf <DecoderFallbackException>()
.With.Property("Index").EqualTo(-2)
.And.Property("BytesUnknown").One.EqualTo(0xed));
// unfinished surrogate sequence at the end
bytes = "abcxyz\xed\xa0".AsBytes();
Assert.That(() => penc.GetChars(bytes),
Throws.TypeOf <DecoderFallbackException>()
.With.Property("Index").EqualTo(6)
.And.Property("BytesUnknown").One.EqualTo(0xed));
dec.Reset();
Assert.That(dec.GetCharCount(bytes, 0, 7, flush: false), Is.EqualTo(6));
Assert.That(dec.GetChars(bytes, 0, 7, chars, 0, flush: false), Is.EqualTo(6));
Assert.That(() => dec.GetCharCount(bytes, 7, 1, flush: true),
Throws.TypeOf <DecoderFallbackException>()
.With.Property("Index").EqualTo(-1)
.And.Property("BytesUnknown").One.EqualTo(0xed));
}
