public int GetTweetLength(string text, bool normalize = true)
{
if (string.IsNullOrEmpty(text))
{
return(0);
}
if (normalize)
{
MiniList <char> normalized;
if (NewSuperNfc.Compose(text, out normalized))
{
text = new string(normalized.InnerArray, 0, normalized.Count); // ポインタ使わせろ！！
}
}
var length = text.Length;
foreach (var x in this._extractor.ExtractUrls(text))
{
length += this.ShortUrlLength - x.Length;
}
// サロゲートペアを削除
var end = text.Length - 1;
for (var i = 0; i < end;)
{
// char.IsSurrogatePair はインライン化されないじゃん？
if (char.IsHighSurrogate(text[i]) && char.IsLowSurrogate(text[i + 1]))
{
length--;
i += 2;
}
else
{
i++;
}
}
return(length);
}
public static void Run()
{
if (!File.Exists(testFile))
{
DownloadTests();
}
var tests = LoadTests();
foreach (var x in tests)
{
// NFC
// c2 == toNFC(c1) == toNFC(c2) == toNFC(c3)
// c4 == toNFC(c4) == toNFC(c5)
MiniList <char> r1;
var s1 = NewSuperNfc.Compose(x[0], out r1) ? ToString(r1) : x[0];
MiniList <char> r2;
var s2 = NewSuperNfc.Compose(x[1], out r2) ? ToString(r2) : x[1];
MiniList <char> r3;
var s3 = NewSuperNfc.Compose(x[2], out r3) ? ToString(r3) : x[2];
MiniList <char> r4;
var s4 = NewSuperNfc.Compose(x[3], out r4) ? ToString(r4) : x[3];
MiniList <char> r5;
var s5 = NewSuperNfc.Compose(x[4], out r5) ? ToString(r5) : x[4];
if (!(s1 == x[1] && s2 == x[1] && s3 == x[1] &&
s4 == x[3] && s5 == x[3]))
{
Debugger.Break();
}
}
var stopwatch = new Stopwatch();
const int ntimes =
#if DEBUG
100 // Debug ビルドだと遅すぎて待ってるのだるい
#else
300
#endif
;
stopwatch.Start();
for (var i = 0; i < ntimes; i++)
{
foreach (var x in tests)
{
foreach (var y in x)
{
MiniList <char> result;
if (NewSuperNfc.Compose(y, out result))
{
ToString(result);
}
}
}
}
stopwatch.Stop();
Console.WriteLine("ToriatamaText.UnicodeNormalization: {0}", stopwatch.Elapsed);
stopwatch.Restart();
for (var i = 0; i < ntimes; i++)
{
foreach (var x in tests)
{
foreach (var y in x)
{
y.Normalize();
}
}
}
stopwatch.Stop();
Console.WriteLine("String.Normalize: {0}", stopwatch.Elapsed);
}