/// <summary>
/// 返回指定字符的宽度，字符串遍历的顺序是从后至前的。
/// </summary>
/// <param name="ch">要获取宽度的字符。</param>
/// <returns><paramref name="ch"/> 的宽度。</returns>
private int GetWidthReversed(char ch)
{
if (!char.IsSurrogate(ch))
{
return(ch.Width());
}
if (char.IsLowSurrogate(ch))
{
surrogate = ch;
return(0);
}
// ch 是高代理项
if (surrogate == NoSurrogate)
{
// 缺失相应低代理项，忽略无效数据。
return(0);
}
var chValue = char.ConvertToUtf32(ch, surrogate);
surrogate = NoSurrogate;
return(CharExt.Width(chValue));
}
/// <summary>
/// 返回指定字符的宽度。
/// </summary>
/// <param name="ch">要获取宽度的字符。</param>
/// <returns><paramref name="ch"/> 的宽度。</returns>
private int GetWidth(char ch)
{
if (!char.IsSurrogate(ch))
{
return(ch.Width());
}
if (char.IsHighSurrogate(ch))
{
surrogate = ch;
return(0);
}
// ch 是低代理项
if (surrogate == NoSurrogate)
{
// 缺失相应高代理项，忽略无效数据。
return(0);
}
var chValue = char.ConvertToUtf32(surrogate, ch);
surrogate = NoSurrogate;
return(CharExt.Width(chValue));
}
/// <summary>
/// 令位置前进指定字符指针所指的部分。
/// </summary>
/// <param name="start">要前进的起始字符指针（包含）。</param>
/// <param name="end">要前进的结束字符指针（包含）。</param>
/// <remarks>不会修改 <c>curIndex</c> 属性。</remarks>
private unsafe void ForwardInternal(char *start, char *end)
{
Contract.Requires(end > start);
Contract.Requires(widths.Count == 0);
var oldHighSurrogate = surrogate;
surrogate = NoSurrogate;
// 最后一个字符需要特殊考虑，否则无法得到当前位置。
var lastWidth = 0;
var lastChar = *end--;
if (!char.IsHighSurrogate(lastChar))
{
if (lastChar == '\n')
{
lastWidth = -1;
}
else if (lastChar == '\t')
{
lastWidth = -2;
}
else if (char.IsLowSurrogate(lastChar))
{
if (char.IsHighSurrogate(*end))
{
// 匹配的代理项对。
lastWidth = CharExt.Width(char.ConvertToUtf32(*end, lastChar));
end--;
}
}
else
{
lastWidth = lastChar.Width();
}
lastChar = NoSurrogate;
}
curCol = nextCol;
curLine = nextLine;
var width = 0;
for (; end >= start; end--)
{
if (*end == '\n')
{
widths.Add(width);
oldHighSurrogate = NoSurrogate;
// 统计之前的行数。
curCol = 1;
curLine++;
for (; end >= start; end--)
{
if (*end == '\n')
{
curLine++;
}
}
break;
}
if (*end == '\t')
{
widths.Add(width);
width = 0;
}
else
{
width += GetWidthReversed(*end);
}
}
if (oldHighSurrogate != NoSurrogate)
{
width += GetWidthReversed(oldHighSurrogate);
}
if (width > 0)
{
widths.Add(width);
}
// 统计列数。
for (var j = widths.Count - 1; j > 0; j--)
{
curCol += widths[j];
curCol = ForwardTab(curCol);
}
surrogate = lastChar;
curCol += widths[0];
widths.Clear();
nextLine = curLine;
if (lastWidth >= 0)
{
nextCol = curCol + lastWidth;
}
else if (lastWidth == -1)
{
nextCol = 1;
nextLine++;
}
else
{
nextCol = ForwardTab(curCol);
}
}
public void TestWidth()
{
// 宽字符范围：
// 0x1100 ~ 0x115F：Hangul Jamo init. consonants
Assert.AreEqual(1, '\u10FF'.Width());
Assert.AreEqual(2, '\u1100'.Width());
Assert.AreEqual(2, '\u1101'.Width());
Assert.AreEqual(2, '\u115E'.Width());
Assert.AreEqual(2, '\u115F'.Width());
Assert.AreEqual(1, '\u1160'.Width());
// 0x2329, 0x232A：左右尖括号〈〉
Assert.AreEqual(1, '\u2328'.Width());
Assert.AreEqual(2, '\u2329'.Width());
Assert.AreEqual(2, '\u232A'.Width());
Assert.AreEqual(1, '\u232B'.Width());
// 0x2E80 ~ 0xA4CF 除了 0x303F：CJK ... YI
Assert.AreEqual(1, '\u2E7F'.Width());
Assert.AreEqual(2, '\u2E80'.Width());
Assert.AreEqual(2, '\u2E81'.Width());
Assert.AreEqual(2, '\u303E'.Width());
Assert.AreEqual(1, '\u303F'.Width());
Assert.AreEqual(2, '\u3040'.Width());
Assert.AreEqual(2, '\uA4CE'.Width());
Assert.AreEqual(2, '\uA4CF'.Width());
Assert.AreEqual(1, '\uA4D0'.Width());
// 0xAC00 ~ 0xD7A3：Hangul Syllables
Assert.AreEqual(1, '\uABFF'.Width());
Assert.AreEqual(2, '\uAC00'.Width());
Assert.AreEqual(2, '\uD7A3'.Width());
Assert.AreEqual(1, '\uD7A4'.Width());
// 0xF900 ~ 0xFAFF：CJK Compatibility Ideographs
Assert.AreEqual(1, '\uF8FF'.Width());
Assert.AreEqual(2, '\uF900'.Width());
Assert.AreEqual(2, '\uFAFF'.Width());
Assert.AreEqual(1, '\uFB00'.Width());
// 0xFE10 ~ 0xFE19：Vertical forms
Assert.AreEqual(0, '\uFE0F'.Width()); // '\uFE0F' 是 NonSpacingMark
Assert.AreEqual(2, '\uFE10'.Width());
Assert.AreEqual(2, '\uFE19'.Width());
Assert.AreEqual(1, '\uFE1A'.Width());
// 0xFE30 ~ 0xFE6F：CJK Compatibility Forms
Assert.AreEqual(1, '\uFE2F'.Width());
Assert.AreEqual(2, '\uFE30'.Width());
Assert.AreEqual(2, '\uFE6F'.Width());
Assert.AreEqual(1, '\uFE70'.Width());
// 0xFF00 ~ 0xFF60：Fullwidth Forms
Assert.AreEqual(0, '\uFEFF'.Width()); // '\uFEFF' 是 Format
Assert.AreEqual(2, '\uFF00'.Width());
Assert.AreEqual(2, '\uFF60'.Width());
Assert.AreEqual(1, '\uFF61'.Width());
// 0xFFE0 ~ 0xFFE6
Assert.AreEqual(1, '\uFFDF'.Width());
Assert.AreEqual(2, '\uFFE0'.Width());
Assert.AreEqual(2, '\uFFE6'.Width());
Assert.AreEqual(1, '\uFFE7'.Width());
// 0x20000 ~ 0x2FFFD
Assert.AreEqual(1, CharExt.Width(char.ConvertFromUtf32(0x1FFFF), 0));
Assert.AreEqual(2, CharExt.Width(char.ConvertFromUtf32(0x20000), 0));
Assert.AreEqual(2, CharExt.Width(char.ConvertFromUtf32(0x2FFFD), 0));
Assert.AreEqual(1, CharExt.Width(char.ConvertFromUtf32(0x2FFFE), 0));
// 0x30000 ~ 0x3FFFD
Assert.AreEqual(1, CharExt.Width(char.ConvertFromUtf32(0x2FFFF), 0));
Assert.AreEqual(2, CharExt.Width(char.ConvertFromUtf32(0x30000), 0));
Assert.AreEqual(2, CharExt.Width(char.ConvertFromUtf32(0x3FFFD), 0));
Assert.AreEqual(1, CharExt.Width(char.ConvertFromUtf32(0x3FFFE), 0));
}