public UnicodeBlockInfo[] Read()
{
if (this.hasReadBeenCalled)
{
throw new InvalidOperationException();
}
this.hasReadBeenCalled = true;
this.m_specialBlockNames = new string[]
{
"CJK Unified Ideographs Extension A",
"CJK Unified Ideographs",
"Hangul Syllables",
"CJK Unified Ideographs Extension B",
};
string allText = Utilities.ReadAllText(FileName, true, Encoding.UTF8);
int specialBlocks_usageFlags = 0;
string[] lineStrings = allText.Split(new string[] { "\r\n", }, StringSplitOptions.RemoveEmptyEntries);
var blockList = new List<UnicodeBlockInfo>();
for (int lineNo = 0; lineNo < lineStrings.Length; ++lineNo)
{
var lineStr = lineStrings[lineNo];
var valueStrs = lineStr.Split(';');
var block = new UnicodeBlockInfo();
ParseRange(valueStrs[0], block);
block.Name = ParseBlockName(valueStrs[1]).ToNullIfEmpty();
int j = this.GetSpecialBlockId(block.Name);
if (0 <= j)
{
if (31 < j)
{
throw new NotSupportedException();
}
if ((specialBlocks_usageFlags & (1 << j)) != 0)
{
throw new InvalidDataException("Data contains multiple blocks with a special block name \"" + block.Name + "\"");
}
specialBlocks_usageFlags |= (1 << j);
block.IsNotSupported = true;
}
blockList.Add(block);
}
if (specialBlocks_usageFlags != (m_specialBlockNames.Length == 32 ? -1 : (1 << m_specialBlockNames.Length) - 1))
{
throw new InvalidDataException("Data does not contain all special blocks.");
}
blockList.Sort((b1, b2) => b1.FirstCodePoint.CompareTo(b2.FirstCodePoint));
return blockList.ToArray();
}
private static UnicodeBlockInfo Find(UnicodeBlockInfo[] blocks, CodePoint c)
{
int i = BinarySearch(blocks, c);
if (i < 0)
{
i = ~i;
--i;
}
var block = blocks[i];
if (block.LastCodePoint < c)
{
return null;
}
return block;
}
private static int BinarySearch(UnicodeBlockInfo[] blocks, CodePoint c)
{
int lo, hi;
lo = 0;
hi = blocks.Length - 1;
while (lo <= hi)
{
int mi = lo + (hi - lo) / 2;
var blockFirstCp = blocks[mi].FirstCodePoint;
if (blockFirstCp < c)
{
lo = mi + 1;
}
else if (c < blockFirstCp)
{
hi = mi - 1;
}
else
{
return mi;
}
}
return ~lo;
}
private static void ParseRange(string s, UnicodeBlockInfo block)
{
var hexNums = s.Split(new string[] { "..", }, StringSplitOptions.None);
if (hexNums.Length != 2)
{
throw new FormatException();
}
int min, max;
try
{
min = Convert.ToInt32(hexNums[0], 16);
max = Convert.ToInt32(hexNums[1], 16);
}
catch (OverflowException)
{
throw new FormatException();
}
if (min > max || 0x10FFFF < max)
{
throw new FormatException();
}
block.FirstCodePoint = new CodePoint(min);
block.LastCodePoint = new CodePoint(max);
}