/// <summary>
/// 状态数据创建器
/// </summary>
/// <param name="values">状态集合</param>
public TmphByteArrayBuilder(TmphKeyValue<byte[], int>[] values)
{
this.values = values;
prefixSize = tableCount = stateCount = tableType = charCount = 0;
state = charsAscii = prefix = table = null;
if (values.Length > 1)
{
byte* chars = stackalloc byte[128 >> 3];
this.chars = new TmphFixedMap(chars, 128 >> 3, 0);
Data = new TmphPointer();
count(0, values.Length, 0);
for (byte* start = chars, end = chars + (128 >> 3); start != end; start += sizeof(int))
{
charCount += (*(uint*)start).bitCount();
}
var size = (1 + (stateCount += tableCount) * 3) * sizeof(int) + 128 + 4 +
(prefixSize & (int.MaxValue - 3));
if (stateCount < 256) size += tableCount * (charCount + 1);
else if (stateCount < 65536)
{
size += tableCount * (charCount + 1) * sizeof(ushort);
tableType = 1;
}
else
{
size += tableCount * (charCount + 1) * sizeof(int);
tableType = 2;
}
Data = TmphUnmanaged.Get(size, true);
*Data.Int = stateCount; //状态数量[int]
state = Data.Byte + sizeof(int); //状态集合[stateCount*(前缀位置[int]+状态位置[int]+名称索引[int])]
charsAscii = state + (stateCount * 3) * sizeof(int); //ascii字符查找表[128*byte]
byte charIndex = 0;
for (byte index = 1; index != 128; ++index)
{
if (this.chars.Get(index)) *(charsAscii + index) = ++charIndex;
}
prefix = charsAscii + 128; //前缀集合
table = prefix + ((prefixSize & (int.MaxValue - 3)) + 4);
//状态矩阵[tableCount*(charCount+1)*[byte/ushort/int]]
*prefix++ = (byte)charCount; //字符数量
stateCount = 0;
create(0, values.Length, 0);
}
else
{
chars = new TmphFixedMap();
var value = values[0].Key;
fixed (byte* valueFixed = value)
{
if (values[0].Key.Length <= 128)
{
Data = TmphUnmanaged.Get(sizeof(int) + sizeof(int) * 3 + 128 + 1, false);
*Data.Int = 1; //状态数量
state = Data.Byte + sizeof(int);
*(int*)state = sizeof(int) * 3; //前缀位置
*(int*)(state + sizeof(int)) = 0; //状态位置
*(int*)(state + sizeof(int) * 2) = values[0].Value; //名称索引
prefix = Data.Byte + sizeof(int) * 4;
Unsafe.TmphMemory.Copy(valueFixed, prefix, value.Length);
*(prefix + value.Length) = *(prefix + 128) = 0;
}
else
{
Data = TmphUnmanaged.Get(sizeof(int) + sizeof(int) * 3 + 128 + 1 + value.Length + 1, true);
*Data.Int = 1; //状态数量
state = Data.Byte + sizeof(int);
*(int*)state = sizeof(int) * 3 + 128 + 1; //前缀位置
*(int*)(state + sizeof(int)) = 0; //状态位置
*(int*)(state + sizeof(int) * 2) = values[0].Value; //名称索引
Unsafe.TmphMemory.Copy(valueFixed, Data.Byte + sizeof(int) * 3 + 128 + 1, value.Length);
}
}
}
}
public unsafe static string toRegex(this string value)
{
if (value == null || value.Length == 0)
{
fixed (char* valueFixed = value)
{
char* end = valueFixed + value.Length;
int count = Unsafe.TmphString.AsciiCount(valueFixed, end, regexEscapeMap.Byte, RegexEscape[0]);
if (count != 0)
{
TmphFixedMap map = new TmphFixedMap(regexEscapeMap);
value = Laurent.Lee.CLB.TmphString.FastAllocateString(count += value.Length);
fixed (char* writeFixed = value)
{
for (char* start = valueFixed, write = writeFixed; start != end; *write++ = *start++)
{
if ((*start & 0xff80) == 0 && map.Get(*start)) *write++ = '\\';
}
}
}
}
}
return value;
}
/// <summary>
/// 状态数据创建器
/// </summary>
/// <param name="values">状态集合</param>
public TmphBuilder(TmphKeyValue<byte[], int>[] values)
{
this.values = values;
prefixSize = tableCount = stateCount = tableType = charCount = 0;
state = bytes = nullPrefix = prefix = table = null;
if (values.Length > 1)
{
byte* chars = stackalloc byte[256 >> 3];
this.chars = new TmphFixedMap(chars, 256 >> 3, 0);
Data = new TmphPointer();
count(0, values.Length, 0);
for (byte* start = chars, end = chars + (256 >> 3); start != end; start += sizeof(int))
{
charCount += (*(uint*)start).bitCount();
}
var size = (1 + (stateCount += tableCount) * 3) * sizeof(int) + 256 + 4 +
((prefixSize + 3) & (int.MaxValue - 3));
if (stateCount < 256) size += tableCount * charCount;
else if (stateCount < 65536)
{
size += tableCount * charCount * sizeof(ushort);
tableType = 1;
}
else
{
size += tableCount * charCount * sizeof(int);
tableType = 2;
}
Data = TmphUnmanaged.Get(size, true);
*Data.Int = stateCount; //状态数量[int]
state = Data.Byte + sizeof(int); //状态集合[stateCount*(前缀位置[int]+状态位置[int]+名称索引[int])]
bytes = state + (stateCount * 3) * sizeof(int); //字节查找表[256*byte]
byte charIndex = 0;
for (var index = 0; index != 256; ++index)
{
if (this.chars.Get(index)) *(bytes + index) = charIndex++;
}
nullPrefix = bytes + 256; //空前缀
table = nullPrefix + (((prefixSize + 3) & (int.MaxValue - 3)) + 4);
//状态矩阵[tableCount*charCount*[byte/ushort/int]]
*(ushort*)nullPrefix = (ushort)charCount; //字符数量
prefix = nullPrefix + sizeof(int) + sizeof(ushort); //前缀集合
nullPrefix += sizeof(int);
stateCount = 0;
create(0, values.Length, 0);
}
else
{
chars = new TmphFixedMap();
var value = values[0].Key;
fixed (byte* valueFixed = value)
{
if (values[0].Key.Length <= 254)
{
Data = TmphUnmanaged.Get(sizeof(int) + sizeof(int) * 3 + 256 + 2, false);
*Data.Int = 1; //状态数量
state = Data.Byte + sizeof(int);
*(int*)state = sizeof(int) * 3 + sizeof(ushort); //前缀位置
*(int*)(state + sizeof(int)) = 0; //状态位置
*(int*)(state + sizeof(int) * 2) = values[0].Value; //名称索引
prefix = Data.Byte + sizeof(int) * 4;
*(ushort*)prefix = (ushort)value.Length;
Unsafe.TmphMemory.Copy(valueFixed, prefix + sizeof(ushort), value.Length);
*(ushort*)(prefix + 256) = 0;
}
else
{
Data = TmphUnmanaged.Get(sizeof(int) + sizeof(int) * 3 + 256 + 4 + 2 + value.Length, true);
*Data.Int = 1; //状态数量
state = Data.Byte + sizeof(int);
*(int*)state = sizeof(int) * 3 + 256 + 4 + sizeof(ushort); //前缀位置
*(int*)(state + sizeof(int)) = 0; //状态位置
*(int*)(state + sizeof(int) * 2) = values[0].Value; //名称索引
prefix = Data.Byte + sizeof(int) * 4 + 256 + 4;
*(ushort*)prefix = (ushort)value.Length;
Unsafe.TmphMemory.Copy(valueFixed, prefix + sizeof(ushort), value.Length);
}
}
}
}
