/// <summary>
/// Build tree
/// </summary>
/// <param name="pl">position length list</param>
/// <param name="count">position length list count</param>
/// <param name="parent">parent node</param>
/// <param name="curIndex">current index of position length list</param>
private void BuildTree(Node parent, int curIndex)
{
//嵌套太多的情况一般很少发生，如果发生，强行中断，以免造成博弈树遍历层次过多
//降低系统效率
if (_LeafNodeList.Count > 8192)
{
return;
}
if (curIndex < _PositionLengthArrCount - 1)
{
if (_PositionLengthArr[curIndex + 1].Position == _PositionLengthArr[curIndex].Position)
{
BuildTree(parent, curIndex + 1);
}
}
int spaceCount = parent.SpaceCount + _PositionLengthArr[curIndex].Position - (parent.PositionLength.Position + parent.PositionLength.Length);
int singleWordCount = parent.SingleWordCount + (_PositionLengthArr[curIndex].Length == 1 ? 1 : 0);
double freqSum = 0;
if (_Options != null)
{
if (_Options.FrequencyFirst)
{
freqSum = parent.FreqSum + _PositionLengthArr[curIndex].WordAttr.Frequency;
}
}
Node curNode = new Node(_PositionLengthArr[curIndex], parent, parent.AboveCount + 1, spaceCount, singleWordCount, freqSum);
int cur = curIndex + 1;
while (cur < _PositionLengthArrCount)
{
if (_PositionLengthArr[cur].Position >= _PositionLengthArr[curIndex].Position + _PositionLengthArr[curIndex].Length)
{
BuildTree(curNode, cur);
break;
}
cur++;
}
if (cur >= _PositionLengthArrCount)
{
curNode.SpaceCount += _InputStringLength - curNode.PositionLength.Position - curNode.PositionLength.Length;
_LeafNodeList.Add(curNode);
}
}
public Framework.AppendList <PositionLength> GetAllMatchs(string text, bool chineseNameIdentify)
{
Framework.AppendList <PositionLength> result = new Framework.AppendList <PositionLength>();
if (text == null && text == "")
{
return(result);
}
string keyText = text;
if (text[0] < 128)
{
keyText = keyText.ToLower();
}
for (int i = 0; i < text.Length; i++)
{
byte[] lenList;
char fst = keyText[i];
List <string> chsNames = null;
if (chineseNameIdentify)
{
chsNames = ChineseName.Match(text, i);
if (chsNames != null)
{
foreach (string name in chsNames)
{
WordAttribute wa = new WordAttribute(name, POS.POS_A_NR, 0);
result.Add(new PositionLength(i, name.Length, wa));
}
}
}
WordAttribute fwa;
if (_FirstCharDict.TryGetValue(fst, out fwa))
{
result.Add(new PositionLength(i, 1, fwa));
}
if (i < keyText.Length - 1)
{
uint doubleChar = ((uint)keyText[i] * 65536) + keyText[i + 1];
if (_DoubleCharDict.TryGetValue(doubleChar, out fwa))
{
result.Add(new PositionLength(i, 2, fwa));
}
}
if (i >= keyText.Length - 2)
{
continue;
}
long tripleChar = ((long)keyText[i]) * 0x100000000 + (uint)(keyText[i + 1] * 65536) + keyText[i + 2];
if (_TripleCharDict.TryGetValue(tripleChar, out lenList))
{
foreach (byte len in lenList)
{
if (len == 0)
{
break;
}
if (i + len > keyText.Length)
{
continue;
}
string key = keyText.Substring(i, len);
WordAttribute wa;
if (_WordDict.TryGetValue(key, out wa))
{
if (chsNames != null)
{
bool find = false;
foreach (string name in chsNames)
{
if (wa.Word == name)
{
find = true;
break;
}
}
if (find)
{
continue;
}
}
result.Add(new PositionLength(i, len, wa));
}
}
}
}
return(result);
}