public PosBinRule(WordPosBuilder pos)
{
_POS = pos;
_PosBinTbl = new Hashtable();
foreach (DualityWordBin bin in _PosBins)
{
_PosBinTbl[bin.HashCode] = true;
}
}
public MatchName(WordPosBuilder pos)
{
_PosBinRule = new PosBinRule(pos);
_Pos = pos;
_FamilyNameTbl = new Hashtable();
foreach (String familyName in FAMILY_NAMES)
{
_FamilyNameTbl[familyName] = true;
}
}
public int ProcessRule(List <string> preWords, int index, List <string> retWords)
{
String word = (String)preWords[index];
bool isReg;
int pos = WordPosBuilder.GetPosFromInnerPosList(_POS.GetPos(word, out isReg));
String num;
if ((pos & (int)PosEnum.POS_A_M) == (int)PosEnum.POS_A_M)
{
num = word;
int i = 0;
for (i = index + 1; i < preWords.Count; i++)
{
String next = (String)preWords[i];
int nextPos = WordPosBuilder.GetPosFromInnerPosList(_POS.GetPos(next, out isReg));
if ((nextPos & (int)PosEnum.POS_A_M) == (int)PosEnum.POS_A_M)
{
num += next;
}
else
{
break;
}
}
if (num == word)
{
return(-1);
}
else
{
retWords.Add(num);
return(i);
}
}
else
{
return(-1);
}
}
/// <summary>
/// 初始化分词组件的工作环境。文件读取异常应在调用时进行处理。
/// </summary>
/// <param name="segwords">主词典的Xml文件</param>
/// <param name="segchsstopwords">中文停止词词典的Xml文件</param>
/// <param name="segchssymbol">中文标点符号的Xml文件</param>
/// <param name="segengstopwords">英文停止词词典的Xml文件</param>
/// <param name="segengsymbol">英文标点符号的Xml文件</param>
public static void Initialize(
string segwords,
string segchsstopwords,
string segchssymbol,
string segengstopwords,
string segengsymbol)
{
SegWords = LoadSegWords(segwords);
SegChsStopwordDic = LoadStopwords(segchsstopwords, SegChsStopwordDic);
SegChsStopwordDic = LoadStopwords(segchssymbol, SegChsStopwordDic);
SegEngStopwordDic = LoadStopwords(segengstopwords, SegEngStopwordDic);
SegEngStopwordDic = LoadStopwords(segengsymbol, SegEngStopwordDic);
ExtractInfo = new ExtractInfo();
_WordPosBuilder = new WordPosBuilder();
ExtractInfo.CompareByPosEvent = CompareByPos;
foreach (WordPos item in SegWords.WordPosList)
{
ExtractInfo.InsertWordToDfa(item.Word);
_WordPosBuilder.AddWordPos(item.Word, item.Pos);
}
SetOwnerRules();
}
public MergeNumRule(WordPosBuilder pos)
{
_POS = pos;
}