//Convert CRFSharp output format to string list
private List <string> ConvertCRFTermOutToStringList(List <List <string> > inbuf, crf_seg_out[] crf_out)
{
StringBuilder sb = new StringBuilder();
for (int i = 0; i < inbuf.Count; i++)
{
sb.Append(inbuf[i][0]);
}
string strText = sb.ToString();
List <string> rstList = new List <string>();
for (int i = 0; i < crf_out.Length; i++)
{
if (crf_out[i] == null)
{
//No more result
break;
}
sb.Clear();
crf_seg_out crf_term_out = crf_out[i];
for (int j = 0; j < crf_term_out.Count; j++)
{
string str = strText.Substring(crf_term_out.tokenList[j].offset, crf_term_out.tokenList[j].length);
string strNE = crf_term_out.tokenList[j].strTag;
sb.Append(str);
if (strNE.Length > 0)
{
sb.Append("[" + strNE + "]");
}
sb.Append(" ");
}
rstList.Add(sb.ToString().Trim());
}
return(rstList);
}
public string Predict(DecoderArgs _property)
{
ParallelOptions parallelOption = new ParallelOptions();
StringWriter sw = null, swSeg = null;
if (_property.outputstyle == 0)
{
sw = new StringWriter();
}
else
{
swSeg = new StringWriter();
}
ConcurrentQueue <List <List <string> > > queueRecords = new ConcurrentQueue <List <List <string> > >();
ConcurrentQueue <List <List <string> > > queueSegRecords = new ConcurrentQueue <List <List <string> > >();
StringReader sr = new StringReader(_property.predictstring);
parallelOption.MaxDegreeOfParallelism = _property.thread;
Parallel.For(0, parallelOption.MaxDegreeOfParallelism, t =>
{
//Create decoder tagger instance. If the running environment is multi-threads, each thread needs a separated instance
SegDecoderTagger tagger = _crfWrapper.CreateTagger();
tagger.set_nbest(_property.nbest);
tagger.set_vlevel(_property.probLevel);
//Initialize result
crf_seg_out[] crf_out = new crf_seg_out[_property.nbest];
for (int i = 0; i < _property.nbest; i++)
{
crf_out[i] = new crf_seg_out();
}
List <List <string> > inbuf = new List <List <string> >();
while (true)
{
lock (rdLocker)
{
if (ReadRecord(inbuf, sr) == false)
{
break;
}
queueRecords.Enqueue(inbuf);
queueSegRecords.Enqueue(inbuf);
}
//Call CRFSharp wrapper to predict given string's tags
if (swSeg != null)
{
_crfWrapper.Segment(crf_out, tagger, inbuf);
}
else
{
_crfWrapper.Segment((crf_term_out[])crf_out, (DecoderTagger)tagger, inbuf);
}
List <List <string> > peek = null;
//Save segmented tagged result into file
if (swSeg != null)
{
List <string> rstList = ConvertCRFTermOutToStringList(inbuf, crf_out);
while (peek != inbuf)
{
queueSegRecords.TryPeek(out peek);
}
foreach (string item in rstList)
{
swSeg.WriteLine(item);
}
queueSegRecords.TryDequeue(out peek);
peek = null;
}
//Save raw tagged result (with probability) into file
if (sw != null)
{
while (peek != inbuf)
{
queueRecords.TryPeek(out peek);
}
OutputRawResult(inbuf, crf_out, tagger, sw);
queueRecords.TryDequeue(out peek);
}
}
});
if (sw != null)
{
sw.Close();
return(sw.ToString());
}
if (swSeg != null)
{
swSeg.Close();
return(swSeg.ToString());
}
return("");
}
int seg_termbuf_build(crf_seg_out term_buf)
{
term_buf.Clear();
//build raw result at first
int iRet = termbuf_build(term_buf);
if (iRet != Utils.ERROR_SUCCESS)
{
return(iRet);
}
//Then build token result
int term_len = 0;
double weight = 0.0;
int num = 0;
for (int i = 0; i < x_.Count; i++)
{
//Adding the length of current token
string strTag = term_buf.result_[i];
term_len += x_[i][0].Length;
weight += term_buf.weight_[i];
num++;
//Check if current term is the end of a token
if ((strTag.StartsWith("B_") == false &&
strTag.StartsWith("M_") == false) ||
i == x_.Count - 1)
{
SegToken tkn = new SegToken();
tkn.length = term_len;
tkn.offset = term_buf.termTotalLength;
int spos = strTag.IndexOf('_');
if (spos < 0)
{
if (strTag == "NOR")
{
tkn.strTag = "";
}
else
{
tkn.strTag = strTag;
}
}
else
{
tkn.strTag = strTag.Substring(spos + 1);
}
term_buf.termTotalLength += term_len;
//Calculate each token's weight
switch (vlevel_)
{
case 0:
tkn.fWeight = 0.0;
break;
case 2:
tkn.fWeight = weight / num;
weight = 0.0;
num = 0;
break;
}
term_buf.tokenList.Add(tkn);
term_len = 0;
}
}
return(Utils.ERROR_SUCCESS);
}