private static InfoLine CreateTcFromXml(IEnumerable <XmlNode> segNodes, string locale, string corpus, string sessionId, string internalId)
{
var list = segNodes.SelectMany(x => x.SelectNodes("w").Cast <XmlNode>().Select(y => y.InnerText));
string trans = StringProcess.CleanupSpace(string.Join(" ", list));
double startTime = 0;
double endTime = 0;
if (segNodes.Where(x => x["time"] != null).Count() > 1)
{
string startStr = string.Empty;
string endStr = string.Empty;
try
{
var startNode = segNodes.First(x => x["time"] != null);
startStr = startNode["time"].Attributes["value"].Value;
startTime = Common.TimeStrToSec(startStr);
}
catch { }
try
{
var endNode = segNodes.Last(x => x["time"] != null);
endStr = endNode["time"].Attributes["value"].Value;
endTime = Common.TimeStrToSec(endStr);
}
catch { }
}
return(new InfoLine(corpus, "U", sessionId, internalId, startTime, endTime, trans, "<NA/>"));
}
private void TaggingSyl()
{
foreach (TextGridItem tgi in ItemDict["SYL"])
{
if (tgi.Text.Contains('/') || tgi.Text.Contains('\\'))
{
tgi.UpdateText("<overlap>");
}
else
{
string tagText = StringProcess.NormXSil(tgi.Text);
string cleanText = StringProcess.CleanupSpace(tagText);
tgi.UpdateText(cleanText);
}
}
}
private void CreateNewData(string cleanDataPath, string tagDataPath, string noEmptyPath, string wbrPath, string outputPath)
{
var noEmptyList = File.ReadLines(cleanDataPath).Select(x => x.Replace(" ", string.Empty));
File.WriteAllLines(noEmptyPath, noEmptyList);
var tagList = File.ReadLines(tagDataPath).Select(x => StringProcess.GetTagPrefixIndices(x));
//RunWordBreak rwb = new RunWordBreak(Cfg);
//rwb.WordBreak(noEmptyPath, wbrPath);
var wbrList = File.ReadLines(wbrPath);
var outputList = wbrList.Zip(tagList, (x, y) => StringProcess.InsertTagToWords(x, " <bi> ", y)).Select(x => StringProcess.CleanupSpace(x));
File.WriteAllLines(outputPath, outputList);
}
private void RefreshTextGridWbr(string cleanDatapath, string tagDataPath, string outputPath, string wbrPath, string tag)
{
string tmpName = Guid.NewGuid().ToString();
string noEmptyPath = Path.Combine(Cfg.TmpFolder, tmpName + ".noEmpty");
var noEmptyList = File.ReadLines(cleanDatapath).Select(x => x.Replace(" ", string.Empty));
File.WriteAllLines(noEmptyPath, noEmptyList);
RunWordBreak rwb = new RunWordBreak(Cfg);
rwb.WordBreak(noEmptyPath, wbrPath);
var tagList = File.ReadLines(tagDataPath).Select(x => StringProcess.GetTagPrefixIndices(x));
var wbrList = File.ReadLines(wbrPath);
var outputList = wbrList.Zip(tagList, (x, y) => StringProcess.InsertTagToWords(x, " " + tag + " ", y)).Select(x => StringProcess.CleanupSpace(x));
File.WriteAllLines(outputPath, outputList);
}
private void CleanupWbr(string wbrPath, string outputPath)
{
var list = File.ReadLines(wbrPath).Select(x => StringProcess.CleanupSpace(x));
File.WriteAllLines(outputPath, list);
}