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 IEnumerable <IInterval> IntervalTransform <T>(IEnumerable <T> collection, IEnumerable <IInterval> intervalList)
{
int i = 0;
var array = collection.ToArray();
foreach (var interval in intervalList)
{
if (!StringProcess.IsTag(interval.Value()))
{
TextGridInterval ti = new TextGridInterval(interval, array[i].ToString());
yield return(ti as IInterval);
i++;
}
}
}
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);
}
public static void RemoveTagsFromFile(string tagPath, string cleanPath)
{
var list = File.ReadLines(tagPath).Select(x => StringProcess.CleanupTag(x));
File.WriteAllLines(cleanPath, list);
}
private static IEnumerable <Tuple <string, string> > MatchPairFiles(string pathLine)
{
string srcFilePath = pathLine.Split('\t')[1];
string tgtFilePath = pathLine.Split('\t')[3];
Console.WriteLine("Processing " + srcFilePath);
var srcList = File.ReadLines(srcFilePath).Select(x => new InfoLine(x).Transcription).Select(x => StringProcess.CleanupEnuString(x));
var tgtList = File.ReadLines(tgtFilePath).Select(x => new InfoLine(x).Transcription).Select(x => StringProcess.CleanupChsString(x));
return(srcList.Zip(tgtList, (x, y) => new Tuple <string, string>(x, y)).Where(ValidPair));
}
private void CleanupWbr(string wbrPath, string outputPath)
{
var list = File.ReadLines(wbrPath).Select(x => StringProcess.CleanupSpace(x));
File.WriteAllLines(outputPath, list);
}