public virtual void Build()
{
LineNumberReader infile = null;
PrintWriter outfile = null;
string currentInfile = string.Empty;
try
{
outfile = new PrintWriter(new BufferedWriter(new OutputStreamWriter(new FileOutputStream(outFileName), "UTF-8")));
foreach (File path in pathsToData)
{
infile = new LineNumberReader(new BufferedReader(new InputStreamReader(new FileInputStream(path), "UTF-8")));
currentInfile = path.GetPath();
while (infile.Ready())
{
List <Word> sent = SentenceUtils.ToUntaggedList(infile.ReadLine().Split("\\s+"));
foreach (Word token in sent)
{
Matcher hasArabic = utf8ArabicChart.Matcher(token.Word());
if (hasArabic.Find())
{
token.SetWord(escaper.Apply(token.Word()));
token.SetWord(lexMapper.Map(null, token.Word()));
}
}
outfile.Println(SentenceUtils.ListToString(sent));
}
toStringBuffer.Append(string.Format(" Read %d input lines from %s", infile.GetLineNumber(), path.GetPath()));
}
infile.Close();
}
catch (UnsupportedEncodingException e)
{
System.Console.Error.Printf("%s: Filesystem does not support UTF-8 output\n", this.GetType().FullName);
Sharpen.Runtime.PrintStackTrace(e);
}
catch (FileNotFoundException)
{
System.Console.Error.Printf("%s: Could not open %s for writing\n", this.GetType().FullName, outFileName);
}
catch (IOException)
{
System.Console.Error.Printf("%s: Error reading from %s (line %d)\n", this.GetType().FullName, currentInfile, infile.GetLineNumber());
}
catch (Exception e)
{
System.Console.Error.Printf("%s: Input sentence from %s contains token mapped to null (line %d)\n", this.GetType().FullName, currentInfile, infile.GetLineNumber());
Sharpen.Runtime.PrintStackTrace(e);
}
finally
{
if (outfile != null)
{
outfile.Close();
}
}
}
