/// <summary>Creates an ArabicTokenizer.</summary>
/// <remarks>
/// Creates an ArabicTokenizer. The default tokenizer
/// is ArabicTokenizer.atbFactory(), which produces the
/// same orthographic normalization as Green and Manning (2010).
/// </remarks>
/// <returns>A TokenizerFactory that produces each Arabic token as a CoreLabel</returns>
private ITokenizerFactory <CoreLabel> GetTokenizerFactory()
{
ITokenizerFactory <CoreLabel> tokFactory = null;
if (!isTokenized)
{
if (tokenizerOptions == null)
{
tokFactory = ArabicTokenizer.AtbFactory();
string atbVocOptions = "removeProMarker,removeMorphMarker,removeLengthening";
tokFactory.SetOptions(atbVocOptions);
}
else
{
if (tokenizerOptions.Contains("removeSegMarker"))
{
throw new Exception("Option 'removeSegMarker' cannot be used with ArabicSegmenter");
}
tokFactory = ArabicTokenizer.Factory();
tokFactory.SetOptions(tokenizerOptions);
}
log.Info("Loaded ArabicTokenizer with options: " + tokenizerOptions);
}
return(tokFactory);
}
/// <summary>A fast, rule-based tokenizer for Modern Standard Arabic (UTF-8 encoding).</summary>
/// <remarks>
/// A fast, rule-based tokenizer for Modern Standard Arabic (UTF-8 encoding).
/// Performs punctuation splitting and light tokenization by default.
/// Orthographic normalization options are available, and can be enabled with
/// command line options.
/// <p>
/// Currently, this tokenizer does not do line splitting. It normalizes non-printing
/// line separators across platforms and prints the system default line splitter
/// to the output.
/// <p>
/// The following normalization options are provided:
/// <ul>
/// <li>
/// <c>useUTF8Ellipsis</c>
/// : Replaces sequences of three or more full stops with \u2026</li>
/// <li>
/// <c>normArDigits</c>
/// : Convert Arabic digits to ASCII equivalents</li>
/// <li>
/// <c>normArPunc</c>
/// : Convert Arabic punctuation to ASCII equivalents</li>
/// <li>
/// <c>normAlif</c>
/// : Change all alif forms to bare alif</li>
/// <li>
/// <c>normYa</c>
/// : Map ya to alif maqsura</li>
/// <li>
/// <c>removeDiacritics</c>
/// : Strip all diacritics</li>
/// <li>
/// <c>removeTatweel</c>
/// : Strip tatweel elongation character</li>
/// <li>
/// <c>removeQuranChars</c>
/// : Remove diacritics that appear in the Quran</li>
/// <li>
/// <c>removeProMarker</c>
/// : Remove the ATB null pronoun marker</li>
/// <li>
/// <c>removeSegMarker</c>
/// : Remove the ATB clitic segmentation marker</li>
/// <li>
/// <c>removeMorphMarker</c>
/// : Remove the ATB morpheme boundary markers</li>
/// <li>
/// <c>removeLengthening</c>
/// : Replace all sequences of three or more identical (non-period) characters with one copy</li>
/// <li>
/// <c>atbEscaping</c>
/// : Replace left/right parentheses with ATB escape characters</li>
/// </ul>
/// </remarks>
/// <param name="args"/>
public static void Main(string[] args)
{
if (args.Length > 0 && args[0].Contains("help"))
{
System.Console.Error.Printf("Usage: java %s [OPTIONS] < file%n", typeof(ArabicTokenizer).FullName);
System.Console.Error.Printf("%nOptions:%n");
log.Info(" -help : Print this message. See javadocs for all normalization options.");
log.Info(" -atb : Tokenization for the parsing experiments in Green and Manning (2010)");
System.Environment.Exit(-1);
}
// Process normalization options
Properties tokenizerOptions = StringUtils.ArgsToProperties(args);
ITokenizerFactory <CoreLabel> tf = tokenizerOptions.Contains("atb") ? ArabicTokenizer.AtbFactory() : ArabicTokenizer.Factory();
foreach (string option in tokenizerOptions.StringPropertyNames())
{
tf.SetOptions(option);
}
// Replace line separators with a token so that we can
// count lines
tf.SetOptions("tokenizeNLs");
// Read the file
int nLines = 0;
int nTokens = 0;
try
{
string encoding = "UTF-8";
ITokenizer <CoreLabel> tokenizer = tf.GetTokenizer(new InputStreamReader(Runtime.@in, encoding));
bool printSpace = false;
while (tokenizer.MoveNext())
{
++nTokens;
string word = tokenizer.Current.Word();
if (word.Equals(ArabicLexer.NewlineToken))
{
++nLines;
printSpace = false;
System.Console.Out.WriteLine();
}
else
{
if (printSpace)
{
System.Console.Out.Write(" ");
}
System.Console.Out.Write(word);
printSpace = true;
}
}
}
catch (UnsupportedEncodingException e)
{
Sharpen.Runtime.PrintStackTrace(e);
}
System.Console.Error.Printf("Done! Tokenized %d lines (%d tokens)%n", nLines, nTokens);
}
public virtual void TestArabicTokenizer()
{
System.Diagnostics.Debug.Assert((untokInputs.Length == tokReferences.Length));
ITokenizerFactory <CoreLabel> tf = ArabicTokenizer.AtbFactory();
tf.SetOptions("removeProMarker");
tf.SetOptions("removeSegMarker");
tf.SetOptions("removeMorphMarker");
for (int i = 0; i < untokInputs.Length; ++i)
{
string line = untokInputs[i];
ITokenizer <CoreLabel> tokenizer = tf.GetTokenizer(new StringReader(line));
IList <CoreLabel> tokens = tokenizer.Tokenize();
string tokenizedLine = SentenceUtils.ListToString(tokens);
string reference = tokReferences[i];
NUnit.Framework.Assert.AreEqual("Tokenization deviates from reference", reference, tokenizedLine);
}
}
public virtual void TestCharOffsets()
{
string untokInput = "إِنَّ- -نا هادِئ+ُونَ .";
int[] beginOffsets = new int[] { 0, 7, 11, 22 };
int[] endOffsets = new int[] { 6, 10, 21, 23 };
ITokenizerFactory <CoreLabel> tf = ArabicTokenizer.AtbFactory();
tf.SetOptions("removeProMarker");
tf.SetOptions("removeSegMarker");
tf.SetOptions("removeMorphMarker");
ITokenizer <CoreLabel> tokenizer = tf.GetTokenizer(new StringReader(untokInput));
IList <CoreLabel> tokens = tokenizer.Tokenize();
NUnit.Framework.Assert.AreEqual("Number of tokens doesn't match reference", tokens.Count, beginOffsets.Length);
for (int i = 0; i < beginOffsets.Length; i++)
{
NUnit.Framework.Assert.AreEqual("Char begin offset deviates from reference", beginOffsets[i], tokens[i].BeginPosition());
NUnit.Framework.Assert.AreEqual("Char end offset deviates from reference", endOffsets[i], tokens[i].EndPosition());
}
}
/// <summary>For debugging.</summary>
/// <param name="args"/>
/// <exception cref="System.IO.IOException"></exception>
public static void Main(string[] args)
{
if (args.Length != 1)
{
System.Console.Error.Printf("Usage: java %s file > output%n", typeof(ArabicDocumentReaderAndWriter).FullName);
System.Environment.Exit(-1);
}
string fileName = args[0];
ITokenizerFactory <CoreLabel> tokFactory = ArabicTokenizer.AtbFactory();
string atbVocOptions = "removeProMarker,removeMorphMarker";
tokFactory.SetOptions(atbVocOptions);
BufferedReader reader = IOUtils.ReaderFromString(fileName);
for (string line; (line = reader.ReadLine()) != null;)
{
string[] toks = line.Split("\\s+");
string delim = Pattern.Quote(tagDelimiter);
bool isStart = true;
foreach (string wordTag in toks)
{
string[] wordTagPair = wordTag.Split(delim);
System.Diagnostics.Debug.Assert(wordTagPair.Length == 2);
string word = wordTagPair[0];
if (tokFactory != null)
{
IList <CoreLabel> lexList = tokFactory.GetTokenizer(new StringReader(word)).Tokenize();
if (lexList.Count == 0)
{
continue;
}
else
{
if (lexList.Count == 1)
{
word = lexList[0].Value();
}
else
{
if (lexList.Count > 1)
{
string secondWord = lexList[1].Value();
if (secondWord.Equals(DefaultSegMarker.ToString()))
{
// Special case for the null marker in the vocalized section
word = lexList[0].Value() + DefaultSegMarker.ToString();
}
else
{
System.Console.Error.Printf("%s: Raw token generates multiple segments: %s%n", typeof(ArabicDocumentReaderAndWriter).FullName, word);
word = lexList[0].Value();
}
}
}
}
}
if (!isStart)
{
System.Console.Out.Write(" ");
}
System.Console.Out.Write(word);
isStart = false;
}
System.Console.Out.WriteLine();
}
}