resources.tokeniser.norm-vowels.jape Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of lang-danish Show documentation

Support for processing Danish documents

The newest version!

// Leon Derczynski
// $id$


Phase: normvowels
Input: Token
Options: control = appelt

// map ae oe and aa to token strings
// store origString as well as string

Rule: Normalise
({Token}):norm
-->
{
  // convert the text to replace Danish vowels with diphthong versions
  AnnotationSet tok = bindings.get("norm");
  Annotation tokAnn = tok.iterator().next();

  FeatureMap tokfeats = (FeatureMap)((SimpleFeatureMapImpl)tokAnn.getFeatures()).clone();

  // preserve original string
  String origString = (String)tokfeats.get("string");
  tokfeats.put("origString", origString);


  String normString = (String)tokfeats.get("string");

  normString = normString.replaceAll("å", "aa");
  normString = normString.replaceAll("æ", "ae");
  normString = normString.replaceAll("ø", "oe");
  normString = normString.replaceAll("Å", "Aa");
  normString = normString.replaceAll("Æ", "Ae");
  normString = normString.replaceAll("Ø", "Oe");

  normString = normString.replaceAll("Kj", "K");

  tokfeats.put("normString", normString);

  // save updated feature map
  tokAnn.setFeatures(tokfeats);

}