resources.tokeniser.faste-ordbindelser.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!
Phase: fasteordbindelser
Input: Token SpaceToken
Options: control = appelt

// From "Vejledning til det danske morfosyntaktisk taggede PAROLE-korpus"
// flerordsforbindelser (multi-word connections)

// 8.5.3 Faste ordforbindelser

Rule: faste
(
 {Token.string == "a"} {SpaceToken} {Token.string == "la"} {SpaceToken} {Token.string == "carte"} |
 {Token.string == "af"} {SpaceToken} {Token.string == "og"} {SpaceToken} {Token.string == "til"} |
 {Token.string == "af"} {SpaceToken} {Token.string == "sted"} |
 {Token.string == "all"} {SpaceToken} {Token.string == "right"} |
 {Token.string == "alt"} {SpaceToken} {Token.string == "imens"} |
 {Token.string == "blandt"} {SpaceToken} {Token.string == "andet"} |
 {Token.string == "blandt"} {SpaceToken} {Token.string == "andre"} |
 {Token.string == "bortset"} {SpaceToken} {Token.string == "fra"} |
 {Token.string == "café"} {SpaceToken} {Token.string == "au"} {SpaceToken} {Token.string == "lait"} |
 {Token.string == "en"} {SpaceToken} {Token.string == "bloc"} |
 {Token.string == "en"} {SpaceToken} {Token.string == "suite"} |
 {Token.string == "for"} {SpaceToken} {Token.string == "nylig"} |
 {Token.string == "for"} {SpaceToken} {Token.string == "tiden"} |
 {Token.string == "for"} {SpaceToken} {Token.string == "øvrigt"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "aften"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "alt"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "bunkevis"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "dag"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "det"} {SpaceToken} {Token.string == "hele"} {SpaceToken} {Token.string == "taget"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "eftermiddag"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "fjor"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "forkøbet"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "forvejen"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "gang"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "går"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "hvert"} {SpaceToken} {Token.string == "fald"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "læssevis "} |
 {Token.string == "i"} {SpaceToken} {Token.string == "morgen"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "månedsvis"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "nat"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "timevis"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "tusindvis"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "ugevis"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "øvrigt"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "år"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "årevis"} |
 {Token.string == "ikke"} {SpaceToken} {Token.string == "desto"} {SpaceToken} {Token.string == "mindre"} |
 {Token.string == "med"} {SpaceToken} {Token.string == "hensyn"} {SpaceToken} {Token.string == "til"} |
 {Token.string == "mere"} {SpaceToken} {Token.string == "eller"} {SpaceToken} {Token.string == "mindre"} |
 {Token.string == "om"} {SpaceToken} {Token.string == "bord"} |
 {Token.string == "om"} {SpaceToken} {Token.string == "end"} |
 {Token.string == "om"} {SpaceToken} {Token.string == "muligt"} |
 {Token.string == "over"} {SpaceToken} {Token.string == "bord"} |
 {Token.string == "på"} {SpaceToken} {Token.string == "grund"} {SpaceToken} {Token.string == "af"} |
 {Token.string == "på"} {SpaceToken} {Token.string == "ny"} |
 {Token.string == "selv"} {SpaceToken} {Token.string == "om"} |
 {Token.string == "simpelt"} {SpaceToken} {Token.string == "hen"} |
 {Token.string == "som"} {SpaceToken} {Token.string == "om"} |
 {Token.string == "stort"} {SpaceToken} {Token.string == "set"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "sidst"}
):match
-->
:match {
   long start = matchAnnots.firstNode().getOffset();
   long end = matchAnnots.lastNode().getOffset();

   // add spanning annotation
   FeatureMap stringfeats = Factory.newFeatureMap();
   stringfeats.put("rule", "faste");
   stringfeats.put("length", (int) end - start);
   String fastestring = gate.Utils.cleanStringFor(doc, matchAnnots);
   fastestring = fastestring.replace(" ", "_");
   stringfeats.put("string", fastestring);

   // do removal of old tokens *before* adding its replacement, else we'll remove the new replacement

   AnnotationSet tokens = gate.Utils.getContainedAnnotations(inputAS, matchAnnots, "Token");
   inputAS.removeAll(tokens);
   AnnotationSet spacetokens = gate.Utils.getContainedAnnotations(inputAS, matchAnnots, "SpaceToken");
   inputAS.removeAll(spacetokens);

   try {
    outputAS.add(start, end, "Token", stringfeats);
   }
   catch (InvalidOffsetException e) {
    e.printStackTrace();
   }

}


// 8.5.4 Fossilerede dativer/genitiver 

Rule: fossil
(
 {Token.string == "af"} {SpaceToken} {Token.string == "syne"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "aftes"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "eftermiddags"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "fredags"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "går"} {SpaceToken} {Token.string == "aftes"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "går"} {SpaceToken} {Token.string == "eftermiddags"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "går"} {SpaceToken} {Token.string == "morges"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "hænde"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "live"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "lørdags"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "mandags"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "onsdags"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "sinde"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "søndags"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "tide"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "tirsdags"} |
 {Token.string == "i"} {SpaceToken} {Token.string == "torsdags"} |
 {Token.string == "med"} {SpaceToken} {Token.string == "rette"} |
 {Token.string == "nu"} {SpaceToken} {Token.string == "til"} {SpaceToken} {Token.string == "dags"} |
 {Token.string == "på"} {SpaceToken} {Token.string == "fode"} |
 {Token.string == "på"} {SpaceToken} {Token.string == "tide"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "bords"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "bunds"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "døde"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "fods"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "fulde"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "gode"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "huse"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "lands"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "orde"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "rette"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "rors"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "stede"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "tops"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "tåls"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "veje"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "vejrs"} |
 {Token.string == "til"} {SpaceToken} {Token.string == "vægs"}
 
):match
-->
:match {
   long start = matchAnnots.firstNode().getOffset();
   long end = matchAnnots.lastNode().getOffset();

   // add spanning annotation
   FeatureMap stringfeats = Factory.newFeatureMap();
   stringfeats.put("rule", "fossil");
   stringfeats.put("length", (int) end - start);
   String fastestring = gate.Utils.cleanStringFor(doc, matchAnnots);
   fastestring = fastestring.replace(" ", "_");
   stringfeats.put("string", fastestring);

   // do removal of old tokens *before* adding its replacement, else we'll remove the new replacement

   AnnotationSet tokens = gate.Utils.getContainedAnnotations(inputAS, matchAnnots, "Token");
   inputAS.removeAll(tokens);
   AnnotationSet spacetokens = gate.Utils.getContainedAnnotations(inputAS, matchAnnots, "SpaceToken");
   inputAS.removeAll(spacetokens);

   try {
    outputAS.add(start, end, "Token", stringfeats);
   }
   catch (InvalidOffsetException e) {
    e.printStackTrace();
   }

}