Mercurial > hg > mpdl-group
view software/mpdl-services/mpiwg-mpdl-lt/bin/de/mpg/mpiwg/berlin/mpdl/lt/text/norm/lang/MpdlNormalizerLexDE.lex @ 25:e9fe3186670c default tip
letzter Stand eingecheckt
author | Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de> |
---|---|
date | Tue, 21 May 2013 10:19:32 +0200 |
parents | e845310098ba |
children |
line wrap: on
line source
/* * Normalization rules for German text * [this is a JFlex specification] * * Wolfgang Schmidle * version 2011-07-12 * */ package de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang; %% %public %class MpdlNormalizerLexDE %type java.lang.String %unicode // German: de, deu, ger %states DISP, DICT, SEARCH %state CELEX, GRIMM %{ private String original = ""; private String normalized = ""; private int problem = 0; private void add (String norm) { original += yytext(); normalized += norm; } private static final String LB = "[\u002d\u00ad] "; %} hyphen = [-\u{00ad}] // hyphen and soft hyphen LB = {hyphen} \u0020 // lb = ({hyphen} \u0020)? END = \n Alphabet = [abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ] %% // jump over empty xml elements "<"[^><]+"/>" { add(yytext()); } "-<"[^><]+"/>" { add(yytext()); } "<"[^><]+"></"[^><]+">" { add(yytext()); } "-<"[^><]+"></"[^><]+">" { add(yytext()); } ſ { add("s"); } // Fraktur <DISP, DICT, SEARCH, GRIMM> { uͦ {add("u"); } aͤ {add("ä"); } oͤ {add("ö"); } uͤ {add("ü"); } } <CELEX> { // normalize ä ö ü ß only for Celex! ä | Ä | aͤ { add("ae"); } ö | Ö | oͤ { add("oe"); } ü | Ü | uͤ { add("ue"); } uͦ {add("u"); } ß { add("ss"); } {Alphabet} { add(yytext()); } . { problem = 1; add(yytext()); } } <GRIMM> { ß { add("sz"); } } // default @ { problem = 1; add(yytext()); } {LB} { add(yytext()); } . { add(yytext()); } <DISP> { {END} { switch (problem) { case 1: return original; default: return normalized; } } } <DICT, CELEX, GRIMM> { {END} { switch (problem) { case 1: return ""; default: return normalized.replaceAll(LB, ""); } } } <SEARCH> { {END} { switch (problem) { case 1: return original; default: return normalized.replaceAll(LB, "").toLowerCase(); } } } /* Annahmen: - die Routine wird wortweise aufgerufen, mit einem \n am Ende des Strings - Zeilenumbrüche innerhalb des Wortes sind als hyphen/soft hyphen plus space markiert TO DO: DE: Trennung von Deutsch und Fraktur? DE: Celex: hyphens weg? */