package org.daisy.pipeline.nlp;
   
   import java.util.Arrays;
   import java.util.HashSet;
   import java.util.Locale;
   import java.util.Set;
   import java.util.regex.Matcher;
   import java.util.regex.Pattern;
   
  public class LanguageUtils {
  	private static Pattern localePattern = Pattern
  	        .compile("(\\p{Alpha}{2})(?:[-_](\\p{Alpha}{2}))?(?:[-_](\\p{Alnum}{1,8}))*");
  
  	public static Locale stringToLanguage(String lang) {
  		//TODO: in Java7 we would use:
  		//return Locale.forLanguageTag(lang)
  		//=> this works with BCP47 tags, and should work with old tags from RFC 3066
  
  		Locale locale = null;
  		if (lang != null) {
  			Matcher m = localePattern.matcher(lang.toLowerCase());
  			if (m.matches()) {
  				locale = new Locale(m.group(1), m.group(2) != null ? m.group(2) : "");
  			}
  		}
  		return locale;
  	}
  
  	//lowercase ISO 639-2/T language codes (www.loc.gov/standards/iso639-2/php/English_list.php)
  	private static Set<String> RightToLeft = new HashSet<String>(Arrays.asList("yid", "ara",
  	        "jpr", "per", "fas", "peo", "heb"));
  
  	public static boolean isRightToLeft(Locale l) {
  		return RightToLeft.contains(l.getISO3Language());
  	}
  
  	// one can customize the punctuation/space symbols depending on the
  	// given language.
  
  	public static String getFullStopSymbol(Locale lang) {
  		return "! ";
  	}
  
  	public static String getCommaSymbol(Locale lang) {
  		return ", ";
  	}
  
  	public static String getWhiteSpaceSymbol(Locale lang) {
  		return " ";
  	}
  }