package org.thdl.tib.dictionary ;

import java.lang.* ;
import java.util.regex.Pattern ;
import java.util.regex.Matcher ;
import java.util.Vector ;
import java.util.Enumeration ;

public class StandardPronounciationEngine
{
	public StandardPronounciationEngine () throws Exception
	{
		setRules () ;
		lastWordOriginal = "" ;
	}

	Rule [] rules ;

	protected int stateFlag ;
	protected int returnFlag ;
	protected String lastWordOriginal ;

	class Rule
	{
		/**
		 * conditions
		 */
		public static final int STARTS_WITH = 0 ;
		public static final int ENDS_WITH = 1 ;
		public static final int ENDS_WITH_FOLLOWS_VOWEL = 2 ;
		public static final int ENDS_WITH_FOLLOWS_CONSONANT = 3 ;
		public static final int STARTS_WITH_BORDERS_VOWEL = 4 ;
		public static final int STARTS_WITH_BORDERS_CONSONANT = 5 ;
		public static final int EQUALS = 6 ;
		public static final int CONTAINS = 7 ;
		public static final int STARTS_WITH_BORDERS_CONSONANTS = 8 ;

		/**
		 * actions
		 */
		public static final int REPLACE_MATCH = 0 ;
		public static final int SET_FLAG = 1 ;
		public static final int REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL = 2 ;
		public static final int REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT = 3 ;
		public static final int REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICELESS_CONSONANT = 4 ;
		public static final int SET_RETURN_FLAG = 5 ;

		/**
		 * flags
		 */
		public static final int FLAG_ENDS_WITH_VOWEL = 1 ;
		public static final int FLAG_ENDS_WITH_CONSONANT_VOICED = 2 ;
		public static final int FLAG_ENDS_WITH_CONSONANT_VOICELESS = 4 ;

		/**
		 * return flags
		 */
		public static final int DO_NOTHING = 0 ;
		public static final int DROP_SUFFIX_AND_NASALIZE = 1 ;

		private int condition ;
		private String condArg ;
		private int action ;
		private Object actionArg ;

		private Pattern pattern ;

		private static final String vowelSet = "aeiou" ;
		private static final String consonantSet = "bBcCdDfgGhjkKlmnNpPrsStTvwXzZ" ;

		/**
		 * constructior
		 */
		Rule ( int condition, String condArg, int action, Object actionArg ) throws Exception
		{
			this.condition = condition ;
			this.action = action ;
			this.condArg = condArg ;
			this.actionArg = actionArg ;

			String patStr = "" ;

			switch ( condition )
			{
				case CONTAINS :
					patStr = condArg ;
					break ;
				case EQUALS :
					patStr = "^" + condArg + "$" ;
					break ;
				case STARTS_WITH :
					patStr = "^" + condArg ;
					break ;
				case ENDS_WITH :
					patStr = condArg + "$" ;
					break ;
				case ENDS_WITH_FOLLOWS_VOWEL :
					patStr = "([" + vowelSet + "]{1,2})" + condArg + "$" ;
					if ( REPLACE_MATCH == action || 
						REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL == action || 
						REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT == action )
						this.actionArg = "$1" + (String)actionArg ;
					break ;
				case ENDS_WITH_FOLLOWS_CONSONANT :
					patStr = "([" + consonantSet + "])" + condArg + "$" ;
					if ( REPLACE_MATCH == action || 
						REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL == action || 
						REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT == action )
						this.actionArg = "$1" + (String)actionArg ;
					break ;
				case STARTS_WITH_BORDERS_VOWEL :
					patStr = "^" + condArg + "([" + vowelSet + "]{1,2})" ;
					if ( REPLACE_MATCH == action || 
						REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL == action || 
						REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT == action )
						this.actionArg = (String)actionArg + "$1" ;
					break ;
				case STARTS_WITH_BORDERS_CONSONANT:
					patStr = "^" + condArg + "(([" + consonantSet + "]))" ;
					if ( REPLACE_MATCH == action || 
						REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL == action || 
						REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT == action )
						this.actionArg = (String)actionArg + "$1" ;
					break ;
				case STARTS_WITH_BORDERS_CONSONANTS:
					patStr = "^" + condArg + "([" + consonantSet + "]{2})" ;
					if ( REPLACE_MATCH == action || 
						REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL == action || 
						REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT == action )
						this.actionArg = (String)actionArg + "$1" ;
					break ;
				default:
					throw new Exception ( "Invalid condition for a rule." ) ;
			}

			pattern = Pattern.compile ( patStr ) ;
		}

		/**
		 * property access
		 */
		public Object getActionArg ()
		{
			return actionArg ;
		}

		public String getConditionArg ()
		{
			return condArg ;
		}

		public int getCondition () 
		{
			return condition ;
		}

		public int getAction () 
		{
			return action ;
		}

		public Pattern getPattern ()
		{
			return pattern ;
		}
	}

	/**
	 * setRules
	 */
	protected void setRules () throws Exception
	{
		//
		// based on http://www.thdl.org/xml/showEssay.php?xml=/collections/langling/THDL_phonetics.xml&l=d1e671
		//
		Rule [] thdlRules = 
		{
			//
			// 6. When ba and bo appear as the final syllable of a word, they are transcribed as "wa" and "wo," respectively. 
			// This also includes ba'i ( > wé, about which see rule 16 below) and bar ( > war) as final syllables, although the latter is 
			// more evident in literary forms
			//
			new Rule ( Rule.EQUALS, "ba", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL, "wa" ), 
			new Rule ( Rule.EQUALS, "ba", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT, "wa" ), 
			new Rule ( Rule.EQUALS, "ba", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICELESS_CONSONANT, "pa" ), 
			new Rule ( Rule.EQUALS, "bo", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL, "wo" ), 
			new Rule ( Rule.EQUALS, "bo", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT, "wo" ), 
			new Rule ( Rule.EQUALS, "bo", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICELESS_CONSONANT, "po" ), 
			new Rule ( Rule.EQUALS, "bar", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL, "war" ), 
			new Rule ( Rule.EQUALS, "bar", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT, "war" ), 
			new Rule ( Rule.EQUALS, "bar", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICELESS_CONSONANT, "par" ), 
			new Rule ( Rule.EQUALS, "bo", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL, "wo" ), 
			new Rule ( Rule.EQUALS, "bo", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT, "wo" ), 
			new Rule ( Rule.EQUALS, "bo", Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICELESS_CONSONANT, "po" ), 

			//
			// 7. The consonant clusters py, phy and by are transcribed respectively as "ch," "ch," and "j." 
			//
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "py", Rule.REPLACE_MATCH, "c" ), 
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "Py", Rule.REPLACE_MATCH, "c" ), 
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "by", Rule.REPLACE_MATCH, "j" ) ,

			//
			// 8. The consonant cluster my is transcribed as "ny." 
			//
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "my", Rule.REPLACE_MATCH, "ny" ) ,

			//
			// 13. When the second syllable of a word begins with the prefix achung ('), nasalization occurs 
			// A. An 'n' is inserted after the first syllable, and the suffix of the first syllable (if there is one) is elided
			// B. If the root letter of the second syllable is pha or ba, an 'm' is inserted after the first syllable, 
			// and the suffix of the first syllable (if there is one) is elided
			new Rule ( Rule.STARTS_WITH, "'", Rule.SET_RETURN_FLAG, new Integer ( Rule.DROP_SUFFIX_AND_NASALIZE ) ),

			//
			// dirty workaround
			//
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANTS, "g", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANTS, "d", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANTS, "b", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANTS, "m", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANTS, "'", Rule.REPLACE_MATCH, "" ),

			//
			// 9. Consonant clusters with r subscripts (which are pronounced as retroflexes) are transcribed with an "r." 
			//
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "kr", Rule.REPLACE_MATCH, "tr" ) ,
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "Kr", Rule.REPLACE_MATCH, "tr" ), 
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "gr", Rule.REPLACE_MATCH, "dr" ) ,
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "nr", Rule.REPLACE_MATCH, "n" ) ,
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "pr", Rule.REPLACE_MATCH, "tr" ) ,
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "Pr", Rule.REPLACE_MATCH, "tr" ), 
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "br", Rule.REPLACE_MATCH, "dr" ) ,

			//
			// For all other consonant clusters in which the r subscript is not pronounced, 
			// such as mr, sr, and so forth, THDL Simplified Phonetics simply drops the "r" 
			// in accordance with the general principle
			//
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "mr", Rule.REPLACE_MATCH, "m" ) ,
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "Sr", Rule.REPLACE_MATCH, "S" ), 
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "sr", Rule.REPLACE_MATCH, "s" ) ,

			//
			// 10. Consonant clusters containing a subscript la are transcribed as "l" 
			// with the exception of zl, which is transcribed as "d." 
			//
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "kl", Rule.REPLACE_MATCH, "l" ) ,
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "gl", Rule.REPLACE_MATCH, "l" ) ,
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "bl", Rule.REPLACE_MATCH, "l" ) ,
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "rl", Rule.REPLACE_MATCH, "l" ) ,
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "zl", Rule.REPLACE_MATCH, "d" ) ,
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "sl", Rule.REPLACE_MATCH, "l" ) ,

			//
			// 11. Consonant clusters with an l superscript and h root letter are rendered "lh." 
			//
			new Rule ( Rule.STARTS_WITH, "lh", Rule.REPLACE_MATCH, "hl" ),
		
			//
			// 12. Consonant clusters with a d prefix and b root letter undergo transformations in the following way, 
			// depending on whether the consonant cluster includes the subscripts y or r:
			//
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "db", Rule.REPLACE_MATCH, "w" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "dby", Rule.REPLACE_MATCH, "y" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_VOWEL, "dbr", Rule.REPLACE_MATCH, "r" ),

			// !!!!! TODO !!!!!
			// Note: there are some exceptions to the form the nasalization takes:
			// skyabs 'gro > kyamdro 
			// rten 'brel > temdrel 
			// lam 'bras > lamdré 

			//
			// http://www.thdl.org/xml/showEssay.php?xml=/collections/langling/THDL_phonetics.xml&l=d1e294
			// The THDL Simplified Phonetic system, in contrast to Wylie, drops all Tibetan letters not 
			// pronounced in a given syllable. This includes the superscribed consonants r, l, and s ;
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANT, "r", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANT, "l", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANT, "s", Rule.REPLACE_MATCH, "" ),
 
			//
			// .... the prefixes g, d, b, m, and ' ; 
			//
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANT, "g", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANT, "d", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANT, "b", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANT, "m", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.STARTS_WITH_BORDERS_CONSONANT, "'", Rule.REPLACE_MATCH, "" ),

			// the suffixes d , ' , and s ; ( THEY ARE HANDLED ELSEWHERE )

			//
			// .... and the post-suffixes s and d.
			//
			new Rule ( Rule.ENDS_WITH_FOLLOWS_CONSONANT, "s", Rule.REPLACE_MATCH, "" ),
			new Rule ( Rule.ENDS_WITH_FOLLOWS_CONSONANT, "d", Rule.REPLACE_MATCH, "" ),

			//
			// 15. When two of the same vowels are connected by an achung, they are transcribed by dropping the achung and 
			// combining the two vowels into one
			//
			new Rule ( Rule.CONTAINS, "a'a", Rule.REPLACE_MATCH, "a" ),
			new Rule ( Rule.CONTAINS, "e'e", Rule.REPLACE_MATCH, "e" ),
			new Rule ( Rule.CONTAINS, "i'i", Rule.REPLACE_MATCH, "i" ),
			new Rule ( Rule.CONTAINS, "o'o", Rule.REPLACE_MATCH, "o" ),
			new Rule ( Rule.CONTAINS, "u'u", Rule.REPLACE_MATCH, "u" ),

			new Rule ( Rule.ENDS_WITH, "as", Rule.REPLACE_MATCH, "é" ) ,
			new Rule ( Rule.ENDS_WITH, "ad", Rule.REPLACE_MATCH, "é" ) ,
			new Rule ( Rule.ENDS_WITH, "an", Rule.REPLACE_MATCH, "én" ) ,
			new Rule ( Rule.ENDS_WITH, "al", Rule.REPLACE_MATCH, "él" ) ,
			new Rule ( Rule.ENDS_WITH, "os", Rule.REPLACE_MATCH, "ö" ) ,
			new Rule ( Rule.ENDS_WITH, "od", Rule.REPLACE_MATCH, "ö" ) ,
			new Rule ( Rule.ENDS_WITH, "on", Rule.REPLACE_MATCH, "ön" ) ,
			new Rule ( Rule.ENDS_WITH, "ol", Rule.REPLACE_MATCH, "öl" ) ,
			new Rule ( Rule.ENDS_WITH, "u'", Rule.REPLACE_MATCH, "u" ) ,
			new Rule ( Rule.ENDS_WITH, "us", Rule.REPLACE_MATCH, "ü" ) ,
			new Rule ( Rule.ENDS_WITH, "ud", Rule.REPLACE_MATCH, "ü" ) ,
			new Rule ( Rule.ENDS_WITH, "un", Rule.REPLACE_MATCH, "ün" ) ,
			new Rule ( Rule.ENDS_WITH, "ul", Rule.REPLACE_MATCH, "ül" ) ,			
			new Rule ( Rule.ENDS_WITH, "es", Rule.REPLACE_MATCH, "e" ) ,
			new Rule ( Rule.ENDS_WITH, "ed", Rule.REPLACE_MATCH, "e" ) ,
			new Rule ( Rule.ENDS_WITH, "en", Rule.REPLACE_MATCH, "en" ) ,
			new Rule ( Rule.ENDS_WITH, "el", Rule.REPLACE_MATCH, "el" ) ,
			new Rule ( Rule.ENDS_WITH, "is", Rule.REPLACE_MATCH, "i" ) ,
			new Rule ( Rule.ENDS_WITH, "id", Rule.REPLACE_MATCH, "i" ) ,

			//
			//	5. The suffixes g and b are devoiced and rendered "k" and "p," respectively, 
			//  since this most closely approximates actual pronunciation. 
			//
			new Rule ( Rule.ENDS_WITH_FOLLOWS_VOWEL, "g", Rule.REPLACE_MATCH, "k" ),
			new Rule ( Rule.ENDS_WITH_FOLLOWS_VOWEL, "b", Rule.REPLACE_MATCH, "p" ),
		
			// these rule must be *last*
			new Rule ( Rule.ENDS_WITH, "[aeiou]", Rule.SET_FLAG, new Integer(Rule.FLAG_ENDS_WITH_VOWEL) ),
			new Rule ( Rule.ENDS_WITH, "(b|d|g|l|m|n|N|r)", Rule.SET_FLAG, new Integer ( Rule.FLAG_ENDS_WITH_CONSONANT_VOICED ) ),
			new Rule ( Rule.ENDS_WITH, "(k|p|s|t)", Rule.SET_FLAG, new Integer ( Rule.FLAG_ENDS_WITH_CONSONANT_VOICELESS ) ),

			new Rule ( Rule.ENDS_WITH, "a'i", Rule.REPLACE_MATCH, "e" ) ,
			new Rule ( Rule.ENDS_WITH, "e'i", Rule.REPLACE_MATCH, "e" ) ,
			new Rule ( Rule.ENDS_WITH, "o'i", Rule.REPLACE_MATCH, "ö" ) ,
			new Rule ( Rule.ENDS_WITH, "u'i", Rule.REPLACE_MATCH, "ü" ),

			//
			// 14. Multiple vowels that have discrete sounds and are connected by an achung (') are transcribed by dropping the achung
			// (at this point all achungs have been removed by previous rules)
			//
			new Rule ( Rule.CONTAINS, "'", Rule.REPLACE_MATCH, "" ),

			//new Rule ( Rule.CONTAINS, "X", Rule.REPLACE_MATCH, "ts" ),
			//new Rule ( Rule.CONTAINS, "T", Rule.REPLACE_MATCH, "t" ),
			//new Rule ( Rule.CONTAINS, "P", Rule.REPLACE_MATCH, "p" ),
		
		} ;

		rules = thdlRules ;
	}

	/**
	 * applyRule
	 */
	protected String applyRule ( Rule rule, String in  ) throws Exception
	{
		switch ( rule.getAction () )
		{
				//
				// Rule.REPLACE_MATCH - if text defined in condArg found, replace it with actionArg unconditionally
				//
			case Rule.REPLACE_MATCH :
				{
				Matcher matcher = rule.getPattern ().matcher ( in ) ;
				return matcher.replaceFirst ( (String)rule.getActionArg () ) ;
			}

				//
				// Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL - if text defined in condArg found, 
				// AND last syllable ends with a vowel - replace it with actionArg
				//
			case Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOWEL :
				if ( 0 != ( stateFlag & Rule.FLAG_ENDS_WITH_VOWEL ) )
				{
					Matcher matcher = rule.getPattern ().matcher ( in ) ;
					return matcher.replaceFirst ( (String)rule.getActionArg () ) ;
				}
				break ;

				//
				// Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT - if text defined in condArg found, 
				// AND last syllable ends with a voiced consonant - replace it with actionArg
				//
			case Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICED_CONSONANT :
				if ( 0 != ( stateFlag & Rule.FLAG_ENDS_WITH_CONSONANT_VOICED ) )
				{
					Matcher matcher = rule.getPattern ().matcher ( in ) ;
					return matcher.replaceFirst ( (String)rule.getActionArg () ) ;
				}
				break ;

				//
				// Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICELESS_CONSONANT - if text defined in condArg found, 
				// AND last syllable ends with a voiceless consonant - replace it with actionArg
				//
			case Rule.REPLACE_IF_LAST_SYLLABLE_ENDED_WITH_VOICELESS_CONSONANT :
				if ( 0 != ( stateFlag & Rule.FLAG_ENDS_WITH_CONSONANT_VOICELESS ) )
				{
					Matcher matcher = rule.getPattern ().matcher ( in ) ;
					return matcher.replaceFirst ( (String)rule.getActionArg () ) ;
				}
				break ;

				//
				// Rule.SET_FLAG - set the flag
				//
			case Rule.SET_FLAG :
				if ( rule.getPattern ().matcher ( in ).find () )
				{
					stateFlag = ((Integer) rule.getActionArg ()).intValue () ;
				}
				break ;

				//
				// Rule.SET_RETURN_FLAG - set the return flag
				//
			case Rule.SET_RETURN_FLAG :
				if ( rule.getPattern ().matcher ( in ).find () )
				{
					returnFlag = ((Integer) rule.getActionArg ()).intValue () ;
				}
				break ;
			
				//
				// shouldn't ever happen
				//
			default:
				throw new Exception ( "Invalid action." ) ;
		}

		return in ;
	}

	/**
	 * processWord
	 */
	protected String processWord ( String in ) throws Exception
	{
		returnFlag = Rule.DO_NOTHING ;
		lastWordOriginal = in ;

		//
		// run the word through all rules
		//
		for ( int i = 0; i < rules.length; i++ )
		{
			in = applyRule ( rules [i], in ) ;
		}

		return in ;
	}

	private static boolean isVowel ( char c )
	{
		return ( -1 != "aeiouéöü".indexOf ( c ) ) ;
	}

	private static String dropSuffix ( String in )
	{
		while ( !isVowel ( in.charAt ( in.length () - 1 ) ) )
			in = in.substring ( 0, in.length () - 1 ) ;

		return in ;
	}

	private String nasalize ( String in, String followingSyllable )
	{
		if ( followingSyllable.startsWith ( "'by" ) || followingSyllable.startsWith ( "'br" ) )
			in = in + "n" ;
		else if ( followingSyllable.startsWith ( "'b" ) || lastWordOriginal.startsWith ( "'ph" ) )
			in = in + "m" ;
		else
			in = in + "n" ;

		return in ;
	}

	public String processWylie ( String in ) throws Exception
	{
		return process ( eliminateDigraphs ( in ) ) ;
	}

	public static String eliminateDigraphs ( String in ) 
	{
		//
		// replace the roman digraphs
		//
		in = in.replaceAll ( "bh", "B" ) ;
		in = in.replaceAll ( "ch", "C" ) ;
		in = in.replaceAll ( "dz", "D" ) ;
		in = in.replaceAll ( "kh", "K" ) ;
		in = in.replaceAll ( "ng", "N" ) ;
		in = in.replaceAll ( "ph", "P" ) ;
		in = in.replaceAll ( "sh", "S" ) ;
		in = in.replaceAll ( "th", "T" ) ;
		in = in.replaceAll ( "ts", "X" ) ;
		in = in.replaceAll ( "tsh", "Q" ) ;
		in = in.replaceAll ( "zh", "Z" ) ;
		
		return in ;
	}

	/**
	 * process
	 */
	public String process ( String in ) throws Exception
	{
		stateFlag = 0 ;

		String out = "" ;

		in = in.trim () ;
		String [] matchWords = in.split ( "[^a-z'A-Z]+" ) ;

		for ( int i = 0; i < matchWords.length; i++ )
		{
			String newWord = processWord ( matchWords [i] ) ;
			if ( i > 0 && Rule.DROP_SUFFIX_AND_NASALIZE == returnFlag )
			{				
				out = dropSuffix ( out ) ;
				out = nasalize ( out, newWord ) ;
			}
			
			if ( out.length () > 0 )
				out += " " ;

			out += newWord ;
		}

		return out ;
	}
}