%% %{ /* * Unicode Romanization to Beta Code back-conversion, * v. 1.0, Malcolm D. Hyman, 2003-06-24 * [this is a jflex specification] */ %} %class BetaCodeTranslitLex %public %implements LexS %yylexthrow LexException %type java.lang.String %unicode %char %state NONINITIAL %% { [aeo][iu] { yybegin(NONINITIAL); return yytext() + ")"; } [AEO][iu] { yybegin(NONINITIAL); return "*" + yytext().toLowerCase() + ")"; } [aeiou] { yybegin(NONINITIAL); return yytext() + ")"; } [AEIOU] { yybegin(NONINITIAL); return "*)" + yytext(); } "\u0113" { yybegin(NONINITIAL); return "h)"; } "\u0112" { yybegin(NONINITIAL); return "*)h"; } "\u014d" { yybegin(NONINITIAL); return "w)"; } "\u014c" { yybegin(NONINITIAL); return "*w)"; } "\u0101" { yybegin(NONINITIAL); return "a)"; } "\u0100" { yybegin(NONINITIAL); return "*a)"; } "\u0101i" { yybegin(NONINITIAL); return "a)|"; } "\u0100i" { yybegin(NONINITIAL); return "*a)|"; } "h\u0101i" { yybegin(NONINITIAL); return "a(|"; } "h\u0100i" { yybegin(NONINITIAL); return "*(a|"; } "h\u0101\u0301i" { yybegin(NONINITIAL); return "a(/|"; } "h\u0100\u0301i" { yybegin(NONINITIAL); return "*(/a|"; } "\u0101\u0300i" { yybegin(NONINITIAL); return "a)\\|"; } "\u0100\u0300i" { yybegin(NONINITIAL); return "*)\\a|"; } "\u00e2i" { yybegin(NONINITIAL); return "a)=|"; } "\u00c2i" { yybegin(NONINITIAL); return "*)=a|"; } "h\u0101\u0301i" { yybegin(NONINITIAL); return "a(/|"; } "h\u0100\u0301i" { yybegin(NONINITIAL); return "*(/a|"; } "h\u0101\u0300i" { yybegin(NONINITIAL); return "a(\\|"; } "h\u0100\u0300i" { yybegin(NONINITIAL); return "*(\\a|"; } "h\u00e2i" { yybegin(NONINITIAL); return "a(=|"; } "h\u00c2i" { yybegin(NONINITIAL); return "*(=a|"; } "\u0113i" { yybegin(NONINITIAL); return "h)|"; } "\u0112i" { yybegin(NONINITIAL); return "*h)|"; } "h\u0113i" { yybegin(NONINITIAL); return "h(|"; } "h\u0112i" { yybegin(NONINITIAL); return "*(h|"; } "h\u0113\u0301i" { yybegin(NONINITIAL); return "h(/|"; } "h\u0112\u0301i" { yybegin(NONINITIAL); return "*(/h|"; } "\u0113\u0300i" { yybegin(NONINITIAL); return "h)\\|"; } "\u0112\u0300i" { yybegin(NONINITIAL); return "*)\\h|"; } "\u00e2i" { yybegin(NONINITIAL); return "h)=|"; } "\u00c2i" { yybegin(NONINITIAL); return "*)=h|"; } "h\u0113\u0301i" { yybegin(NONINITIAL); return "h(/|"; } "h\u0112\u0301i" { yybegin(NONINITIAL); return "*(/h|"; } "h\u0113\u0300i" { yybegin(NONINITIAL); return "h(\\|"; } "h\u0112\u0300i" { yybegin(NONINITIAL); return "*(\\h|"; } "h\u00e2i" { yybegin(NONINITIAL); return "h(=|"; } "h\u00c2i" { yybegin(NONINITIAL); return "*(=h|"; } "\u014di" { yybegin(NONINITIAL); return "w)|"; } "\u014ci" { yybegin(NONINITIAL); return "*w)|"; } "h\u014di" { yybegin(NONINITIAL); return "w(|"; } "h\u014ci" { yybegin(NONINITIAL); return "*(w|"; } "h\u014d\u0301i" { yybegin(NONINITIAL); return "w(/|"; } "h\u014c\u0301i" { yybegin(NONINITIAL); return "*(/w|"; } "\u014d\u0300i" { yybegin(NONINITIAL); return "w)\\|"; } "\u014c\u0300i" { yybegin(NONINITIAL); return "*)\\w|"; } "\u00e2i" { yybegin(NONINITIAL); return "w)=|"; } "\u00c2i" { yybegin(NONINITIAL); return "*)=w|"; } "h\u014d\u0301i" { yybegin(NONINITIAL); return "w(/|"; } "h\u014c\u0301i" { yybegin(NONINITIAL); return "*(/w|"; } "h\u014d\u0300i" { yybegin(NONINITIAL); return "w(\\|"; } "h\u014c\u0300i" { yybegin(NONINITIAL); return "*(\\w|"; } "h\u00e2i" { yybegin(NONINITIAL); return "w(=|"; } "h\u00c2i" { yybegin(NONINITIAL); return "*(=w|"; } "ha" { yybegin(NONINITIAL); return "a("; } "Ha" { yybegin(NONINITIAL); return "*(a"; } "he" { yybegin(NONINITIAL); return "e("; } "He" { yybegin(NONINITIAL); return "*(e"; } "h\u0113" { yybegin(NONINITIAL); return "h("; } "H\u0113" { yybegin(NONINITIAL); return "*(h"; } "hi" { yybegin(NONINITIAL); return "i("; } "Hi" { yybegin(NONINITIAL); return "*(i"; } "ho" { yybegin(NONINITIAL); return "o("; } "Ho" { yybegin(NONINITIAL); return "*(o"; } "hu" { yybegin(NONINITIAL); return "u("; } "Hu" { yybegin(NONINITIAL); return "*(u"; } "h\u014d" { yybegin(NONINITIAL); return "w("; } "H\u014d" { yybegin(NONINITIAL); return "*(w"; } "\u00e1" { yybegin(NONINITIAL); return "a)/"; } "\u00c1" { yybegin(NONINITIAL); return "*)/a"; } "\u00e9" { yybegin(NONINITIAL); return "e)/"; } "\u00c9" { yybegin(NONINITIAL); return "*)/e"; } "\u0113\u0301" { yybegin(NONINITIAL); return "h)/"; } "\u0112\u0301" { yybegin(NONINITIAL); return "*)/h"; } "\u00ed" { yybegin(NONINITIAL); return "i)/"; } "\u00cd" { yybegin(NONINITIAL); return "*)/i"; } "\u00f3" { yybegin(NONINITIAL); return "o)/"; } "\u00d3" { yybegin(NONINITIAL); return "*)/o"; } "\u00fa" { yybegin(NONINITIAL); return "u)/"; } "\u00da" { yybegin(NONINITIAL); return "*)/u"; } "\u014d\u0301" { yybegin(NONINITIAL); return "w)/"; } "\u014c\u0301" { yybegin(NONINITIAL); return "*)/w"; } "\u00e0" { yybegin(NONINITIAL); return "a)\\"; } "\u00c0" { yybegin(NONINITIAL); return "*)\\a"; } "\u00e8" { yybegin(NONINITIAL); return "e)\\"; } "\u00c8" { yybegin(NONINITIAL); return "*)\\e"; } "\u0113\u0300" { yybegin(NONINITIAL); return "h)\\"; } "\u0112\u0300" { yybegin(NONINITIAL); return "*)\\h"; } "\u00ec" { yybegin(NONINITIAL); return "i)\\"; } "\u00cc" { yybegin(NONINITIAL); return "*)\\i"; } "\u00f2" { yybegin(NONINITIAL); return "o)\\"; } "\u00d2" { yybegin(NONINITIAL); return "*)\\o"; } "\u00f9" { yybegin(NONINITIAL); return "u)\\"; } "\u00d9" { yybegin(NONINITIAL); return "*)\\u"; } "\u014d\u0301" { yybegin(NONINITIAL); return "w)\\"; } "\u014c\u0301" { yybegin(NONINITIAL); return "*)\\w"; } "\u00e2" { yybegin(NONINITIAL); return "a)="; } "\u00c2" { yybegin(NONINITIAL); return "*)=a"; } "\u00ea" { yybegin(NONINITIAL); return "h)="; } "\u00ca" { yybegin(NONINITIAL); return "*)=h"; } "\u00ee" { yybegin(NONINITIAL); return "i)="; } "\u00ce" { yybegin(NONINITIAL); return "*)=i"; } "\u00fb" { yybegin(NONINITIAL); return "u)="; } "\u00db" { yybegin(NONINITIAL); return "*)=u"; } "\u00f4" { yybegin(NONINITIAL); return "w)="; } "\u00d4" { yybegin(NONINITIAL); return "*)=w"; } "h\u00e1" { yybegin(NONINITIAL); return "a(/"; } "H\u00e1" { yybegin(NONINITIAL); return "*(/a"; } "h\u00e9" { yybegin(NONINITIAL); return "e(/"; } "H\u00e9" { yybegin(NONINITIAL); return "*(/e"; } "h\u0113\u0301" { yybegin(NONINITIAL); return "h(/"; } "H\u0113\u0301" { yybegin(NONINITIAL); return "*(/h"; } "h\u00ed" { yybegin(NONINITIAL); return "i(/"; } "H\u00ed" { yybegin(NONINITIAL); return "*(/i"; } "h\u00fe" { yybegin(NONINITIAL); return "o(/"; } "H\u00fe" { yybegin(NONINITIAL); return "*(/o"; } "h\u00fa" { yybegin(NONINITIAL); return "u(/"; } "H\u00fa" { yybegin(NONINITIAL); return "*(/u"; } "h\u014d\u0301" { yybegin(NONINITIAL); return "w(/"; } "H\u014d\u0301" { yybegin(NONINITIAL); return "*(/w"; } "h\u00e0" { yybegin(NONINITIAL); return "a(\\"; } "H\u00e0" { yybegin(NONINITIAL); return "*(\\a"; } "h\u00e8" { yybegin(NONINITIAL); return "e(\\"; } "H\u00e8" { yybegin(NONINITIAL); return "*(\\e"; } "h\u0113\u0300" { yybegin(NONINITIAL); return "h(\\"; } "H\u0113\u0300" { yybegin(NONINITIAL); return "*(\\h"; } "h\u00ec" { yybegin(NONINITIAL); return "i(\\"; } "H\u00ec" { yybegin(NONINITIAL); return "*(\\i"; } "h\u00f2" { yybegin(NONINITIAL); return "o(\\"; } "H\u00f2" { yybegin(NONINITIAL); return "*(\\o"; } "h\u00f9" { yybegin(NONINITIAL); return "u(\\"; } "H\u00f9" { yybegin(NONINITIAL); return "*(\\u"; } "h\u014d\u0301" { yybegin(NONINITIAL); return "w(\\"; } "H\u014d\u0301" { yybegin(NONINITIAL); return "*(\\w"; } "h\u00e2" { yybegin(NONINITIAL); return "a(="; } "H\u00e2" { yybegin(NONINITIAL); return "*(=a"; } "h\u00ea" { yybegin(NONINITIAL); return "h(="; } "H\u00ea" { yybegin(NONINITIAL); return "*(=h"; } "h\u00ee" { yybegin(NONINITIAL); return "i(="; } "H\u00ee" { yybegin(NONINITIAL); return "*(=i"; } "h\u00fb" { yybegin(NONINITIAL); return "u(="; } "H\u00fb" { yybegin(NONINITIAL); return "*(=u"; } "h\u00f4" { yybegin(NONINITIAL); return "w(="; } "H\u00f4" { yybegin(NONINITIAL); return "*(=w"; } "rh" { yybegin(NONINITIAL); return "r("; } "hai" { yybegin(NONINITIAL); return "ai("; } "Hai" { yybegin(NONINITIAL); return "*ai("; } "ha\u00ed" { yybegin(NONINITIAL); return "ai(/"; } "Ha\u00ed" { yybegin(NONINITIAL); return "*ai(/"; } "ha\u00ec" { yybegin(NONINITIAL); return "ai(\\"; } "Ha\u00ec" { yybegin(NONINITIAL); return "*ai(\\"; } "ha\u00ee" { yybegin(NONINITIAL); return "ai(="; } "Ha\u00ee" { yybegin(NONINITIAL); return "*ai(="; } "hei" { yybegin(NONINITIAL); return "ei("; } "Hei" { yybegin(NONINITIAL); return "*ei("; } "he\u00ed" { yybegin(NONINITIAL); return "ei(/"; } "He\u00ed" { yybegin(NONINITIAL); return "*ei(/"; } "he\u00ec" { yybegin(NONINITIAL); return "ei(\\"; } "He\u00ec" { yybegin(NONINITIAL); return "*ei(\\"; } "he\u00ee" { yybegin(NONINITIAL); return "ei(="; } "He\u00ee" { yybegin(NONINITIAL); return "*ei(="; } "hoi" { yybegin(NONINITIAL); return "oi("; } "Hoi" { yybegin(NONINITIAL); return "*oi("; } "ho\u00ed" { yybegin(NONINITIAL); return "oi(/"; } "Ho\u00ed" { yybegin(NONINITIAL); return "*oi(/"; } "ho\u00ec" { yybegin(NONINITIAL); return "oi(\\"; } "Ho\u00ec" { yybegin(NONINITIAL); return "*oi(\\"; } "ho\u00ee" { yybegin(NONINITIAL); return "oi(="; } "Ho\u00ee" { yybegin(NONINITIAL); return "*oi(="; } "hau" { yybegin(NONINITIAL); return "au("; } "Hau" { yybegin(NONINITIAL); return "*au("; } "ha\u00fa" { yybegin(NONINITIAL); return "au(/"; } "Ha\u00fa" { yybegin(NONINITIAL); return "*au(/"; } "ha\u00f9" { yybegin(NONINITIAL); return "au(\\"; } "Ha\u00f9" { yybegin(NONINITIAL); return "*au(\\"; } "ha\u00fb" { yybegin(NONINITIAL); return "au(="; } "Ha\u00fb" { yybegin(NONINITIAL); return "*au(="; } "heu" { yybegin(NONINITIAL); return "eu("; } "Heu" { yybegin(NONINITIAL); return "*eu("; } "he\u00fa" { yybegin(NONINITIAL); return "eu(/"; } "He\u00fa" { yybegin(NONINITIAL); return "*eu(/"; } "he\u00f9" { yybegin(NONINITIAL); return "eu(\\"; } "He\u00f9" { yybegin(NONINITIAL); return "*eu(\\"; } "he\u00fb" { yybegin(NONINITIAL); return "eu(="; } "He\u00fb" { yybegin(NONINITIAL); return "*eu(="; } "hou" { yybegin(NONINITIAL); return "ou("; } "Hou" { yybegin(NONINITIAL); return "*ou("; } "ho\u00fa" { yybegin(NONINITIAL); return "ou(/"; } "Ho\u00fa" { yybegin(NONINITIAL); return "*ou(/"; } "ho\u00f9" { yybegin(NONINITIAL); return "ou(\\"; } "Ho\u00f9" { yybegin(NONINITIAL); return "*ou(\\"; } "ho\u00fb" { yybegin(NONINITIAL); return "ou(="; } "Ho\u00fb" { yybegin(NONINITIAL); return "*ou(="; } "ai" { yybegin(NONINITIAL); return "ai)"; } "Ai" { yybegin(NONINITIAL); return "*ai)"; } "a\u00ed" { yybegin(NONINITIAL); return "ai)/"; } "A\u00ed" { yybegin(NONINITIAL); return "*ai)/"; } "a\u00ec" { yybegin(NONINITIAL); return "ai)\\"; } "A\u00ec" { yybegin(NONINITIAL); return "*ai)\\"; } "a\u00ee" { yybegin(NONINITIAL); return "ai)="; } "A\u00ee" { yybegin(NONINITIAL); return "*ai)="; } "ei" { yybegin(NONINITIAL); return "ei)"; } "Ei" { yybegin(NONINITIAL); return "*ei)"; } "e\u00ed" { yybegin(NONINITIAL); return "ei)/"; } "E\u00ed" { yybegin(NONINITIAL); return "*ei)/"; } "e\u00ec" { yybegin(NONINITIAL); return "ei)\\"; } "E\u00ec" { yybegin(NONINITIAL); return "*ei)\\"; } "e\u00ee" { yybegin(NONINITIAL); return "ei)="; } "E\u00ee" { yybegin(NONINITIAL); return "*ei)="; } "oi" { yybegin(NONINITIAL); return "oi)"; } "Oi" { yybegin(NONINITIAL); return "*oi)"; } "o\u00ed" { yybegin(NONINITIAL); return "oi)/"; } "O\u00ed" { yybegin(NONINITIAL); return "*oi)/"; } "o\u00ec" { yybegin(NONINITIAL); return "oi)\\"; } "O\u00ec" { yybegin(NONINITIAL); return "*oi)\\"; } "o\u00ee" { yybegin(NONINITIAL); return "oi)="; } "O\u00ee" { yybegin(NONINITIAL); return "*oi)="; } "au" { yybegin(NONINITIAL); return "au)"; } "Au" { yybegin(NONINITIAL); return "*au)"; } "a\u00fa" { yybegin(NONINITIAL); return "au)/"; } "A\u00fa" { yybegin(NONINITIAL); return "*au)/"; } "a\u00f9" { yybegin(NONINITIAL); return "au)\\"; } "A\u00f9" { yybegin(NONINITIAL); return "*au)\\"; } "a\u00fb" { yybegin(NONINITIAL); return "au)="; } "A\u00fb" { yybegin(NONINITIAL); return "*au)="; } "eu" { yybegin(NONINITIAL); return "eu)"; } "Eu" { yybegin(NONINITIAL); return "*eu)"; } "e\u00fa" { yybegin(NONINITIAL); return "eu)/"; } "E\u00fa" { yybegin(NONINITIAL); return "*eu)/"; } "e\u00f9" { yybegin(NONINITIAL); return "eu)\\"; } "E\u00f9" { yybegin(NONINITIAL); return "*eu)\\"; } "e\u00fb" { yybegin(NONINITIAL); return "eu)="; } "E\u00fb" { yybegin(NONINITIAL); return "*eu)="; } "ou" { yybegin(NONINITIAL); return "ou)"; } "Ou" { yybegin(NONINITIAL); return "*ou)"; } "o\u00fa" { yybegin(NONINITIAL); return "ou)/"; } "O\u00fa" { yybegin(NONINITIAL); return "*ou)/"; } "o\u00f9" { yybegin(NONINITIAL); return "ou)\\"; } "O\u00f9" { yybegin(NONINITIAL); return "*ou)\\"; } "o\u00fb" { yybegin(NONINITIAL); return "ou)="; } "O\u00fb" { yybegin(NONINITIAL); return "*ou)="; } } { [aeiou] { return yytext(); } [AEIOU] { return "*" + yytext(); } "\u0113" { return "h"; } "\u0112" { return "*h"; } "\u014d" { return "w"; } "\u014c" { return "*w"; } "\u0101" { return "a"; } "\u0100" { return "*a"; } "\u0101i" { return "a|"; } "\u0100i" { return "*a|"; } "\u0101\u0300i" { return "a\\|"; } "\u0100\u0300i" { return "*\\a|"; } "\u00e2i" { return "a=|"; } "\u00c2i" { return "*=a|"; } "\u0113i" { return "h|"; } "\u0112i" { return "*h|"; } "\u0113\u0300i" { return "h\\|"; } "\u0112\u0300i" { return "*\\h|"; } "\u00e2i" { return "h=|"; } "\u00c2i" { return "*=h|"; } "\u014di" { return "w|"; } "\u014ci" { return "*w|"; } "\u014d\u0300i" { return "w\\|"; } "\u014c\u0300i" { return "*\\w|"; } "\u00e2i" { return "w=|"; } "\u00c2i" { return "*=w|"; } "\u00e1" { return "a/"; } "\u00c1" { return "*/a"; } "\u00e9" { return "e/"; } "\u00c9" { return "*/e"; } "\u0113\u0301" { return "h/"; } "\u0112\u0301" { return "*/h"; } "\u00ed" { return "i/"; } "\u00cd" { return "*/i"; } "\u00f3" { return "o/"; } "\u00d3" { return "*/o"; } "\u00fa" { return "u/"; } "\u00da" { return "*/u"; } "\u014d\u0301" { return "w/"; } "\u014c\u0301" { return "*/w"; } "\u00e0" { return "a\\"; } "\u00c0" { return "*\\a"; } "\u00e8" { return "e\\"; } "\u00c8" { return "*\\e"; } "\u0113\u0300" { return "h\\"; } "\u0112\u0300" { return "*\\h"; } "\u00ec" { return "i\\"; } "\u00cc" { return "*\\i"; } "\u00f2" { return "o\\"; } "\u00d2" { return "*\\o"; } "\u00f9" { return "u\\"; } "\u00d9" { return "*\\u"; } "\u014d\u0301" { return "w\\"; } "\u014c\u0301" { return "*\\w"; } "\u00e2" { return "a="; } "\u00c2" { return "*=a"; } "\u00ea" { return "h="; } "\u00ca" { return "*=h"; } "\u00ee" { return "i="; } "\u00ce" { return "*=i"; } "\u00fb" { return "u="; } "\u00db" { return "*=u"; } "\u00f4" { return "w="; } "\u00d4" { return "*=w"; } "\u00ef" { return "i+"; } "\u00cf" { return "*i+"; } "\u00ef\u0301" { return "i/+"; } "\u00ef\u0300" { return "i\\+"; } "\u00ef\u0302" { return "i=+"; } "\u00fc" { return "u+"; } "\u00dc" { return "*u+"; } "\u00dc\u0301" { return "u/+"; } "\u00dc\u0300" { return "u\\+"; } "\u00dc\u0302" { return "u=+"; } } [bgdzklmnprst] { yybegin(NONINITIAL); return yytext(); } [BGDZKLMNPRST] { yybegin(NONINITIAL); return "*" + yytext().toLowerCase(); } [Tt]"h" { yybegin(NONINITIAL); return (Character.isUpperCase(yytext().charAt(0)) ? "*q" : "q"); } [Xx] { yybegin(NONINITIAL); return (Character.isUpperCase(yytext().charAt(0)) ? "*c" : "c"); } [Pp]"h" { yybegin(NONINITIAL); return (Character.isUpperCase(yytext().charAt(0)) ? "*f" : "f"); } [Kk]"h" { yybegin(NONINITIAL); return (Character.isUpperCase(yytext().charAt(0)) ? "*x" : "x"); } [Pp]"s" { yybegin(NONINITIAL); return (Character.isUpperCase(yytext().charAt(0)) ? "*y" : "y"); } " "|\t|\n { yybegin(YYINITIAL); return yytext(); } . { return yytext(); }