diff software/mpdl-services/mpiwg-mpdl-lt/src/de/mpg/mpiwg/berlin/mpdl/lt/text/transcode/Unicode2Betacode.lex @ 19:4a3641ae14d2

Erstellung
author Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
date Wed, 09 Nov 2011 15:32:05 +0100
parents
children
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/software/mpdl-services/mpiwg-mpdl-lt/src/de/mpg/mpiwg/berlin/mpdl/lt/text/transcode/Unicode2Betacode.lex	Wed Nov 09 15:32:05 2011 +0100
@@ -0,0 +1,319 @@
+package de.mpg.mpiwg.berlin.mpdl.lt.general;
+
+%%
+
+%class Unicode2BetacodeLex
+%public
+%type java.lang.String
+%unicode
+%%
+
+
+"<"[^>]\u+">" { return yytext(); }
+
+"H"    { return "*j"; }
+"h"     { return "j"; }
+"F"    { return "*v"; }
+"f"    { return "v"; }
+"\u03a3"    { return "*s"; }
+
+"." 	{ return "!"; }
+"\u00B7"   { return ":"; }  /* MPDL update  */
+
+"\u1F00"	{ return "a)"; }
+"\u1F01"	{ return "a("; }
+"\u1F02"	{ return "a)\\"; }
+"\u1F03"	{ return "a(\\"; }
+"\u1F04"	{ return "a)/"; }
+"\u1F05"	{ return "a(/"; }
+"\u1F06"	{ return "a)="; }
+"\u1F07"	{ return "a(="; }
+"\u1F08"	{ return "*)a"; }
+"\u1F09"	{ return "*(a"; }
+"\u1F0A"	{ return "*)\\a"; }
+"\u1F0B"	{ return "*(\\a"; }
+"\u1F0C"	{ return "*)/a"; }
+"\u1F0D"	{ return "*(/a"; }
+"\u1F0E"	{ return "*)=a"; }
+"\u1F0F"	{ return "*(=a"; }
+"\u1F10"	{ return "e)"; }
+"\u1F11"	{ return "e("; }
+"\u1F12"	{ return "e)\\"; }
+"\u1F13"	{ return "e(\\"; }
+"\u1F14"	{ return "e)/"; }
+"\u1F15"	{ return "e(/"; }
+"\u1F18"	{ return "*)e"; }
+"\u1F19"	{ return "*(e"; }
+"\u1F1A"	{ return "*)\\e"; }
+"\u1F1B"	{ return "*(\\e"; }
+"\u1F1C"	{ return "*)/e"; }
+"\u1F1D"	{ return "*(/e"; }
+"\u1F20"	{ return "h)"; }
+"\u1F21"	{ return "h("; }
+"\u1F22"	{ return "h)\\"; }
+"\u1F23"	{ return "h(\\"; }
+"\u1F24"	{ return "h)/"; }
+"\u1F25"	{ return "h(/"; }
+"\u1F26"	{ return "h)="; }
+"\u1F27"	{ return "h(="; }
+"\u1F28"	{ return "*)h"; }
+"\u1F29"	{ return "*(h"; }
+"\u1F2A"	{ return "*)\\h"; }
+"\u1F2B"	{ return "*(\\h"; }
+"\u1F2C"	{ return "*)/h"; }
+"\u1F2D"	{ return "*(/h"; }
+"\u1F2E"	{ return "*)=h"; }
+"\u1F2F"	{ return "*(=h"; }
+"\u1F30"	{ return "i)"; }
+"\u1F31"	{ return "i("; }
+"\u1F32"	{ return "i)\\"; }
+"\u1F33"	{ return "i(\\"; }
+"\u1F34"	{ return "i)/"; }
+"\u1F35"	{ return "i(/"; }
+"\u1F36"	{ return "i)="; }
+"\u1F37"	{ return "i(="; }
+"\u1F38"	{ return "*)i"; }
+"\u1F39"	{ return "*(i"; }
+"\u1F3A"	{ return "*)\\i"; }
+"\u1F3B"	{ return "*(\\i"; }
+"\u1F3C"	{ return "*)/i"; }
+"\u1F3D"	{ return "*(/i"; }
+"\u1F3E"	{ return "*)=i"; }
+"\u1F3F"	{ return "*(=i"; }
+"\u1F40"	{ return "o)"; }
+"\u1F41"	{ return "o("; }
+"\u1F42"	{ return "o)\\"; }
+"\u1F43"	{ return "o(\\"; }
+"\u1F44"	{ return "o)/"; }
+"\u1F45"	{ return "o(/"; }
+"\u1F48"	{ return "*)o"; }
+"\u1F49"	{ return "*(o"; }
+"\u1F4A"	{ return "*)\\o"; }
+"\u1F4B"	{ return "*(\\o"; }
+"\u1F4C"	{ return "*)/o"; }
+"\u1F4D"	{ return "*(/o"; }
+"\u1F50"	{ return "u)"; }
+"\u1F51"	{ return "u("; }
+"\u1F52"	{ return "u)\\"; }
+"\u1F53"	{ return "u(\\"; }
+"\u1F54"	{ return "u)/"; }
+"\u1F55"	{ return "u(/"; }
+"\u1F56"	{ return "u)="; }
+"\u1F57"	{ return "u(="; }
+"\u1F59"	{ return "*(u"; }
+"\u1F5B"	{ return "*(\\u"; }
+"\u1F5D"	{ return "*(/u"; }
+"\u1F5F"	{ return "*(=u"; }
+"\u1F60"	{ return "w)"; }
+"\u1F61"	{ return "w("; }
+"\u1F62"	{ return "w)\\"; }
+"\u1F63"	{ return "w(\\"; }
+"\u1F64"	{ return "w)/"; }
+"\u1F65"	{ return "w(/"; }
+"\u1F66"	{ return "w)="; }
+"\u1F67"	{ return "w(="; }
+"\u1F68"	{ return "*)w"; }
+"\u1F69"	{ return "*(w"; }
+"\u1F6A"	{ return "*)\\w"; }
+"\u1F6B"	{ return "*(\\w"; }
+"\u1F6C"	{ return "*)/w"; }
+"\u1F6D"	{ return "*(/w"; }
+"\u1F6E"	{ return "*)=w"; }
+"\u1F6F"	{ return "*(=w"; }
+"\u1F70"	{ return "a\\"; }
+"\u1F71"	{ return "a/"; }
+"\u1F72"	{ return "e\\"; }
+"\u1F73"	{ return "e/"; }
+"\u1F74"	{ return "h\\"; }
+"\u1F75"	{ return "h/"; }
+"\u1F76"	{ return "i\\"; }
+"\u1F77"	{ return "i/"; }
+"\u1F78"	{ return "o\\"; }
+"\u1F79"	{ return "o/"; }
+"\u1F7A"	{ return "u\\"; }
+"\u1F7B"	{ return "u/"; }
+"\u1F7C"	{ return "w\\"; }
+"\u1F7D"	{ return "w/"; }
+"\u1F80"	{ return "a)|"; }
+"\u1F81"	{ return "a(|"; }
+"\u1F82"	{ return "a)\\|"; }
+"\u1F83"	{ return "a(\\|"; }
+"\u1F84"	{ return "a)/|"; }
+"\u1F85"	{ return "a(/|"; }
+"\u1F86"	{ return "a)=|"; }
+"\u1F87"	{ return "a(=|"; }
+"\u1F88"	{ return "*)|a"; }
+"\u1F89"	{ return "*(|a"; }
+"\u1F8A"	{ return "*)\\|a"; }
+"\u1F8B"	{ return "*(\\|a"; }
+"\u1F8C"	{ return "*)/|a"; }
+"\u1F8D"	{ return "*(/|a"; }
+"\u1F8E"	{ return "*)=|a"; }
+"\u1F8F"	{ return "*(=|a"; }
+"\u1F90"	{ return "h)|"; }
+"\u1F91"	{ return "h(|"; }
+"\u1F92"	{ return "h)\\|"; }
+"\u1F93"	{ return "h(\\|"; }
+"\u1F94"	{ return "h)/|"; }
+"\u1F95"	{ return "h(/|"; }
+"\u1F96"	{ return "h)=|"; }
+"\u1F97"	{ return "h(=|"; }
+"\u1F98"	{ return "*)|h"; }
+"\u1F99"	{ return "*(|h"; }
+"\u1F9A"	{ return "*)\\|h"; }
+"\u1F9B"	{ return "*(\\|h"; }
+"\u1F9C"	{ return "*)/|h"; }
+"\u1F9D"	{ return "*(/|h"; }
+"\u1F9E"	{ return "*)=|h"; }
+"\u1F9F"	{ return "*(=|h"; }
+"\u1FA0"	{ return "w)|"; }
+"\u1FA1"	{ return "w(|"; }
+"\u1FA2"	{ return "w)\\|"; }
+"\u1FA3"	{ return "w(\\|"; }
+"\u1FA4"	{ return "w)/|"; }
+"\u1FA5"	{ return "w(/|"; }
+"\u1FA6"	{ return "w)=|"; }
+"\u1FA7"	{ return "w(=|"; }
+"\u1FA8"	{ return "*)|w"; }
+"\u1FA9"	{ return "*(|w"; }
+"\u1FAA"	{ return "*)\\|w"; }
+"\u1FAB"	{ return "*(\\|w"; }
+"\u1FAC"	{ return "*)/|w"; }
+"\u1FAD"	{ return "*(/|w"; }
+"\u1FAE"	{ return "*)=|w"; }
+"\u1FAF"	{ return "*(=|w"; }
+"\u1FB0"	{ return "a^"; }
+"\u1FB1"	{ return "a_"; }
+"\u1FB2"	{ return "a\\|"; }
+"\u1FB3"	{ return "a|"; }
+"\u1FB4"	{ return "a/|"; }
+"\u1FB6"	{ return "a="; }
+"\u1FB7"	{ return "a=|"; }
+"\u1FB8"	{ return "*a^"; }
+"\u1FB9"	{ return "*a_"; }
+"\u1FBA"	{ return "*a\\"; }
+"\u1FBB"	{ return "*a/"; }
+"\u1FBC"	{ return "*a|"; }
+"\u1FC2"	{ return "h\\|"; }
+"\u1FC3"	{ return "h|"; }
+"\u1FC4"	{ return "h/|"; }
+"\u1FC6"	{ return "h="; }
+"\u1FC7"	{ return "h=|"; }
+"\u1FC8"	{ return "*e\\"; }
+"\u1FC9"	{ return "*e/"; }
+"\u1FCA"	{ return "*h\\"; }
+"\u1FCB"	{ return "*h/"; }
+"\u1FCC"	{ return "*h|"; }
+"\u1FD0"	{ return "i^"; }
+"\u1FD1"	{ return "i_"; }
+"\u1FD2"	{ return "i+\\"; }
+"\u1FD3"	{ return "i+/"; }
+"\u1FD6"	{ return "i="; }
+"\u1FD7"	{ return "i+="; }
+"\u1FD8"	{ return "*i^"; }
+"\u1FD9"	{ return "*i_"; }
+"\u1FDA"	{ return "*i\\"; }
+"\u1FDB"	{ return "*i/"; }
+"\u1FE0"	{ return "u^"; }
+"\u1FE1"	{ return "u_"; }
+"\u1FE2"	{ return "u+\\"; }
+"\u1FE3"	{ return "u+/"; }
+"\u1FE4"	{ return "r)"; }
+"\u1FE5"	{ return "r("; }
+"\u1FE6"	{ return "u="; }
+"\u1FE7"	{ return "u+="; }
+"\u1FE8"	{ return "*u^"; }
+"\u1FE9"	{ return "*u_"; }
+"\u1FEA"	{ return "*u\\"; }
+"\u1FEB"	{ return "*u/"; }
+"\u1FEC"	{ return "*(r"; }
+"\u1FF2"	{ return "w\\|"; }
+"\u1FF3"	{ return "w|"; }
+"\u1FF4"	{ return "w/|"; }
+"\u1FFA"	{ return "*w\\"; }
+"\u1FFB"	{ return "*w/"; }
+"\u1FFC"	{ return "*w|"; }
+"\u1FF6"	{ return "w="; }
+"\u1FF7"	{ return "w=|"; }
+"\u1FF8"	{ return "*o\\"; }
+"\u1FF9"	{ return "*o/"; }
+
+"\u0300"	{ return "\\"; }
+"\u0301"		{ return "/"; }
+"\u0304"		{ return "_"; }
+"\u0306"		{ return "^"; }
+"\u0308"		{ return "+"; }
+"\u0302"		{ return "="; }
+"\u0313"		{ return ")"; }
+"\u0314"		{ return "("; }
+"\u0323"		{ return "?"; }
+"\u0345"		{ return "|"; }
+
+"\u03b1"		 { return "a"; }  /* MPDL update  */
+"\u0391"   { return "*a"; }  /* MPDL update  */
+"\u03b2"		{ return "b"; }   /* MPDL update  */
+"\u0392"   { return "*b"; }  /* MPDL update  */
+"\u03b3"		{ return "g"; }   /* MPDL update  */
+"\u0393"   { return "*g"; }  /* MPDL update  */
+"\u03b4"		{ return "d"; }   /* MPDL update  */
+"\u0394"   { return "*d"; }  /* MPDL update  */
+"\u03b5"		{ return "e"; }   /* MPDL update  */
+"\u0395"   { return "*e"; }  /* MPDL update  */
+"\u03b6"		{ return "z"; }   /* MPDL update  */
+"\u0396"   { return "*z"; }  /* MPDL update  */
+"\u03b7"		{ return "h"; }   /* MPDL update  */
+"\u0397"   { return "*h"; }  /* MPDL update  */
+"\u03b8"		{ return "q"; }   /* MPDL update  */
+"\u0398"   { return "*q"; }  /* MPDL update  */
+"\u03b9"		{ return "i"; }   /* MPDL update  */
+"\u0399"   { return "*i"; }  /* MPDL update  */
+"\u03ba"		{ return "k"; }   /* MPDL update  */
+"\u039a"   { return "*k"; }  /* MPDL update  */
+"\u03bb"		{ return "l"; }   /* MPDL update  */
+"\u039b"   { return "*l"; }  /* MPDL update  */
+"\u03bc"		{ return "m"; }   /* MPDL update  */
+"\u039c"   { return "*m"; }  /* MPDL update  */
+"\u03bd"		{ return "n"; }   /* MPDL update  */
+"\u039d"   { return "*n"; }  /* MPDL update  */
+"\u03be"		{ return "c"; }   /* MPDL update  */
+"\u039e"   { return "*c"; }  /* MPDL update  */
+"\u03bf"		{ return "o"; }   /* MPDL update  */
+"\u039f"   { return "*o"; }  /* MPDL update  */
+"\u03c0"		{ return "p"; }   /* MPDL update  */
+"\u03a0"   { return "*p"; }  /* MPDL update  */
+"\u03c1"		{ return "r"; }   /* MPDL update  */
+"\u03a1"   { return "*r"; }  /* MPDL update  */
+
+"\u03a3"  { return "*s"; }  /* MPDL update  */
+"\u03c3"	{ return "s1"; } /* mdh 2002-01-07 */
+"\u03c2"/\-\-	{ return "s"; }   
+"\u03c3"/\&gt; }[a-z\?\!0-9*=\/()\'\-] { return "s"; }  /* MPDL update  */
+"\u03c2"/\&lt; { return "s"; }  /* MPDL update  */
+"\u03c3"/[\[\]][a-z\?\!0-9*=\/()\'\-] { return "s"; }  /* MPDL update  */
+"\u03c2"/\??[^a-z0-9*=\/()\'\-\[\?] { return "s"; }
+"\u03c3"		{ return "s"; }  /* MPDL update  */
+
+"\u03c4"		{ return "t"; }   /* MPDL update  */
+"\u03a4"   { return "*t"; }  /* MPDL update  */
+"\u03c5"		{ return "u"; }   /* MPDL update  */
+"\u03a5"   { return "*u"; }  /* MPDL update  */
+"\u03c6"		{ return "f"; }   /* MPDL update  */
+"\u03a6"   { return "*f"; }  /* MPDL update  */
+"\u03c7"		{ return "x"; }   /* MPDL update  */
+"\u03a7"   { return "*x"; }  /* MPDL update  */
+"\u03c8"		{ return "y"; }   /* MPDL update  */
+"\u03a8"   { return "*y"; }  /* MPDL update  */
+"\u03c9"		{ return "w"; }   /* MPDL update  */
+"\u03a9"   { return "*w"; }  /* MPDL update  */
+
+[\&_]"vert;"   { return "|"; }
+[\&_]"lpar;"   { return "("; }
+[\&_]"rpar;"   { return ")"; }
+[\_\&]"lt;"    { return "&lt;"; }
+[\_\&]"gt;"    { return "&gt;"; }
+"&#039;"       { return "'"; }  /* MPDL update  */
+
+"&"[a-zA-Z]+";"  { return yytext(); }
+
+.       { return yytext(); }
+\n      { return yytext(); }
\ No newline at end of file