[sword-svn] r256 - trunk/source/data/translit/crosswire

[chrislit at crosswire.org]

Author: chrislit
Date: 2014-02-06 01:26:30 -0700 (Thu, 06 Feb 2014)
New Revision: 256

Modified:
   trunk/source/data/translit/crosswire/Greek_Latin_Beta.txt
Log:
updated, checked, tested Greek/Beta transliterator


Modified: trunk/source/data/translit/crosswire/Greek_Latin_Beta.txt
===================================================================
--- trunk/source/data/translit/crosswire/Greek_Latin_Beta.txt	2014-02-06 07:46:55 UTC (rev 255)
+++ trunk/source/data/translit/crosswire/Greek_Latin_Beta.txt	2014-02-06 08:26:30 UTC (rev 256)
@@ -1,6 +1,6 @@
 # ***************************************************************************
 # *
-# *  Copyright (C) 2002-2013, CrossWire Bible Society
+# *  Copyright (C) 2002-2014, CrossWire Bible Society
 # *  All Rights Reserved.
 # *
 # ***************************************************************************
@@ -9,39 +9,55 @@
 
 # Based on Beta code for CCAT/CATSS/TLG:
 # http://ccat.sas.upenn.edu/gopher/text/religion/biblical/0-betacode.txt
+# and:
+# http://www.tlg.uci.edu/encoding/BCM2013.pdf
+
 # Greek-Latin/Beta
+:: [[:Greek:][̓ʼ᾽᾿̔ʽ῞´ˊ΄´͂̂\^ˆ῀̀\`ˋ`̈¨ͅιͺ̣··;’‐—]];
 :: NFD (NFC);
-:: [:Greek:] upper();
-# Uppercase
-Α<>'*A';
-Β<>'*B';
-Γ<>'*G';
-Δ<>'*D';
-Ε<>'*E';
-Ζ<>'*Z';
-Η<>'*H';
-Θ<>'*Q';
-Ι<>'*I';
-Κ<>'*K';
-Λ<>'*L';
-Μ<>'*M';
-Ν<>'*N';
-Ξ<>'*C';
-Ο<>'*O';
-Φ<>'*F';
-Ψ<>'*Y';
-Π<>'*P';
-Ρ<>'*R';
-Σ<>'*S';
-Τ<>'*T';
-Υ<>'*U';
-Χ<>'*X';
-Ω<>'*W';
-# Archaic
-Ϝ<>'*V';
-Ϟ<>'*#3';
-Ϡ<>'*#5';
-# Lowercase
+
+$betaMarks = [|():/=\\+_];
+
+|\*$2$1<\*($betaMarks)([A-Z]);
+|$2$1<[A-Z]{($betaMarks)([A-Z]);
+
+# Letters
+#  Uppercase
+Α<>\*A;
+Β<>\*B;
+Γ<>\*G;
+Δ<>\*D;
+Ε<>\*E;
+Ζ<>\*Z;
+Η<>\*H;
+Θ<>\*Q;
+Ι<>\*I;
+Κ<>\*K;
+Λ<>\*L;
+Μ<>\*M;
+Ν<>\*N;
+Ξ<>\*C;
+Ο<>\*O;
+Φ<>\*F;
+Ψ<>\*Y;
+Π<>\*P;
+Ρ<>\*R;
+Ϲ<>\*S3;
+Σ<\*S1;
+Σ<\*S2;
+Σ<\*J;
+Σ<>\*S;
+Τ<>\*T;
+Υ<>\*U;
+Χ<>\*X;
+Ω<>\*W;
+
+#  Uppercase Archaic
+Ϝ<>\*V;
+Ϟ<>\*\#3;
+Ϡ<>\*\#5;
+
+#  Lowercase
 α<>A;
 β<>B;
 γ<>G;
@@ -61,37 +77,75 @@
 ψ<>Y;
 π<>P;
 ρ<>R;
-ς<>J;
+ϲ<S3;
+ς<S2;
+ς<J;
+ς>S;
+σ<S1;
+ς<S}[^[:L:][:M:]];
 σ<>S;
 τ<>T;
 υ<>U;
 χ<>X;
 ω<>W;
-# Archaic
+
+#  Lowercase Archaic
 ϝ<>V;
-ϟ<>'#3';
-ϡ<>'#5';
+ϟ<>\#3;
+ϡ<>\#5;
+
 # Non-letter characters
-ͅ<>'|';
-ͺ>'|';
-̔<>'(';
-ʽ>'(';
-·<>':';
-̓<>')';
-ʼ>')';
-́<>'/';
-´>'/';
-ˊ>'/';
-΄>'/';
-͂<>'=';
-̂>'=';
-'^'>'=';
-ˆ>'=';
-̀<>'\\';
-'`'>'\\';
-ˋ>'\\';
-̈<>'+';
-¨>'+';
-̄<>'_';   # Perseus LSJ uses _ for long vowels
+#  smooth breathing
+̓<>\);
+ʼ>\);
+᾽>\);
+᾿>\);
+#  rough breathing
+̔<>\(;
+ʽ>\(;
+῾>\(;
+#  accent acute
+́<>\/;
+´>\/;
+ˊ>\/;
+΄>\/;
+´>\/;
+#  circumflex
+͂<>\=;
+̂>\=;
+\^>\=;
+ˆ>\=;
+῀>\=;
+#  accent grave
+̀<>\\;
+\`>\\;
+ˋ>\\;
+`>\\;
+#  diaeresis
+̈<>\+;
+¨>\+;
+#  iota subscript
+ͅ<>\|;
+ι>\|;
+ͺ>\|;
+#  dot below
+̣<\?;
+
+#  ignore period & comma
+
+#  middle dot
+·<>\:;
+·>\:;
+#  semicolon (question mark) -- use a semicolon in both scripts
+;>\;;
+#  apostrophe
+’<>\';
+#  hyphen
+‐<>\-;
+—<>\_;
+
+# ̄<>\_;	# Perseus LSJ uses _ for long vowels
+
 :: ([a-z] upper);
 :: NFC (NFD);
+:: ([\*A-Za-z\#%0-9\)\(\/\=\\\+\|\?\:\;\'\-\_]);