From 025223cd801e62eeae6085dcd1a516975bd88260 Mon Sep 17 00:00:00 2001
From: Philipp Gesang <pgesang@ix.urz.uni-heidelberg.de>
Date: Tue, 10 May 2011 15:09:40 +0200
Subject: yet better exception handling

---
 .../third/transliterator/transliterator.tex        |  2 ++
 .../third/transliterator/trans_tables_sr.lua       | 24 ++++++++++++++--------
 2 files changed, 18 insertions(+), 8 deletions(-)

diff --git a/doc/context/third/transliterator/transliterator.tex b/doc/context/third/transliterator/transliterator.tex
index b11cb3b..a758dbb 100644
--- a/doc/context/third/transliterator/transliterator.tex
+++ b/doc/context/third/transliterator/transliterator.tex
@@ -730,6 +730,8 @@ alphabets are \type{sr_tolt} and \type{sr_tocy}.
 \trlex{sr_tocy}{hr}{sr}{computer-modern-unicode}{%
   Transliteration latinica \rightarrow\ ćirilica.%
 }{%
+  nadživeti, Nadživeti, NADŽIVETI
+
   Srpski jezik je jedan od slovenskih jezika iz porodice
   indoevropskih jezika. Prvi pisani spomenici u srpskoj
   redakciji staroslovenskog jezika potiču iz XI i XII veka.
diff --git a/tex/context/third/transliterator/trans_tables_sr.lua b/tex/context/third/transliterator/trans_tables_sr.lua
index 0f9eae4..8e27168 100644
--- a/tex/context/third/transliterator/trans_tables_sr.lua
+++ b/tex/context/third/transliterator/trans_tables_sr.lua
@@ -92,23 +92,30 @@ if not translit.done_serbian then
     translit.sr_tocy_upper = translit.make_add_dict(__inverse_tab(translit.sr_tolt_upper))
 
     local except = {
-        ["nadživeti"] = "надживети",
+        ["nadživ"] = "наджив", -- nadživeti and derivatives
     }
 
     local P = lpeg.P
     local sub, upper = unicode.utf8.sub, unicode.utf8.upper
 
-    local sre, p_sre   = {}
+    local sre, srei, p_sre, p_sre_i = { }, { }
     for lat, cyr in next, except do
         local Lat = upper(sub(lat, 1, 1)) .. sub(lat, 2)
         local Cyr = upper(sub(cyr, 1, 1)) .. sub(cyr, 2)
         local LAT, CYR = upper(lat), upper(cyr)
-        p_sre = p_sre and p_sre + P(lat) + Lat + LAT or P(lat) + Lat + LAT
-        sre[lat], sre[Lat], sre[LAT] = cyr, Cyr, CYR
+        local p_lat    = P" " * (P(lat) + Lat + LAT)
+
+        p_sre_i = p_sre_i and p_sre_i + P(lat) + Lat + LAT or P(lat) + Lat + LAT
+        p_sre   = p_sre   and p_sre + p_lat                or p_lat
+
+        srei[lat], srei[Lat], srei[LAT]             = cyr, Cyr, CYR
+        sre[" "..lat], sre[" "..Lat], sre[" "..LAT] = " "..cyr, " "..Cyr, " "..CYR
     end
 
-    translit.serbian_exceptions   = sre
-    translit.p_serbian_exceptions = p_sre
+    translit.serbian_exceptions        = sre
+    translit.serbian_exceptions_init   = srei
+    translit.p_serbian_exceptions      = p_sre
+    translit.p_serbian_exceptions_init = p_sre_i
 
     translit.done_serbian = true
 end
@@ -122,7 +129,8 @@ local t = translit
 local function sr (mode, text)
     local P, R, Cs = lpeg.P, lpeg.R, lpeg.Cs
     local utfchar  = translit.utfchar
-    local _p_sre   = translit.p_serbian_exceptions / translit.serbian_exceptions
+    local _p_sre   = translit.p_serbian_exceptions      / translit.serbian_exceptions
+    local _p_sre_i = translit.p_serbian_exceptions_init / translit.serbian_exceptions_init
 
     local trl_sr   = translit.make_add_dict{}
     trl_sr         = t[mode.."_upper"] + t[mode.."_lower"]
@@ -130,7 +138,7 @@ local function sr (mode, text)
     -- transliteration from latin script requires macro handling … 
     local _p_macro = P[[\]] * R("az", "AZ")^1
     local _p_sr    = translit.addrules (trl_sr, _p_sr)
-    local p_sr     = Cs((_p_macro + _p_sre + (_p_sr / trl_sr) + utfchar)^0)
+    local p_sr     = Cs(_p_sre_i^-1 * (_p_macro + _p_sre + (_p_sr / trl_sr) + utfchar)^0)
 
     return p_sr:match(text)
 end
-- 
cgit v1.2.3