using context’s builtin pattern for utf8 characters

5 files changed, 21 insertions, 59 deletions

diff --git a/tex/context/third/transliterator/trans_tables_gr.lua b/tex/context/third/transliterator/trans_tables_gr.lua
index bed775a..084478d 100644
--- a/tex/context/third/transliterator/trans_tables_gr.lua
+++ b/tex/context/third/transliterator/trans_tables_gr.lua
@@ -644,23 +644,8 @@ translit.tables["Greek transliteration archaic characters"] = translit.gr_other
 
 function translit.dogreek (mode, text)
     local P, R, S, V, Cs = lpeg.P, lpeg.R, lpeg.S, lpeg.V, lpeg.Cs
-
-
-    -- http://lua-users.org/lists/lua-l/2009-06/msg00343.html
-    local utfchar = R("\000\127") + 
-        R("\194\223") * R("\128\191") + 
-        R("\224\240") * R("\128\191") * R("\128\191") + 
-        R("\241\244") * R("\128\191") * R("\128\191") * R("\128\191")
-
-    -- Add keys of a dictionary to a ruleset.
-    function addrules (dict, rules)
-        for i, _ in pairs(dict) do
-            if rules == nil then rules = P(i)
-            else rules = rules + P(i)
-            end
-        end
-        return rules
-    end
+    local addrules = translit.addrules
+    local utfchar = lpeg.patterns.utf8char
 
     if mode == "gr" or mode == "gr_n" then
 
diff --git a/tex/context/third/transliterator/trans_tables_iso9.lua b/tex/context/third/transliterator/trans_tables_iso9.lua
index e5fa980..4057a0e 100644
--- a/tex/context/third/transliterator/trans_tables_iso9.lua
+++ b/tex/context/third/transliterator/trans_tables_iso9.lua
@@ -251,23 +251,10 @@ translit.tables["cyrillic other uppercase ISO~9"] = translit.non_ru_upp
 
 function translit.iso9 (mode, text)
     local P, R, S, V, Cs = lpeg.P, lpeg.R, lpeg.S, lpeg.V, lpeg.Cs
-    local loc = lpeg.locale ()
-
-    -- http://lua-users.org/lists/lua-l/2009-06/msg00343.html
-    local utfchar = R("\000\127") + 
-        R("\194\223") * R("\128\191") + 
-        R("\224\240") * R("\128\191") * R("\128\191") + 
-        R("\241\244") * R("\128\191") * R("\128\191") * R("\128\191")
+    local addrules = translit.addrules
+    local utfchar = lpeg.patterns.utf8char
 
     -- Add keys of a dictionary to a ruleset.
-    function addrules (dict, rules)
-        for i, _ in pairs(dict) do
-            if rules == nil then rules = P(i)
-            else rules = rules + P(i)
-            end
-        end
-        return rules
-    end
 
     local iso9 = translit.make_add_dict{}
     iso9 = translit.ru_upp + translit.ru_low
diff --git a/tex/context/third/transliterator/trans_tables_scntfc.lua b/tex/context/third/transliterator/trans_tables_scntfc.lua
index 7132fc8..f54eb8e 100644
--- a/tex/context/third/transliterator/trans_tables_scntfc.lua
+++ b/tex/context/third/transliterator/trans_tables_scntfc.lua
@@ -201,13 +201,7 @@ translit.tables["OCS \\quotation{scientific} transliteration additional uppercas
 
 function translit.scientific (mode, text)
     local P, R, S, Cs = lpeg.P, lpeg.R, lpeg.S, lpeg.Cs
-    local loc = lpeg.locale ()
-
-    -- http://lua-users.org/lists/lua-l/2009-06/msg00343.html
-    local utfchar = R("\000\127") + 
-        R("\194\223") * R("\128\191") + 
-        R("\224\240") * R("\128\191") * R("\128\191") + 
-        R("\241\244") * R("\128\191") * R("\128\191") * R("\128\191")
+    local utfchar = lpeg.patterns.utf8char
 
     local cyr = translit.make_add_dict{}
     local cyruk, p_cyruk, p_cyr
diff --git a/tex/context/third/transliterator/trans_tables_trsc.lua b/tex/context/third/transliterator/trans_tables_trsc.lua
index cad3567..0458539 100644
--- a/tex/context/third/transliterator/trans_tables_trsc.lua
+++ b/tex/context/third/transliterator/trans_tables_trsc.lua
@@ -487,17 +487,10 @@ translit.tables["Czech transcription for OCS and pre-1918 uppercase"] = translit
 
 function translit.transcript (mode, text)
     local P, R, S, V, Cs = lpeg.P, lpeg.R, lpeg.S, lpeg.V, lpeg.Cs
-    local loc = lpeg.locale ()
+    local addrules = translit.addrules
+    local utfchar = lpeg.patterns.utf8char
 
-    -- http://lua-users.org/lists/lua-l/2009-06/msg00343.html
-    local utfchar = R("\000\127") + 
-        R("\194\223") * R("\128\191") + 
-        R("\224\240") * R("\128\191") * R("\128\191") + 
-        R("\241\244") * R("\128\191") * R("\128\191") * R("\128\191")
-
-    local trsc_parser, p_rules, capt
-
-    local p_de
+    local trsc_parser, p_rules, capt, p_de
 
     function tab_subst (s, ...)
         local p_tmp, tmp = nil, translit.make_add_dict{}
@@ -509,16 +502,6 @@ function translit.transcript (mode, text)
         return fp:match(s)
     end
 
-    -- Add keys of a dictionary to a ruleset.
-    function addrules (dict, rules)
-        for i, _ in pairs(dict) do
-            if rules == nil then rules = P(i)
-            else rules = rules + P(i)
-            end
-        end
-        return rules
-    end
-
     -- The following is needed becaus lpeg.S doesn't work with utf.
     local vow, con, iy
     for _,v in ipairs (translit.ru_vowels) do
diff --git a/tex/context/third/transliterator/transliterator.lua b/tex/context/third/transliterator/transliterator.lua
index 4b64d82..110d862 100644
--- a/tex/context/third/transliterator/transliterator.lua
+++ b/tex/context/third/transliterator/transliterator.lua
@@ -80,6 +80,19 @@ end
 -- Auxiliary Functions
 --------------------------------------------------------------------------------
 
+-- Generate a rule pattern from hash table.
+do
+    local P = lpeg.P
+    function translit.addrules (dict, rules)
+        for i, _ in pairs(dict) do
+            if rules == nil then rules = P(i)
+            else rules = rules + P(i)
+            end
+        end
+        return rules
+    end
+end
+
 -- We might want to have all the table data nicely formatted by \CONTEXT\ 
 -- itself, here's how we'll do it.  \type{translit.show_tab(t)} handles a
 -- single table \type{t}, builds a Natural TABLE out of its content and