almost done implementing Buckwalter scheme; testing romanization

1 files changed, 52 insertions, 14 deletions

diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index b295254..836ee4b 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -72,26 +72,64 @@ raw = {
 }
 
 buckwalter = {
-   -- trigraphs
-   {a="^Aal", b="al-"},
-   {a="(%W)Aal", b="%1al-"},
-   {a="([%-%s])Al", b="%1\'l-"},
---   {a="([uai])Al", b="%1-\'l-"},
-   {a="(al%-[%a])(%~)", b="%1"},
-   {a="(\'l%-[%a])(%~)", b="%1"},
-   -- šadda
-   {a="([%a])(%~)", b="%1%1"},
-   -- digraphs begin
-   {a="aA", b="A"},
-   {a="uw", b="U"},
-   {a="iy", b="I"},
-   -- digraphs end
+   --- madda: hold it for now
+   {a="%|", b="@"},
+   -- prevent any unwanted šadda from being generated
+   {a="bb", b="b|b"},
+   {a="tt", b="t|t"},
+   {a="vv", b="v|v"},
+   {a="jj", b="j|j"},
+   {a="HH", b="H|H"},
+   {a="xx", b="x|x"},
+   {a="dd", b="d|d"},
+   {a="%*|%*", b="*|*"},
+   {a="rr", b="r|r"},
+   {a="ss", b="s|s"},
+   {a="%$%$", b="$|$"},
+   {a="SS", b="S|S"},
+   {a="DD", b="D|D"},
+   {a="TT", b="T|T"},
+   {a="ZZ", b="Z|Z"},
+   {a="EE", b="E|E"},
+   {a="gg", b="g|g"},
+   {a="ff", b="f|f"},
+   {a="qq", b="q|q"},
+   {a="kk", b="k|k"},
+   {a="ll", b="l|l"},
+   {a="mm", b="m|m"},
+   {a="nn", b="n|n"},
+   {a="hh", b="h|h"},
+   {a="ww", b="w|w"},
+   {a="yy", b="y|y"},
    -- hamza begin
+   {a="\'", b="|\"\'"},
    {a="%>", b="a\"\'"},
    {a="%&", b="w\"\'"},
    {a="%<", b="i\"\'"},
    {a="%]", b="y\"\'"},
    -- hamza end
+   -- trigraphs
+   {a="^Aal%-?", b="al-"},
+   {a="(%W)Aal%-?", b="%1al-"},
+   {a="([%-%s])Al%-?", b="%1\'l-"},
+--   {a="([fw]a?)([tw]a?)Al", b="%1-%2-\'l-"},
+--   {a="([fw]a?)(bi?)Al", b="%1-%2-\'l-"},
+--   {a="([fw]a?)(l[ai]?l)Al", b="%1-%2-l-"},
+--   {a="(al%-[%a])(%~)", b="%1"},
+--   {a="([fltw]a?)Al%-?", b="%1-\'l-"},
+--   {a="(bi?)Al%-?", b="%1-'l-"},
+--   {a="(li?)Al%-?", b="%1-l-"},
+   {a="(\'l%-[%g])(%~)", b="%1"},
+   -- digraphs begin
+   {a="([^a])A", b="%1\"A"},
+   {a="aA", b="A"},
+   {a="uw([^%~])", b="U%1"},
+   {a="iy([^%~])", b="I%1"},
+   -- digraphs end
+   -- madda: get it back
+   {a="%@", b="A\"\'"},
+   -- šadda:
+   {a="([%g])(%~)", b="%1%1"},
    {a="v", b="_t"},
    {a="H", b=".h"},
    {a="%*", b="_d"},