corrected final hamza; added quoting before hamza

4 files changed, 54 insertions, 10 deletions

diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index 23f640b..a56a33f 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -24,6 +24,13 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
 
 -- this is new
 hamzafv = {
+   -- hard coded hamza
+   {a="|\"'", b="ء"},
+   {a="A\"'", b="آ"},
+   {a="[au]\"'", b="أ"},
+   {a="w\"'", b="ؤ"},
+   {a="i\"'", b="إ"},
+   {a="y\"'", b="ئ"},
    -- initial long u
    {a="%'%_U", b="أU"},
    -- madda (historic writing below)
@@ -38,12 +45,19 @@ hamzafv = {
    {a="(%W)(')([ua])", b="%1أ%3"},
    {a="(%W)(')(i)", b="%1إ%3"},
    -- final
-   {a="([Iy])(')(aN)$", b="%1ئ%3"},
-   {a="([Iy])(')(aN)(%W)", b="%1ئ%3%4"},
+--   {a="([Iy])(')(aN)$", b="%1ئ%3"},
+--   {a="([Iy])(')(aN)(%W)", b="%1ئ%3%4"},
+--   {a="([^uai])(')([uai]N?)$", b="%1ء%3"},
+--   {a="([^uai])(')([uai]N?)(%W)", b="%1ء%3%4"},
+--   {a="([UI])(')([uai])$", b="%1ء%3"},
+--   {a="([UI])(')([uai])(%W)", b="%1ء%3%4"},
+--   {a="([UI])(')([uai]N?)$", b="%1ء%3"},
+--   {a="([UI])(')([uai]N?)(%W)", b="%1ء%3%4"},
+   -- ^say'aN and .zim'aN are special orthographies
+   {a="(%^say)(%')(aN)", b="%1ئ%3"},
+   {a="(.zim)(%')(aN)", b="%1ئ%3"},
    {a="([^uai])(')([uai]N?)$", b="%1ء%3"},
-   {a="([^uai])(')([uai]N?)(%W)", b="%1ء%3%4"},
-   {a="([UI])(')([uai])$", b="%1ء%3"},
-   {a="([UI])(')([uai])(%W)", b="%1ء%3%4"},
+   {a="([^uai])(')([uai]N?)(%W)", b="%1ء%3%4"},   
    -- middle
    {a="(U)(')", b="%1ء"},
    {a="([Iy])(')", b="%1ئ"},
diff --git a/arabluatex_novoc.lua b/arabluatex_novoc.lua
index ef8b0ff..d0c9710 100644
--- a/arabluatex_novoc.lua
+++ b/arabluatex_novoc.lua
@@ -26,6 +26,7 @@ tanwinnv = {
         {a="uN", b=""},
         {a="(aN)(_A)", b="ى"},
         {a="(T)(aN)", b="%1"},
+        {a="(ء)(aN)", b="%1"},
         {a="([^TA])(aN)", b="%1ا"},
         {a="iN", b=""}
 }
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index d9b4418..2c2be73 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -31,6 +31,13 @@ nulltr = {
 -- dmg
 
 hamzatrdmg = {
+   -- hard coded hamza
+   {a="|\"'", b="'"},
+   {a="A\"'", b="'"},
+   {a="[au]\"'", b="'"},
+   {a="w\"'", b="'"},
+   {a="i\"'", b="'"},
+   {a="y\"'", b="'"},
    -- initial long u
    {a="%'%_U", b="ʾU"},
    -- madda (historic writing below)
@@ -184,6 +191,13 @@ shortvtrdmg = {
 -- loc
 
 hamzatrloc = {
+   -- hard coded hamza
+   {a="|\"'", b="'"},
+   {a="A\"'", b="'"},
+   {a="[au]\"'", b="'"},
+   {a="w\"'", b="'"},
+   {a="i\"'", b="'"},
+   {a="y\"'", b="'"},
    -- initial long u
    {a="%'%_U", b="U"},
    -- madda (historic writing below)
diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index face931..1dc7a0c 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -65,6 +65,13 @@ raw = {
 }
 
 hamza = {
+   -- hard coded hamza
+   {a="|\"'", b="ء"},
+   {a="A\"'", b="آ"},
+   {a="[au]\"'", b="أ"},
+   {a="w\"'", b="ؤ"},
+   {a="i\"'", b="إ"},
+   {a="y\"'", b="ئ"},
    -- initial long u
    {a="%'%_U", b="أU"},
    -- madda (historic writing below)
@@ -79,12 +86,19 @@ hamza = {
    {a="(%W)(')([ua])", b="%1أ%3"},
    {a="(%W)(')(i)", b="%1إ%3"},
    -- final
-   {a="([Iy])(')(aN)$", b="%1ئ%3"},
-   {a="([Iy])(')(aN)(%W)", b="%1ئ%3%4"},
+--   {a="([Iy])(')(aN)$", b="%1ئ%3"},
+--   {a="([Iy])(')(aN)(%W)", b="%1ئ%3%4"},
+--   {a="([^uai])(')([uai]N?)$", b="%1ء%3"},
+--   {a="([^uai])(')([uai]N?)(%W)", b="%1ء%3%4"},
+--   {a="([UI])(')([uai])$", b="%1ء%3"},
+--   {a="([UI])(')([uai])(%W)", b="%1ء%3%4"},
+--   {a="([UI])(')([uai]N?)$", b="%1ء%3"},
+--   {a="([UI])(')([uai]N?)(%W)", b="%1ء%3%4"},
+   -- ^say'aN and .zim'aN are special orthographies
+   {a="(%^say)(%')(aN)", b="%1ئ%3"},
+   {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
    {a="([^uai])(')([uai]N?)$", b="%1ء%3"},
-   {a="([^uai])(')([uai]N?)(%W)", b="%1ء%3%4"},
-   {a="([UI])(')([uai])$", b="%1ء%3"},
-   {a="([UI])(')([uai])(%W)", b="%1ء%3%4"},
+   {a="([^uai])(')([uai]N?)(%W)", b="%1ء%3%4"},   
    -- middle
    {a="(U)(')", b="%1ء"},
    {a="([Iy])(')", b="%1ئ"},
@@ -149,6 +163,7 @@ tanwin = {
         {a="(aN)(_A)", b="ًى"},
         {a="(aN)(Y)", b="ًى"},
         {a="(T)(aN)", b="%1ً"},
+        {a="(ء)(aN)", b="%1ً"},
         {a="([^TA])(aN)", b="%1ًا"},
         {a="iN", b="ٍ"}
 }