3 files changed, 66 insertions, 30 deletions

diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index 99efb75..8379693 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -78,11 +78,14 @@ hamzafv = {
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
-   {a="(%W)(')([ua])", b="%1أ%3"},
-   {a="(%W)(')(i)", b="%1إ%3"},
+   -- consider replacing initial %W with [%s%(%[%<]:
+   --   {a="(%W)(')([ua])", b="%1أ%3"},
+   --   {a="(%W)(')(i)", b="%1إ%3"},
+   {a="([%s%(%[%<])(')([ua])", b="%1أ%3"},
+   {a="([%s%(%[%<])(')(i)", b="%1إ%3"},
    -- final
    -- ^say'aN and .zim'aN are special orthographies
-   {a="(%^say)(%')(aN)", b="%1ئ%3"},
+   {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"},
    {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
    {a="(')(I%p?)$", b="إ%2"}, -- affix pr. 1st person sg.
    {a="(')(I)(%p?%s)", b="إ%2%3"}, -- affix pr. 1st person sg.
@@ -190,11 +193,14 @@ hamzafveasy = { -- differences marked below with 'easy'
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
-   {a="(%W)(')([ua])", b="%1أ%3"},
-   {a="(%W)(')(i)", b="%1إ%3"},
+   -- consider replacing initial %W with [%s%(%[%<]:
+   --   {a="(%W)(')([ua])", b="%1أ%3"},
+   --   {a="(%W)(')(i)", b="%1إ%3"},
+   {a="([%s%(%[%<])(')([ua])", b="%1أ%3"},
+   {a="([%s%(%[%<])(')(i)", b="%1إ%3"},
    -- final
    -- ^say'aN and .zim'aN are special orthographies
-   {a="(%^say)(%')(aN)", b="%1ئ%3"},
+   {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"},
    {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
    {a="(')(I%p?)$", b="إ%2"}, -- affix pr. 1st person sg.
    {a="(')(I)(%p?%s)", b="إ%2%3"}, -- affix pr. 1st person sg.
@@ -565,6 +571,7 @@ digraphsfvidgham = {
    {a="hh", b="هّ"},
    {a="ww", b="وّ"},
    {a="yy", b="يّ"},
+   {a="%.y%.y", b="ىّ"},
    -- additional characters + šaddah (begin)
    {a="pp", b="پّ"},
    {a="vv", b="ڤّ"},
@@ -575,10 +582,10 @@ digraphsfvidgham = {
    -- additional characters + šaddah (end)
    -- sukūn begin
    -- first, take out hyphen if any:
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-([uaiUAI])", b="%1%2"},
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy]%p?)$", b="%1ْ"},
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])(%p?%s)", b="%1ْ%2"},
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIYًٌٍ])", b="%1ْ%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])%-([uaiUAI])", b="%1%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY]%p?)$", b="%1ْ"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])(%p?%s)", b="%1ْ%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])([%_]?[^%_uaiUAIYًٌٍ])", b="%1ْ%2"},
    -- take out sukūn in cases of assimilation
    {a="(n)(ْ)(%s)(ر)", b="%1%3%4"},
    {a="(n)(ْ)(%s)(و)", b="%1%3%4"},
@@ -599,6 +606,7 @@ digraphsfvidgham = {
    {a="%.t", b="ط"},
    {a="%.z", b="ظ"},
    {a="%.g", b="غ"},
+   {a="%.y", b="ى"},
    -- additional characters (begin)
    {a="%^c", b="چ"},
    {a="%^z", b="ژ"},
@@ -688,6 +696,7 @@ digraphsfv = {
    {a="hh", b="هّ"},
    {a="ww", b="وّ"},
    {a="yy", b="يّ"},
+   {a="%.y%.y", b="ىّ"},
    -- additional characters + šaddah (begin)
    {a="pp", b="پّ"},
    {a="vv", b="ڤّ"},
@@ -698,10 +707,10 @@ digraphsfv = {
    -- additional characters + šaddah (end)
    -- sukūn begin
    -- first, take out hyphen if any:
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-([uaiUAI])", b="%1%2"},
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy]%p?)$", b="%1ْ"},
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])(%p?%s)", b="%1ْ%2"},
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIYًٌٍ])", b="%1ْ%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])%-([uaiUAI])", b="%1%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY]%p?)$", b="%1ْ"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])(%p?%s)", b="%1ْ%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])([%_]?[^%_uaiUAIYًٌٍ])", b="%1ْ%2"},
    -- take out sukūn in cases of assimilation
 --   {a="(n)(ْ)(%s)(ر)", b="%1%3%4"},
 --   {a="(n)(ْ)(%s)(و)", b="%1%3%4"},
@@ -722,6 +731,7 @@ digraphsfv = {
    {a="%.t", b="ط"},
    {a="%.z", b="ظ"},
    {a="%.g", b="غ"},
+   {a="%.y", b="ى"},
    -- additional characters (begin)
    {a="%^c", b="چ"},
    {a="%^z", b="ژ"},
@@ -811,6 +821,7 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
    {a="hh", b="هّ"},
    {a="ww", b="وّ"},
    {a="yy", b="يّ"},
+   {a="%.y%.y", b="ىّ"},
    -- additional characters + šaddah (begin)
    {a="pp", b="پّ"},
    {a="vv", b="ڤّ"},
@@ -824,10 +835,10 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
    -- so there be no need to edit them:
    {a="ْ", b=""}, 
    -- first, take out hyphen if any:
---   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-([uaiUAI])", b="%1%2"},
---   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy]%p?)$", b="%1ْ"},
---   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])(%p?%s)", b="%1ْ%2"},
---   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIًٌٍ])", b="%1ْ%2"},
+--   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])%-([uaiUAI])", b="%1%2"},
+--   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY]%p?)$", b="%1ْ"},
+--   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])(%p?%s)", b="%1ْ%2"},
+--   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])([%_]?[^%_uaiUAIًٌٍ])", b="%1ْ%2"},
    -- take out sukūn in cases of assimilation
 --   {a="(n)(ْ)(%s)(ر)", b="%1%3%4"},
 --   {a="(n)(ْ)(%s)(و)", b="%1%3%4"},
@@ -848,6 +859,7 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
    {a="%.t", b="ط"},
    {a="%.z", b="ظ"},
    {a="%.g", b="غ"},
+   {a="%.y", b="ى"},
    -- additional characters (begin)
    {a="%^c", b="چ"},
    {a="%^z", b="ژ"},
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index bffc1ae..31e00b2 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -108,8 +108,11 @@ hamzatrdmg = {
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="ʾ%2"},
    {a="^(')(i)", b="ʾ%2"},
-   {a="(%W)(')([ua])", b="%1ʾ%3"},
-   {a="(%W)(')(i)", b="%1ʾ%3"},
+   -- consider replacing initial %W with [%s%(%[%<]:
+   --   {a="(%W)(')([ua])", b="%1ʾ%3"},
+   --   {a="(%W)(')(i)", b="%1ʾ%3"},
+   {a="([%s%(%[%<])(')([ua])", b="%1ʾ%3"},
+   {a="([%s%(%[%<])(')(i)", b="%1ʾ%3"},
    -- final
    {a="([Iy])(')(aN)$", b="%1ʾ%3"},
    {a="([Iy])(')(aN)(%W)", b="%1ʾ%3%4"},
@@ -365,6 +368,7 @@ digraphstrdmg = {
    {a="%.t", b="ṭ"},
    {a="%.z", b="ẓ"},
    {a="%.g", b="ġ"},
+   {a="%.y", b="y"},
    -- additional characters (begin)
    {a="%^c", b="č"},
    {a="%^z", b="ž"},
@@ -419,6 +423,7 @@ longvtrdmg = {
    {a="A", b="ā"},
    {a="U", b="ū"},
    {a="I", b="ī"},
+   {a="aY", b="ay"},
    {a="iY", b="ī"},
    {a="Y", b="ạ̄"}
 }
@@ -469,8 +474,11 @@ hamzatrloc = {
    -- initial (needs both ^ and %W patterns)
    {a="^(')([ua])", b="%2"},
    {a="^(')(i)", b="%2"},
-   {a="(%W)(')([ua])", b="%1%3"},
-   {a="(%W)(')(i)", b="%1%3"},
+   -- consider replacing initial %W with [%s%(%[%<]:
+   --   {a="(%W)(')([ua])", b="%1%3"},
+   --   {a="(%W)(')(i)", b="%1%3"},
+   {a="([%s%(%[%<])(')([ua])", b="%1%3"},
+   {a="([%s%(%[%<])(')(i)", b="%1%3"},
    -- final
    {a="([Iy])(')(aN)$", b="%1ʾ%3"},
    {a="([Iy])(')(aN)(%W)", b="%1ʾ%3%4"},
@@ -611,6 +619,7 @@ digraphstrloc = {
    {a="%.t", b="ṭ"},
    {a="%.z", b="ẓ"},
    {a="%.g", b="gh"},
+   {a="%.y", b="y"},
    -- additional characters (begin)
    {a="%^c", b="ch"},
    {a="%^z", b="zh"},
@@ -664,6 +673,7 @@ longvtrloc = {
    {a="A", b="ā"},
    {a="U", b="ū"},
    {a="I", b="ī"},
+   {a="aY", b="ay"},
    {a="iY", b="ī"},
    {a="Y", b="á"},
 }
@@ -719,8 +729,11 @@ hamzatrarabica = { -- ≠ from hamzatrloc: initial hamza has to be held
    -- hold it for now (see below, beginning of digraphs table)
    {a="^(')([ua])", b="@%2"},
    {a="^(')(i)", b="@%2"},
-   {a="(%W)(')([ua])", b="%1@%3"},
-   {a="(%W)(')(i)", b="%1@%3"},
+   -- consider replacing initial %W with [%s%(%[%<]:
+   --   {a="(%W)(')([ua])", b="%1@%3"},
+   --   {a="(%W)(')(i)", b="%1@%3"},
+   {a="([%s%(%[%<])(')([ua])", b="%1@%3"},
+   {a="([%s%(%[%<])(')(i)", b="%1@%3"},
    -- final
    {a="([Iy])(')(aN)$", b="%1ʾ%3"},
    {a="([Iy])(')(aN)(%W)", b="%1ʾ%3%4"},
@@ -848,6 +861,7 @@ digraphstrarabica = {
    {a="%.t", b="ṭ"},
    {a="%.z", b="ẓ"},
    {a="%.g", b="ġ"},
+   {a="%.y", b="y"},
    -- additional characters (begin)
    {a="%^c", b="č"},
    {a="%^z", b="ž"},
@@ -898,6 +912,7 @@ singletrarabica = {
 }
 
 longvtrarabica = {
+   {a="aY", b="ay"},
    {a="iY", b="ī"},
    {a="[AY]", b="ā"},
    {a="U", b="ū"},
diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index 9182e52..9251ee9 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -214,11 +214,14 @@ hamza = {
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
-   {a="(%W)(')([ua])", b="%1أ%3"},
-   {a="(%W)(')(i)", b="%1إ%3"},
+   -- consider replacing initial %W with [%s%(%[%<]:
+   --   {a="(%W)(')([ua])", b="%1أ%3"},
+   --   {a="(%W)(')(i)", b="%1إ%3"},
+   {a="([%s%(%[%<])(')([ua])", b="%1أ%3"},
+   {a="([%s%(%[%<])(')(i)", b="%1إ%3"},
    -- final
    -- ^say'aN and .zim'aN are special orthographies
-   {a="(%^say)(%')(aN)", b="%1ئ%3"},
+   {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"},
    {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
    {a="(')(I%p?)$", b="إ%2"}, -- affix pr. 1st person sg.
    {a="(')(I)(%p?%s)", b="إ%2%3"}, -- affix pr. 1st person sg.
@@ -317,11 +320,14 @@ hamzaeasy = { -- differences marked below with 'easy'
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
-   {a="(%W)(')([ua])", b="%1أ%3"},
-   {a="(%W)(')(i)", b="%1إ%3"},
+   -- consider replacing initial %W with [%s%(%[%<]:
+   --   {a="(%W)(')([ua])", b="%1أ%3"},
+   --   {a="(%W)(')(i)", b="%1إ%3"},
+   {a="([%s%(%[%<])(')([ua])", b="%1أ%3"},
+   {a="([%s%(%[%<])(')(i)", b="%1إ%3"},
    -- final
    -- ^say'aN and .zim'aN are special orthographies
-   {a="(%^say)(%')(aN)", b="%1ئ%3"},
+   {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"},
    {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
    {a="(')(I%p?)$", b="إ%2"}, -- affix pr. 1st person sg.
    {a="(')(I)(%p?%s)", b="إ%2%3"}, -- affix pr. 1st person sg.
@@ -695,6 +701,7 @@ digraphs = {
    {a="hh", b="هّ"},
    {a="ww", b="وّ"},
    {a="yy", b="يّ"},
+   {a="%.y%.y", b="ىّ"},
    -- additional characters + šaddah (begin)
    {a="pp", b="پّ"},
    {a="vv", b="ڤّ"},
@@ -714,6 +721,7 @@ digraphs = {
    {a="%.t", b="ط"},
    {a="%.z", b="ظ"},
    {a="%.g", b="غ"},
+   {a="%.y", b="ى"},
    -- additional characters (begin)
    {a="%^c", b="چ"},
    {a="%^z", b="ژ"},
@@ -776,6 +784,7 @@ longv = {
    {a="A", b="َا"},
    {a="U", b="ُو"},
    {a="I", b="ِي"},
+   {a="aY", b="aى"},
    {a="iY", b="iى"},
    {a="Y", b="aى"}
 }