arbnulls inserted in fullvoc table; some additional corrections in the other tables

4 files changed, 94 insertions, 39 deletions

diff --git a/arabluatex.lua b/arabluatex.lua
index 564e08d..7c49693 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -526,6 +526,7 @@ end
 
 function processfullvoc(str, rules, scheme)
    str = "\\arb{".. str.."}"
+   str = processarbnull(str, scheme)
    str = takeoutcapetc(str)
    str = protectarb(str)
    str = breakcmd(str)
diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index 9891319..32f53f7 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -66,7 +66,11 @@ hamzafv = {
    {a="(A)(')(u)", b="aآؤ%3"}, -- historic madda
    {a="(A)(')", b="aآء"}, -- historic madda
    -- initial (needs both ^ and %W patterns)
-   -- 'aw: the diphthong is to be resoved into 'awi' (next 4 lines)
+   -- 'aw: the diphthong is to be resolved into 'awi' (next 8 lines)
+   {a="^('aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
+   {a="(%W)('aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1%2i"},
+   {a="^('aw)(o)([%(%[%|%<]?\"?[uai])([%S]-o)", b="%1i"},
+   {a="(%W)('aw)(o)([%(%[%|%<]?\"?[uai])([%S]-o)", b="%1%2i"},
    {a="^('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
    {a="(%W)('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
    {a="^('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
@@ -172,7 +176,11 @@ hamzafveasy = { -- differences marked below with 'easy'
    {a="(A)(')", b="aاء"}, -- historic madda
    --easy (end)
    -- initial (needs both ^ and %W patterns)
-   -- 'aw: the diphthong is to be resoved into 'awi' (next 4 lines)
+   -- 'aw: the diphthong is to be resolved into 'awi' (next 8 lines)
+   {a="^('aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
+   {a="(%W)('aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1%2i"},
+   {a="^('aw)(o)([%(%[%|%<]?\"?[uai])([%S]-o)", b="%1i"},
+   {a="(%W)('aw)(o)([%(%[%|%<]?\"?[uai])([%S]-o)", b="%1%2i"},
    {a="^('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
    {a="(%W)('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
    {a="^('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
@@ -233,12 +241,13 @@ hamzafveasy = { -- differences marked below with 'easy'
 }
 
 tanwinfv = { -- with assimilations (\SetArbDflt*)
-   {a="%-?(uNU)(%s)([uai])", b="%1%2ٱ"},
-   {a="%-?(aNU)(%s)([uai])", b="%1%2ٱ"},
-   {a="%-?(iNU)(%s)([uai])", b="%1%2ٱ"},
    -- assimilations (begin)
+   {a="(o[%S]-)(%-?[uai]N[UI]?)(o)([rlmnwy])", b="%4%4"},
    {a="(%-?[uai]NU)(%s)([rlmnwy])", b="%1%2%3%3"},
    -- assimilations (end)
+   {a="%-?(uNU)(%s)([uai])", b="%1%2ٱ"},
+   {a="%-?(aNU)(%s)([uai])", b="%1%2ٱ"},
+   {a="%-?(iNU)(%s)([uai])", b="%1%2ٱ"},
    {a="%-?uNU", b="ٌو"},
    {a="%-?aNU", b="ًوا"},
    {a="%-?iNU", b="ٍو"},
@@ -246,6 +255,8 @@ tanwinfv = { -- with assimilations (\SetArbDflt*)
    {a="%-?(aN[%_]?[AY])(%s)([uai])", b="%1%2ٱ"},
    -- assimilations (begin)
    {a="%-?(uN)(%s)([rlmnwy])", b="ٌ%2%3%3"},
+   {a="(o[%S]-)(%-?aN)(_A)(o)([rlmnwy])", b="%5%5"},
+   {a="(o[%S]-)(%-?aN)(Y)(o)([rlmnwy])", b="%5%5"},
    {a="%-?(aN)(_A)(%s)([rlmnwy])", b="ًى%3%4%4"},
    {a="%-?(aN)(Y)(%s)([rlmnwy])", b="ًى%3%4%4"},
    {a="(T)%-?(aN)(%s)([rlmnwy])", b="%1ً%3%4%4"},
@@ -274,12 +285,13 @@ tanwinfv = { -- with assimilations (\SetArbDflt*)
 }
 
 tanwinfveasy = { -- no assimilations (see below)
-   {a="%-?(uNU)(%s)([uai])", b="%1%2ٱ"},
-   {a="%-?(aNU)(%s)([uai])", b="%1%2ٱ"},
-   {a="%-?(iNU)(%s)([uai])", b="%1%2ٱ"},
    -- assimilations (begin)
+--easy   {a="(o[%S]-)(%-?[uai]N[UI]?)(o)([rlmnwy])", b="%4%4"},
 --easy   {a="(%-?[uai]NU)(%s)([rlmnwy])", b="%1%2%3%3"},
    -- assimilations (end)
+   {a="%-?(uNU)(%s)([uai])", b="%1%2ٱ"},
+   {a="%-?(aNU)(%s)([uai])", b="%1%2ٱ"},
+   {a="%-?(iNU)(%s)([uai])", b="%1%2ٱ"},
    {a="%-?uNU", b="ٌو"},
    {a="%-?aNU", b="ًوا"},
    {a="%-?iNU", b="ٍو"},
@@ -287,6 +299,8 @@ tanwinfveasy = { -- no assimilations (see below)
    {a="%-?(aN[%_]?[AY])(%s)([uai])", b="%1%2ٱ"},
    -- assimilations (begin)
 --easy   {a="%-?(uN)(%s)([rlmnwy])", b="ٌ%2%3%3"},
+--easy   {a="(o[%S]-)(%-?aN)(_A)(o)([rlmnwy])", b="%5%5"},
+--easy   {a="(o[%S]-)(%-?aN)(Y)(o)([rlmnwy])", b="%5%5"},
 --easy   {a="%-?(aN)(_A)(%s)([rlmnwy])", b="ًى%3%4%4"},
 --easy   {a="%-?(aN)(Y)(%s)([rlmnwy])", b="ًى%3%4%4"},
 --easy   {a="(T)%-?(aN)(%s)([rlmnwy])", b="%1ً%3%4%4"},
@@ -318,7 +332,11 @@ trigraphsfv = { -- trigraphs or more
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="ٱلّ%1%2"},
    {a="([%(%[%|%<%s%-])'ll(a)([%_]?[dt])", b="%1ٱلّ%2%3"}, --p
-   -- law: the diphthong is to be resoved into 'awi' (next 4 lines)
+   -- law: the diphthong is to be resolved into 'awi' (next 8 lines)
+   {a="^(law)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
+   {a="(%W)(law)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1%2i"},
+   {a="^(law)(o)(\"?[uai])([%S]-o)", b="%1i"},
+   {a="(%W)(law)(o)(\"?[uai])([%S]-o)", b="%1%2i"},
    {a="^(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
    {a="(%W)(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
    {a="^(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
@@ -341,6 +359,9 @@ trigraphsfv = { -- trigraphs or more
    {a="^(a)l%-", b="ا%1لْ"},
    {a="([%(%[%|%<%s%-o])(a)l%-", b="%1ا%2لْ"}, --p
    -- diphthongs to be resolved before ʾalif conjunctionis
+   {a="(aW)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="awuا"},
+   {a="(aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1u"},
+   {a="(ay)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
    {a="(aW)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awuا%2%3"},
    {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
    {a="(ay)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
@@ -380,7 +401,11 @@ trigraphsfveasy = { -- trigraphs or more (see 'easy' tag below for the diffs)
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="ٱلّ%1%2"},
    {a="([%(%[%|%<%s%-])'ll(a)([%_]?[dt])", b="%1ٱلّ%2%3"}, --p
-   -- law: the diphthong is to be resoved into 'awi' (next 4 lines)
+   -- law: the diphthong is to be resolved into 'awi' (next 8 lines)
+   {a="^(law)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
+   {a="(%W)(law)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1%2i"},
+   {a="^(law)(o)(\"?[uai])([%S]-o)", b="%1i"},
+   {a="(%W)(law)(o)(\"?[uai])([%S]-o)", b="%1%2i"},
    {a="^(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
    {a="(%W)(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
    {a="^(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
@@ -403,6 +428,9 @@ trigraphsfveasy = { -- trigraphs or more (see 'easy' tag below for the diffs)
    {a="^(a)l%-", b="ا%1لْ"},
    {a="([%(%[%|%<%s%-o])(a)l%-", b="%1ا%2لْ"}, --p
    -- diphthongs to be resolved before ʾalif conjunctionis
+   {a="(aW)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="awuا"},
+   {a="(aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1u"},
+   {a="(ay)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
    {a="(aW)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awuا%2%3"},
    {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
    {a="(ay)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
@@ -452,14 +480,20 @@ digraphsfvidgham = {
    {a="^\"[uai]", b="ٱ"},
    {a="([%(%[%|%<%s%-])\"[uai]", b="%1ٱ"}, --p
    -- diphthongs to be resolved before ʾalif conjunctionis
-   {a="(aW)(%s)(\"?[uai])", b="awuا%2ٱ"},
-   {a="(aw)(%s)(\"?[uai])", b="%1u%2ٱ"},
-   {a="(ay)(%s)(\"?[uai])", b="%1i%2ٱ"},
-   {a="([uai]%-)(\"?[uai])", b="%1ٱ"}, -- hyphen + initial alif without hamza
+   {a="(aW)(o)(\"?[uai])([%S]-o)", b="awuا"},
+   {a="(aW)(%s)([%(%[%|%<]?)(\"?[uai])", b="awuا%2%3ٱ"}, --p
+   {a="(aw)(o)(\"?[uai])([%S]-o)", b="%1u"},
+   {a="(aw)(%s)([%(%[%|%<]?)(\"?[uai])", b="%1u%2%3ٱ"}, --p
+   {a="(ay)(o)(\"?[uai])([%S]-o)", b="%1i"},
+   {a="(ay)(%s)([%(%[%|%<]?)(\"?[uai])", b="%1i%2%3ٱ"}, --p
+   -- hyphen + initial alif without hamza
+   --   {a="([uai]%-)(\"?[uai])", b="%1ٱ"},
+   {a="([uai]%-)(\"?[uai])([%^%_%.%`]?)(%a)", b="%1ٱ%3%4"},
    -- initial alif without hamza
+   {a="(o[%S]-)([uaiUAIY])(o)(\"?[uai])", b="ٱ"},
    {a="([%_]?[uaiUAIY])(%s[%(%[%|%<]?)(\"?[uai])", b="%1%2ٱ"}, --p
-   {a="^([%(%[%|%<]?)([uai])", b="%1ا%2"}, -- initial alif without hamza --p
-   {a="(%s)([%(%[%|%<]?)([uai])", b="%1%2ا%3"}, -- initial alif without hamza --p
+   {a="^([%(%[%|%<]?)(\"?[uai])", b="%1ا%2"}, -- initial alif without hamza --p
+   {a="(%s)([%(%[%|%<]?)(\"?[uai])", b="%1%2ا%3"}, -- initial alif without hamza --p
    {a="%-%-", b="ـ"},
    {a="ؤؤ", b="ؤّ"},
    {a="أأ", b="أّ"},
@@ -554,14 +588,20 @@ digraphsfv = {
    {a="^\"[uai]", b="ٱ"},
    {a="([%(%[%|%<%s%-])\"[uai]", b="%1ٱ"}, --p
    -- diphthongs to be resolved before ʾalif conjunctionis
-   {a="(aW)(%s)(\"?[uai])", b="awuا%2ٱ"},
-   {a="(aw)(%s)(\"?[uai])", b="%1u%2ٱ"},
-   {a="(ay)(%s)(\"?[uai])", b="%1i%2ٱ"},
-   {a="([uai]%-)(\"?[uai])", b="%1ٱ"}, -- hyphen + initial alif without hamza
+   {a="(aW)(o)(\"?[uai])([%S]-o)", b="awuا"},
+   {a="(aW)(%s)([%(%[%|%<]?)(\"?[uai])", b="awuا%2%3ٱ"}, --p
+   {a="(aw)(o)(\"?[uai])([%S]-o)", b="%1u"},
+   {a="(aw)(%s)([%(%[%|%<]?)(\"?[uai])", b="%1u%2%3ٱ"}, --p
+   {a="(ay)(o)(\"?[uai])([%S]-o)", b="%1i"},
+   {a="(ay)(%s)([%(%[%|%<]?)(\"?[uai])", b="%1i%2%3ٱ"}, --p
+   -- hyphen + initial alif without hamza
+   --   {a="([uai]%-)(\"?[uai])", b="%1ٱ"},
+   {a="([uai]%-)(\"?[uai])([%^%_%.%`]?)(%a)", b="%1ٱ%3%4"},
    -- initial alif without hamza
+   {a="(o[%S]-)([uaiUAIY])(o)(\"?[uai])", b="ٱ"},
    {a="([%_]?[uaiUAIY])(%s[%(%[%|%<]?)(\"?[uai])", b="%1%2ٱ"}, --p
-   {a="^([%(%[%|%<]?)([uai])", b="%1ا%2"}, -- initial alif without hamza --p
-   {a="(%s)([%(%[%|%<]?)([uai])", b="%1%2ا%3"}, -- initial alif without hamza --p
+   {a="^([%(%[%|%<]?)(\"?[uai])", b="%1ا%2"}, -- initial alif without hamza --p
+   {a="(%s)([%(%[%|%<]?)(\"?[uai])", b="%1%2ا%3"}, -- initial alif without hamza --p
    {a="%-%-", b="ـ"},
    {a="ؤؤ", b="ؤّ"},
    {a="أأ", b="أّ"},
@@ -656,14 +696,20 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
    {a="^\"[uai]", b="ٱ"},
    {a="([%(%[%|%<%s%-])\"[uai]", b="%1ٱ"}, --p
    -- diphthongs to be resolved before ʾalif conjunctionis
-   {a="(aW)(%s)(\"?[uai])", b="awuا%2ٱ"},
-   {a="(aw)(%s)(\"?[uai])", b="%1u%2ٱ"},
-   {a="(ay)(%s)(\"?[uai])", b="%1i%2ٱ"},
-   {a="([uai]%-)(\"?[uai])", b="%1ٱ"}, -- hyphen + initial alif without hamza
+   {a="(aW)(o)(\"?[uai])([%S]-o)", b="awuا"},
+   {a="(aW)(%s)([%(%[%|%<]?)(\"?[uai])", b="awuا%2%3ٱ"}, --p
+   {a="(aw)(o)(\"?[uai])([%S]-o)", b="%1u"},
+   {a="(aw)(%s)([%(%[%|%<]?)(\"?[uai])", b="%1u%2%3ٱ"}, --p
+   {a="(ay)(o)(\"?[uai])([%S]-o)", b="%1i"},
+   {a="(ay)(%s)([%(%[%|%<]?)(\"?[uai])", b="%1i%2%3ٱ"}, --p
+   -- hyphen + initial alif without hamza
+   -- {a="([uai]%-)(\"?[uai])", b="%1ٱ"},
+   {a="([uai]%-)(\"?[uai])([%^%_%.%`]?)(%a)", b="%1ٱ%3%4"},
    -- initial alif without hamza
+   {a="(o[%S]-)([uaiUAIY])(o)(\"?[uai])", b="ٱ"},
    {a="([%_]?[uaiUAIY])(%s[%(%[%|%<]?)(\"?[uai])", b="%1%2ٱ"}, --p
-   {a="^([%(%[%|%<]?)([uai])", b="%1ا%2"}, -- initial alif without hamza --p
-   {a="(%s)([%(%[%|%<]?)([uai])", b="%1%2ا%3"}, -- initial alif without hamza --p
+   {a="^([%(%[%|%<]?)(\"?[uai])", b="%1ا%2"}, -- initial alif without hamza --p
+   {a="(%s)([%(%[%|%<]?)(\"?[uai])", b="%1%2ا%3"}, -- initial alif without hamza --p
    {a="%-%-", b="ـ"},
    {a="ؤؤ", b="ؤّ"},
    {a="أأ", b="أّ"},
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index b29ae37..c420f6f 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -262,8 +262,8 @@ trigraphstrdmg = { -- trigraphs or more
 idghamtrdmg = {
    -- assimilations
    {a="(n)(}?)(%s)([rlmnwy])", b="%4%2%3%4"},
-   {a="(n)(o)([rlmnwy])([%S]-o)", b="%3"}
-} 
+   {a="(n)(}?)(o)([rlmnwy])([%S]-o)", b="%4%2"}
+}
 
 digraphstrdmg = {
    {a="([uai]%-)(\"?[uai])", b="%1'"}, -- hyphen + initial alif without hamza
@@ -276,6 +276,10 @@ digraphstrdmg = {
    {a="(%W)(\"?[uai])", b="%1%2"},      -- initial alif without hamza
    -- this is not necessary, take out for now:
 -- {a="([%_]?[uaiUAIY])(%s)([uai])", b="%1%2'"}, -- initial alif without hamza
+   {a="(aw)(o)(\"?[uai])([%S]-o)", b="%1u"},
+   {a="(aw)(%s)([%(%[%|%<]?)(\"?[uai])", b="%1u%2%3'"}, --p
+   {a="(ay)(o)(\"?[uai])([%S]-o)", b="%1i"},
+   {a="(ay)(%s)([%(%[%|%<]?)(\"?[uai])", b="%1i%2%3'"}, --p
    {a="(aW)(o)(\"?[uai])([%S]-o)", b="awu"},
    {a="(UA)(o)(\"?[uai])([%S]-o)", b="u"},
    {a="(%_A)(o)(\"?[uai])([%S]-o)", b="ạ"},
diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index fb53b5a..57ab34c 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -202,7 +202,7 @@ hamza = {
    {a="(A)(')(u)", b="aآؤ%3"}, -- historic madda
    {a="(A)(')", b="aآء"}, -- historic madda
    -- initial (needs both ^ and %W patterns)
-   -- 'aw: the diphthong is to be resoved into 'awi' (next 8 lines)
+   -- 'aw: the diphthong is to be resolved into 'awi' (next 8 lines)
    {a="^('aw)(o)('[%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
    {a="(%W)('aw)(o)('[%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1%2i"},
    {a="^('aw)(o)(\"?[uai])([%S]-o)", b="%1i"},
@@ -482,13 +482,9 @@ trigraphs = { -- trigraphs or more
    {a="(aW)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="awuا"},
    {a="(aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1u"},
    {a="(ay)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
-   {a="(aw)(o)(\"?[uai])([%S]-o)", b="%1u"},
-   {a="(ay)(o)(\"?[uai])([%S]-o)", b="%1i"},
    {a="(aW)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awuا%2%3"},
    {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
    {a="(ay)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
-   {a="(aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1u%2%3"}, --p
-   {a="(ay)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"}, --p
    -- art. with waṣla + lām
    {a="'l%-(l)", b="ال%1%1"},
    -- art. with waṣla + solar consonant
@@ -613,12 +609,20 @@ digraphs = {
    -- initial straight double quote gives a connective ʾalif
    {a="^\"[uai]", b="ٱ"},
    {a="([%(%[%|%<%s%-])\"[uai]", b="%1ٱ"}, --p
-   {a="(aW)(o)([uai]([%S]-o))", b="awuا"},
+   -- diphthongs to be resolved before ʾalif conjunctionis
+   {a="(aW)(o)(\"?[uai])([%S]-o)", b="awuا"},
    {a="(aW)(%s)([%(%[%|%<]?)([uai])", b="awuا%2%3%4"}, --p
+   {a="(aw)(o)(\"?[uai])([%S]-o)", b="%1u"},
+   {a="(aw)(%s)([%(%[%|%<]?)(\"?[uai])", b="%1u%2%3ا"}, --p
+   {a="(ay)(o)(\"?[uai])([%S]-o)", b="%1i"},
+   {a="(ay)(%s)([%(%[%|%<]?)(\"?[uai])", b="%1i%2%3ا"}, --p
    -- hyphen + initial alif without hamza:
-   {a="(%-)([uai])([%^%_%.%`]?)(%a)", b="%1ا%3%4"},
-   {a="^([%(%[%|%<]?)([uai])", b="%1ا%2"}, -- initial alif without hamza --p
-   {a="(%s)([%(%[%|%<]?)([uai])", b="%1%2ا"}, -- initial alif without hamza --p
+   {a="([uai]%-)(\"?[uai])([%^%_%.%`]?)(%a)", b="%1ا%3%4"},
+   -- initial alif without hamza
+   {a="^([%(%[%|%<]?)(\"?[uai])", b="%1ا%2"}, --p
+   -- initial alif without hamza
+   {a="(o[%S]-)([uaiUAIY])(o)(\"?[uai])", b="ا"},
+   {a="(%s)([%(%[%|%<]?)(\"?[uai])", b="%1%2ا"}, --p
    {a="%-%-", b="ـ"},
    {a="ؤؤ", b="ؤّ"},
    {a="أأ", b="أّ"},