arbnulls inserted in voc table

1 files changed, 64 insertions, 23 deletions

diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index 2f5e445..fb53b5a 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -202,7 +202,11 @@ 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 4 lines)
+   -- 'aw: the diphthong is to be resoved 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"}, --p
@@ -299,7 +303,11 @@ hamzaeasy = { -- 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 resoved 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"},
@@ -360,11 +368,17 @@ hamzaeasy = { -- differences marked below with 'easy'
 }
 
 tanwin = {
+   -- 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", b="ٌو"},
    {a="%-?aNU", b="ًوا"},
    {a="%-?iNU", b="ٍو"},
    -- 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"},
@@ -393,11 +407,17 @@ tanwin = {
 }
 
 tanwineasy = { -- 'easy' requires some lines to be taken out:
+   -- 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", b="ٌو"},
    {a="%-?aNU", b="ًوا"},
    {a="%-?iNU", b="ٍو"},
    -- 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"},
@@ -429,35 +449,46 @@ trigraphs = { -- trigraphs or more
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="الّ%1%2"},
    {a="([%(%[%|%<%s%-])'ll(a)([%_]?[dt])", b="%1الّ%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"},
-   {a="(%W)(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
+   {a="^(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"}, --p
+   {a="(%W)(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"}, --p
    -- al- + lām
    {a="^(a)l%-(l)", b="ا%1ل%2%2"},
-   {a="([%(%[%|%<%s%-])(a)l%-(l)", b="%1ا%2ل%3%3"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-(l)", b="%1ا%2ل%3%3"}, --p
    -- al- + solar consonant
    {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="ا%1ل%2%2"},
-   {a="([%(%[%|%<%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
    -- assim. art. + solar consonant
    {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل%2"},
-   {a="([%(%[%|%<%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%3"}, --p
+   {a="([%(%[%|%<%s%-o])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%3"}, --p
    -- al- + initial unstable hamza
    {a="^(a)l%-(\")([uai])", b="ا%1ل%3ٱ"},
-   {a="([%(%[%|%<%s%-])(a)l%-(\")([uai])", b="%1ا%2ل%4ٱ"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-(\")([uai])", b="%1ا%2ل%4ٱ"}, --p
    {a="^(a)l%-([uai])", b="ا%1ل%2ا"},
-   {a="([%(%[%|%<%s%-])(a)l%-([uai])", b="%1ا%2ل%3ا"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-([uai])", b="%1ا%2ل%3ا"}, --p
    -- li-/la- + art. + initial unstable hamza is a special orthography
    {a="l([ai])%-l%-(\")([uai])", b="ل%1ل%3ٱ"},
    {a="l([ai])%-l%-([uai])", b="ل%1ل%2ا"},
    -- al- + lunar consonant (i.e. what remains)
    {a="^(a)l%-", b="ا%1ل"},
-   {a="([%(%[%|%<%s%-])(a)l%-", b="%1ا%2ل"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-", b="%1ا%2ل"}, --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)(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
@@ -490,7 +521,8 @@ trigraphs = { -- trigraphs or more
 
 idgham = {
    -- assimilations
-   {a="(n)(%s)([rlmnwy])", b="%1%2%3%3"}
+   {a="(n)(%s)([rlmnwy])", b="%1%2%3%3"},
+   {a="(n)(o)([rlmnwy])([%S]-o)", b="%3"}
 }
 
 trigraphseasy = { -- differences marked below with 'easy'
@@ -499,32 +531,39 @@ trigraphseasy = { -- differences marked below with 'easy'
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="الّ%1%2"},
    {a="([%(%[%|%<%s%-])'ll(a)([%_]?[dt])", b="%1الّ%2%3"}, --p
-   -- law: the diphthong is to be resoved into 'awi' (next 4 lines)
+   -- law: the diphthong is to be resloved 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"},
-   {a="(%W)(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
+   {a="^(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"}, --p
+   {a="(%W)(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"}, --p
    -- al- + lām (easy)
    {a="^(a)l%-(l)", b="ا%1ل%2"},
-   {a="([%(%[%|%<%s%-])(a)l%-(l)", b="%1ا%2ل%3"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-(l)", b="%1ا%2ل%3"}, --p
    -- al- + solar consonant (easy)
    {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="ا%1ل%2"},
-   {a="([%(%[%|%<%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3"}, --p
    -- assim. art. + solar consonant (easy)
    {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل"},
-   {a="([%(%[%|%<%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل"}, --p
+   {a="([%(%[%|%<%s%-o])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل"}, --p
    -- al- + initial unstable hamza
    {a="^(a)l%-(\")([uai])", b="ا%1ل%3ٱ"},
-   {a="([%(%[%|%<%s%-])(a)l%-(\")([uai])", b="%1ا%2ل%4ٱ"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-(\")([uai])", b="%1ا%2ل%4ٱ"}, --p
    {a="^(a)l%-([uai])", b="ا%1ل%2ا"},
-   {a="([%(%[%|%<%s%-])(a)l%-([uai])", b="%1ا%2ل%3ا"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-([uai])", b="%1ا%2ل%3ا"}, --p
    -- li-/la- + art. + initial unstable hamza is a special orthography
    {a="l([ai])%-l%-(\")([uai])", b="ل%1ل%3ٱ"},
    {a="l([ai])%-l%-([uai])", b="ل%1ل%2ا"},
    -- al- + lunar consonant (i.e. what remains)
    {a="^(a)l%-", b="ا%1ل"},
-   {a="([%(%[%|%<%s%-])(a)l%-", b="%1ا%2ل"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-", b="%1ا%2ل"}, --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"},
@@ -574,7 +613,8 @@ digraphs = {
    -- initial straight double quote gives a connective ʾalif
    {a="^\"[uai]", b="ٱ"},
    {a="([%(%[%|%<%s%-])\"[uai]", b="%1ٱ"}, --p
-   {a="(aW)(%s)([uai])", b="awuا%2%3"},
+   {a="(aW)(o)([uai]([%S]-o))", b="awuا"},
+   {a="(aW)(%s)([%(%[%|%<]?)([uai])", b="awuا%2%3%4"}, --p
    -- hyphen + initial alif without hamza:
    {a="(%-)([uai])([%^%_%.%`]?)(%a)", b="%1ا%3%4"},
    {a="^([%(%[%|%<]?)([uai])", b="%1ا%2"}, -- initial alif without hamza --p
@@ -716,5 +756,6 @@ punctuation = {
 
 null = {
    {a="%|", b=""},
-   {a="([^0-9])(%-)", b="%1"}
+   {a="([^0-9])(%-)", b="%1"},
+   {a="o[%S]-o", b=""}
 }