1 files changed, 53 insertions, 25 deletions

diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index bb928e4..efe26c0 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -66,6 +66,12 @@ 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)
+   {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"},
+   {a="(%W)('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
+   -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
    {a="(%W)(')([ua])", b="%1أ%3"},
@@ -166,6 +172,12 @@ 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)
+   {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"},
+   {a="(%W)('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
+   -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
    {a="(%W)(')([ua])", b="%1أ%3"},
@@ -220,7 +232,10 @@ hamzafveasy = { -- differences marked below with 'easy'
    {a="(i)(')([^uaiUAI])", b="%1ئْ%3"}
 }
 
-tanwinfv = {
+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"},
    {a="%-?uNU", b="ٌو"},
    {a="%-?aNU", b="ًوا"},
    {a="%-?iNU", b="ٍو"},
@@ -256,6 +271,9 @@ tanwinfv = {
 }
 
 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"},
    {a="%-?uNU", b="ٌو"},
    {a="%-?aNU", b="ًوا"},
    {a="%-?iNU", b="ٍو"},
@@ -293,24 +311,29 @@ tanwinfveasy = { -- no assimilations (see below)
 trigraphsfv = { -- trigraphs or more
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="ٱلّ%1%2"},
-   {a="([%p%s%-])'ll(a)([%_]?[dt])", b="%1ٱلّ%2%3"}, --p
+   {a="([%(%[%|%<%s%-])'ll(a)([%_]?[dt])", b="%1ٱلّ%2%3"}, --p
+   -- law: the diphthong is to be resoved into 'awi' (next 4 lines)
+   {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"},
    -- al- + lām
    {a="^(a)l%-(l)", b="ا%1ل%2%2"},
-   {a="([%p%s%-])(a)l%-(l)", b="%1ا%2ل%3%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-(l)", b="%1ا%2ل%3%3"}, --p
    -- al- + solar consonant
    {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="ا%1ل%2%2"},
-   {a="([%p%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
    -- assim. art. + solar consonant
    {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل%2"},
-   {a="([%p%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%3"}, --p
    -- al- + initial unstable hamza
    {a="^(a)l%-(\"?[uai])", b="ا%1ل%2ٱ"},
-   {a="([%p%s%-])(a)l%-(\"?[uai])", b="%1ا%2ل%3ٱ"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-(\"?[uai])", b="%1ا%2ل%3ٱ"}, --p
    -- li-/la + art. + initial unstable hamza is a special orthography
    {a="l([ai])%-l%-(\"?[uai])", b="ل%1ل%2ٱ"},
    -- al- + lunar consonant (i.e. what remains)
    {a="^(a)l%-", b="ا%1لْ"},
-   {a="([%p%s%-])(a)l%-", b="%1ا%2لْ"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-", b="%1ا%2لْ"}, --p
    -- diphthongs to be resolved before ʾalif conjunctionis
    {a="(aW)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awuا%2%3"},
    {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
@@ -350,24 +373,29 @@ trigraphsfv = { -- trigraphs or more
 trigraphsfveasy = { -- trigraphs or more (see 'easy' tag below for the diffs)
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="ٱلّ%1%2"},
-   {a="([%p%s%-])'ll(a)([%_]?[dt])", b="%1ٱلّ%2%3"}, --p
+   {a="([%(%[%|%<%s%-])'ll(a)([%_]?[dt])", b="%1ٱلّ%2%3"}, --p
+   -- law: the diphthong is to be resoved into 'awi' (next 4 lines)
+   {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"},
    -- al- + lām
    {a="^(a)l%-(l)", b="ا%1ل%2%2"},
-   {a="([%p%s%-])(a)l%-(l)", b="%1ا%2ل%3%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-(l)", b="%1ا%2ل%3%3"}, --p
    -- al- + solar consonant
    {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="ا%1ل%2%2"},
-   {a="([%p%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
    -- assim. art. + solar consonant
    {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل%2"},
-   {a="([%p%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%3"}, --p
    -- al- + initial unstable hamza
    {a="^(a)l%-(\"?[uai])", b="ا%1ل%2ٱ"},
-   {a="([%p%s%-])(a)l%-(\"?[uai])", b="%1ا%2ل%3ٱ"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-(\"?[uai])", b="%1ا%2ل%3ٱ"}, --p
    -- li-/la + art. + initial unstable hamza is a special orthography
    {a="l([ai])%-l%-(\"?[uai])", b="ل%1ل%2ٱ"},
    -- al- + lunar consonant (i.e. what remains)
    {a="^(a)l%-", b="ا%1لْ"},
-   {a="([%p%s%-])(a)l%-", b="%1ا%2لْ"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-", b="%1ا%2لْ"}, --p
    -- diphthongs to be resolved before ʾalif conjunctionis
    {a="(aW)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awuا%2%3"},
    {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
@@ -416,16 +444,16 @@ digraphsfvidgham = {
    -- ʾiʿrāb (end)
    -- initial straight double quote gives a connective ʾalif
    {a="^\"[uai]", b="ٱ"},
-   {a="([%p%s%-])\"[uai]", b="%1ٱ"}, --p
+   {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
    -- initial alif without hamza
-   {a="([%_]?[uaiUAIY])(%s)(\"?[uai])", b="%1%2ٱ"},
-   {a="^([uai])", b="ا%1"},      -- initial alif without hamza
-   {a="([%p%s])([uai])", b="%1ا%2"}, -- initial alif without hamza --p
+   {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="%-%-", b="ـ"},
    {a="ؤؤ", b="ؤّ"},
    {a="أأ", b="أّ"},
@@ -518,16 +546,16 @@ digraphsfv = {
    -- ʾiʿrāb (end)
    -- initial straight double quote gives a connective ʾalif
    {a="^\"[uai]", b="ٱ"},
-   {a="([%p%s%-])\"[uai]", b="%1ٱ"}, --p
+   {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
    -- initial alif without hamza
-   {a="([%_]?[uaiUAIY])(%s)(\"?[uai])", b="%1%2ٱ"},
-   {a="^([uai])", b="ا%1"},      -- initial alif without hamza
-   {a="([%p%s])([uai])", b="%1ا%2"}, -- initial alif without hamza --p
+   {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="%-%-", b="ـ"},
    {a="ؤؤ", b="ؤّ"},
    {a="أأ", b="أّ"},
@@ -620,16 +648,16 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
    -- ʾiʿrāb (end)
    -- initial straight double quote gives a connective ʾalif
    {a="^\"[uai]", b="ٱ"},
-   {a="([%p%s%-])\"[uai]", b="%1ٱ"}, --p
+   {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
    -- initial alif without hamza
-   {a="([%_]?[uaiUAIY])(%s)(\"?[uai])", b="%1%2ٱ"},
-   {a="^([uai])", b="ا%1"},      -- initial alif without hamza
-   {a="([%p%s])([uai])", b="%1ا%2"}, -- initial alif without hamza --p
+   {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="%-%-", b="ـ"},
    {a="ؤؤ", b="ؤّ"},
    {a="أأ", b="أّ"},