1 files changed, 66 insertions, 21 deletions

diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index 073b23f..6c63a88 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -471,12 +471,16 @@ trigraphs = { -- trigraphs or more
    -- al- + lām
    {a="^(a)l%-(l)", b="ا%1ل%2%2"},
    {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%-o])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
-   -- assim. art. + solar consonant
-   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل%2"},
-   {a="([%(%[%|%<%s%-o])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%3"}, --p
+   -- al- + solar consonant ('c' and '^n' are additional characters)
+   {a="^(a)l%-(%^n)", b="ا%1ل%2"}, -- ^n is lunar
+   {a="([%(%[%|%<%s%-o])(a)l%-(%^n)", b="%1ا%2ل%3"},-- ^n is lunar --p
+   {a="^(a)l%-([%_%^%.]?[tdrzsnc])", b="ا%1ل%2%2"},
+   {a="([%(%[%|%<%s%-o])(a)l%-([%_%^%.]?[tdrzsnc])", b="%1ا%2ل%3%3"}, --p
+   -- assim. art. + solar consonant ('c' and '^n' are additional characters)
+   {a="^(a)(%^n)%-", b="ا%1ل"}, -- ^n is lunar
+   {a="([%(%[%|%<%s%-o])(a)(%^n)%-", b="%1ا%2ل"},-- ^n is lunar --p
+   {a="^(a)([%_%^%.]?[tdrzsnc])%-", b="ا%1ل%2"},
+   {a="([%(%[%|%<%s%-o])(a)([%_%^%.]?[tdrzsnc])%-", b="%1ا%2ل%3"}, --p
    -- al- + initial unstable hamza
    {a="^(a)l%-(\")([uai])", b="ا%1ل%3ٱ"},
    {a="([%(%[%|%<%s%-o])(a)l%-(\")([uai])", b="%1ا%2ل%4ٱ"}, --p
@@ -498,15 +502,23 @@ trigraphs = { -- trigraphs or more
    -- art. with waṣla + lām
    {a="'l%-(l)", b="ال%1%1"},
    -- art. with waṣla + solar consonant
-   {a="'l%-([%_%^%.]?[tdrzsn])", b="ال%1%1"},
+   -- ('c' and '^n' are additional characters)
+   {a="'l%-(%^n)", b="ال%1"}, -- ^n is lunar
+   {a="'l%-([%_%^%.]?[tdrzsnc])", b="ال%1%1"},
    -- li-/la- + art. + lām
    {a="l([ai])%-l%-(l)", b="ل%1%2%2"},
-   -- assim. art. with waṣla + solar consonant
-   {a="'([%_%^%.]?[tdrzsn])%-", b="ال%1"},
+   -- assim. art. with waṣla + solar consonant ('c' and '^n' are
+   -- additional characters)   
+   {a="'(%^n)%-", b="ال"}, -- ^n is lunar
+   {a="'([%_%^%.]?[tdrzsnc])%-", b="ال%1"},
    -- li-/la- + art. + solar consonant is a special orthography
-   {a="l([ai])%-l%-([%_%^%.]?[tdrzsn])", b="ل%1ل%2%2"},
+   -- ('c' and '^n' are additional characters)
+   {a="l([ai])%-l%-(%^n)", b="ل%1ل%2"}, -- '^n' is lunar
+   {a="l([ai])%-l%-([%_%^%.]?[tdrzsnc])", b="ل%1ل%2%2"},
    -- li-/la + assim. art. + solar consonant is a special orthography
-   {a="l([ai])%-([%_%^%.]?[tdrzsn])%-([%_%^%.]?[tdrzsn])", b="ل%1ل%3%3"},
+   -- ('c' and '^n' are additional characters)
+   {a="l([ai])%-(%^n)%-(%^n)", b="ل%1ل%3"}, -- ^n is lunar
+   {a="l([ai])%-([%_%^%.]?[tdrzsnc])%-([%_%^%.]?[tdrzsnc])", b="ل%1ل%3%3"},
    -- art. with waṣla + initial unstable hamza
    {a="'l%-(\")([uai])", b="ال%2ٱ"},
    {a="'l%-([uai])", b="ال%1ا"},   
@@ -549,12 +561,17 @@ trigraphseasy = { -- differences marked below with 'easy'
    -- al- + lām (easy)
    {a="^(a)l%-(l)", b="ا%1ل%2"},
    {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%-o])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3"}, --p
-   -- assim. art. + solar consonant (easy)
-   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل"},
-   {a="([%(%[%|%<%s%-o])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل"}, --p
+   -- al- + solar consonant (easy) ('c' and '^n' are additional characters)
+   {a="^(a)l%-(%^n)", b="ا%1ل%2"}, -- ^n is lunar
+   {a="([%(%[%|%<%s%-o])(a)l%-(%^n)", b="%1ا%2ل%3"}, -- ^n is lunar --p
+   {a="^(a)l%-([%_%^%.]?[tdrzsnc])", b="ا%1ل%2"},
+   {a="([%(%[%|%<%s%-o])(a)l%-([%_%^%.]?[tdrzsnc])", b="%1ا%2ل%3"}, --p
+   -- assim. art. + solar consonant (easy) ('c' and '^n' are
+   -- additional characters)
+   {a="^(a)(%^n)%-", b="ا%1ل"}, -- ^n is lunar
+   {a="([%(%[%|%<%s%-o])(a)(%^n)%-", b="%1ا%2ل"}, -- ^n is lunar --p
+   {a="^(a)([%_%^%.]?[tdrzsnc])%-", b="ا%1ل"},
+   {a="([%(%[%|%<%s%-o])(a)([%_%^%.]?[tdrzsnc])%-", b="%1ا%2ل"}, --p
    -- al- + initial unstable hamza
    {a="^(a)l%-(\")([uai])", b="ا%1ل%3ٱ"},
    {a="([%(%[%|%<%s%-o])(a)l%-(\")([uai])", b="%1ا%2ل%4ٱ"}, --p
@@ -576,15 +593,23 @@ trigraphseasy = { -- differences marked below with 'easy'
    -- art. with waṣla + lām (easy)
    {a="'l%-(l)", b="ال%1"},
    -- art. with waṣla + solar consonant (easy)
-   {a="'l%-([%_%^%.]?[tdrzsn])", b="ال%1"},
+   -- ('c' and '^n' are additional characters)
+   {a="'l%-(%^n)", b="ال%1"}, -- ^n is lunar
+   {a="'l%-([%_%^%.]?[tdrzsnc])", b="ال%1"},
    -- li-/la- + art. + lām (easy)
    {a="l([ai])%-l%-(l)", b="ل%1%2"},
    -- assim. art. with waṣla + solar consonant (easy)
-   {a="'([%_%^%.]?[tdrzsn])%-", b="ال"},
+   -- ('c' and '^n' are additional characters)
+   {a="'(%^n)%-", b="ال"}, -- ^n is lunar
+   {a="'([%_%^%.]?[tdrzsnc])%-", b="ال"},
    -- li-/la- + art. + solar consonant is a special orthography (easy)
-   {a="l([ai])%-l%-([%_%^%.]?[tdrzsn])", b="ل%1ل%2"},
+   -- ('c' and '^n' are additional characters)
+   {a="l([ai])%-l%-(%^n)", b="ل%1ل%2"}, -- ^n is lunar
+   {a="l([ai])%-l%-([%_%^%.]?[tdrzsnc])", b="ل%1ل%2"},
    -- li-/la + assim. art. + solar consonant is a special orthography (easy)
-   {a="l([ai])%-([%_%^%.]?[tdrzsn])%-([%_%^%.]?[tdrzsn])", b="ل%1ل%3"},
+   -- ('c' and '^n' are additional characters)
+   {a="l([ai])%-(%^n)%-(%^n)", b="ل%1ل%3"}, -- ^n is lunar
+   {a="l([ai])%-([%_%^%.]?[tdrzsnc])%-([%_%^%.]?[tdrzsnc])", b="ل%1ل%3"},
    -- art. with waṣla + initial unstable hamza
    {a="'l%-(\")([uai])", b="ال%2ٱ"},
    {a="'l%-([uai])", b="ال%1ا"},
@@ -640,6 +665,8 @@ digraphs = {
    {a="bb", b="بّ"},
    {a="BB", b="ـّ"},
    {a="([%_%^%.])([tghdsz])([tghdsz])", b="%1%2|%3"},
+   -- same as above for additional characters:
+   {a="([%_%^%.])([cn])([cn])", b="%1%2|%3"},
    {a="tt", b="تّ"},
    {a="%_t%_t", b="ثّ"},
    {a="jj", b="جّ"},
@@ -668,6 +695,14 @@ digraphs = {
    {a="hh", b="هّ"},
    {a="ww", b="وّ"},
    {a="yy", b="يّ"},
+   -- additional characters + šaddah (begin)
+   {a="pp", b="پّ"},
+   {a="vv", b="ڤّ"},
+   {a="gg", b="گّ"},
+   {a="%^c%^c", b="چّ"},
+   {a="%^z%^z", b="ژّ"},
+   {a="%^n%^n", b="ڭّ"},
+   -- additional characters + šaddah (end)
    {a="_t", b="ث"},
    {a="%^g", b="ج"},
    {a="%.h", b="ح"},
@@ -679,6 +714,11 @@ digraphs = {
    {a="%.t", b="ط"},
    {a="%.z", b="ظ"},
    {a="%.g", b="غ"},
+   -- additional characters (begin)
+   {a="%^c", b="چ"},
+   {a="%^z", b="ژ"},
+   {a="%^n", b="ڭ"},
+   -- additional characters (end)
    {a="(U)(A)", b="%1ا"},
    {a="WA", b="وا"},
    {a="(a)W\"", b="%1وْا"},
@@ -716,6 +756,11 @@ single = {
    {a="w", b="و"},
    {a="y", b="ي"},
    {a="T", b="ة"},
+   -- additional characters (begin)
+   {a="p", b="پ"},
+   {a="v", b="ڤ"},
+   {a="g", b="گ"},
+   -- additional characters (end)
    {a="\"$", b="ْ"},
    {a="\"(%W)", b="ْ%1"},
    {a="\"([^uaiUAI])", b="ْ%1"},