1 files changed, 99 insertions, 20 deletions
diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index 6d63192..99efb75 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -354,12 +354,16 @@ trigraphsfv = { -- 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
+   -- al- + solar consonant ('c' and '^n' are additional characters)
-   {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="ا%1ل%2%2"},
+   {a="^(a)l%-(%^n)", b="ا%1ل%2"}, -- ^n is lunar
-   {a="([%(%[%|%<%s%-o])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-(%^n)", b="%1ا%2ل%3"}, -- ^n is lunar --p
-   -- assim. art. + solar consonant
+   {a="^(a)l%-([%_%^%.]?[tdrzsnc])", b="ا%1ل%2%2"},
-   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل%2"},
+   {a="([%(%[%|%<%s%-o])(a)l%-([%_%^%.]?[tdrzsnc])", b="%1ا%2ل%3%3"}, --p
-   {a="([%(%[%|%<%s%-o])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%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%2ٱ"},
   {a="([%(%[%|%<%s%-o])(a)l%-(\"?[uai])", b="%1ا%2ل%3ٱ"}, --p
@@ -378,15 +382,23 @@ trigraphsfv = { -- trigraphs or more
   -- art. with waṣla + lām
   {a="'l%-(l)", b="ٱل%1%1"},
   -- art. with waṣla + solar consonant
-   {a="'l%-([%_%^%.]?[tdrzsn])", b="ٱل%1%1"},
+   -- ('c' and '^n' are additional characters)
+   {a="'l%-(%^n)", b="ٱل%1"}, -- ^n is lunar
+   {a="'l%-([%_%^%.]?[tdrzsnc])", b="ٱل%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="ٱل%1"},
+   -- ('c' and '^n' are additional characters)
+   {a="'(%^n)%-", b="ٱل"}, -- ^n is lunar
+   {a="'([%_%^%.]?[tdrzsnc])%-", b="ٱل%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="ٱل%1ٱ"},
   -- art. with waṣla + lunar consonant (i.e. what remains)
@@ -423,12 +435,16 @@ trigraphsfveasy = { -- trigraphs or more (see 'easy' tag below for the diffs)
   -- 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
+   -- al- + solar consonant ('c' and '^n' are additional characters)
-   {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="ا%1ل%2%2"},
+   {a="^(a)l%-(%^n)", b="ا%1ل%2"}, -- ^n is lunar
-   {a="([%(%[%|%<%s%-o])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
+   {a="([%(%[%|%<%s%-o])(a)l%-(%^n)", b="%1ا%2ل%3"}, -- ^n is lunar --p
-   -- assim. art. + solar consonant
+   {a="^(a)l%-([%_%^%.]?[tdrzsnc])", b="ا%1ل%2%2"},
-   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل%2"},
+   {a="([%(%[%|%<%s%-o])(a)l%-([%_%^%.]?[tdrzsnc])", b="%1ا%2ل%3%3"}, --p
-   {a="([%(%[%|%<%s%-o])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%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%2ٱ"},
   {a="([%(%[%|%<%s%-o])(a)l%-(\"?[uai])", b="%1ا%2ل%3ٱ"}, --p
@@ -447,15 +463,23 @@ trigraphsfveasy = { -- trigraphs or more (see 'easy' tag below for the diffs)
   -- art. with waṣla + lām
   {a="'l%-(l)", b="ٱل%1%1"},
   -- art. with waṣla + solar consonant
-   {a="'l%-([%_%^%.]?[tdrzsn])", b="ٱل%1%1"},
+   -- ('c' and '^n' are additional characters)
+   {a="'l%-(%^n)", b="ٱل%1"}, -- ^n is lunar
+   {a="'l%-([%_%^%.]?[tdrzsnc])", b="ٱل%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="ٱل%1"},
+   -- ('c' and '^n' are is an additional characters)
+   {a="'(%^n)%-", b="ٱل"}, -- ^n is lunar
+   {a="'([%_%^%.]?[tdrzsnc])%-", b="ٱل%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="ٱل%1ٱ"},   
   -- art. with waṣla + lunar consonant (i.e. what remains)
@@ -511,6 +535,8 @@ digraphsfvidgham = {
   {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"},
@@ -539,6 +565,14 @@ digraphsfvidgham = {
   {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)
   -- sukūn begin
   -- first, take out hyphen if any:
   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-([uaiUAI])", b="%1%2"},
@@ -565,6 +599,11 @@ digraphsfvidgham = {
   {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="وْا"},
   {a="(a)W\"", b="%1وا"},
@@ -619,6 +658,8 @@ digraphsfv = {
   {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"},
@@ -647,6 +688,14 @@ digraphsfv = {
   {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)
   -- sukūn begin
   -- first, take out hyphen if any:
   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-([uaiUAI])", b="%1%2"},
@@ -673,6 +722,11 @@ digraphsfv = {
   {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="وْا"},
   {a="(a)W\"", b="%1وا"},
@@ -727,6 +781,8 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
   {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"},
@@ -755,6 +811,14 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
   {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)
   -- sukūn begin ('easy' needs these rules to be taken out); but
   -- first take out every previously generated sukūn by hamza rules,
   -- so there be no need to edit them:
@@ -784,6 +848,11 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
   {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="وْا"},
   {a="(a)W\"", b="%1وا"},
@@ -821,6 +890,11 @@ singlefv = {
   {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="\"$", b=""},
   {a="\"(%W)", b="%1"},
   {a="\"([^uaiUAI])", b="%1"},
@@ -849,6 +923,11 @@ singlefveasy = { -- see the differences under 'easy' tag below
   {a="w", b="و"},
   {a="y", b="ي"},
   {a="T", b="ة"},
+   -- additional characters (begin)
+   {a="p", b="پ"},
+   {a="v", b="ڤ"},
+   {a="g", b="گ"},
+   -- additional characters (end)
   -- easy (begin): \" needs to put back the sukūn
   {a="\"$", b="ْ"},
   {a="\"(%W)", b="ْ%1"},