1 files changed, 111 insertions, 8 deletions

diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index ff2ef35..898aecd 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -341,11 +341,12 @@ trigraphsfv = { -- trigraphs or more
    -- words ending in -āT with silent wāw/yāʾ
    {a="(_a)UA", b="%1وا"},
    {a="(_a)U", b="%1و"},
-   {a="(_a)I", b="%1ي"},
-   -- assimilations
-   {a="(n)(%s)([rlmnwy])", b="%1%2%3%3"}
+   {a="(_a)I", b="%1ي"}
 }
 
+-- idgham/assimilation rules for trigraphs apply here : see voc lua
+-- file.
+
 trigraphsfveasy = { -- trigraphs or more (see 'easy' tag below for the diffs)
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="ٱلّ%1%2"},
@@ -397,12 +398,10 @@ trigraphsfveasy = { -- trigraphs or more (see 'easy' tag below for the diffs)
    -- words ending in -āT with silent wāw/yāʾ
    {a="(_a)UA", b="%1وا"},
    {a="(_a)U", b="%1و"},
-   {a="(_a)I", b="%1ي"},
-   -- assimilations
---easy   {a="(n)(%s)([rlmnwy])", b="%1%2%3%3"}
+   {a="(_a)I", b="%1ي"}
 }
 
-digraphsfv = {
+digraphsfvidgham = {
    -- iʿrāb: straight double quote must be discarded
    {a="(%-)(\"?[UI]na)(%p?%s)", b="%2%3"},
    {a="(%-)(\"?[UI]na)(%p?)$", b="%2%3"},
@@ -504,6 +503,108 @@ digraphsfv = {
    {a="%^d", b="ڊ"}
 }
 
+digraphsfv = {
+   -- iʿrāb: straight double quote must be discarded
+   {a="(%-)(\"?[UI]na)(%p?%s)", b="%2%3"},
+   {a="(%-)(\"?[UI]na)(%p?)$", b="%2%3"},
+   {a="(%-)(\"?At[ui])(%p?%s)", b="%2%3"},
+   {a="(%-)(\"?At[ui])(%p?)$", b="%2%3"},
+   {a="(%-)(\"?Ani)(%p?%s)", b="%2%3"},
+   {a="(%-)(\"?Ani)(%p?)$", b="%2%3"},
+   {a="(%-)(\"?ayni)(%p?%s)", b="%2%3"},
+   {a="(%-)(\"?ayni)(%p?)$", b="%2%3"},
+   {a="(%-)(\"?[uai])(%p?%s)", b="%2%3"},
+   {a="(%-)(\"?[uai])(%p?)$", b="%2%3"},
+   -- ʾiʿrāb (end)
+   -- initial straight double quote gives a connective ʾalif
+   {a="^\"[uai]", b="ٱ"},
+   {a="([%s%-])\"[uai]", b="%1ٱ"},
+   -- diphthongs to be resolved before ʾalif conjunctionis
+   {a="(aW)(%s)(\"?[uai])", b="awuا%2ٱ"},
+   {a="(aw)(%s)(\"?[uai])", b="%1u%2ٱ"},
+   {a="(ay)(%s)(\"?[uai])", b="%1i%2ٱ"},
+   {a="([uai]%-)(\"?[uai])", b="%1ٱ"}, -- hyphen + initial alif without hamza
+   -- initial alif without hamza
+   {a="([%_]?[uaiUAIY])(%s)(\"?[uai])", b="%1%2ٱ"},
+   {a="^([uai])", b="ا%1"},      -- initial alif without hamza
+   {a="(%s)([uai])", b="%1ا%2"}, -- initial alif without hamza
+   {a="%-%-", b="ـ"},
+   {a="ؤؤ", b="ؤّ"},
+   {a="أأ", b="أّ"},
+   {a="ئئ", b="ئّ"},
+   {a="bb", b="بّ"},
+   {a="BB", b="ـّ"},
+   {a="(%_)([thd])([thd])", b="%1%2|%3"},
+   {a="tt", b="تّ"},
+   {a="%_t%_t", b="ثّ"},
+   {a="jj", b="جّ"},
+   {a="%^g%^g", b="جّ"},
+   {a="%.h%.h", b="حّ"},
+   {a="xx", b="خّ"},
+   {a="%_h%_h", b="خّ"},
+   {a="dd", b="دّ"},
+   {a="%_d%_d", b="ذّ"},
+   {a="rr", b="رّ"},
+   {a="zz", b="زّ"},
+   {a="ss", b="سّ"},
+   {a="%^s%^s", b="شّ"},
+   {a="%.s%.s", b="صّ"},
+   {a="%.d%.d", b="ضّ"},
+   {a="%.t%.t", b="طّ"},
+   {a="%.z%.z", b="ظّ"},
+   {a="%`%`", b="عّ"},
+   {a="%.g%.g", b="غّ"},
+   {a="ff", b="فّ"},
+   {a="qq", b="قّ"},
+   {a="kk", b="كّ"},
+   {a="ll", b="لّ"},
+   {a="mm", b="مّ"},
+   {a="nn", b="نّ"},
+   {a="hh", b="هّ"},
+   {a="ww", b="وّ"},
+   {a="yy", b="يّ"},
+   -- sukūn begin
+   -- first, take out hyphen if any:
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-([uaiUAI])", b="%1%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])$", b="%1ْ"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%s])", b="%1ْ%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIYًٌٍ])", b="%1ْ%2"},
+   -- take out sukūn in cases of assimilation
+--   {a="(n)(ْ)(%s)(ر)", b="%1%3%4"},
+--   {a="(n)(ْ)(%s)(و)", b="%1%3%4"},
+--   {a="(n)(ْ)(%s)(ي)", b="%1%3%4"},
+--   {a="(n)(ْ)(%s)(ل)", b="%1%3%4"},
+--   {a="(n)(ْ)(%s)(م)", b="%1%3%4"},
+--   {a="(n)(ْ)(%s)(ن)", b="%1%3%4"},
+--   {a="ْ\"", b="\""},
+   -- sukūn end
+   {a="_t", b="ث"},
+   {a="%^g", b="ج"},
+   {a="%.h", b="ح"},
+   {a="_h", b="خ"},
+   {a="_d", b="ذ"},
+   {a="%^s", b="ش"},
+   {a="%.s", b="ص"},
+   {a="%.d", b="ض"},
+   {a="%.t", b="ط"},
+   {a="%.z", b="ظ"},
+   {a="%.g", b="غ"},
+   {a="(U)(A)", b="%1ا"},
+   {a="WA", b="وْا"},
+   {a="(a)W\"", b="%1وا"},
+   {a="(a)W", b="%1وْا"},
+   {a="_A", b="aى"},
+   {a="_u", b="ٗ"},
+   {a="_a", b="ٰ"},
+   {a="_i", b="ٖ"},
+   {a="%.b", b="ٮ"},
+   {a="%.f", b="ڡ"},
+   {a="%.q", b="ٯ"},
+   {a="%.k", b="ک"},
+   {a="%.n", b="ں"},
+   {a="%^d", b="ڊ"}
+}
+
 digraphsfveasy = { -- see the differences under 'easy' marker below
    -- iʿrāb: straight double quote must be discarded
    {a="(%-)(\"?[UI]na)(%p?%s)", b="%2%3"},
@@ -574,7 +675,9 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
 --   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%s])", b="%1ْ%2"},
 --   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIًٌٍ])", b="%1ْ%2"},
    -- take out sukūn in cases of assimilation
---   {a="(n)(ْ)(%s)([روي])", b="%1%3%4"},
+--   {a="(n)(ْ)(%s)(ر)", b="%1%3%4"},
+--   {a="(n)(ْ)(%s)(و)", b="%1%3%4"},
+--   {a="(n)(ْ)(%s)(ي)", b="%1%3%4"},
 --   {a="(n)(ْ)(%s)([ل])", b="%1%3%4"},
 --   {a="(n)(ْ)(%s)([م])", b="%1%3%4"},
 --   {a="(n)(ْ)(%s)([ن])", b="%1%3%4"},