assimilations [Wright i §14 (b)]; no more euphonic tašdīd in novoc mode

5 files changed, 90 insertions, 9 deletions

diff --git a/arabluatex.lua b/arabluatex.lua
index 8ebc30b..de33bb3 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -204,8 +204,8 @@ local function novoc(str)
         for i = 1,#tanwinnv do
           inside  = string.gsub(inside, tanwinnv[i].a, tanwinnv[i].b)
         end
-        for i = 1,#trigraphs do
-                inside = string.gsub(inside, trigraphs[i].a, trigraphs[i].b)
+        for i = 1,#trigraphsnv do
+                inside = string.gsub(inside, trigraphsnv[i].a, trigraphsnv[i].b)
         end
         for i = 1,#digraphs do
                 inside = string.gsub(inside, digraphs[i].a, digraphs[i].b)
diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index 472583e..b0e00d4 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -164,7 +164,9 @@ 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"}
+   {a="(_a)I", b="%1ي"},
+   -- assimilations
+   {a="(n)(%s)([rlmnwy])", b="%1%2%3%3"}
 }
 
 -- this is new
@@ -210,6 +212,11 @@ digraphsfv = {
    {a="([%_%^%.]?[btjghxdrzs%`fqklmnwy])$", b="%1ْ"},
    {a="([%_%^%.]?[btjghxdrzs%`fqklmnwy])([%s])", b="%1ْ%2"},
    {a="([%_%^%.]?[btjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIًٌٍ])", b="%1ْ%2"},
+   -- take out sukūn in cases of assimilation
+   {a="(n)(ْ)(%s)([روي])", 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"},
diff --git a/arabluatex_novoc.lua b/arabluatex_novoc.lua
index 4710ccc..4f08d68 100644
--- a/arabluatex_novoc.lua
+++ b/arabluatex_novoc.lua
@@ -22,18 +22,79 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
 02111-1307, USA.
 --]]
 
+trigraphsnv = { -- trigraphs or more
+   -- Allah
+   {a="l%-l_ah", b="l-ll_ah"},
+   -- 'llatI / 'llad_I
+   {a="^'ll(a)([%_]?[dt])", b="ال%1%2"},
+   {a="(%s)'ll(a)([%_]?[dt])", b="%1ال%2%3"},
+   -- al- + lām
+   {a="^(a)l%-(l)", b="ا%1ل%2"},
+   {a="(%s)(a)l%-(l)", b="%1ا%2ل%3"},
+   -- al- + solar consonant
+   {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="ا%1ل%2"},
+   {a="(%s)(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3"},
+   -- assim. art. + solar consonant
+   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل"},
+   {a="(%s)(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل"},
+   -- al- + initial unstable hamza
+   {a="^(a)l%-([uai])", b="ا%1لا%2"},
+   {a="(%s)(a)l%-([uai])", b="%1ا%2لا%3"},
+   -- li-/la- + art. + initial unstable hamza is a special orthography
+   {a="l([ai])%-l%-([uai])", b="ل%1لا%2"},
+   -- al- + lunar consonant (i.e. what remains)
+   {a="^(a)l%-", b="ا%1ل"},
+   {a="(%s)(a)l%-", b="%1ا%2ل"},
+   -- art. with waṣla + lām
+   {a="'l%-(l)", b="ال%1"},
+   -- art. with waṣla + solar consonant
+   {a="'l%-([%_%^%.]?[tdrzsn])", b="ال%1"},
+   -- li-/la- + art. + lām
+   {a="l([ai])%-l%-(l)", b="ل%1%2"},
+   -- assim. art. with waṣla + solar consonant
+   {a="'([%_%^%.]?[tdrzsn])%-", b="ال"},
+   -- li-/la- + art. + solar consonant is a special orthography
+   {a="l([ai])%-l%-([%_%^%.]?[tdrzsn])", b="ل%1ل%2"},
+   -- li-/la + assim. art. + solar consonant is a special orthography
+   {a="l([ai])%-([%_%^%.]?[tdrzsn])%-([%_%^%.]?[tdrzsn])", b="ل%1ل%3"},
+   -- art. with waṣla + initial unstable hamza
+   {a="'l%-([uai])", b="الا%1"},   
+   -- art. with waṣla + lunar consonant (i.e. what remains)
+   {a="'l%-", b="ال"},
+   -- the silent wāw
+   {a="uU$", b="uو"},
+   {a="uU(%W)", b="uو%1"},
+   {a="aU$", b="aو"},
+   {a="aU(%W)", b="aو%1"},
+   {a="iU$", b="iو"},
+   {a="iU(%W)", b="iو%1"},
+   -- words ending in -āT with silent wāw/yāʾ
+   {a="(_a)UA", b="%1وا"},
+   {a="(_a)U", b="%1و"},
+   {a="(_a)I", b="%1ي"}
+}
+
 tanwinnv = {
    {a="uNU", b="و"},
    {a="aNU", b="وا"},
    {a="iNU", b="و"},
-   {a="uN", b=""},
+   -- assimilations (begin)
+--   {a="(uN)(%s)([rlmnwy])", b="|%2%3%3"},
+--   {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="%1%3%4%4"},
+--   {a="(ء)(aN)(%s)([rlmnwy])", b="%1%3%4%4"},
+--   {a="([^TA])(aN)(%s)([rlmnwy])", b="%1ا%3%4%4"},
+--   {a="(iN)(%s)([rlmnwy])", b="|%2%3%3"},
+   --assimilations (end)
+   {a="(uN)", b=""},
    {a="(B)(aN)", b="%1"},
    {a="(aN)(_A)", b="ى"},
    {a="(aN)(Y)", b="ى"},
    {a="(T)(aN)", b="%1"},
    {a="(ء)(aN)", b="%1"},
    {a="([^TA])(aN)", b="%1ا"},
-   {a="iN", b=""}
+   {a="(iN)", b=""}
 }
 
 longvnv = {
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index 3a69463..1e2d89a 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -135,7 +135,9 @@ trigraphstrdmg = { -- trigraphs or more
    -- words ending in -āT with silent wāw/yāʾ
    {a="(_a)UA", b="A"},
    {a="(_a)U", b="A"},
-   {a="(_a)I", b="A"}
+   {a="(_a)I", b="A"},
+   -- assimilations
+   {a="(n)(%s)([rlmnwy])", b="%3%2%3"}
 }
 
 tanwintrdmg = {
diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index a3c9b62..b6dbd12 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -195,21 +195,32 @@ trigraphs = { -- 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"}
+   {a="(_a)I", b="%1ي"},
+   -- assimilations
+   {a="(n)(%s)([rlmnwy])", b="%1%2%3%3"}
 }
 
 tanwin = {
    {a="uNU", b="ٌو"},
    {a="aNU", b="ًوا"},
    {a="iNU", b="ٍو"},
-   {a="uN", b="ٌ"},
+   -- assimilations (begin)
+   {a="(uN)(%s)([rlmnwy])", b="ٌ%2%3%3"},
+   {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"},
+   {a="(ء)(aN)(%s)([rlmnwy])", b="%1ً%3%4%4"},
+   {a="([^TA])(aN)(%s)([rlmnwy])", b="%1ًا%3%4%4"},
+   {a="(iN)(%s)([rlmnwy])", b="ٍ%2%3%3"},
+   -- assimilations (end)
+   {a="(uN)", b="ٌ"},
    {a="(B)(aN)", b="%1ً"},
    {a="(aN)(_A)", b="ًى"},
    {a="(aN)(Y)", b="ًى"},
    {a="(T)(aN)", b="%1ً"},
    {a="(ء)(aN)", b="%1ً"},
    {a="([^TA])(aN)", b="%1ًا"},
-   {a="iN", b="ٍ"}
+   {a="(iN)", b="ٍ"}
 }
 
 digraphs = {