new option \SetArbDflt* for applying the assimilation rules for these are not applied by default anymore

5 files changed, 183 insertions, 126 deletions

diff --git a/arabluatex.dtx b/arabluatex.dtx
index 0c5d2a1..4bd9057 100644
--- a/arabluatex.dtx
+++ b/arabluatex.dtx
@@ -2751,17 +2751,28 @@ wa-ya.sIru ta.hta 'l-jild-i
 %   \arb[trans]{sukUn} that is generated, while the starred version
 %   \cs{SetArbEasy*} takes it away. Default complex rules can be set
 %   back at any point of the document with \cs{SetArbDflt}.
+% \begin{macro}{\SetArbDflt*}
+%   \changes{v1.6}{2016/11/30}{This starred version applies the
+%   assimilation rules (\arb[trans]{al-'id.gAm}) in addition to what
+%   \cs{SetArbDflt} already does.} As of v1.6, \package{arabluatex}
+%   does not applies any more the assimilation rules that are laid on
+%   \vref{ref:assimilation}; a new starred version \cs{SetArbDflt*} is
+%   now available to the user should he wish to apply them.
 %    \begin{macrocode}
 \def\al@arb@rules{dflt}
 \NewDocumentCommand{\SetArbEasy}{s}{%
   \IfBooleanTF{#1}
   {\def\al@arb@rules{easynosukun}}
   {\def\al@arb@rules{easy}}}
-\NewDocumentCommand{\SetArbDflt}{}{\def\al@arb@rules{dflt}}
+\NewDocumentCommand{\SetArbDflt}{s}{%
+  \IfBooleanTF{#1}
+  {\def\al@arb@rules{idgham}}
+  {\def\al@arb@rules{dflt}}}
 %    \end{macrocode}
 % \end{macro}
 % \end{macro}
 % \end{macro}
+% \end{macro}
 % \begin{macro}{\SetTranslitFont}
 %   \changes{v1.4}{2016/07/05}{For selecting a specific font for
 %   transliterated texts} By default, the font that is used for
diff --git a/arabluatex.lua b/arabluatex.lua
index 38fe87b..07968c0 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -146,18 +146,29 @@ local function takeoutcapetc(str)
    return str
 end
 
-local function voc(str)
+local function voc(str, rules)
    str = string.gsub(str, "\\arb(%b{})", function(inside)
    inside = string.sub(inside, 2, -2)
    for i = 1,#hamza do
       inside = string.gsub(inside, hamza[i].a, hamza[i].b)
    end
-   for i = 1,#tanwin do
-      inside  = string.gsub(inside, tanwin[i].a, tanwin[i].b)
+   if rules == "idgham" then
+      for i = 1,#tanwin do
+         inside  = string.gsub(inside, tanwin[i].a, tanwin[i].b)
+      end
+   else
+      for i = 1,#tanwineasy do
+         inside  = string.gsub(inside, tanwineasy[i].a, tanwineasy[i].b)
+      end
    end
    for i = 1,#trigraphs do
       inside = string.gsub(inside, trigraphs[i].a, trigraphs[i].b)
    end
+   if rules == "idgham" then
+      for i = 1,#idgham do
+         inside = string.gsub(inside, idgham[i].a, idgham[i].b)
+      end
+   end
    for i = 1,#digraphs do
       inside = string.gsub(inside, digraphs[i].a, digraphs[i].b)
    end
@@ -218,20 +229,37 @@ local function voceasy(str)
 return str
 end
 
-local function fullvoc(str)
+local function fullvoc(str, rules)
    str = string.gsub(str, "\\arb(%b{})", function(inside)
    inside = string.sub(inside, 2, -2)
    for i = 1,#hamzafv do
       inside = string.gsub(inside, hamzafv[i].a, hamzafv[i].b)
    end
-   for i = 1,#tanwinfv do
-      inside  = string.gsub(inside, tanwinfv[i].a, tanwinfv[i].b)
+   if rules == "idgham" then
+      for i = 1,#tanwinfv do
+         inside  = string.gsub(inside, tanwinfv[i].a, tanwinfv[i].b)
+      end
+   else
+      for i = 1,#tanwinfveasy do
+         inside  = string.gsub(inside, tanwinfveasy[i].a, tanwinfveasy[i].b)
+      end
    end
    for i = 1,#trigraphsfv do
       inside = string.gsub(inside, trigraphsfv[i].a, trigraphsfv[i].b)
    end
-   for i = 1,#digraphsfv do
-      inside = string.gsub(inside, digraphsfv[i].a, digraphsfv[i].b)
+   if rules == "idgham" then
+      for i = 1,#idgham do
+         inside = string.gsub(inside, idgham[i].a, idgham[i].b)
+      end
+   end
+   if rules == "idgham" then
+      for i = 1,#digraphsfvidgham do
+         inside = string.gsub(inside, digraphsfvidgham[i].a, digraphsfvidgham[i].b)
+      end
+   else
+      for i = 1,#digraphsfv do
+         inside = string.gsub(inside, digraphsfv[i].a, digraphsfv[i].b)
+      end
    end
    for i = 1,#singlefv do
       inside = string.gsub(inside, singlefv[i].a, singlefv[i].b)
@@ -254,7 +282,7 @@ local function fullvoc(str)
 return str
 end
 
-local function fullvoceasy(str, opt)
+local function fullvoceasy(str, rules)
    str = string.gsub(str, "\\arb(%b{})", function(inside)
    inside = string.sub(inside, 2, -2)
    for i = 1,#hamzafveasy do
@@ -266,7 +294,7 @@ local function fullvoceasy(str, opt)
    for i = 1,#trigraphsfveasy do
       inside = string.gsub(inside, trigraphsfveasy[i].a, trigraphsfveasy[i].b)
    end
-   if opt == "nosukun" then
+   if rules == "nosukun" then
       for i = 1,#digraphsfveasy do
          inside = string.gsub(inside, digraphsfveasy[i].a, digraphsfveasy[i].b)
       end
@@ -368,7 +396,7 @@ local function novoceasy(str)
 return str
 end
 
-local function transdmg(str)
+local function transdmg(str, rules)
    str = string.gsub(str, "\\arb(%b{})", function(inside)
    inside = string.sub(inside, 2, -2)
    for i = 1,#hamzatrdmg do
@@ -380,40 +408,10 @@ local function transdmg(str)
    for i = 1,#trigraphstrdmg do
       inside = string.gsub(inside, trigraphstrdmg[i].a, trigraphstrdmg[i].b)
    end
-   for i = 1,#digraphstrdmg do
-      inside = string.gsub(inside, digraphstrdmg[i].a, digraphstrdmg[i].b)
-   end
-   for i = 1,#singletrdmg do
-      inside = string.gsub(inside, singletrdmg[i].a, singletrdmg[i].b)
-   end
-   for i = 1,#longvtrdmg do
-      inside = string.gsub(inside, longvtrdmg[i].a, longvtrdmg[i].b)
-   end
-   for i = 1,#shortvtrdmg do
-      inside = string.gsub(inside, shortvtrdmg[i].a, shortvtrdmg[i].b)
-   end
-   for i = 1,#punctuationtr do
-      inside = string.gsub(inside, punctuationtr[i].a, punctuationtr[i].b)
-   end
-   for i = 1,#nulltr do
-      inside = string.gsub(inside, nulltr[i].a, nulltr[i].b)
-   end
-   return string.format("\\txtrans{%s}", inside)
-   end)
-return str
-end
-
-local function transdmgeasy(str)
-   str = string.gsub(str, "\\arb(%b{})", function(inside)
-   inside = string.sub(inside, 2, -2)
-   for i = 1,#hamzatrdmg do
-      inside = string.gsub(inside, hamzatrdmg[i].a, hamzatrdmg[i].b)
-   end
-   for i = 1,#tanwintrdmg do
-      inside  = string.gsub(inside, tanwintrdmg[i].a, tanwintrdmg[i].b)
-   end
-   for i = 1,#trigraphstrdmgeasy do
-  inside = string.gsub(inside, trigraphstrdmgeasy[i].a, trigraphstrdmgeasy[i].b)
+   if rules == "idgham" then
+      for i = 1,#idghamtrdmg do
+         inside = string.gsub(inside, idghamtrdmg[i].a, idghamtrdmg[i].b)
+      end
    end
    for i = 1,#digraphstrdmg do
       inside = string.gsub(inside, digraphstrdmg[i].a, digraphstrdmg[i].b)
@@ -498,8 +496,8 @@ function processvoc(str, rules, scheme)
       else end
    if rules == "easy" or rules == "easynosukun" then
       str = voceasy(str)
-   elseif rules == "dflt" then
-      str = voc(str)
+   elseif rules == "dflt" or rules == "idgham" then
+      str = voc(str, rules)
       else end
    str = unprotectarb(str)
 return str
@@ -518,8 +516,8 @@ function processfullvoc(str, rules, scheme)
       str = fullvoceasy(str, "sukun")
    elseif rules == "easynosukun" then
       str = fullvoceasy(str, "nosukun")
-   elseif rules == "dflt" then
-      str = fullvoc(str)
+   elseif rules == "dflt" or rules == "idgham" then
+      str = fullvoc(str, rules)
       else end
    str = unprotectarb(str)
 return str
@@ -536,7 +534,7 @@ function processnovoc(str, rules, scheme)
       else end
    if rules == "easy" or rules == "easynosukun" then
       str = novoceasy(str)
-   elseif rules == "dflt" then
+   elseif rules == "dflt" or rules == "idgham" then
       str = novoc(str)
       else end
    str = unprotectarb(str)
@@ -551,16 +549,12 @@ function processtrans(str, mode, rules, scheme)
    str = holdcmd(str)
    if scheme == "buckwalter" then
       str = processbuckw(str)
-      else end
+   end
    if mode == "dmg" then
-      if rules == "easy" or rules == "easynosukun" then
-         str = transdmgeasy(str)
-      elseif rules == "dflt" then
-         str = transdmg(str)
-         else end
+      str = transdmg(str, rules)
    elseif mode == "loc" then
       str = transloc(str)
-      else end
+   end
    str = unprotectarb(str)
 return str
 end
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"},
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index 87c85e9..981810f 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -205,65 +205,13 @@ 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"},
-   -- assimilations
-   {a="(n)(}?)(%s)([rlmnwy])", b="%4%2%3%4"}
+   {a="(_a)I", b="A"}
 }
 
-trigraphstrdmgeasy = { -- see the differences below under 'easy' tag
-   -- 'llatI / 'llad_I
-   {a="^'ll(a)([%_]?[dt])", b="'ll%1%2"},
-   {a="(%s)'ll(a)([%_]?[dt])", b="%1'll%2%3"},
-   -- al- + lām
-   {a="^(a)l%-(l)", b="%1l-%2"},
-   {a="([%s%-])(a)l%-(l)", b="%1%2l-%3"},
-   -- al- + solar consonant
-   {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="%1%2-%2"},
-   {a="([%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1%2%3-%3"},
-   -- assim. art. + solar consonant
-   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="%1%2-"},
-   {a="([%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1%2%3-"},
-   -- al- + initial unstable hamza
-   {a="^(a)l%-([uai])", b="%1l-%2"},
-   {a="([%s%-])(a)l%-([uai])", b="%1%2l-%3"},
-   -- li-/la- + art. + initial unstable hamza is a special orthography
-   {a="l([ai])%-l%-([uai])", b="l%1-l-%2"},
-   -- al- + lunar consonant (i.e. what remains)
-   {a="^(a)l%-", b="%1l-"},
-   {a="([%s%-])(a)l%-", b="%1%2l-"},
-   -- diphthongs to be resolved before ʾalif conjunctionis
-   {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
-   {a="(ay)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
-   -- art. with waṣla + lām
-   {a="'l%-(l)", b="'l-%1"},
-   -- art. with waṣla + solar consonant
-   {a="'l%-([%_%^%.]?[tdrzsn])", b="'%1-%1"},
-   -- li-/la- + art. + lām
-   {a="l([ai])%-l%-(l)", b="l%1-%2%2"},
-   -- assim. art. with waṣla + solar consonant
-   {a="'([%_%^%.]?[tdrzsn])%-", b="'%1-"},
-   -- li-/la- + art. + solar consonant is a special orthography
-   {a="l([ai])%-l%-([%_%^%.]?[tdrzsn])", b="l%1-%2-%2"},
-   -- li-/la- + assim. art. + solar consonant is a special orthography
-   {a="l([ai])%-([%_%^%.]?[tdrzsn])%-([%_%^%.]?[tdrzsn])", b="l%1-%2-%3"},
-   -- art. with waṣla + initial unstable hamza
-   {a="'l%-([uai])", b="'l-%1"},   
-   -- art. with waṣla + lunar consonant (i.e. what remains)
-   {a="'l%-", b="'l-"},
-   -- the silent wāw
-   {a="uU$", b="u"},
-   {a="uU(%W)", b="u%1"},
-   {a="aU$", b="a"},
-   {a="aU(%W)", b="a%1"},
-   {a="iU$", b="i"},
-   {a="iU(%W)", b="i%1"},
-   -- words ending in -āT with silent wāw/yāʾ
-   {a="(_a)UA", b="A"},
-   {a="(_a)U", b="A"},
-   {a="(_a)I", b="A"}
+idghamtrdmg = {
    -- assimilations
---easy   {a="(n)(%s)([rlmnwy])", b="%3%2%3"}
-}
+   {a="(n)(}?)(%s)([rlmnwy])", b="%4%2%3%4"}
+} 
 
 digraphstrdmg = {
    {a="([uai]%-)(\"?[uai])", b="%1'"}, -- hyphen + initial alif without hamza
diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index 8f71e52..ddd8fda 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -467,7 +467,10 @@ 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"}
+}
+
+idgham = {
    -- assimilations
    {a="(n)(%s)([rlmnwy])", b="%1%2%3%3"}
 }
@@ -529,9 +532,7 @@ trigraphseasy = { -- differences marked below with 'easy'
    -- 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"}
 }
 
 digraphs = {