arbnulls inserted in trans table

4 files changed, 80 insertions, 19 deletions

diff --git a/arabluatex.dtx b/arabluatex.dtx
index 5ec4cbe..efac2c3 100644
--- a/arabluatex.dtx
+++ b/arabluatex.dtx
@@ -3324,7 +3324,6 @@ wa-ya.sIru ta.hta 'l-jild-i
   \ifdefined\spewnotes\spewnotes\else\fi%
 }
 %    \end{macrocode}
-%   
 % \end{macro} 
 % \end{environment}
 % \begin{macro}{\abjad} \cs{abjad}\marg{number} expresses its argument
@@ -3345,6 +3344,16 @@ wa-ya.sIru ta.hta 'l-jild-i
 \fi}
 %    \end{macrocode}
 % \end{macro}
+% \begin{macro}{\arbnull}
+%   \changes{v1.7}{2016/12/23}{New \cs{arbnull} command for putting
+%   back on any contextual analysis broken by other commands.} The
+%   \cs{arbnull} command does nothing by itself. It is processed only
+%   if it is found in Arabic context so as to put back on contextual
+%   analysis in case it has been broken by other commands.
+%    \begin{macrocode}
+\NewDocumentCommand{\arbnull}{m}{\relax}
+%    \end{macrocode}
+% \end{macro}
 % \begin{macro}{\abraces}
 %   \cs{abraces}\marg{Arabic text} puts its argument between
 %   braces. This macro is written in Lua and is dependent on the
diff --git a/arabluatex.lua b/arabluatex.lua
index 527be3b..294e75f 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -128,6 +128,11 @@ local function indnum(str)
    return str
 end
 
+local function processdiscretionary(str)
+   str = string.gsub(str, "\\%-", "\\-{}")
+   return str
+end
+
 local function processarbnull(str, scheme)
    if scheme == "buckwalter" then
       str = string.gsub(str, "(\\arbnull.?)(%b{})", function(tag, body)
@@ -558,6 +563,7 @@ end
 
 function processtrans(str, mode, rules, scheme)
    str = "\\arb{".. str.."}"
+   str = processdiscretionary(str)
    str = processarbnull(str, scheme)
    str = takeoutabjad(str)
    str = protectarb(str)
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index bc44229..0868262 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -31,7 +31,7 @@ punctuationtr = {
 
 nulltr = {
    {a="%|", b=""},
-   {a="o.-o", b=""}
+   {a="o[%S]-o", b=""}
 }
 
 -- cap
@@ -97,14 +97,14 @@ hamzatrdmg = {
    {a="(A)(')", b="%1ʾ"}, -- historic madda
    -- initial (needs both ^ and %W patterns)
    -- 'aw: the diphthong is to be resolved into 'awi' (next 8 lines)
-   {a="^('aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)(.-o)", b="%1i"},
-   {a="(%W)('aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)(.-o)", b="%1%2i"},
-   {a="^('aw)(o)([%(%[%|%<]?\"?[uai])(.-o)", b="%1i"},
-   {a="(%W)('aw)(o)([%(%[%|%<]?\"?[uai])(.-o)", b="%1%2i"},
+   {a="^('aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
+   {a="(%W)('aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1%2i"},
+   {a="^('aw)(o)(\"?[uai])([%S]-o)", b="%1i"},
+   {a="(%W)('aw)(o)(\"?[uai])([%S]-o)", b="%1%2i"},
    {a="^('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
    {a="(%W)('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
-   {a="^('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
-   {a="(%W)('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
+   {a="^('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"}, --p
+   {a="(%W)('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"}, --p
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="ʾ%2"},
    {a="^(')(i)", b="ʾ%2"},
@@ -138,12 +138,19 @@ hamzatrdmg = {
 }
 
 tanwintrdmg = {
+   {a="%-?([uai]NU)(o)([ui])([%S]-o)", b="\\arbup{un%3}"},
    {a="%-?([uai]NU)(%s)([ui])", b="\\arbup{un%3}%2'"},
    {a="%-?uNU", b="\\arbup{un}"},
    {a="%-?aNU", b="\\arbup{an}"},
    {a="%-?iNU", b="\\arbup{in}"},
    {a="%-?iNI", b="i\\arbup{n}"},
    -- tanwīn preceding ʾalif conjunctionis
+   {a="%-?(uN)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="\\arbup{uni}"},
+   {a="%-?(aN)(_A)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="ạ\\arbup{ni}"},
+   {a="%-?(aN)(Y)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="ạ\\arbup{ni}"},
+   {a="(T)%-?(aN)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="t\\arbup{ani}"},
+   {a="([^TA])%-?(aN)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1\\arbup{ani}"},
+   {a="%-?(iN)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="\\arbup{ini}"},
    {a="%-?(uN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="\\arbup{uni}%2%3"},
    {a="%-?(aN)(_A)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ạ\\arbup{ni}%3%4"},
    {a="%-?(aN)(Y)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ạ\\arbup{ni}%3%4"},
@@ -151,6 +158,12 @@ tanwintrdmg = {
    {a="([^TA])%-?(aN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1\\arbup{ani}%3%4"},
    {a="%-?(iN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="\\arbup{ini}%2%3"},
    -- tanwīn preceding 'lla_dI/'llatI
+   {a="%-?(uN)(o)('lla[%_]?[dt])([%S]-o)", b="\\arbup{uni}"},
+   {a="%-?(aN)(_A)(o)('lla[%_]?[dt])([%S]-o)", b="ạ\\arbup{ni}"},
+   {a="%-?(aN)(Y)(o)('lla[%_]?[dt])([%S]-o)", b="ạ\\arbup{ni}"},
+   {a="(T)%-?(aN)(o)('lla[%_]?[dt])([%S]-o)", b="t\\arbup{ani}"},
+   {a="([^TA])%-?(aN)(o)('lla[%_]?[dt])([%S]-o)", b="%1\\arbup{ani}"},
+   {a="%-?(iN)(o)('lla[%_]?[dt])([%S]-o)", b="\\arbup{ini}"},
    {a="%-?(uN)(%s)('lla[%_]?[dt])", b="\\arbup{uni}%2%3"},
    {a="%-?(aN)(_A)(%s)('lla[%_]?[dt])", b="ạ\\arbup{ni}%3%4"},
    {a="%-?(aN)(Y)(%s)('lla[%_]?[dt])", b="ạ\\arbup{ni}%3%4"},
@@ -158,12 +171,19 @@ tanwintrdmg = {
    {a="([^TA])%-?(aN)(%s)('lla[%_]?[dt])", b="%1\\arbup{ani}%3%4"},
    {a="%-?(iN)(%s)('lla[%_]?[dt])", b="\\arbup{ini}%2%3"},
    -- tanwīn + alif without hamza and kasra (ibn) or dhamma (uhrub)
+   {a="%-?(uN)(o)([ui])([%S]-o)", b="\\arbup{un%3}"},
+   {a="%-?(aN)(_A)(o)([ui])([%S]-o)", b="ạ\\arbup{n%4}"},
+   {a="%-?(aN)(Y)(o)([ui])([%S]-o)", b="ạ\\arbup{n%4}"},
+   {a="(T)%-?(aN)(o)([ui])([%S]-o)", b="t\\arbup{an%4}"},
+   {a="([^TA])%-?(aN)(o)([ui])([%S]-o)", b="%1\\arbup{an%4}"},
+   {a="%-?(iN)(o)([ui])([%S]-o)", b="\\arbup{in%3}"},
+   {a="(o[%S]-)([uai]N)(o)(\"?[ui])", b="'"},
    {a="%-?(uN)(%s)([ui])", b="\\arbup{un%3}%2'"},
-   {a="%-?(aN)(_A)(%s)([ui])", b="ạ\\arbup{n%4}%3%'"},
-   {a="%-?(aN)(Y)(%s)([ui])", b="ạ\\arbup{n%4}%3%'"},
-   {a="(T)%-?(aN)(%s)([ui])", b="t\\arbup{an%4}%3%'"},
-   {a="([^TA])%-?(aN)(%s)([ui])", b="%1\\arbup{an%4}%3%'"},
-   {a="%-?(iN)(%s)([ui])", b="\\arbup{in%3}%2%'"},
+   {a="%-?(aN)(_A)(%s)([ui])", b="ạ\\arbup{n%4}%3'"},
+   {a="%-?(aN)(Y)(%s)([ui])", b="ạ\\arbup{n%4}%3'"},
+   {a="(T)%-?(aN)(%s)([ui])", b="t\\arbup{an%4}%3'"},
+   {a="([^TA])%-?(aN)(%s)([ui])", b="%1\\arbup{an%4}%3'"},
+   {a="%-?(iN)(%s)([ui])", b="\\arbup{in%3}%2'"},
    --
 -- {a="uN", b="\\arbup{un}"}, (now included in the last line of this table)
    {a="%-?(\"?At)%-?([ui])N", b="\\arbup{%1%2n}"},
@@ -178,11 +198,15 @@ trigraphstrdmg = { -- trigraphs or more
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="'ll%1%2"},
    {a="([%(%[%|%<%s])'ll(a)([%_]?[dt])", b="%1'll%2%3"}, --p
-   -- law: the diphthong is to be resoved into 'awi' (next 4 lines)
+   -- law: the diphthong is to be resolved into 'awi' (next 8 lines)
+   {a="^(law)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
+   {a="(%W)(law)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1%2i"},
+   {a="^(law)(o)(\"?[uai])([%S]-o)", b="%1i"},
+   {a="(%W)(law)(o)(\"?[uai])([%S]-o)", b="%1%2i"},
    {a="^(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
    {a="(%W)(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
-   {a="^(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
-   {a="(%W)(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
+   {a="^(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"}, --p
+   {a="(%W)(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"}, --p
    -- al- + lām
    {a="^(a)l%-(l)", b="%1l-%2"},
    {a="([%(%[%|%<%s%-])(a)l%-(l)", b="%1%2l-%3"}, --p
@@ -201,8 +225,14 @@ trigraphstrdmg = { -- trigraphs or more
    {a="^(a)l%-", b="%1l-"},
    {a="([%(%[%|%<%s%-])(a)l%-", b="%1%2l-"}, --p
    -- diphthongs to be resolved before ʾalif conjunctionis
+   {a="(aw)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1u"},
+   {a="(ay)(o)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="%1i"},
+   {a="(aw)(o)(\"?[uai])([%S]-o)", b="%1u"},
+   {a="(ay)(o)(\"?[uai])([%S]-o)", b="%1i"},
    {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
    {a="(ay)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
+   {a="(aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1u%2%3"}, --p
+   {a="(ay)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"}, --p
    -- art. with waṣla + lām
    {a="'l%-(l)", b="'l-%1"},
    -- art. with waṣla + solar consonant
@@ -235,7 +265,7 @@ trigraphstrdmg = { -- trigraphs or more
 idghamtrdmg = {
    -- assimilations
    {a="(n)(}?)(%s)([rlmnwy])", b="%4%2%3%4"},
-   {a="(n)(o)([rlmnwy])(.-o)", b="%3"}
+   {a="(n)(o)([rlmnwy])([%S]-o)", b="%3"}
 } 
 
 digraphstrdmg = {
@@ -249,6 +279,15 @@ digraphstrdmg = {
    {a="(%W)(\"?[uai])", b="%1%2"},      -- initial alif without hamza
    -- this is not necessary, take out for now:
 -- {a="([%_]?[uaiUAIY])(%s)([uai])", b="%1%2'"}, -- initial alif without hamza
+   {a="(aW)(o)(\"?[uai])([%S]-o)", b="awu"},
+   {a="(UA)(o)(\"?[uai])([%S]-o)", b="u"},
+   {a="(%_A)(o)(\"?[uai])([%S]-o)", b="ạ"},
+   {a="(Y)(o)(\"?[uai])([%S]-o)", b="ạ"},
+   {a="(%_a)(o)(\"?[uai])([%S]-o)", b="a"},
+   {a="(A)(o)(\"?[uai])([%S]-o)", b="a"},
+   {a="([%_]?[Uu])(o)(\"?[uai])([%S]-o)", b="u"},
+   {a="([%_]?[Ii])(o)(\"?[uai])([%S]-o)", b="i"},
+   {a="(o[%S]-)([UAIYWuaiyw])(o)(\"?[uai])", b="'"},
    {a="(aW)(%s)([%(%[%|%<]?)(\"?[uai])", b="awu%2%3%4"}, --p
    {a="(UA)(%s)([%(%[%|%<]?)(\"?[uai])", b="u%2%3'"}, --p
    {a="([^%_][uai])(%s)([%(%[%|%<]?)(\"?[uai])", b="%1%2%3'"}, --p
@@ -272,6 +311,13 @@ digraphstrdmg = {
    -- ʾiʿrāb hyphen (end)
    -- shorten long vowels preceding ʾalif conjunctionis
    {a="(U)(A)", b="U"},
+   {a="(aW)(o)('[%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="awu"},
+   {a="(%_a)(o)('[%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="a"},
+   {a="(%_A)(o)('[%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="ạ"},
+   {a="(A)(o)('[%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="a"},
+   {a="(Y)(o)('[%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="ạ"},
+   {a="([%_]?[Uu])(o)('[%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="u"},
+   {a="([%_]?[Ii])(o)('[%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)([%S]-o)", b="i"},
    --p (next 7 lines, just after %s)
    {a="(aW)(%s)([%(%[%|%<]?['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awu%2%3"},
    {a="(%_a)(%s)([%(%[%|%<]?['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="a%2%3"},
diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index c856818..2f5e445 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -205,8 +205,8 @@ hamza = {
    -- 'aw: the diphthong is to be resoved into 'awi' (next 4 lines)
    {a="^('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
    {a="(%W)('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
-   {a="^('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
-   {a="(%W)('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
+   {a="^('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"}, --p
+   {a="(%W)('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"}, --p
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},