first implementation of \arbup viz. superscript tanwīn in |dmg| mode

5 files changed, 44 insertions, 26 deletions

diff --git a/arabluatex.dtx b/arabluatex.dtx
index 995be49..2e3dc7e 100644
--- a/arabluatex.dtx
+++ b/arabluatex.dtx
@@ -2144,8 +2144,8 @@ wa-ya.sIru ta.hta 'l-jildi
 % \end{macro}
 % \end{macro}
 % \begin{macro}{\SetTranslitStyle} By default any transliterated
-% Arabic text is printed in italics. This can be changed globally in
-% the preamble or at any point of the document:
+%   Arabic text is printed in italics. This can be changed either
+%   globally in the preamble or at any point of the document:
 %    \begin{macrocode}
 \def\al@trans@style{\itshape}%
 \NewDocumentCommand{\SetTranslitStyle}{m}{\def\al@trans@style{#1}}
@@ -2159,6 +2159,22 @@ wa-ya.sIru ta.hta 'l-jildi
 \NewDocumentCommand{\SetTranslitConvention}{m}{\def\al@trans@convention{#1}}
 %    \end{macrocode}
 % \end{macro}
+% \begin{macro}{\arbup}
+% \begin{macro}{\SetArbUp}
+%   By default, \cs{arbup} is set to \cs{textsuperscript}. This is how
+%   the \arb[trans]{tanwIn} that takes place at the end of a word
+%   should be displayed in |dmg| mode. \cs{SetArbUp} may be used
+%   either in the preamble or at any point of the document to change
+%   the default definition.
+%    \begin{macrocode}
+\NewDocumentCommand{\al@arbup@dflt}{m}{\textsuperscript{\thinspace#1}}%
+\NewDocumentCommand{\al@arbup}{m}{\al@arbup@dflt{#1}}
+\NewDocumentCommand{\arbup}{m}{\al@arbup{#1}}
+\NewDocumentCommand{\ArbUpDflt}{}{\let\al@arbup=\al@arbup@dflt}
+\NewDocumentCommand{\NoArbUp}{}{\RenewDocumentCommand{\al@arbup}{m}{##1}}
+%    \end{macrocode}
+% \end{macro}
+% \end{macro}
 % \begin{macro}{\cap} Proper Arabic names or book titles should be
 % passed to the command \cs{cap} so that they have their first letters
 % uppercased. \cs{cap} is actually coded in Lua.
diff --git a/arabluatex.lua b/arabluatex.lua
index 0bb4756..eb0d55c 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -123,11 +123,12 @@ local function takeoutabjad(str)
    return str
 end
 
-local function takeoutcap(str)
+local function takeoutcapetc(str)
    str = string.gsub(str, "(\\cap.?)(%b{})", function(tag, body)
                         body = string.sub(body, 2, -2)
                         return string.format("%s", body)
    end)
+   str = string.gsub(str, "\\linebreak", "")
    return str
 end
 
@@ -421,7 +422,7 @@ end
 
 function processvoc(str, rules)
    str = "\\arb{".. str.."}"
-   str = takeoutcap(str)
+   str = takeoutcapetc(str)
    str = protectarb(str)
    str = breakcmd(str)
    str = holdcmd(str)
@@ -436,7 +437,7 @@ end
 
 function processfullvoc(str, rules)
    str = "\\arb{".. str.."}"
-   str = takeoutcap(str)
+   str = takeoutcapetc(str)
    str = protectarb(str)
    str = breakcmd(str)
    str = holdcmd(str)
@@ -451,7 +452,7 @@ end
 
 function processnovoc(str)
    str = "\\arb{".. str.."}"
-   str = takeoutcap(str)
+   str = takeoutcapetc(str)
    str = protectarb(str)
    str = breakcmd(str)
    str = holdcmd(str)
diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index a2676e0..577fff7 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -238,7 +238,7 @@ tanwinfv = {
    {a="(T)(aN)", b="%1ً"},
    {a="(ء)(aN)", b="%1ً"},
    {a="([^TA])(aN)", b="%1ًا"},
-   {a="(iN)", b="ٍ"}
+   {a="(iNI?)", b="ٍ"}
 }
 
 tanwinfveasy = { -- no assimilations (see below)
@@ -273,7 +273,7 @@ tanwinfveasy = { -- no assimilations (see below)
    {a="(T)(aN)", b="%1ً"},
    {a="(ء)(aN)", b="%1ً"},
    {a="([^TA])(aN)", b="%1ًا"},
-   {a="(iN)", b="ٍ"}
+   {a="(iNI?)", b="ٍ"}
 }
 
 trigraphsfv = { -- trigraphs or more
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index c96a9da..fbea94b 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -127,23 +127,24 @@ hamzatrdmg = {
 }
 
 tanwintrdmg = {
-   {a="uNU", b="un"},
-   {a="aNU", b="an"},
-   {a="iNU", b="in"},
+   {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)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="uni%2%3"},
-   {a="(aN)(_A)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ạni%3%4"},
-   {a="(aN)(Y)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ạni%3%4"},
-   {a="(T)(aN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="tani%3%4"},
-   {a="([^TA])(aN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1ani%3%4"},
-   {a="(iN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ini%2%3"},
+   {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"},
+   {a="(T)(aN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="t\\arbup{ani}%3%4"},
+   {a="([^TA])(aN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1\\arbup{ani}%3%4"},
+   {a="(iN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="\\arbup{ini}%2%3"},
    --
-   {a="uN", b="un"},
-   {a="(aN)(_A)", b="ạn"},
-   {a="(aN)(Y)", b="ạn"},
-   {a="(T)(\"?aN)", b="tan"},
-   {a="([^TA])(\"?aN)", b="%1an"},
-   {a="iN", b="in"}
+   {a="uN", b="\\arbup{un}"},
+   {a="(aN)(_A)", b="ạ\\arbup{n}"},
+   {a="(aN)(Y)", b="ạ\\arbup{n}"},
+   {a="(T)(\"?aN)", b="t\\arbup{an}"},
+   {a="([^TA])(\"?aN)", b="%1\\arbup{an}"},
+   {a="iN", b="\\arbup{in}"}
 }
 
 trigraphstrdmg = { -- trigraphs or more
@@ -198,7 +199,7 @@ trigraphstrdmg = { -- trigraphs or more
    {a="(_a)U", b="A"},
    {a="(_a)I", b="A"},
    -- assimilations
-   {a="(n)(%s)([rlmnwy])", b="%3%2%3"}
+   {a="(n)(})(%s)([rlmnwy])", b="%4%2%3%4"}
 }
 
 trigraphstrdmgeasy = { -- see the differences below under 'easy' tag
diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index 4f5399f..878f351 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -267,7 +267,7 @@ tanwin = {
    {a="(T)(aN)", b="%1ً"},
    {a="(ء)(aN)", b="%1ً"},
    {a="([^TA])(aN)", b="%1ًا"},
-   {a="(iN)", b="ٍ"}
+   {a="(iNI?)", b="ٍ"}
 }
 
 tanwineasy = { -- 'easy' requires some lines to be taken out:
@@ -300,7 +300,7 @@ tanwineasy = { -- 'easy' requires some lines to be taken out:
    {a="(T)(aN)", b="%1ً"},
    {a="(ء)(aN)", b="%1ً"},
    {a="([^TA])(aN)", b="%1ًا"},
-   {a="(iN)", b="ٍ"}
+   {a="(iNI?)", b="ٍ"}
 }
 
 trigraphs = { -- trigraphs or more