started implementing Buckwalter input scheme

3 files changed, 43 insertions, 9 deletions

diff --git a/arabluatex.dtx b/arabluatex.dtx
index 8554fc3..d7a4adb 100644
--- a/arabluatex.dtx
+++ b/arabluatex.dtx
@@ -2262,17 +2262,29 @@ wa-ya.sIru ta.hta 'l-jild-i
   {\RenewDocumentCommand{\aemph}{m}{$\overline{\text{#1}}$}}
   {\NewDocumentCommand{\aemph}{m}{$\overline{\text{#1}}$}}}
 %    \end{macrocode}
+% \begin{macro}{\SetInputScheme}
+%   \changes{v1.4}{2016/06/30}{\cs{SetInputScheme} may be used to
+%   process other input schemes such as \enquote*{Buckwalter}}
+%   \package{arabluatex} is designed for processing Arab\TeX\ input
+%   notation. \cs{SetInputScheme} may be used in the preamble or at
+%   any point of the document should the user wish to use a different
+%   notation such as the \enquote*{Buckwalter scheme}.
+%    \begin{macrocode}
+\def\al@input@scheme{arabtex}
+\NewDocumentCommand{\SetInputScheme}{m}{\def\al@input@scheme{#1}}
+%    \end{macrocode}
+% \end{macro}
 % \end{macro}
 % \begin{macro}{\SetArbEasy}
 %   \changes{v1.2}{2016/05/09}{New \cs{SetArbEasy}/\cs{SetArbDflt} for
 %   \enquote*{modern} or \enquote*{classic} Arabic styles.}
 % \begin{macro}{\SetArbDflt}
 %   By default, \package{arabluatex} applies complex rules to generate
-%   euphonic \arb[trans]{ta^sdId}, \arb[trans]{'alif mamdUdaT}
-%   \linebreak and \arb[trans]{sukUn} depending on the modes which are
-%   selected, either |voc|, |fullvoc| or |trans|. Such refinements can
-%   be discarded with \cs{SetArbEasy}, either globally in the preamble
-%   or at any point of the document. Default complex rules can be set
+%   euphonic \arb[trans]{ta^sdId}, \arb[trans]{'alif mamdUdaT} and
+%   \arb[trans]{sukUn} depending on the modes which are selected,
+%   either |voc|, |fullvoc| or |trans|. Such refinements can be
+%   discarded with \cs{SetArbEasy}, either globally in the preamble or
+%   at any point of the document. Default complex rules can be set
 %   back at any point of the document with \cs{SetArbDflt}.
 %    \begin{macrocode}
 \def\al@arb@rules{dflt}
@@ -2289,7 +2301,6 @@ wa-ya.sIru ta.hta 'l-jild-i
 \def\al@trans@font{\rmfamily}%
 \NewDocumentCommand{\SetTranslitFont}{m}{\def\al@trans@font{#1}}
 %    \end{macrocode}
-%   
 % \end{macro}
 % \end{macro}
 % \begin{macro}{\SetTranslitStyle} By default any transliterated
@@ -2369,7 +2380,7 @@ wa-ya.sIru ta.hta 'l-jild-i
   \ifx\@tempa\al@mode@voc%
   \bgroup\textdir TRT\arabicfont%
   \luadirect{tex.sprint(processvoc(\luastringN{#2},
-    \luastringO{\al@arb@rules}))}\egroup%
+    \luastringO{\al@arb@rules}, \luastringO{\al@input@scheme}))}\egroup%
   \else%
   \ifx\@tempa\al@mode@fullvoc%
   \bgroup\textdir TRT\arabicfont%
@@ -2399,7 +2410,7 @@ wa-ya.sIru ta.hta 'l-jild-i
   \ifx\@tempa\al@mode@voc%
   \bgroup\pardir TRT\textdir TRT\arabicfont%
   \luadirect{tex.sprint(processvoc(\luastringO{\BODY},
-    \luastringO{\al@arb@rules}))}\egroup%
+    \luastringO{\al@arb@rules}, \luastringO{\al@input@scheme}))}\egroup%
   \else%
   \ifx\@tempa\al@mode@fullvoc%
   \bgroup\pardir TRT\textdir TRT\arabicfont%
diff --git a/arabluatex.lua b/arabluatex.lua
index 3e574cf..9134ce5 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -456,12 +456,26 @@ local function transloc(str)
 return str
 end
 
-function processvoc(str, rules)
+local function processbuckw(str)
+   str = string.gsub(str, "\\arb(%b{})", function(inside)
+   inside = string.sub(inside, 2, -2)
+   for i = 1,#buckwalter do
+      inside = string.gsub(inside, buckwalter[i].a, buckwalter[i].b)
+   end
+   return string.format("\\arb{%s}", inside)
+   end)
+return str
+end
+
+function processvoc(str, rules, scheme)
    str = "\\arb{".. str.."}"
    str = takeoutcapetc(str)
    str = protectarb(str)
    str = breakcmd(str)
    str = holdcmd(str)
+   if scheme == "buckwalter" then
+      str = processbuckw(str)
+      else end
    if rules == "easy" then
       str = voceasy(str)
    elseif rules == "dflt" then
diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index e47ef96..1a64109 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -71,6 +71,15 @@ raw = {
    {a="i", b="ِ"}
 }
 
+buckwalter = {
+--   {a="\'", b="|\"\'"},
+   {a="%>", b="a\"\'"},
+   {a="%&", b="w\"\'"},
+   {a="%<", b="i\"\'"},
+   {a="%}", b="y\"\'"},
+   {a="p", b="T"}
+}
+
 hamza = {
    -- next line for ʾiʿrāb hyphen
    {a="(')(%-)([uaiUAI])", b="%1%3"},