add \cap{} command; correct transliteration tables

3 files changed, 76 insertions, 8 deletions

diff --git a/arabluatex.dtx b/arabluatex.dtx
index 1dbfd3d..09b848c 100644
--- a/arabluatex.dtx
+++ b/arabluatex.dtx
@@ -225,6 +225,8 @@
 \def\al@trans@style{\itshape}%
 \NewDocumentCommand{\SetTranslitStyle}{m}{\def\al@trans@style{#1}}
 \NewDocumentCommand{\SetTranslitConvention}{m}{\def\al@trans@convention{#1}}
+\DeclareDocumentCommand{\cap}{m}%
+  {\luadirect{tex.sprint(cap(\luastringN{#1}))}}
 \def\al@trans@convention{dmg}
 \DeclareDocumentCommand{\txarb}{+m}{\bgroup\textdir
   TRT\arabicfont#1\egroup}
diff --git a/arabluatex.lua b/arabluatex.lua
index d76a284..df23ddc 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -108,6 +108,14 @@ local function indnum(str)
    return str
 end
 
+local function takeoutcap(str)
+   str = string.gsub(str, "(\\cap.?)(%b{})", function(tag, body)
+                        body = string.sub(body, 2, -2)
+                        return string.format("%s", body)
+   end)
+   return str
+end
+
 local function voc(str)
    str = string.gsub(str, "\\arb(%b{})", function(inside)
    inside = string.sub(inside, 2, -2)
@@ -270,6 +278,7 @@ end
 
 function processvoc(str)
    str = "\\arb{".. str.."}"
+   str = takeoutcap(str)
    str = protectarb(str)
    str = breakcmd(str)
    str = holdcmd(str)
@@ -280,6 +289,7 @@ end
 
 function processfullvoc(str)
    str = "\\arb{".. str.."}"
+   str = takeoutcap(str)
    str = protectarb(str)
    str = breakcmd(str)
    str = holdcmd(str)
@@ -290,6 +300,7 @@ end
 
 function processnovoc(str)
    str = "\\arb{".. str.."}"
+   str = takeoutcap(str)
    str = protectarb(str)
    str = breakcmd(str)
    str = holdcmd(str)
@@ -312,6 +323,27 @@ function processtrans(str, mode)
 return str
 end
 
+function cap(str)
+   str = string.gsub(str, "(\\txtrans.?)(%b{})", function(tag, body)
+                        body = string.sub(body, 2, -2)
+                        return string.format("%s", body)
+   end)
+   if string.find(str, "%-['`ʾʿ]") then
+      str = string.gsub(str, "(%-['`])", "%1\\MakeUppercase ")
+      str = string.gsub(str, "(%-ʿ)", "%1\\MakeUppercase ")
+      str = string.gsub(str, "(%-ʾ)", "%1\\MakeUppercase ")
+   elseif string.find(str, "%-[^'`ʾʿ]") then
+      str = string.gsub(str, "(%-)", "%1\\MakeUppercase ")
+   elseif string.find(str, "^['`ʾʿ]") then
+      str = string.gsub(str, "^(['`])", "%1\\MakeUppercase ")
+      str = string.gsub(str, "^(ʿ)", "%1\\MakeUppercase ")
+      str = string.gsub(str, "^(ʾ)", "%1\\MakeUppercase ")
+   else
+      str = "\\MakeUppercase "..str
+   end
+return str
+end
+
 --print(processvoc("'at_A .sadIquN 'il_A ju.hA ya.tlubu min-hu .himAra-hu li-yarkiba-hu fI safraTiN qa.sIraTiN wa-qAla la-hu sawfa 'u`Idu-hu 'ilay-ka fI 'l-masA'i wa-'adfa`u la-ka 'ujraTaN 'at_A .sadIquN 'il_A ju.hA ya.tlubu min-hu .himAra-hu li-yarkiba-hu fI safraTiN qa.sIraTiN wa-qAla la-hu sawfa 'u`Idu-hu 'ilay-ka fI 'l-masA'i wa-'adfa`u la-ka 'ujraTaN"))
 --print(arbnum("bop 0123456789"))
 --print(processvoc("daxala \\arb[novoc]{mubtasimaN}"))
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index 8a9aee8..19411f1 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -70,14 +70,14 @@ trigraphstrdmg = { -- trigraphs or more
    {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%2"},
-   {a="(%s)(a)l%-(l)", b="%1%2l-%3%3"},
+   {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"},
+   {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="%1%2-"},
+   {a="(%s)(a)l%-([%_%^%.]?[tdrzsn])", b="%1%2%3-"},
    -- assim. art. + solar consonant
-   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="%1%2-%2"},
-   {a="(%s)(a)([%_%^%.]?[tdrzsn])%-", b="%1%2%3-%3"},
+   {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"},
@@ -231,8 +231,8 @@ trigraphstrloc = { -- trigraphs or more
    {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="%1l-%2"},
    {a="(%s)(a)l%-([%_%^%.]?[tdrzsn])", b="%1%2l-%3"},
    -- assim. art. + solar consonant
-   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="%1l-%2"},
-   {a="(%s)(a)([%_%^%.]?[tdrzsn])%-", b="%1%2l-%3"},
+   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="%1l-"},
+   {a="(%s)(a)([%_%^%.]?[tdrzsn])%-", b="%1%2l-"},
    -- al- + initial unstable hamza
    {a="^(a)l%-([uai])", b="%1l-%2"},
    {a="(%s)(a)l%-([uai])", b="%1%2l-%3"},
@@ -339,6 +339,40 @@ finaltrloc = {
    {a="ʾ", b="'"},
 }
 
+-- cap
+captr = {
+   -- dmg (defaut); loc as well
+   {a="ā", b="Ā"},
+   {a="b", b="B"},
+   {a="t", b="T"},
+   {a="ṯ", b="Ṯ"},
+   {a="ǧ", b="Ǧ"},
+   {a="ḥ", b="Ḥ"},
+   {a="ḫ", b="Ḫ"},
+   {a="d", b="D"},
+   {a="ḏ", b="Ḏ"},
+   {a="r", b="R"},
+   {a="z", b="Z"},
+   {a="s", b="S"},
+   {a="š", b="Š"},
+   {a="ṣ", b="Ṣ"},
+   {a="ḍ", b="Ḍ"},
+   {a="ṭ", b="Ṭ"},
+   {a="ẓ", b="Ẓ"},
+   {a="ġ", b="Ġ"},
+   {a="f", b="F"},
+   {a="q", b="Q"},
+   {a="k", b="K"},
+   {a="l", b="L"},
+   {a="m", b="M"},
+   {a="n", b="N"},
+   {a="h", b="H"},
+   {a="w", b="W"},
+   {a="ū", b="Ū"},
+   {a="y", b="Y"},
+   {a="ī", b="Ī"}
+}
+
 -- return {
 -- raw = raw,
 -- hamza = hamza,