arabluatex.lua: newtable 'outofarb' with an option in mkarbbreak() for inserting new entries in this table. added LPeg spcenc^-1 in holdcmd(). new function takeoutarb().

1 files changed, 92 insertions, 41 deletions

diff --git a/arabluatex.lua b/arabluatex.lua
index d77aa51..b8ef9df 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -65,9 +65,17 @@ local function unprotectarb(str)
    return str
 end
 
--- commands the arguments of which must not be processed by
--- arabluatex.  'albrkcmds' is what is set by default.  'brkcmds' is
--- what may be set in the preamble as argument of \MkArbBreak{}
+-- the following is to be taken out of \arb{}
+local outofarb = {
+   "LRfootnote",
+   "RLfootnote",
+   "edtext",
+   "pstart",
+   "pend"
+}
+-- commands the arguments of which must not be processed by arabluatex
+-- inside \arb{}.  'albrkcmds' is what is set by default.  'brkcmds'
+-- collects the commands set in the preamble with \MkArbBreak{}
 local albrkcmds = {
    "begin",
    "end",
@@ -75,23 +83,31 @@ local albrkcmds = {
    "LRmarginpar",
    "Footnote",
    "Marginpar",
-   "edtext",
-   "pstart",
-   "pend",
-   "arbmark"
+   "arbmark",
+   "textcolor",
+   "abjad"
 }
 local brkcmds = {}
 
-function mkarbbreak(str)
+function mkarbbreak(str, opt)
    str = str ..","
    str = string.gsub(str, "%s+", "")
    local fieldstart = 1
-   repeat
-      local nexti = string.find(str, "%,", fieldstart)
-      table.insert(brkcmds, string.sub(str, fieldstart, nexti-1))
-      fieldstart = nexti +1
-   until fieldstart > string.len(str)
-   return brkcmds
+   if opt == "dflt" then
+      repeat
+         local nexti = string.find(str, "%,", fieldstart)
+         table.insert(brkcmds, string.sub(str, fieldstart, nexti-1))
+         fieldstart = nexti +1
+      until fieldstart > string.len(str)
+      return brkcmds
+   elseif opt == "out" then
+      repeat
+         local nexti = string.find(str, "%,", fieldstart)
+         table.insert(outofarb, string.sub(str, fieldstart, nexti-1))
+         fieldstart = nexti +1
+      until fieldstart > string.len(str)
+      return outofarb
+   end
 end
 
 local function breakcmd(str)
@@ -103,29 +119,37 @@ local function breakcmd(str)
         end)
         str = string.gsub(str, "(\\item)(%s+)", "%1{}%2")
    -- \textcolor
-   str = string.gsub(str, "\\(textcolor%s?)(%b{})(%b{})",
-    function(tag, bodycolor, bodytext)
-    bodycolor = string.sub(bodycolor, 2, -2)
-    bodytext = string.sub(bodytext, 2, -2)
-    return string.format("\\al@brk{\\%s{%s}{\\arb{%s}}}", tag, bodycolor, bodytext)
-   end)
+   -- str = string.gsub(str, "\\(textcolor%s?)(%b{})(%b{})",
+   --  function(tag, bodycolor, bodytext)
+   --  bodycolor = string.sub(bodycolor, 2, -2)
+   --  bodytext = string.sub(bodytext, 2, -2)
+   --  return string.format("\\al@brk{\\%s{%s}{\\arb{%s}}}", tag, bodycolor, bodytext)
+   -- end)
+        str = string.gsub(str, "(.)(\\arbcolor%s?)(%b[])(%b{})(.)",
+          function (li, tag, opt, body, lii)
+             li = string.gsub(li, "%+", "‍") -- ^^^^200d
+             lii = string.gsub(lii, "%+", "‍") -- ^^^^200d
+             body = string.sub(body, 2, -2)
+             body = string.gsub(body, "%+", "‍") -- ^^^^200d
+             return string.format("%s%s%s{%s}%s", li, tag, opt, body, lii)
+        end)
    -- Abjad
-   str = string.gsub(str, "\\(abjad%s?)(%b{})",
-                     function(tag, body)
-                        body = string.sub(body, 2, -2)
-                        return string.format("\\al@brk{\\%s{%s}}", tag, body)
-   end)
-   -- LR/RL footnotes
-   str = string.gsub(str, "\\(RLfootnote%s?)(%b{})",
-        function(tag, body)
-        body = string.sub(body, 2, -2)
-        return string.format("}\\%s{%s}\\arb{", tag, body)
-   end)
-   str = string.gsub(str, "\\(LRfootnote%s?)(%b{})",
-        function(tag, body)
-        body = string.sub(body, 2, -2)
-        return string.format("}\\%s{%s}\\arb{", tag, body)
-   end)
+   -- str = string.gsub(str, "\\(abjad%s?)(%b{})",
+   --                function(tag, body)
+   --                   body = string.sub(body, 2, -2)
+   --                   return string.format("\\al@brk{\\%s{%s}}", tag, body)
+   -- end)
+---[[
+   -- to be taken out of \arb{}, as LR/RL footnotes and the like:
+   for i = 1,#outofarb do
+      str = gsub(str, dblbkslash * lpeg.Cs(outofarb[i]) * cmdargs, "}%1%2%3\\arb{")
+      -- str = gsub(str, dblbkslash * lpeg.Cs(outofarb[i]) * spcenc^-1 * bcbraces,
+      --                 function(prefix, tag, body)
+      --                    body = string.sub(body, 2, -2)
+      --                    return string.format("}%s%s{%s}\\arb{", prefix, tag, body)
+      -- end)
+   end
+--]]
    -- commands set by default in albrkcmds
    for i = 1,#albrkcmds do
       str = gsub(str, dblbkslash * lpeg.Cs(albrkcmds[i]) * cmdargs, "\\al@brk{%1%2%3}")
@@ -144,7 +168,7 @@ end
 local function holdcmd(str)
    str = gsub(str, lpeg.Cs("\\arb") * bcbraces, function(tag, body)
         body = string.sub(body, 2, -2)
-        body = gsub(body, cmd * bsqbracketsii * spcenc^-1 * bcbraces, function(btag, bopt, bbody)
+        body = gsub(body, cmd * spcenc^-1 * bsqbracketsii * spcenc^-1 * bcbraces, function(btag, bopt, bbody)
         bbody = string.sub(bbody, 2, -2)
         if string.find(btag, "@") then
            return holdcmd(string.format("}%s%s{%s}\\arb{", btag, bopt, bbody))
@@ -218,6 +242,32 @@ local function takeoutcapetc(str)
    return str
 end
 
+local function takeoutarb(str)
+   for i = 1,#outofarb do
+      str = gsub(str, dblbkslash * lpeg.Cs(outofarb[i]) * cmdargs,
+                 function(prefix, tag, body)
+--                  body = string.sub(body, 2, -2)
+                    body = gsub(body, lpeg.P("\\arb"), "\\@rb")
+                    return string.format("%s%s%s", prefix, tag, body)
+      end)
+   end
+   -- for i = 1,#outofarb do
+   --    str = gsub(str, dblbkslash * lpeg.Cs(outofarb[i]) * spcenc^-1 * bcbraces,
+   --            function(prefix, tag, body)
+   --               body = string.sub(body, 2, -2)
+   --               body = gsub(body, lpeg.P("\\arb"), "\\@rb")
+   --               return string.format("%s%s{%s}", prefix, tag, body)
+   --    end)
+   -- end
+   str = string.gsub(str, "(\\arb%s?)(%b{})", function(tag, body)
+                        body = string.sub(body, 2, -2)
+                        return string.format("%s", body)
+   end)
+   str = string.gsub(str, "\\@rb", "\\arb")
+   str = "\\arb{"..str.."}"
+   return str
+end
+
 local function voc(str, rules)
    str = string.gsub(str, "\\arb(%b{})", function(inside)
    inside = string.sub(inside, 2, -2)
@@ -599,7 +649,7 @@ return str
 end
 
 function processvoc(str, rules, scheme)
-   str = "\\arb{".. str.."}"
+   str = takeoutarb(str)
    str = processarbnull(str, scheme)
    str = takeoutcapetc(str)
    str = protectarb(str)
@@ -618,7 +668,7 @@ return str
 end
 
 function processfullvoc(str, rules, scheme)
-   str = "\\arb{".. str.."}"
+   str = takeoutarb(str)
    str = processarbnull(str, scheme)
    str = takeoutcapetc(str)
    str = protectarb(str)
@@ -639,7 +689,7 @@ return str
 end
 
 function processnovoc(str, rules, scheme)
-   str = "\\arb{".. str.."}"
+   str = takeoutarb(str)
    str = processarbnull(str, scheme)
    str = takeoutcapetc(str)
    str = protectarb(str)
@@ -658,7 +708,7 @@ return str
 end
 
 function processtrans(str, mode, rules, scheme)
-   str = "\\arb{".. str.."}"
+   str = takeoutarb(str)
    str = processdiscretionary(str)
    str = processarbnull(str, scheme)
    str = takeoutabjad(str)
@@ -682,6 +732,7 @@ end
 function newarbmark(abbr, rtlmk, ltrmk)
    rtlmk = "\\arabicfont "..rtlmk
    table.insert(arbmarks, {a = abbr, b = rtlmk, c = ltrmk})
+   table.sort(arbmarks, function(a ,b) return(#a.a > #b.a) end)
    return true
 end