huge commit: improved LPeg patterns, (un)protectarb() now operates on \arb[]{}, LR{}/LR{} and any command found in (al)brkcmds tables, which withholds the execution of the 'breaking' mechanism until holdcmd() function operates.

improved holdcmd() with LPeg patterns: commands with up to two optional arguments are now processed. moved back \par into albrkcmds.

1 files changed, 41 insertions, 46 deletions

diff --git a/arabluatex.lua b/arabluatex.lua
index 6a08a33..d77aa51 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -34,14 +34,35 @@ local function gsub(s, patt, repl)
   return lpeg.match(patt, s)
 end
 
+-- some basic patterns:
+local ascii = lpeg.R("az", "AZ", "@@")
+local dblbkslash = lpeg.Cs("\\")
+local bsqbrackets = lpeg.Cs{ "[" * ((1 - lpeg.S"[]") + lpeg.V(1))^0 * "]" }
+local bcbraces = lpeg.Cs{ "{" * ((1 - lpeg.S"{}") + lpeg.V(1))^0 * "}" }
+local spce = lpeg.Cs(" ")
+local spcenc = lpeg.P(" ")
+local cmdstar = lpeg.Cs(spce * lpeg.P("*"))
+local bsqbracketsii = lpeg.Cs(bsqbrackets^-2)
+local bcbracesii = lpeg.Cs(bcbraces^-2)
+local cmd = lpeg.Cs(dblbkslash * ascii^1 * cmdstar^-1)
+local cmdargs = lpeg.Cs(spce * bsqbracketsii * bcbracesii * bsqbrackets^-1)
+
 local function protectarb(str)
    str = string.gsub(str, "(\\arb%s?)(%[.-%])(%b{})", "\\@arb%2%3")
-return str
+   str = string.gsub(str, "(\\LR%s?)(%b{})", "\\@LR%2")
+   str = string.gsub(str, "(\\RL%s?)(%b{})", "\\@RL%2")
+   return str
 end
 
 local function unprotectarb(str)
    str = string.gsub(str, "(\\@arb)(%[.-%])(%b{})", "\\arb%2%3")
-return str
+   str = string.gsub(str, "(\\@LR)(%b{})", "\\LR%2")
+   str = string.gsub(str, "(\\@RL)(%b{})", "\\RL%2")
+   str = gsub(str, lpeg.Cs("\\al@brk") * bcbraces, function(tag, body)
+                 body = string.sub(body, 2, -2)
+                 return string.format("%s", body)
+   end)
+   return str
 end
 
 -- commands the arguments of which must not be processed by
@@ -50,10 +71,7 @@ end
 local albrkcmds = {
    "begin",
    "end",
-   "RL",
-   "LR",
-   "RLfootnote",
-   "LRfootnote",
+   "par",
    "LRmarginpar",
    "Footnote",
    "Marginpar",
@@ -76,25 +94,12 @@ function mkarbbreak(str)
    return brkcmds
 end
 
--- some basic patterns:
-local ascii = lpeg.R("az", "AZ", "@@")
-local dblbkslash = lpeg.Cs("\\")
-local bsqbrackets = lpeg.Cs{ "[" * ((1 - lpeg.S"[]") + lpeg.V(1))^0 * "]" }
-local bcbraces = lpeg.Cs{ "{" * ((1 - lpeg.S"{}") + lpeg.V(1))^0 * "}" }
-local spce = lpeg.Cs(" ")
-local spcenc = lpeg.P(" ")
-local cmdstar = lpeg.Cs(spce * lpeg.P("*"))
-local bsqbracketsii = lpeg.Cs(bsqbrackets^-2)
-local bcbracesii = lpeg.Cs(bcbraces^-2)
-local cmd = lpeg.Cs(dblbkslash * ascii^1 * cmdstar^-1)
-local cmdargs = lpeg.Cs(spce * bsqbracketsii * bcbracesii * bsqbrackets^-1)
-
 local function breakcmd(str)
    -- process \item[], then \item[]
         str = string.gsub(str, "\\(item.?)(%b[])",
           function(tag, body)
           body = string.sub(body, 2, -2)
-          return string.format("}\\item[\\arb{%s}] \\arb{", body)
+          return string.format("\\al@brk{\\item[\\arb{%s}] }", body)
         end)
         str = string.gsub(str, "(\\item)(%s+)", "%1{}%2")
    -- \textcolor
@@ -102,30 +107,40 @@ local function breakcmd(str)
     function(tag, bodycolor, bodytext)
     bodycolor = string.sub(bodycolor, 2, -2)
     bodytext = string.sub(bodytext, 2, -2)
-    return string.format("}\\%s{%s}{\\arb{%s}}\\arb{", tag, bodycolor, bodytext)
+    return string.format("\\al@brk{\\%s{%s}{\\arb{%s}}}", tag, bodycolor, bodytext)
    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("}\\aemph{\\%s{%s}}\\arb{", tag, body)
+        return string.format("}\\%s{%s}\\arb{", tag, body)
    end)
    -- commands set by default in albrkcmds
    for i = 1,#albrkcmds do
-      str = gsub(str, dblbkslash * lpeg.Cs(albrkcmds[i]) * cmdargs, "}%1%2%3\\arb{")
+      str = gsub(str, dblbkslash * lpeg.Cs(albrkcmds[i]) * cmdargs, "\\al@brk{%1%2%3}")
    end
    -- user commands
    if next(brkcmds) == nil then
       -- nothing to do
    else
       for i = 1,#brkcmds do
-         str = gsub(str, dblbkslash * lpeg.Cs(brkcmds[i]) * cmdargs, "}%1%2%3\\arb{")
+         str = gsub(str, dblbkslash * lpeg.Cs(brkcmds[i]) * cmdargs, "\\al@brk{%1%2%3}")
       end
    end
    return str
 end
 
----[[
 local function holdcmd(str)
    str = gsub(str, lpeg.Cs("\\arb") * bcbraces, function(tag, body)
         body = string.sub(body, 2, -2)
@@ -142,26 +157,6 @@ local function holdcmd(str)
    str = string.gsub(str, "\\arb{}", "")
    return str
 end
---]]
-
---[[
-local function holdcmd(str)
-        str = string.gsub(str, "\\(arb)(%b{})", function(tag, body)
-        body = string.sub(body, 2, -2)
-        body = string.gsub(body, "\\(.-)(%b{})", function(btag, bbody)
-        bbody = string.sub(bbody, 2, -2)
-        if string.find(btag, "@") then
-        return holdcmd(string.format("}\\%s{%s}\\arb{", btag, bbody))
-        else
-        return holdcmd(string.format("}\\%s{\\arb{%s}}\\arb{", btag, bbody))
-        end
-        end)
-        return string.format("\\%s{%s}", tag, body)
-        end)
-        str = string.gsub(str, "\\arb{}", "")
-return str
-end
---]]
 
 local function arbnum(str)
    str = string.gsub(str, "([0-9%,%-%/]+)", function(num)
@@ -220,7 +215,6 @@ local function takeoutcapetc(str)
    end)
    str = string.gsub(str, "\\linebreak", "")
    str = string.gsub(str, "\\%-", "")
-   str = string.gsub(str, "\\(@-)par", "\\%1par{}")
    return str
 end
 
@@ -686,6 +680,7 @@ return str
 end
 
 function newarbmark(abbr, rtlmk, ltrmk)
+   rtlmk = "\\arabicfont "..rtlmk
    table.insert(arbmarks, {a = abbr, b = rtlmk, c = ltrmk})
    return true
 end