kct-uni-2.lua

#!/usr/bin/env texlua
local src = {}
src[0] = io.open(arg[1],'r')
src[1] = io.open(arg[2],'r')
src[2] = io.open(arg[3],'r')
src[3] = io.open(arg[4],'r')
src[4] = io.open(arg[5],'r')
src[5] = io.open(arg[6],'r')

local texout = io.open(arg[7],'w')
tprint = function (a) texout:write(a .. '\n') end
local txtout = io.open(arg[8],'w')
dprint = function (a) txtout:write(a .. '\n') end

tprint('\\documentclass[uplatex]{jsarticle}')
tprint('\\usepackage[dvipdfmx]{graphicx,xcolor}')
tprint('\\definecolor{green}{rgb}{0,0.5,0}\n')
tprint('\\usepackage[width=44zw,lines=45]{geometry}')
tprint('\\usepackage{supertabular,booktabs,metalogo,lmodern}')
tprint('\\begin{document}\\fboxsep0pt{\\centering')

tprint('\\tablehead{%')
tprint('\\toprule')
tprint('&UCS&JIS&\\pTeX&u\\pTeX&\\XeTeX&Lua\\TeX&Lua\\TeX-ja&\\\\')
tprint('\\midrule')
tprint('}%')
tprint('\\tabletail{\\bottomrule}')
tprint('\\begin{supertabular}{ccccccccp{5em}}')

dprint('## kct-uni-out.txt: \\[k]catcode table of JIS X 0213 non-Kanji characters.')
dprint('## This is generated by kct-uni.sh.')
dprint('##')
dprint("## JIS04 characters which can't be represent by one UCS character")
dprint("## are omitted from this table.")
dprint('##')
dprint('## UCS\tJIS\t区\t点\tpTeX\tupTeX\tXeTeX\tLuaTeX\tLuaTeX-ja\tNote')

local cxt = function(n)
  return '\\textcolor{' .. 
         (((n==16 or n==17 or n==11) and  'green')
           or ((n==15) and 'blue' or 'red'))
     .. '}{' .. tostring(n) .. '}'
end
local cyt = function(n, x)
   return cxt(n) .. ' (\\textcolor{'
      .. (x=='0' and 'black}{\\tt\\bfseries JA}' or 'blue}{\\tt AL}' )
      .. ')'
end
local czt = function(n, x)
   return (x=='N' and ('(' .. cxt(n) .. ')') or cxt(n) )
end

local cxd = function(n)
  return tostring(n) .. '\t'
end
local cyd = function(n, x)
   return cxd(n) .. (x==0 and 'J\t' or 'A\t' )
end
local czd = function(n, x)
   return (x=='N' and '-' or tostring(n) ) .. '\t'
end


local moji = function(c)
  return '\\fcolorbox{cyan}{white}{\\Large\\kchar"'.. c .. '}'
end

local z = {}
local zu, zx, zlc, zlj
for il in src[0]:lines() do
    zu = src[1]:read('*n')
    zx = src[2]:read('*n')
    zl = src[3]:read('*n')
    zp = src[5]:read('*n')
    zlc, zlj = (src[4]:read('*l')):match('(%w*)-(%w*)')
    zlc = tonumber(zlc)
    local za, zb, zc, zd = il:match('(%w*) (%w*) (.)(.)')
--    if (zlc==11 and zu==18)or(zlc==12 and zu~=18) then
    tprint(
      moji(zb) .. '&\\tt U+' .. zb .. '&\\tt 0x' .. za .. '&'
	 .. czt(zp, zd) .. '&' .. cxt(zu).. '&' .. cxt(zx) .. '&'
         .. cxt(zl) .. '&' .. cyt(zlc, zlj)  
	 .. (zc=='F' and '&Fullwidth' or (zc=='W' and '&Windows' or ''))
     .. '\\\\')
    dprint('U+' .. zb .. '\t0x' .. za .. '\t'
         .. (tonumber(za,16)//256-32) .. '\t'
         .. (tonumber(za,16)%256-32) .. '\t'
         .. czd(zp,zd) 
	 .. cxd(zu).. cxd(zx).. cxd(zl) .. cyd(zlc, zlj)
              .. (zd=='N' and '[JIS04] ' or '')
              .. (zc=='F' and 'Fullwidth' or (zc=='W' and 'Windows' or ''))
    )
--    end
end
for k=0,5 do z[k]=src[k]:close() end

tprint('\\end{supertabular}\\par}\\medskip')
tprint('\\pTeX の列の括弧囲みは，その文字がJIS~X~0208:1990に規定されていないことを示す．\\\\')
tprint('Unicodeで合成文字で表さないと表せない文字は省略した．')
tprint('\\end{document}')

texout:close()
txtout:close()