#!/usr/bin/env lua5.1
-- This file: (find-angg "LUA/texcatcodes.lua")
-- See:       (find-angg "LUA/oldglyphs.lua")
--            (find-LATEXfile "2009unilog-dnc.tex" "\\usepackage{edrx08}")
--            (find-angg "dednat4/examples/edrx08.sty" "glyphs")
--            (find-LATEXgrep "grep --color -nH --null -e raw-text *.tex")
--            (find-eevfile "eev-math-glyphs.el" "defun eev-math-glyphs-edrx")
--            (find-TH "glyphs")
--            http://angg.twu.net/glyphs.html
--            (find-LATEX "istanbulglyphs.el")
--
-- Most of my .tex files until 2014 were written in raw-text-unix with
-- one-byte glyphs. This file implements SOME functions to convert
-- files in that format to the new format with unibyte glyphs, but
-- this is a mess!
--
--   (find-LATEX     "2009unilog-dnc.tex")
--   (find-LATEXfile "2009unilog-dnc.tex" "\\usepackage{edrx08}")
--   (find-angg "dednat4/examples/edrx08.sty" "glyphs")
--
-- (defun e () (interactive) (find-fline "~/LUA/texcatcodes.lua"))
--
-- (find-sh "./texcatcodes.lua arg1 arg2")
--
-- «.getdef»		(to "getdef")
-- «.bslify»		(to "bslify")
-- «.Triples»		(to "Triples")
-- «.translation»	(to "translation")



-- «getdef»  (to ".getdef")
getdef = function (li)
    if not li:match"^\\catcode" then return end
    local c,def = li:match"\\def([^{}]+)(%b{})"
    if not c then return end
    return c, def:sub(2,-2)
  end
getdefs = function (bigstr)
    local A = {}
    for li in bigstr:gmatch"([^\n]+)" do
      if getdef(li) then table.insert(A, {getdef(li)}) end
    end
    return A
  end
gendefs = function (bigstr)
    return cow(function ()
        for li in bigstr:gmatch"([^\n]+)" do
          if getdef(li) then coy(getdef(li)) end
        end
      end)
  end

-- «bslify»  (to ".bslify")
-- backslashify
bslify1 = function (c) return format("\\%d", string.byte(c, 1)) end
bslify  = function (str) return '"'..(str:gsub("[\1-\9\11-\31\127-\255]", bslify1))..'"' end
string.bslify = bslify
-- = bslify("\5")

-- «Triples»  (to ".Triples")
-- (find-angg "LUA/lua50init.lua" "SetL")

Triples = SetL.new()
Triples.entry = function (def)
    if not Triples:has(def) then Triples:add(def, {}) end
    return Triples:val(def)
  end
Triples.goodchars0 = function ()
    local cs = Set.new()
    for k,v in Triples:gen() do if v.old and v.new then cs:add(v.old) end end
    return cs:ks()
  end
Triples.goodchars = function () return table.concat(Triples.goodchars0()) end
Triples.substs = function ()
    local ss = {}
    for k,v in Triples:gen() do if v.old and v.new then ss[v.old] = v.new end end
    return ss
  end

Triples.getdefault = function ()
    -- (find-LATEX "edrxchars.tex")
    -- (find-angg "dednat4/examples/edrx08.sty")
    bigstrnew = ee_readfile "~/LATEX/edrxchars.tex"
    bigstrold = ee_readfile "~/dednat4/examples/edrx08.sty"
    for c,def in gendefs(bigstrold) do Triples.entry(def).old = c end
    for c,def in gendefs(bigstrnew) do Triples.entry(def).new = c end
    -- Triples.entry("\\pi").old = "\21"
    -- Triples.entry("\\epsilon").old = "\5"
    --
    oldchars = Triples.goodchars()
    substs = Triples.substs()
  end
Triples.subst = function (bigstr)
    return bigstr:gsub("["..oldchars.."]", substs)
  end

--[[
* (eepitch-lua51)
* (eepitch-kill)
* (eepitch-lua51)
dofile "texcatcodes.lua"
Triples.getdefault()
= Triples:ksc(" ")
= bslify(Triples:goodchars())

= bslify(oldchars)

-- PPV(Triples.substs())
PPV(Triples:val("\\sigma"))
PPV(Triples:val("\\pi"))
PPV(Triples.entry("\\pi"))

--]]


Translate = Class {
  type    = "Translate",
  __index = {
    translatables = function () end,
    nontranslatables = function () end,
  },
}




sortchars = function (str)
    local cs = SetL.new()
    for c in str:gmatch(".") do cs:add(c) end
    return cs:ksc("")
  end
non_translatable_chars_in = function (bigstr)
    local bigstr = bigstr:gsub("[^\1-\8\11-\31\127-\255]", "") -- del non-ascii
    local bigstr = bigstr:gsub("["..oldchars.."]", "")     -- del translatables
    return sortchars(bigstr)
  end


--[[
* (eepitch-lua51)
* (eepitch-kill)
* (eepitch-lua51)
dofile "texcatcodes.lua"
= sortchars "Hello! Bye!"

-- (find-angg "LATEX/2008hyp.tex")
bigstr = ee_readfile "~/LATEX/2008hyp.tex"
bigstr = ee_readfile "~/LATEX/2008notations.tex"
= non_translatable_chars_in(bigstr)

--]]






-- «translation»  (to ".translation")
--
AB = {}
BC = {}
AC = {}
badAs  = SetL.new()
goodAs = SetL.new()
BC["\\varepsilon"] = "ε"
BC["\\underline"]  = "▁"


oldchars = "\4\5\14\16\17\18\20\21\23\28\31\130\162\164\165\166\167\168\169\172"..
  "\173\174\175\179\180\183\184\197\198\203\204\206\207\208\209\215\216\219\221"..
  "\222\229\238\240\241\247\253\254"
translation = {
  [  "\4"] = "δ", [  "\5"] = "ε", [ "\14"] = "σ", [ "\16"] = "⟧",
  [ "\17"] = "⟦", [ "\18"] = "ρ", [ "\20"] = "θ", [ "\21"] = "π",
  [ "\23"] = "⊓", [ "\28"] = "τ", [ "\31"] = "∨", ["\130"] = "∞",
  ["\162"] = "∘", ["\164"] = "⊗", ["\165"] = "⊕", ["\166"] = "𝐛",
  ["\167"] = "⊤", ["\168"] = "⠆", ["\169"] = "≥", ["\172"] = "¬",
  ["\173"] = "▁", ["\174"] = "⊥", ["\175"] = "𝐫", ["\179"] = "¹",
  ["\180"] = "∧", ["\183"] = "·", ["\184"] = "π", ["\197"] = "∫",
  ["\198"] = "Σ", ["\203"] = "𝐭", ["\204"] = "∩", ["\206"] = "∃",
  ["\207"] = "ω", ["\208"] = "𝐬", ["\209"] = "⅋", ["\215"] = "×",
  ["\216"] = "Ω", ["\219"] = "ν", ["\221"] = "∈", ["\222"] = "Θ",
  ["\229"] = "Π", ["\238"] = "♮", ["\240"] = "λ", ["\241"] = "◻",
  ["\247"] = "⊔", ["\253"] = "∀", ["\254"] = "∪"
}

translation["\182"] = "<<limp>>"
translation["\182"] = "→"
translation["\171"] = "«"
translation["\187"] = "»"

oldchars2 = table.concat(sorted(keys(translation)), "")

--[[
* (eepitch-lua51)
* (eepitch-kill)
* (eepitch-lua51)
dofile "texcatcodes.lua"
= oldchars
= oldchars2

--]]



-- Command-line arguments:
a,b,c = ...
if a == "-trans" then 
  bigstrold = ee_readfile(b)
  bigstrnew, nc = bigstrold:gsub("["..oldchars.."]", translation)
  bigstrnew, nc = bigstrold:gsub("[\1-\8\11-\31\127-\255]", translation)
  bigstrnew2,nu = bigstrnew:gsub("raw%-text%-unix", "utf-8-unix")
  print(nc, nu)
  ee_writefile(c, bigstrnew2)
end

-- (find-fline "~/LATEX/2010kockdiff.tex")
-- (find-sh  "~/LUA/texcatcodes.lua -trans ~/LATEX/2010kockdiff.tex")
-- (find-sh0 "~/LUA/texcatcodes.lua -trans ~/LATEX/2010kockdiff.tex > /tmp/o")
-- (find-LATEXsh "~/LUA/texcatcodes.lua -trans 2010kockdiff.tex /tmp/foo")
-- (find-fline "/tmp/foo")

--[[
;; (find-eevfile "eev-math-glyphs.el" "920")
(find-einsert '(920 928 931 937 948 949 952 955 957 960 961 963 964
   969 8868 8869 8743 8744 8835 8704 8707 9633 9744 8758 8712 9675
   8745 8746 8747 8711 8734 8805 185 8851 8852 8523 8853 8855 8594
   8888 9480 9838 9623 9629 8988 8989))

--]]


--[[
* (eepitch-lua51)
* (eepitch-kill)
* (eepitch-lua51)
dofile "texcatcodes.lua"

OldCatcode = Class {
  type = "OldCatcode",
  new  = function (c, d)
      return OldCatcode {c=c, d=d}
    end,
  fromline = function (li)
      if not li:match"^\\catcode" then return end
      local c,def = li:match"\\def([^{}]+)(%b{})"
      if not c then return end
      return OldCatcode {c=c, d=def:sub(2,-2)}
    end,
  __index = {
    tolatex = function (occ)
        return format("\\catcode`%s=13 \\def%s{%s}", occ.c, occ.c, occ.d)
      end,
  },
}

= OldCatcode.new("@", "At"):tolatex()

--]]



-- (find-es "lua5" "cow-and-coy")

--[[
* (eepitch-lua51)
* (eepitch-kill)
* (eepitch-lua51)
dofile "texcatcodes.lua"
bigstrnew = ee_readfile "~/LATEX/edrxchars.tex"
bigstrold = ee_readfile "~/dednat4/examples/edrx08.sty"

for c,def in gendefs(bigstrnew) do BC[def] = c end
for c,def in gendefs(bigstrold) do
  local a = c
  if BC[def] then
    table.insert(AB, {a, BC[def]})
    goodAs:add(a)
    AC[a] = BC[def]
  else
    badAs:add(a)
    print("No translation for "..bslify(a).." -> "..def)
  end
end
-- PPV(AB)
-- PPV(badAs)
= badAs:ksc(""):bslify()
= goodAs:ksc(""):bslify()
for c in goodAs:ksc(""):gmatch"." do
  printf("  [%6s] = \"%s\",\n", c:bslify(), AC[c])
end


for c,def in gendefs(bigstrold) do print(c:bslify(), def) end



PPV(AC)


PPV(getdefs(bigstrnew))
PPV(getdefs(bigstrold))

for c,def in gendefs(bigstrold) do Triples.entry(def).old = c end
for c,def in gendefs(bigstrnew) do Triples.entry(def).new = c end

oldchars = Triples.goodchars()
substs = Triples.substs()

f = function (c) return format("\\%d", string.byte(c, 1)) end
= oldchars:gsub(".", f)
-- PPV(substs)




--  _____ ___  ____   ___  
-- |_   _/ _ \|  _ \ / _ \ 
--   | || | | | | | | | | |
--   | || |_| | |_| | |_| |
--   |_| \___/|____/ \___/ 
--                         
Translations = SetL.new()





-- (find-LATEXgrep "grep -niH -e kock *")
-- (find-fline "~/LATEX/edrxchars.tex")
-- (find-fline "~/dednat4/examples/")
-- (find-fline "~/dednat4/examples/edrx08.sty")
-- (find-sh0 "lua51 -e 'print(tonumber(\"2293\", 16))'")
-- (find-sh0 "lua51 -e 'print(tonumber(\"214B\", 16))'")

-- (insert 8523)




= f("@")
gcs = map(f, Triples.goodchars())
PPV(gcs)
= table.concat(gcs)




Triples.texs = function (f)
    local ks = Set.new()
    f = f or function (v) return true end
    for k,v in Triples:gen() do if f(v) then ks:add(k) end end
    return ks:ks()
  end

PPV(map(f, Triples.goodchars()))
PPV(table.concat(map(f, Triples.goodchars())))

= Triples.keys
PPV(Triples.texs())
PPV(Triples.texs(function (v) return v.new end))
PPV(Triples.texs(function (v) return v.old end))

print("\64")



--]]