Module:Chem2

local getArgs = require('Module:Arguments').getArgs
local p = {} -- module's table

-- Elements with wiki links
local am = {
 H = "Hydrogen",
 He = "Helium",
 Li = "Lithium",
 Be = "Beryllium",
 B = "Boron",
 C = "Carbon",
 N = "Nitrogen",
 O = "Oxygen",
 F = "Fluorine",
 Ne = "Neon",
 Na = "Sodium",
 Mg = "Magnesium",
 Al = "Aluminium",
 Si = "Silicon",
 P = "Phosphorus",
 S = "Sulfur",
 Cl = "Chlorine",
 Ar = "Argon",
 K = "Potassium",
 Ca = "Calcium",
 Sc = "Scandium",
 Ti = "Titanium",
 V = "Vanadium",
 Cr = "Chromium",
 Mn = "Manganese",
 Fe = "Iron",
 Co = "Cobalt",
 Ni = "Nickel",
 Cu = "Copper",
 Zn = "Zinc",
 Ga = "Gallium",
 Ge = "Germanium",
 As = "Arsenic",
 Se = "Selenium",
 Br = "Bromine",
 Kr = "Krypton",
 Rb = "Rubidium",
 Sr = "Strontium",
 Y = "Yttrium",
 Zr = "Zirconium",
 Nb = "Niobium",
 Mo = "Molybdenum",
 Tc = "Technetium",
 Ru = "Ruthenium",
 Rh = "Rhodium",
 Pd = "Palladium",
 Ag = "Silver",
 Cd = "Cadmium",
 In = "Indium",
 Sn = "Tin",
 Sb = "Antimony",
 Te = "Tellurium",
 I = "Iodine",
 Xe = "Xenon",
 Cs = "Caesium",
 Ba = "Barium",
 La = "Lanthanum",
 Ce = "Cerium",
 Pr = "Praseodymium",
 Nd = "Neodymium",
 Pm = "Promethium",
 Sm = "Samarium",
 Eu = "Europium",
 Gd = "Gadolinium",
 Tb = "Terbium",
 Dy = "Dysprosium",
 Ho = "Holmium",
 Er = "Erbium",
 Tm = "Thulium",
 Yb = "Ytterbium",
 Lu = "Lutetium",
 Hf = "Hafnium",
 Ta = "Tantalum",
 W = "Tungsten",
 Re = "Rhenium",
 Os = "Osmium",
 Ir = "Iridium",
 Pt = "Platinum",
 Au = "Gold",
 Hg = "Mercury (element)",
 Tl = "Thallium",
 Pb = "Lead",
 Bi = "Bismuth",
 Po = "Polonium",
 At = "Astatine",
 Rn = "Radon",
 Fr = "Francium",
 Ra = "Radium",
 Ac = "Actinium",
 Th = "Thorium",
 Pa = "Protactinium",
 U = "Uranium",
 Np = "Neptunium",
 Pu = "Plutonium",
 Am = "Americium",
 Cm = "Curium",
 Bk = "Berkelium",
 Cf = "Californium",
 Es = "Einsteinium",
 Fm = "Fermium",
 Md = "Mendelevium",
 No = "Nobelium",
 Lr = "Lawrencium",
 Rf = "Rutherfordium",
 Db = "Dubnium",
 Sg = "Seaborgium",
 Bh = "Bohrium",
 Hs = "Hassium",
 Mt = "Meitnerium",
 Ds = "Darmstadtium",
 Rg = "Roentgenium",
 Cp = "Copernicium",
 Nh = "Nihonium",
 Fl = "Flerovium",
 Mc = "Moscovium",
 Lv = "Livermorium",
 Ts = "Tennessine",
 Og = "Oganesson",
 
 -- Groups etc with element-like names
 Bn = 'Benzyl group',
 Bz = 'Benzoyl group',
 D = 'Deuterium',
 Et = 'Ethyl group',
 Ln = 'Lanthanide',
 Nu = 'Nucleophile',
 Ph = 'Phenyl group',
 R = 'Substituent',
 T = 'Tritium',
 Tf = 'Trifluoromethylsulfonyl group',
 X = 'Halogen',
}

-- Groups which are redirected from their normal target if wikilinked; never
-- autolinked.
local groups = {
 CH3 = 'Methyl group',
 CO3 = 'Carbonate',
 COOH = 'Carboxyl group',
 ClO = 'Hypochlorite',
 ClO2 = 'Chlorite',
 ClO3 = 'Chlorate',
 ClO4 = 'Perchlorate',
 H2O = 'Water of crystallization',
 H3O = 'Hydronium',
 NH2 = 'Amine group',
 NH4 = 'Ammonium',
 NO3 = 'Nitrate',
 PO3 = 'Phosphite',
 PO4 = 'Phosphate',
 SH = 'Thiol group',
 SO3 = 'Sulfite',
 SO4 = 'Sulfate',
 SeH = 'Selenol group'
}

local T_ELEM = 0         -- token types
local T_NUM = 1          -- number
local T_OPEN = 2         -- open '('
local T_CLOSE = 3        -- close ')'
local T_PM_CHARGE = 4    -- + or −
local T_WATER = 6        -- .xH2O x number
local T_CRYSTAL = 9      -- .x
local T_CHARGE = 8       -- charge (x+), (x-)
local T_SUF_CHARGE = 10  -- suffix and charge e.g. 2+ from H2+
local T_SUF_CHARGE2 = 12 -- suffix and (charge) e.g. 2(2+) from He2(2+)
local T_SPECIAL = 14     -- starting with \ e.g. \d for double bond (=)
local T_SPECIAL2 = 16    -- starting with \y{x} e.g. \i{12} for isotope with mass number 12
local T_ARROW_R = 17     -- match: ->
local T_ARROW_EQ = 18    -- match: <->
local T_UNDERSCORE = 19  -- _{ ... }
local T_CARET = 20       -- ^{ ... }
local T_LINKOPEN = 21    -- Opening of link, always like "[[target|" even if the source wasn't
local T_NOCHANGE = 30    -- Anything else like ☃

function su(up, down)
 if up == "" then
  return ('<sub class="template-chem2-sub">%s</sub>'):format(down)
 end
 if down == "" then
  return ('<sup class="template-chem2-sup">%s</sup>'):format(up)
 end
    return ('<span class="template-chem2-su"><span>%s</span><span>%s</span></span>'):format(up, down)
end

function DotIt()
    return '&middot;'
end

function item(f) -- (iterator) returns one token (type, value) at a time from the formula 'f'
    local i = 1

    return function ()
        local t, x = nil, nil

        if (i == 1) and f:match('^[0-9]', i) then 
            x = f:match('^[%d.]+', i); t = T_NOCHANGE; i = i + x:len();   -- matching coefficient (need a space first)

        elseif i <= f:len() then
            x = f:match('^%s+[%d.]+', i); t = T_NOCHANGE;  -- matching coefficient (need a space first)
            if not x then x = f:match('^%s[+]', i); t = T_NOCHANGE; end       -- matching + (H2O + H2O)
            if not x then x = f:match('^%&%#[%w%d]+%;', i); t = T_NOCHANGE; end       -- &#...;
            if not x then x = f:match('^%<%-%>', i); t = T_ARROW_EQ; end       -- matching <->
            if not x then x = f:match('^%-%>', i); t = T_ARROW_R; end       -- matching ->
            if not x then x = f:match('^%u%l*', i); t = T_ELEM; end        -- matching symbols like Aaaaa
            if not x then x = f:match('^%d+[+-]', i); t = T_SUF_CHARGE; end        -- matching x+, x-
            if not x then x = f:match('^%d+%(%d*[+-]%)', i); t = T_SUF_CHARGE2; end        -- matching x(y+/-), x(+/-)
            if not x then x = f:match('^%(%d*[+-]%)', i); t = T_CHARGE; end        -- matching (x+) (xx+), (x-) (xx-)
            if not x then x = f:match('^[%d.]+', i); t = T_NUM; end        -- matching number
            if not x and (f:match('^%[%[%[[^[]', i) or f:match('^%[[^[]', i)) then
             i = i + 1; return T_OPEN, '&#91;' end -- escape [[[X or [X (relevant to auto-linking)
            if not x and f:sub(i, i + 1) == '[[' then
             x = f:match('^%[%[([^]|]*)', i) -- link target
             local len = x:len() + 3
             x = '[[' .. (groups[x] or am[x] or x) .. '|'  -- override link target for common groups

             if f:sub(len + i, len + i) == ']' then
              -- We're going to read the link twice, once as target and once as
              -- chemical markup, e.g. [[CH3]] => "[[CH3|", "CH3]]"
              i = i + 2
             else
              i = i + len
             end
             return T_LINKOPEN, x
            end
            if not x then x = f:match('^[(|{|%[]', i); t = T_OPEN; end     -- matching ({[
            if not x then x = f:match('^[)|}|%]]', i); t = T_CLOSE; end    -- matching )}]
            if not x then x = f:match('^[+-]', i); t = T_PM_CHARGE; end    -- matching + or -
            if not x then x = f:match('^%*[%d.]*H2O', i); t = T_WATER; end -- Crystal water
            if not x then x = f:match('^%*[%d.]*', i); t = T_CRYSTAL; end  -- Crystal
            if not x then x = f:match('^[\\].{%d+}', i); t = T_SPECIAL2; end -- \y{x}
            if not x then x = f:match('^[\\].', i); t = T_SPECIAL; end -- \x
            if not x then x = f:match('^_{[^}]*}', i); t = T_UNDERSCORE; end -- _{...}
            if not x then x = f:match('^^{[^}]*}', i); t = T_CARET; end   -- ^{...}
            if not x then x = f:match('^.', i); t = T_NOCHANGE; end  --the rest - one by one
            if x then i = i + x:len(); else i = i + 999; error("Invalid character in formula! : "..f) end
        end
        return t, x
    end
end

function p._chem(args)
 
    local f = args[1] or ''

 f = mw.text.decode( f, true ) -- handle entity input (like &minus;): decode right away
    f = string.gsub(f, "–", "-")  -- replace – with - (hyphen not ndash)
    f = string.gsub(f, "−", "-")  -- replace – with - (hyphen not minus sign)

    local formula = ''
    local t, x

    local link = args['link'] or ""
    local auto = args['auto'] or ""
    local seen = {}
    local _debug = false

    if not (link == '') then formula = formula .. "[[" .. link .. "|"; end   -- wikilink start [[link|
 
    for t, x in item(f) do 
     if _debug then
      formula = ("%s\n* %d %s"):format(formula, t, x)
        elseif t == T_ELEM then
            if (auto == '') or (not am[x]) or seen[x] then formula = formula .. x
            else formula = ("%s[[%s|%s]]"):format(formula, am[x], x); seen[x] = true
            end
        elseif t == T_COEFFICIENT then formula = formula .. x
        elseif t == T_NUM        then formula = formula .. su("", x);
        elseif t == T_LINKOPEN   then formula = formula .. x;      -- [[Link|
        elseif t == T_OPEN       then formula = formula .. x;          -- ([{
        elseif t == T_CLOSE      then formula = formula .. x;          -- )]}
        elseif t == T_PM_CHARGE  then formula = formula .. su(x:gsub("-", "−"), "");
        elseif t == T_SUF_CHARGE then 
            formula = formula .. su(x:match("[+-]"):gsub("-", "−"), x:match("%d+"), "");
        elseif t == T_SUF_CHARGE2 then 
            formula = formula .. su(x:match("%(%d*[+-]"):gsub("-", "−"):sub(2, -1), x:match("%d+"))
        elseif t == T_CHARGE then
         formula = formula .. "<sup>"
         if x:match("%d+") then formula = formula .. x:match("%d+"); end
         formula = formula .. x:match("[%+-]"):gsub("-", "−") .. "</sup>";
         -- Cannot concatenat a nil value from x:match("%d+");
        elseif t == T_CRYSTAL then formula = formula .. DotIt() .. string.gsub( x, "*", '', 1 );
        elseif t == T_SPECIAL then
            parameter = x:sub(2, 2) -- x fra \x  
            if     parameter == "s" then formula = formula .. "−"   -- single bond
            elseif parameter == "d" then formula = formula .. "="   -- double bond
            elseif parameter == "t" then formula = formula .. "≡"   -- tripple bond
            elseif parameter == "q" then formula = formula .. "≣"   -- Quadruple bond
            elseif parameter == "h" then formula = formula .. "η"   -- η, hapticity
            elseif parameter == "*" then formula = formula .. "*"   -- *, normal *
            elseif parameter == "-" then formula = formula .. "-"   -- -
            elseif parameter == "\\" then formula = formula .. "\\"   -- \
            elseif parameter == "\'" then formula = formula .. "&#39;"   -- html-code for '
            end
        elseif t == T_SPECIAL2 then  -- \y{x}
            parameter = x:sub(2, 2) -- y fra \y{x} 
            if parameter  == "h" then --[[Hapticity]]
                if (auto == '') then formula = formula .. "η<sup>" .. x:match('%d+') .. "</sup>-"
                else
                    formula = formula .. "[[Hapticity|η<sup>" .. x:match('%d+') .. "</sup>]]-"
                end
            elseif parameter == "m" then formula = formula .. "μ<sub>" .. x:match('%d+') .. "</sub>-" -- mu ([[bridging ligand]])
            end
        elseif t == T_WATER then 
            if x:match("^%*[%d.]") then 
                formula = formula .. DotIt() .. x:match("%f[%.%d]%d*%.?%d*%f[^%.%d%]]") .. "H<sub>2</sub>O";
            else
                formula = formula .. DotIt() .. "H<sub>2</sub>O";
            end
        elseif t == T_UNDERSCORE  then formula = formula .. su("", x:gsub("-", "−"):sub(3, -2)) -- x contains _{string}
        elseif t == T_CARET       then formula = formula .. su(x:gsub("-", "−"):sub(3, -2), "") -- x contains ^{string}
        elseif t == T_ARROW_R     then formula = formula .. " → "
        elseif t == T_ARROW_EQ    then formula = formula .. " ⇌ "
        elseif t == T_NOCHANGE    then formula = formula .. x;  -- The rest - everything which isn't captured by the regular expresions.
        else error('unreachable - ???') end -- in fact, unreachable
    end

    if not (link == nil or link == '') then formula = formula .. "]]"; end   -- wikilink closing ]]
    formula = mw.getCurrentFrame():preprocess('<templatestyles src="Module:Chem2/styles.css"/>') ..
        '<span class="chemf nowrap">' .. formula .. '</span>'
    if args[2] or args[3] or args[4] then
        formula = formula .. require('Module:If preview')._warning{
'&#123;&#123;chem2&#125;&#125; was called with multiple positional arguments. It should have just one, e.g. &#123;&#123;chem2&#124;H2O&#125;&#125;.'
        }
    end
    return formula
end

function p.chem(frame)
    local args = getArgs(frame)
    return p._chem(args)
end

-- PRIVATE function to generate documentation.
function p._autodoc(frame)
 local TableTools = require('Module:TableTools') -- we don't want to load this on articles for no reason
 local result = {
  '===Elements and element-style symbols===\nThese may be automatically linked or used as if they were redirects.\n',
  '{| class="wikitable"\n! Symbol !! Link target\n'
 }
 for symbol, target in TableTools.sortedPairs(am) do
  result[#result + 1] = ('|-\n| %s || [[%s]]\n'):format(symbol, target)
 end
 result[#result + 1] = '|}\n===Groups===\nThese must be linked manually; they work as if they were redirects.\n'
 result[#result + 1] = '{| class="wikitable"\n! Symbol !! Link target\n'
 for symbol, target in TableTools.sortedPairs(groups) do
  result[#result + 1] = ('|-\n| %s || [[%s]]\n'):format(symbol, target)
 end
 result[#result + 1] = '|}'
 return table.concat(result)
end

return p