Module:Chem2: Difference between revisions

Content deleted Content added
Inline
Revision as of 12:01, 12 December 2021

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

-- Elements with wiki links
local am = {
H = "[[Hydrogen|H]]",
He = "[[Helium|He]]",
Li = "[[Lithium|Li]]",
Be = "[[Beryllium|Be]]",
B = "[[Boron|B]]",
C = "[[Carbon|C]]",
N = "[[Nitrogen|N]]",
O = "[[Oxygen|O]]",
F = "[[Fluorine|F]]",
Ne = "[[Neon|Ne]]",
Na = "[[Sodium|Na]]",
Mg = "[[Magnesium|Mg]]",
Al = "[[Aluminium |Al]]",
Si = "[[Silicon|Si]]",
P = "[[Phosphorus|P]]",
S = "[[Sulfur|S]]",
Cl = "[[Chlorine|Cl]]",
Ar = "[[Argon|Ar]]",
K = "[[Potassium|K]]",
Ca = "[[Calcium|Ca]]",
Sc = "[[Scandium|Sc]]",
Ti = "[[Titanium|Ti]]",
V = "[[Vanadium|V]]",
Cr = "[[Chromium|Cr]]",
Mn = "[[Manganese|Mn]]",
Fe = "[[Iron|Fe]]",
Co = "[[Cobalt|Co]]",
Ni = "[[Nickel|Ni]]",
Cu = "[[Copper|Cu]]",
Zn = "[[Zinc|Zn]]",
Ga = "[[Gallium|Ga]]",
Ge = "[[Germanium|Ge]]",
As = "[[Arsenic|As]]",
Se = "[[Selenium|Se]]",
Br = "[[Bromine|Br]]",
Kr = "[[Krypton|Kr]]",
Rb = "[[Rubidium|Rb]]",
Sr = "[[Strontium|Sr]]",
Y = "[[Yttrium|Y]]",
Zr = "[[Zirconium|Zr]]",
Nb = "[[Niobium|Nb]]",
Mo = "[[Molybdenum|Mo]]",
Tc = "[[Technetium|Tc]]",
Ru = "[[Ruthenium|Ru]]",
Rh = "[[Rhodium|Rh]]",
Pd = "[[Palladium|Pd]]",
Ag = "[[Silver|Ag]]",
Cd = "[[Cadmium|Cd]]",
In = "[[Indium|In]]",
Sn = "[[Tin|Sn]]",
Sb = "[[Antimony|Sb]]",
Te = "[[Tellurium|Te]]",
I = "[[Iodine|I]]",
Xe = "[[Xenon|Xe]]",
Cs = "[[Caesium|Cs]]",
Ba = "[[Barium|Ba]]",
La = "[[Lanthanum|La]]",
Ce = "[[Cerium|Ce]]",
Pr = "[[Praseodymium|Pr]]",
Nd = "[[Neodymium|Nd]]",
Pm = "[[Promethium|Pm]]",
Sm = "[[Samarium|Sm]]",
Eu = "[[Europium|Eu]]",
Gd = "[[Gadolinium|Gd]]",
Tb = "[[Terbium|Tb]]",
Dy = "[[Dysprosium|Dy]]",
Ho = "[[Holmium|Ho]]",
Er = "[[Erbium|Er]]",
Tm = "[[Thulium|Tm]]",
Yb = "[[Ytterbium|Yb]]",
Lu = "[[Lutetium|Lu]]",
Hf = "[[Hafnium|Hf]]",
Ta = "[[Tantalum|Ta]]",
W = "[[Tungsten|W]]",
Re = "[[Rhenium|Re]]",
Os = "[[Osmium|Os]]",
Ir = "[[Iridium|Ir]]",
Pt = "[[Platinum|Pt]]",
Au = "[[Gold|Au]]",
Hg = "[[Mercury (element)|Hg]]",
Tl = "[[Thallium|Tl]]",
Pb = "[[Lead|Pb]]",
Bi = "[[Bismuth|Bi]]",
Po = "[[Polonium|Po]]",
At = "[[Astatine|At]]",
Rn = "[[Radon|Rn]]",
Fr = "[[Francium|Fr]]",
Ra = "[[Radium|Ra]]",
Ac = "[[Actinium|Ac]]",
Th = "[[Thorium|Th]]",
Pa = "[[Protactinium|Pa]]",
U = "[[Uranium|U]]",
Np = "[[Neptunium|Np]]",
Pu = "[[Plutonium|Pu]]",
Am = "[[Americium|Am]]",
Cm = "[[Curium|Cm]]",
Bk = "[[Berkelium|Bk]]",
Cf = "[[Californium|Cf]]",
Es = "[[Einsteinium|Es]]",
Fm = "[[Fermium|Fm]]",
Md = "[[Mendelevium|Md]]",
No = "[[Nobelium|No]]",
Lr = "[[Lawrencium|Lr]]",
Rf = "[[Rutherfordium|Rf]]",
Db = "[[Dubnium|Db]]",
Sg = "[[Seaborgium|Sg]]",
Bh = "[[Bohrium|Bh]]",
Hs = "[[Hassium|Hs]]",
Mt = "[[Meitnerium|Mt]]",
Ds = "[[Darmstadtium|Ds]]",
Rg = "[[Roentgenium|Rg]]",
Cp = "[[Copernicium|Cp]]",
Nh = "[[Nihonium|Nh]]",
Fl = "[[Flerovium|Fl]]",
Mc = "[[Moscovium|Mc]]",
Lv = "[[Livermorium|Lv]]",
Ts = "[[Tennessine|Ts]]",
Og = "[[Oganesson|Og]]",
}

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_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
    local first = true

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

        if (first 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 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
        first = false
        return t, x
    end
end

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

    f = string.gsub(f, "–", "-")  -- replace – with - (hyphen not ndash)
    f = string.gsub(f, "−", "-")  -- replace – with - (hyphen not minus sign)

    local sumO = 0
    local formula = ''
    local t, x

    local link = args['link'] or ""
    local auto = args['auto'] or ""

    if not (link == '') then formula = formula .. "[[" .. link .. "|"; end   -- wikilink start [[link|
 
    for t, x in item(f) do 
        if t == T_ELEM then
              if (auto == '') then formula = formula .. x
              elseif am[x] then formula = formula .. am[x]; am[x] = x
              else formula = formula .. x
            end
        elseif t == T_COEFFICIENT then formula = formula .. x
        elseif t == T_NUM   then formula = formula .. su("", x);
        elseif t == T_OPEN  then formula = formula .. x; sumO = sumO + 1;        -- ( {
        elseif t == T_CLOSE then formula = formula .. x; sumO = sumO -1;         -- ) }
        elseif t == T_PM_CHARGE    then formula = formula .. su(string.gsub(x, "-", "−"), "");
        elseif t == T_SUF_CHARGE then 
            formula = formula .. su(string.gsub(string.match(x, "[+-]"), "-", "−"), string.match(x, "%d+"), "");
        elseif t == T_SUF_CHARGE2 then 
            formula = formula .. su(string.sub(string.gsub(string.match(x, "%(%d*[+-]"), "-", "−"), 2, -1), string.match(x, "%d+"))
        elseif t == T_CHARGE then formula = formula .. "<sup>"; if string.match(x, "%d+") then formula = formula .. string.match(x, "%d+"); end formula = formula .. string.gsub(string.match(x, "[%+-]"), "-", "−") .. "</sup>";  -- can not concatenat a nil value from string.match(x, "%d+");
        elseif t == T_CRYSTAL then formula = formula .. DotIt() .. string.gsub( x, "*", '', 1 );
        elseif t == T_SPECIAL then
            parameter = string.sub(x, 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 = string.sub(x, 2, 2) -- y fra \y{x} 
            if parameter  == "h" then --[[Hapticity]]
                if (auto == '') then formula = formula .. "η<sup>" .. string.match(x, '%d+') .. "</sup>-"
                else
                    formula = formula .. "[[Hapticity|η<sup>" .. string.match(x, '%d+') .. "</sup>]]-"
                end
            elseif parameter == "m" then formula = formula .. "μ<sub>" .. string.match(x, '%d+') .. "</sub>-" -- mu ([[bridging ligand]])
            end
        elseif t == T_WATER then 
            if string.match(x, "^%*[%d.]") then 
                formula = formula .. DotIt() .. string.match(x, "%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("", string.sub(x,3,-2)) -- x contains _{string}
        elseif t == T_CARET then formula = formula .. su(string.sub(x,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. E.g. wikilinks and pipes
     
        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

return p