local is_plant = { Tree = true, Grass = true, Bush = true }

local function vivify(map,key)
    if not map[key] then map[key] = {} end
    return map[key]
end

local function count_pop(ninfo, pop, div)
    ninfo.count = (ninfo.count or 0) + 1
    if pop.count_min ~= 10000001 then
        ninfo.min_sum = (ninfo.min_sum or 0) + pop.count_min
        ninfo.min_sum_d = (ninfo.min_sum_d or 0) + pop.count_min/div
    else
        ninfo.min_inf = (ninfo.min_inf or 0) + 1
    end
    if pop.count_max ~= 10000001 then
        ninfo.max_sum = (ninfo.max_sum or 0) + pop.count_max
        ninfo.max_sum_d = (ninfo.max_sum_d or 0) + pop.count_max/div
    else
        ninfo.max_inf = (ninfo.max_inf or 0) + 1
    end
end

local function report_pop(ninfo)
    return string.format(
        "count %d, sum %d, avg %f (%f), max sum %d, avg %f (%f)",
        (ninfo.count or 0),
        (ninfo.min_sum or 0),
        ((ninfo.min_sum or 0) / ((ninfo.count or 1) - (ninfo.min_inf or 0))),
        ((ninfo.min_sum_d or 0) / ((ninfo.count or 1) - (ninfo.min_inf or 0))),
        (ninfo.max_sum or 0),
        ((ninfo.max_sum or 0) / ((ninfo.count or 1) - (ninfo.max_inf or 0))),
        ((ninfo.max_sum_d or 0) / ((ninfo.count or 1) - (ninfo.max_inf or 0)))
    )
end

local function sort_keys(tab)
    local kt = {}
    for k,v in pairs(tab) do table.insert(kt,k) end
    table.sort(kt)
    return ipairs(kt)
end

function classify_population(info, pop, key, div)
    local ptype = df.world_population_type[pop.type]
    local tinfo = vivify(info, ptype)
    local name
    if is_plant[ptype] then
        name = df.plant_raw.find(pop.plant).id
    else
        name = df.creature_raw.find(pop.race).creature_id
    end
    local ninfo = vivify(tinfo, name)
    count_pop(ninfo, pop, div)
    local kinfo = vivify(vivify(ninfo, 'subcat'), key)
    count_pop(kinfo, pop, div)
end

local info = {}

for _,r in ipairs(df.global.world.world_data.regions) do
    local rtype = df.world_region.T_type[r.type]
    for _,p in ipairs(r.population) do
        classify_population(info, p, rtype, #r.region_coords.x)
    end
end

for _,poptype in sort_keys(info) do
    local tinfo = info[poptype]
    print(poptype..':')
    for _,name in sort_keys(tinfo) do
        local ninfo = tinfo[name]
        print('',name..':')
        for _,key in sort_keys(ninfo.subcat) do
            print('','',key,report_pop(ninfo.subcat[key]))
        end
        print('','','TOTAL',report_pop(ninfo))
    end
end