#!/usr/bin/lua

function parseline(line)
	local word
	local words = {}
	for word in string.gmatch(line, "%S+") do
		words[#words+1] = word
	end

	return {
		logdatetime=words[1],
		elapsed=words[2],
		clientip=words[3],
		code=string.match(words[4] or "", "^[^/]*"),
		status=string.match(words[4] or "", "[^/]*$"),
		bytes=tonumber(words[5]),
		method=words[6],
		URL=words[7],
		site = string.match(words[7], "^(%a+://[^/]*)"),
		clientuserid=words[8],
		peerstatus=string.match(words[9] or "", "^[^/]*"),
		peerhost=string.match(words[9] or "", "[^/]*$"),
		squarkcategory=string.match(words[11] or "", "^[^,]*"),
		squarkaction=string.match(words[11] or "", ",([^,]+)$")
	}

end

function add_by_subkey(tbl, key, entry)
	if key == nil then
		return tbl
	end
	if tbl[key] == nil then
		tbl[key] = {
			lines = entry.lines,
			bytes = entry.bytes,
			elapsed = entry.elapsed,
		}
		return tbl
	end
	tbl[key].lines = (tbl[key].lines or 0) + 1
	tbl[key].bytes = tbl[key].bytes + entry.bytes
	tbl[key].elapsed = tbl[key].elapsed + entry.elapsed
	return tbl
end


function add_stats(total, subkeys, entry)
	local _,k
	for _,k in pairs(subkeys) do
		if entry[k] ~= nil then
			add_by_subkey(total[k], entry[k], entry)
		end
	end

	total.elapsed = total.elapsed + entry.elapsed
	total.bytes = total.bytes + entry.bytes
	total.lines = total.lines + 1
	return total
end

local function newstats(groupname, subkeys)
	local t = {
		bytes = 0, lines = 0, elapsed = 0,
		group = groupname,
	}
	local _,k
	for _,k in pairs(subkeys) do
		t[k] = {}
	end
	return t
end

--		code = {}, category = {}, action = {},
local function add_stats_by_key(tbl, key, subkeys, entry)
	if key == nil then
		return tbl
	end
	if tbl[key] == nil then
		tbl[key] = newstats(key, subkeys)
	end
	add_stats(tbl[key], subkeys, entry)
	return tbl
end

function collect_stats(stats, keytbl, subkeys, entry)
	local _,k
	add_stats(stats.total, subkeys, entry)
	for _,k in pairs(keytbl) do
		add_stats_by_key(stats[k], entry[k], subkeys, entry)
	end
	return stats
end

function parsefile(file, keytbl, subkeys)
	local numlines = 0
	local f = assert(io.open(file))
	local stats = {}
	subkeys = subkeys or {"code", "squarkcategory", "squarkaction" }
	stats.total = newstats("total", subkeys)
	local _,k
	for _,k in pairs(keytbl or {"site"}) do
		stats[k] = {}
	end

	local line
	for line in f:lines() do
		collect_stats(stats, keytbl, subkeys, parseline(line))
	end
	f:close()
	return stats
end

function dump_tbl(header, tbl)
	io.write(header..":\n")
	local k,v
	for k,v in pairs(tbl) do
		print("\t"..k..":", v.lines, v.bytes, v.elapsed)
	end
end

function print_totals(header, tbl)
	local k, v
	print("Totals for "..header..":")
	for k,v in pairs(tbl) do
		if type(v) == "table" then
			dump_tbl(k, v)
		else
			io.write(k..": "..v.."\n")
		end
	end
	io.write("\n")
end

function sort_by_func(tbl, func)
	local t = {}
	for k, v in pairs(tbl) do
		table.insert(t, v)
	end
	table.sort(t, func)
	return t
end

function sort_by_key(tbl, key)
	return sort_by_func(tbl, function(a,b)
		return a[key] > b[key]
	end)
end

function top_n_by_func(tbl, n, sortfunc, callback)
	sorted = sort_by_func(tbl, sortfunc)
	local i
	for i = 1, #sorted do
		callback(sorted[i])
		if i > n then
			break
		end
	end
end

function top_n_by_key(tbl, n, statskey)
	top_n_by_func(tbl, n,
		function(a, b)
			return a[statskey] > b[statskey]
		end,
		function(rec)
			print(rec.group, rec[statskey])
		end)
end


stats = parsefile(arg[1] or "/var/log/squid/access.log",
	{"site"},
	{"code"})


print("= Top 10 sites by hits =")
top_n_by_key(stats.site, 10, "lines")

print("= Top 10 sites by TCP_HIT bytes =")
top_n_by_func(stats.site, 10,
	function(a, b)
		local an, bn = 0, 0
		if a.code.TCP_HIT and a.code.TCP_HIT.bytes then
			an = a.code.TCP_HIT.bytes
		end
		if b.code.TCP_HIT and b.code.TCP_HIT.bytes then
			bn = b.code.TCP_HIT.bytes
		end
		print(an, bn)
		return an > bn
	end,
	function(rec)
		print(rec.group, rec.code.TCP_HIT.bytes)
	end)