initial commit

1 files changed, 174 insertions, 0 deletions

diff --git a/squlogan b/squlogan
new file mode 100644
index 0000000..b337351
--- /dev/null
+++ b/squlogan
@@ -0,0 +1,174 @@
+#!/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 "", "[^/]*$"),
+		bytes=words[5],
+		method=words[6],
+		URL=words[7],
+		site = string.match(words[7], "^(%a+://[^/]*)"),
+		clientuserid=words[8],
+		peerstatus=string.match(words[9] or "", "^[^/]*"),
+		peerhost=string.match(words[9] or "", "[^/]*$"),
+		squarkcategory=string.match(words[11] or "", "^[^,]*"),
+		squarkaction=string.match(words[11] or "", ",([^,]+)$")
+	}
+
+end
+
+function add_by_key(tbl, key, entry)
+	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, entry)
+	add_by_key(total.code, entry.code, entry)
+	add_by_key(total.category, entry.squarkcategory, entry)
+	if entry.squarkaction then
+		add_by_key(total.action, entry.squarkaction, entry)
+	end
+--	inc_key(total.squark, entry.squark)
+
+	total.elapsed = total.elapsed + entry.elapsed
+	total.bytes = total.bytes + entry.bytes
+	total.lines = total.lines + 1
+	return total
+end
+
+local function newstats(keyname)
+	return {
+		bytes = 0, lines = 0, elapsed = 0,
+		code = {}, category = {}, action = {},
+		key = keyname,
+	}
+end
+
+local function add_stats_by_key(tbl, key, entry)
+	if tbl[key] == nil then
+		tbl[key] = newstats(key)
+	end
+	add_stats(tbl[key], entry)
+	return tbl
+end
+
+function collect_stats(stats, entry)
+	add_stats(stats.total, entry)
+	add_stats_by_key(stats.user, entry.clientuserid, entry)
+	add_stats_by_key(stats.code, entry.code, entry)
+	add_stats_by_key(stats.site, entry.site, entry)
+	return stats
+end
+
+function parsefile(file)
+	local numlines = 0
+	local f = assert(io.open(file))
+	local stats = {
+		total = newstats(),
+		user = {},
+		site = {},
+		code = {},
+	}
+	local line
+	for line in f:lines() do
+		collect_stats(stats, 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_key(tbl, key)
+	local t = {}
+	for k, v in pairs(tbl) do
+		table.insert(t, v)
+	end
+	table.sort(t, function(a,b)
+		return a[key] > b[key]
+	end)
+	return t
+end
+
+function top_n_by_key(tbl, n, key)
+	sorted = sort_by_key(tbl, key)
+	local i
+	for i = 1, #sorted do
+		print(sorted[i].key, sorted[i][key])
+		if i > n then
+			break
+		end
+	end
+	print()
+end
+
+stats = parsefile(arg[1] or "/var/log/squid/access.log")
+
+--[[
+numusers = 0
+for user,tbl in pairs(stats.user) do
+	print_totals("user "..user, tbl)
+	numusers = numusers + 1
+end
+--]]
+
+
+--[[
+numsites = 0
+for site, tbl in pairs(stats.site) do
+	print_totals("site "..site, tbl)
+	numsites = numsites + 1
+end
+--]]
+
+print("Top 10 sites by hits")
+top_n_by_key(stats.site, 10, "lines")
+print()
+
+
+print("Top 10 sites by bytes")
+top_n_by_key(stats.site, 10, "bytes")
+print()
+
+print("Top 5 users by bytes")
+top_n_by_key(stats.user, 5, "bytes")
+print()