Quick and dirty glassfish log parser and stat analyzer

glassparse.rb

require 'date'
require 'colorize'
# Parse glassfish default log files 
# "X.X.X.X" "NULL-AUTH-USER" "22/Jul/2013:09:39:33 +0000" "POST /url/address/id HTTP/1.1" 200 7 

# group things inside "" (?<=")(.*?)(?=")
# group things inside whitespaces (?<=\s)(.*?)(?=\s)
# last two digits parser

class LogLine 
	attr_accessor :ip,:timestamp,:action,:post,:reply_code,:reply_size,:user	
	def set_time timestring
		@timestamp = DateTime.strptime(timestring,"%d/%b/%Y:%H:%M:%S %z")
	end
end


def self.incr_hash_value hash,key
	if hash.include? key 
		v = hash[key]
		hash[key] = (v+1)
	else
		hash[key] = 1
	end
end


def self.readfile filename
	log_lines =[]
	lines = 0
	IO.foreach(filename) do |log_line|
		line_sections_raw = log_line.scan(/(?<=")(.*?)(?=")/i)
		reply_part = log_line.scan(/".*"\s{1}(\d*)\s{1}(\d*)/)
		line_sections = []
		line_sections_raw.each do |line_section|		
			if line_section.first.size >= 2
				line_sections << line_section.first
			end
		end
		ll = LogLine.new()
		ll.ip = line_sections[0]
		ll.user = line_sections[1]
		ll.set_time line_sections[2]
		ll.action = line_sections[3]
		ll.reply_code = reply_part[0][0]
		ll.reply_size = reply_part[0][1]
		log_lines << ll
		lines += 1
		print "Reading line:#{lines} from #{filename}\r".white
	end
	puts
	return log_lines
end

def self.hits_per_min log_lines
	hash_ip_minutes = {}

	log_lines.each do |log_line|
		key = log_line.timestamp.strftime("%d%m%Y%H%M")
		self.incr_hash_value hash_ip_minutes,key
	end
	per_min_hits = []
	hash_ip_minutes.each_key do |key|
		tstamp = DateTime.strptime(key,"%d%m%Y%H%M")
		hits = hash_ip_minutes[key]
		str = "TIME: #{tstamp}".yellow 
		str += " HITS: #{hits}".red
		per_min_hits << hits
		puts str
	end
	size = per_min_hits.size
	avg_hits = per_min_hits.inject{ |sum, el| sum + el }.to_f / size
	puts "\n >>> AVG HITS x MINUTE: #{avg_hits.round} <<<".white
	hash_ip_minutes = hash_ip_minutes.sort_by{|k,v| v.to_i}.reverse
	puts "\n*** TOP 10 HITTED MINUTES***".yellow 
	if hash_ip_minutes.size > 10 
		10.times do |i|
			time_hits = hash_ip_minutes[i]
			time = DateTime.strptime(time_hits[0],"%d%m%Y%H%M")
			msg = "#{i+1} -> TIME: #{time} ".white 
			msg += "HITS: #{time_hits[1]}".yellow
			puts msg
		end
	else
		i = 1
		hash_ip_minutes.each do |time_hit|
			time = DateTime.strptime(time_hits[0],"%d%m%Y%H%M")
			msg = "#{i} -> TIME: #{time} ".white 
			msg += "HITS: #{time_hit[1]}".yellow
			i += 1
			puts msg
		end	
	end
end

def self.hits_per_ip log_lines
	hash_ip = {}
	log_lines.each do |log_line|
		key = log_line.ip
		self.incr_hash_value hash_ip,key
	end

	hash_ip.each_key do |key|		
		hits = hash_ip[key]
		str = "IP: #{key}".yellow 
		str += " HITS: #{hits}".red
		puts str
	end		
end

def self.reply_code_analysis log_lines
	bad_replies = {}
	good_replies = {}	
	log_lines.each do |ll|
		key = ll.ip
		if ll.reply_code == "200"
			self.incr_hash_value good_replies,key
		else 
			self.incr_hash_value bad_replies,key
		end
	end
	t_good = 0
	t_bad = 0	
	bad_replies.each_key do |key|
		t_bad += bad_replies[key]
	end
	good_replies.each_key do |key|
		t_good += good_replies[key]
	end
	total_reqs = t_good+t_bad
	per_good = (t_good*100.0)/total_reqs
	per_bad = (t_bad*100.0)/total_reqs
	msg = "TOTAL KO: #{t_bad}\t".red
	msg += "TOTAL OK: #{t_good}\t".green
	msg += "PERC KO: #{per_bad}%".yellow
	bad_replies = bad_replies.sort_by{|k,v| v.to_i}
	good_replies = good_replies.sort_by{|k,v| v.to_i}
	bad_replies = bad_replies.reverse
	good_replies = good_replies.reverse

	puts "** TOP TEN OK CLIENTS: **".green
	puts
	if good_replies.size <= 10
		i = 1
		good_replies.each do |rep|
			puts "#{i} -> IP: #{rep[0]} REQS: #{rep[1]}".green
		end
	else
		10.times do |x|
			rep = good_replies[x]
			puts "#{x+1} -> IP: #{rep[0]} OK REQS: #{rep[1]}".green
		end
	end
	puts
	puts "** TOP TEN KO CLIENTS: **".red
	puts
	if bad_replies.size <= 10
		i = 1
		bad_replies.each do |rep|
			puts "#{i} -> IP: #{rep[0]} REQS: #{rep[1]}".red
		end
	else
		10.times do |x|
			rep = bad_replies[x]
			puts "#{x+1} -> IP: #{rep[0]} REQS: #{rep[1]}".red
		end		
	end
	puts
	puts msg
end


if ARGV.size == 0
	puts "No arguments, dude? Are you fucking crazy?! C'mon!".upcase.red
	puts "I need an access filename to work.".white
	exit
end

@columns = `/usr/bin/env tput cols`.to_i
breakline = "=".green*@columns
filename = ARGV.first


#while true
	log_lines = self.readfile filename
	puts "HITS per MINUTE".blue
	puts breakline
	self.hits_per_min log_lines
	puts
	#puts "IPs and HITS"
	#puts breakline
	#self.hits_per_ip log_lines
	puts "REPLIES".blue
	puts breakline
	self.reply_code_analysis log_lines
	#sleep 60
#end