142 lines
3.2 KiB
Ruby
142 lines
3.2 KiB
Ruby
# frozen_string_literal: true
|
|
|
|
require 'date'
|
|
require_relative 'utils/http_codes'
|
|
require_relative 'utils/date_for_humans'
|
|
|
|
args = ARGV
|
|
|
|
if args[0] == '--parse' || args[0] == '-p'
|
|
puts("[!] Opening log file at #{args[1]}")
|
|
|
|
begin
|
|
logfile = File.open(args[1])
|
|
rescue Errno::ENOENT
|
|
warn "[x] The file #{args[1]} was not found."
|
|
exit 1
|
|
end
|
|
|
|
else
|
|
puts('[!] Log file path not provided, assuming default access.log')
|
|
|
|
begin
|
|
logfile = File.open('access.log')
|
|
rescue Errno::ENOENT
|
|
warn '[x] Default file access.log not found. Specify an alternative log file to parse with -p or --parse.'
|
|
exit 1
|
|
end
|
|
|
|
end
|
|
|
|
data = logfile.read
|
|
lines = data.split("\n")
|
|
|
|
|
|
# Gets the IP address from the string
|
|
# Note: sometimes the function can fetch the UA version as an IP address
|
|
def get_line_ip(line)
|
|
pattern = /\b(?:\d{1,3}\.){3}\d{1,3}\b/
|
|
line.match(pattern)[0] if line.match(pattern)
|
|
end
|
|
|
|
# TODO: Implement visits per month using this method
|
|
# Returns the date in the current line
|
|
def get_line_date(line)
|
|
pattern = /\[(.*?)\]/
|
|
match = line.match(pattern)[0] if line.match(pattern)
|
|
|
|
# [12/Apr/2023:13:56:41 +0100] -> 12/Apr/2023:13:56:41 +0100
|
|
Date.parse(match.gsub('[', '').gsub(']', ''))
|
|
end
|
|
|
|
# Gets the HTTP status code of the given log line
|
|
def get_line_code(line)
|
|
pattern = /\s(\d{3})\s/
|
|
line.match(pattern)[1] if line.match(pattern)
|
|
end
|
|
|
|
# Gets the user agent of the given log line
|
|
def get_line_browser(line)
|
|
pattern = /"([^"]*)"$/
|
|
user_agent = line.match(pattern)[0] if line.match(pattern)
|
|
|
|
user_agent.gsub('"', '')
|
|
end
|
|
|
|
# Gets the number of times an IP contacted the site
|
|
def times_appeared_single(ips, ip_to_check)
|
|
counter = 0
|
|
ips.each do |ip|
|
|
counter += 1 if ip == ip_to_check
|
|
end
|
|
counter
|
|
end
|
|
|
|
# Returns all unique IPs in a given list of IP addresses with duplicates
|
|
def sort_unique_ip(ips)
|
|
|
|
seen_ips = {}
|
|
unique_ips = []
|
|
|
|
ips.each do |ip|
|
|
unless seen_ips[ip]
|
|
unique_ips << ip
|
|
seen_ips[ip] = true
|
|
end
|
|
end
|
|
unique_ips
|
|
end
|
|
|
|
visit_counter = {}
|
|
client_errors = {}
|
|
browsers = {}
|
|
all_ips = []
|
|
|
|
lines.each do |line|
|
|
all_ips << get_line_ip(line)
|
|
end
|
|
|
|
unique_ips = sort_unique_ip(all_ips)
|
|
|
|
puts("There were a total of #{unique_ips.length} unique IPs who connected to our site.")
|
|
|
|
unique_ips.each do |ip|
|
|
visit_counter[ip] = times_appeared_single(all_ips, ip)
|
|
puts("IP #{ip} contacted our site #{times_appeared_single(all_ips, ip)} times.")
|
|
end
|
|
|
|
lines.each do |line|
|
|
ip = get_line_ip(line).to_s
|
|
code = get_line_code(line)
|
|
browser = get_line_browser(line)
|
|
# refactored from "ua" to "browser"
|
|
|
|
if is_client_err?(code.to_i)
|
|
if client_errors[ip]
|
|
client_errors[ip] += 1
|
|
else
|
|
client_errors[ip] = 1
|
|
end
|
|
end
|
|
|
|
# refactored from: "user_agents[ua]" to "browsers[browser]"
|
|
if browsers[browser]
|
|
browsers[browser] += 1
|
|
else
|
|
browsers[browser] = 1
|
|
end
|
|
|
|
end
|
|
|
|
top_browsers = browsers.sort_by { |_ua, count| -count }.first(5)
|
|
puts 'Top 5 browsers contacting the site:'
|
|
top_browsers.each do |ua, count|
|
|
puts "#{ua}: #{count} visits"
|
|
end
|
|
|
|
top_client_errors = client_errors.sort_by { |_ip, count| -count }.first(5)
|
|
puts 'Top 5 IPs with most client errors (400-499):'
|
|
top_client_errors.each do |ip, count|
|
|
puts "#{ip}: #{count} errors"
|
|
end
|