A logfile stats aggregator hack

mywebapp_stats.rb

unless ARGV.length == 2
  raise <<EOF


Usage:   mywebapp_stats.rb folder filepattern

Example: mywebapp_stats.rb . AccountWeb.log*

EOF
end

$user_hash = {}
$page_hash = {}
$report_hash = {}

def main
  log_dir = ARGV[0]
  files_pattern = ARGV[1]

  if File.directory?(log_dir)
    Dir.chdir log_dir
  else
    raise "#{log_dir} is not a directory"
  end

  Dir.glob(files_pattern).each { |f| parse_file f }
  
  report
end

### ANALYSE FILE STUFF ###############################################

def parse_file filepath
  puts "Analysing file #{filepath}"
  File.open(filepath, "r") do |f|
    while (line = f.gets)      
      reg_count(/Loading report "(.*)"/, line, $report_hash)
      reg_count(/\| (\d+)\/(.*)\/(PROD[123])/, line, $user_hash) do |m| 
        m[0][1].ljust(20) + m[0][2] + "  ID:" + m[0][0]
      end
      reg_count(/ ([A-Za-z]+)\.aspx \(loading\.\.\)/, line, $page_hash)      
    end
  end
end

def reg_count pattern, line, hash, &block
  match = line.scan pattern
  if match.length > 0
    key = block.call(match) if block
    inc_hash! key || match[0][0], hash 
  end
end 

def inc_hash! k, h
  n = h[k] || 0
  h[k] = n + 1
end

### REPORT OUTPUT STUFF ###############################################

def report
  report_hash "Page", "Hits", $page_hash
  report_hash "User", "Hits", $user_hash
  report_hash "Report", "Hits", $report_hash
end

def report_hash col_a, col_b, h
  report_header col_a, col_b
  h.
    sort{|a,b| a[1] <=> b[1]}.
    reverse.
    each { |key, value| report_line(key, value) }
end

def report_header col_a, col_b
  puts
  report_line col_a, col_b  
  puts "-" * 60
end

def report_line a, b
  puts a.ljust(50) + b.to_s.rjust(10)
end

main # run the program