har2csv.rb

#!/usr/bin/env ruby
#--
# Copyright 2016 by Fernando Ike <fike@midstorm.org>.
# All rights reserved.
# This script is under LGPL-2
#
# har2csv exports some har headers and values to csv.  
#++

require 'addressable/uri'
require 'json'
require 'yaml'
require 'csv'

if ARGV.length != 1
  puts "#{__FILE__} /path/to/archive.har"
  exit 1
end

archive = nil

csv_file = (ARGV[0]).gsub(/\.har$/, '.csv')

begin
  archive = JSON.parse(File.read(ARGV.shift))
rescue => err
  puts "could not parse archive file: #{err.to_s}"
end

entries = archive['log']['entries']

File.open(csv_file, 'a+') do |f|
  if f.tell() == 0
    f << "host,url,content-type,content-encoding,cache-control,expires,size\n"
  end

  urls = entries.map do |entry|

    cache_control = []
    expires = []
    content_type = []
    content_encoding = []
    content_length = []
    host = Addressable::URI.parse(entry['request']['url']).host
    url = entry['request']['url']
    res_headers = entry['response']['headers'].map do |res_header|
      if res_header['name'].match(/[Cc]ache\-[Cc]ontrol/)
        cache_control = res_header['value']
      elsif res_header['name'].match(/[Ee]xpires/)
        expires = res_header['value']
      elsif res_header['name'].match(/[Cc]ontent\-[Tt]ype/)
        content_type = res_header['value']
      elsif res_header['name'].match(/[Cc]ontent\-[Ee]ncoding/)
        content_encoding = res_header['value']
      elsif res_header['name'].match(/[Cc]ontent\-[Ll]ength/)
        content_length = res_header['value']
      end

    end
    res_headers.reject! { |url| url.nil? }
    f << "\"#{host}\",\"#{url}\",\"#{content_type}\",\"#{content_encoding}\",\"#{cache_control}\",\"#{expires}\",\"#{content_length}\"\n"
  end
end