#!/usr/bin/env ruby

#############################################################################
# This example script shows custom parsing of API requests from a log file.
#############################################################################

require "log_stats"
require "log_stats/config"
require "json"
require "uri"

parse_time = Proc.new { |line| line[/\b20\d\d-\d\d-\d\dT\d\d:\d\d:\d\d/] }

def endpoint_pattern(url)
  Regexp.new('^' + url.gsub(%r{/:id/}, '/[^/]+/').gsub(%r{/:id$}, '/[^/]+') + '$')
end

ENDPOINTS = [
  'restapi.example.(se|no|dk)/api/user/:id/password',
  'restapi.example.(se|no|dk)/api/user/:id/password',
  'restapi.example.(se|no|dk)/api/user/find/(userName|email)/:id',
  'restapi.example.(se|no|dk)/api/order/extUserId/:id',
  'restapi.example.(se|no|dk)/api/voucher/.+'
].map do |url|
  {url: url, pattern: endpoint_pattern(url)}
end

def endpoint(api_call)
  uri = URI(api_call[:url])
  # NOTE: don't include digits in the path as then we will have too many unique paths
  path_without_digits = uri.path.gsub(%r{/\d+}, '/:id')
  url = uri.host + path_without_digits
  if endpoint = ENDPOINTS.detect { |endpoint| endpoint[:pattern].match(url) }
    endpoint[:url]
  else
    url
  end
end

def response_time_95(item)
  -1 * item[:fields][:response_time][:percentiles][0.95]
end

custom_config = {
  events: {
    requests: {
      # NOTE: matches Heroku router lines. Also matches Papertrails slightly modified lines.
      line_pattern: /\s(heroku\/router)|(heroku\[router\]:)\s/,
      fields: [
        {name: :time, parse: parse_time},
        {name: :method},
        {name: :host},
        {name: :path},
        {name: :status, numeric: true},
        {name: :code, optional: true},
        {name: :service, numeric: true}
      ],
      top_list_limit: 100,
      apdex: {tolerating: 500, frustrated: 2000},
      apdex_goal: 0.9,
      stats: false # Skip listing stats per request path to keep output size manageable
    },
    api_calls: {
      # 2017-02-19T06:21:25.522274+00:00 app[worker.2]: [WARN] [Vac::Request] Slow response time for url=http://sumore02.example.dk/api/search/categories/160145/assets/ method=get status=304 size= response_time=141
      line_pattern: /\s\[Vac::Request\] Slow response time\s/,
      fields: [
        {name: :time, parse: parse_time},
        {name: :url},
        {name: :method},
        {name: :response_time, numeric: true, events: true}
      ],
      group_by: {
        hostname: {
          id: Proc.new { |api_call| URI(api_call[:url]).host },
          sort_by: method(:response_time_95)
        },
        endpoint: {
          id: method(:endpoint),
          sort_by: method(:response_time_95)
        }
        # NOTE: Grouping by HTTP method doesn't really add much at the moment, so commenting out for now
        # method: {
        #   id: Proc.new { |api_call| URI(api_call[:method]) }
        # }
      },
      events: false, # Skip listing all individual API calls to keep output size manageable
      limit: 5
    }
  },
  verbose: true
}
config = LogStats::Config.default_config.
          merge(custom_config).
          merge(LogStats::Config.env_config)

log_file_data = LogStats::Logger.elapsed(config, "Reading file") { ARGF.readlines }
data = LogStats.get_data(log_file_data, config)
puts JSON.pretty_generate(data)