clean data, present the data with graph and chart(graphviz,scruffy) #practical ruby for system administrator

Ruby: clean data, present the data with graph and chart(graphviz,scruffy).rb

# attempting to present data without thought as to the analysis often gives rise to 
# presentations that are nothing more than stylish gloss on a stinky foundation.
# 其实这一章没有太多好用的类库介绍，基本是使用。ruby的使用
# 回去可以整理下。

# Listing 9-8. Storing the Array of Logins in a Timestamped File for Later Retrieval
require "fileutils"
require "remote_host"
require "yaml"
FileUtiles.mkdir_p("data")
Dir.chdir("data")
hosts_to_monitor = ["fishbrain.local", "fishwife.local"]
hosts_to_monitor.each do |host|  logins = RemoteHost.new(host).last_logins
  dump_file = host + Time.now.strftime(".%Y%m%dT%H%M%S.yaml")
  File.open(dump_file, "w") { |f| YAML.dump(logins, f) }
end

# Listing 9-9. Parsing Login Fields into an Event, Particularly to Coerce the Date Information
DAY_INDEX, NOW = {}, Time.now
["Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"].each_with_index do |day, i|
  DAY_INDEX[day] = i + 1
end
def parse_event(login_fields, host)
  user, device, source, date_info = login_fields
  date_info =~ /^(...) (.){12}/
  wday, date = DAY_INDEX[$1], Time.parse($2)
  while wday != date.wday or date > NOW
    date = Time.local(date.year - 1, date.month, date.day, date.hour, date.min)
  end
  duration = nil
  if date_info =~ /\((\d+\+)?(\d+):(\d+)\)$/
    days, hours, mins = $1, $2, $3
    duration = mins + hours * 60 + days.to_i * 24 * 60
  end
  {:user => user, :device => device, :source => source,
   :date => date, :duration => duration, :host => host}
end


# Listing 9-10. Assigning Events to a Unique 24-Hour Event Bin
def assign_event(event, bin_hash)
  date_key = event[:date].strftime("%Y%m%d")
  bin = (bin_hash[date_key] ||= {})
  event_key = event.values_at(:user, :device, :source, :date, :host).join(";")
  bin[event_key] = event
end


#Listing 9-11. The Final Event Processing Script
require "yaml"
data_dir = (ARGV[0] || "data")
fail "couldn't find directory #{data_dir}" unless File.directory?(data_dir)
Dir.chdir(data_dir)
Dir.mkdir("processed") unless File.directory?("processed")
# code from Listings 9-9 and 9-10 would go here
data_files, periods = DIR["*.yaml"], {}
data_files.each do |file|
  host = File.basename(file).split(".").first
  logins = YAML.load_file(file)
  logins.each do |login|
    event = parse_event(login, host)
    assign_event(event, periods)
  end
end
periods.each do |date_string, events|
  event_file = File.join("processed", date_string + ".yaml")
  if File.exist?(event_file)
    old_events = YAML.load_file(event_file)
    events = old_events.merge(events)
  end  File.open(event_file, "w") { |f| YAML.dump(events, f) }
end
File.delete(data_files)

#Listing 9-12. Summarizing the Total Session Duration Amounts by User and Then by Date
require "yaml"
source_dir = ARGV[0]
fail "couldn't find directory #{source_dir}" unless File.directory?(source_dir)
users = {}
Dir[File.join(source_dir, "*")].each do |file|
  date_string = File.basename(file, ".yaml")
  events = YAML.load_file(file)
  events.each do |key, event|
    duration = event[:duration].to_i
    next unless duration > 0
    user = (users[event[:user]] ||= Hash.new(0))
    user[date_string] += duration
  end
end


# using Scruffy Charts
# SVG Format
# For rendering a graph, scruffy relies upon the RMagick library (itself a Ruby interface to the  ImageMagick library). 


require "scruffy"
include Scruffy
graph = Graph.new(:title => "Average Flatulence per Capita")

graph.add(:line, "Cows", [150, 200, 250, 240])
graph.add(:area, "Sheep", [100, 110, 120, 140])
graph.add(:bar, "Man", [50, 65, 78, 66])
graph.render(:as => "PNG", :size => [500, 400], :to => "/tmp/flatulence.png")


require "scruffy"
include Scruffy
# code from Listing 9-12 would go here
exit if users.empty?

all_dates = users.map { |user, dates| dates.keys }.flatten.uniq.sort
graph = Graph.new(:point_markers => all_dates)
graph.title = "Total Login Minutes per User"

users.each do |user, dates|
  all_values = all_dates.map { |date| dates[date].to_i }
  graph.add(:line, user, all_values)
end

graph.render(:as => "PNG", :width => 1024, :to => "/tmp/user_accounting.png")

# CSS Charts in Rails
#  The CSS box model allows for doing pretty basic rectangular drawing, including background, border, and alignment facilities.
# $ script/plugin install http://topfunky.net/svn/plugins/css_graphs
# $ script/generate css_graphs
# edit rhtml templates using the following pattern:
=begin
<h2>Horizontal Bar Chart (Complex)</h2>
<%= complex_bar_graph(["Mail", 58], ["Web", 38], ["FTP", 29], ["P2P", 25], ["Other", 24]) %>
=end

# Graph
# Using graphviz language - dot
# the Ruby wrapper - ruby-graphviz
$ tar -xzf ruby-graphviz_0.6.0.tar.gz
$ cd ruby-graphviz
$ ruby extconf.rb

require "graphviz"
# code from Listing 9-12 modified as discussed (for indexing by host) would go here
exit if users.empty?
graph = Graphviz.new("G")
users.each do |user, hosts|
  graph.add_node(user)
  hosts.each do |host, usage|
    graph.add_node(host)
    graph.add_edge(user, host, "weight" => usage)
  end
end
graph.output("output" => "png", "file" => "/tmp/user_relationships.png")