org chart scrape-builder from FM

gistfile1.rb

require 'rubygems'
require 'set'
require 'hpricot'
require 'mechanize'


USER_ACCOUNT   = "[email protected]"
PASSWORD       = ARGV[0]
MAX_LINKS      = 200
SLEEP_INTERVAL = 0.5


###########
# Helpers #
###########

module Enumerable
  def uniq_by
    h = {}
    inject([]) {|a,x| h[yield(x)] ||= a << x}
  end
end

class OmniscientHash < Hash
  def produce key  # default_proc for < 1.9
    self[key] = Person.new key unless self.has_key? key
    self[key]
  end
end

# Thought it would be neat to visually weight number of descendant
# reports to get a sense of departmental distribution.  End result
# not smooth curve as hoped -- report counts distributed too chunkily.
def make_heat_map max
  $colors = %w(magenta hotpink maroon crimson red orange yellow greenyellow steelblue cornflowerblue)
  quanta = (1..max-1).step(max / $colors.count.to_f).to_a.reverse  # e.g. [0.0, 9.16, 18.33...]
  quanta.map!{|x| x.floor}.reverse  # e.g. [165, 155, 146...]
  $heat_map = quanta.zip $colors
  $heat_map.push([0, "grey"])  # With zero reports, you get a boring color.
end

def render
  `dot -Tsvg graph.out > temp.svg`
  # git clone https://github.com/vidarh/diagram-tools
  `xsltproc diagram-tools/notugly.xsl temp.svg > graph.svg`
  #`rm graph.out temp.svg`
end


#########################################################################
# Screen-scrape the two worksimple pages that have direct links to each #
# FM employee. Stash reporting graph data in Person::PEOPLE.            #
#########################################################################

def harvest
  visited = []
  find_targets.uniq_by {|x| x.href}.each do |link|
    p 'Visited max links!' and break if visited.length > MAX_LINKS
    visited.push link

    this_guy, his_manager = search_for_people link

    # GOALLLLLLLLLL
    Person::PEOPLE.produce(this_guy).reports_to(his_manager)
    
    sleep SLEEP_INTERVAL
  end
  Person::PEOPLE.produce("Tom Davies").reports_to("Nate Perry-Thistle") # Tom's not in yet.
end

def find_targets
  agent = Mechanize.new
  page = agent.get 'https://federatedmedia.getworksimple.com/session/new'
  page = page.form_with :action => '/session' do |f|
    f.field_with(:name => "user_session[email]").value    = USER_ACCOUNT
    f.field_with(:name => "user_session[password]").value = PASSWORD
    f.field_with(:name => "authenticity_token").value     = f.authenticity_token
  end.submit

  page = page.links.select {|x| x.text == "Teams" }.first.click # Page 1 of listing.
  targets = page.links.select {|x| x.href.match "/users/" }

  page = page.links.select {|x| x.text.match "Next" }.first.click # Page 2 of listing.
  targets.push(page.links.select {|x| x.href.match "/users/" }).flatten!

  targets.reject! {|x| x.text.match "Profile" } # Snip my account edit pages.
end

def search_for_people link
  h = Hpricot link.click.content

  # \nThis  Guy\n => This Guy
  this_guy = h.search("div.name").innerHTML.split("\n").last.squeeze(" ")

  # \n\nHis Manager\n => His Manager
  his_manager = h.search('div.reviewer a.user').text.split("\n").last

  # Workaround Amy Yeh's zombied entry ("/users/7894").
  his_manager = "Charlie Speight" if his_manager == "Amy Yeh"

  puts "#{this_guy}'s manager is #{his_manager}"
  [this_guy, his_manager]
end


#######################################################################
# Model for people, with each person holding a list of their reports. #
#######################################################################

class Person
  attr_accessor :name, :pk, :reports
  PEOPLE = OmniscientHash.new

  def initialize name
    @name        = name
    @pk          = name.downcase.gsub(/[ '-]/, "_")  # Punctuated names break DOT. :(
    @reports     = []
    PEOPLE[name] = self
  end

  def reports_to name
    PEOPLE.produce(name).reports << self if name
  end

  # Recurse over person's reports, return total top-down count.
  def self.count_sub_reports person
    return 0 if person.reports.count == 0
    count = 0
    # Wants to be a list comprehension so bad.
    person.reports.each do |r|
      count += 1 + count_sub_reports(r)
    end
    count
  end
end


##########################################################################
# Generate a graphviz chart showing the reporting structure described in #
# Person::PEOPLE. The three basic DOT concerns -- points, edges and      #
# boxes -- fail to parse in our top-down rendering if intermixed,        #
# requiring that we sort by type.                                        #
##########################################################################

class OrgChart
  def initialize root_person
    @seen = Set.new

    @points, @edges, @boxes = [], [], []

    node root_person
    walk_reports_of root_person

    emit_DOT
  end

  def node person
    return false if @seen.member? person.pk
    @seen << person.pk

    color = apply_heat_map person
    shape = person.reports.empty? ? "box" : "oval"
    
    @boxes << %Q{ p#{person.pk}  [ shape = #{shape}, style="filled", fillcolor="#{color}", label="#{person.name}" ] }
    @boxes << %Q{ p#{person.pk+"1"}  [ shape = #{shape}, style="filled", fillcolor="#{color}", label="#{person.name}" ] }
    @boxes << %Q{ p#{person.pk+"2"}  [ shape = #{shape}, style="filled", fillcolor="#{color}", label="#{person.name}" ] }
    @boxes << %Q{ p#{person.pk+"3"}  [ shape = #{shape}, style="filled", fillcolor="#{color}", label="#{person.name}" ] }
    @boxes << %Q{ p#{person.pk+"4"}  [ shape = #{shape}, style="filled", fillcolor="#{color}", label="#{person.name}" ] }
  end

  def walk_reports_of person
    person.reports.each do |r|
      if [email protected]? r
        @points << " p#{person.pk} [ shape = point ]"
        @edges  << " p#{person.pk} -> p#{r.pk}"
      end

      walk_reports_of r if node r
    end
  end

  def apply_heat_map person
    count = Person.count_sub_reports person
    num, hex = $heat_map.find { |num, hex| count >= num }

    hex
  end

  def emit_DOT
    open("graph.out", "w") do |f|
      f.write "digraph orgchart {"
      [@points, @edges, @boxes].map {|x| f.write(x.join("\n")) }
      f.write "}"
    end
  end
  
end


#######
# Go. #
#######

def main
  # harvest
  # make_heat_map Person.count_sub_reports(Person::PEOPLE["John Battelle"])
  # OrgChart.new Person::PEOPLE["John Battelle"]
  p = YAML::load(open("marshalled.txt"))
  make_heat_map Person.count_sub_reports(p["John Battelle"])
  OrgChart.new p["John Battelle"]
  
  render
end


# ruby graph.rb [PASSWORD]
main