gh_archive_challenge

gh_repo_stats.rb

#!/usr/bin/env ruby
#
# Gist Leaderboard Challenge
# By Chris Dobler 2016
#
# TODO: add defaults for when no args are sent, add storage of event type for multiple event types, 
#       use big query to increase speed, move request to seperate thread,

require 'optparse'
require 'open-uri'
require 'zlib'
require 'json'
require 'yajl'
require 'time'


##
## 	 Gh_archive
##
## 	 This object can connect to the github archive, and store a list of events. This object can also
##   display a list of formatted events.
##################################

class Gh_archive
	attr_reader :options, :counter, :gz, :combos

	def initialize
		@combos = Array.new
		@events = Array.new
		@gz = nil
		@counter = 0
		@options = {}

		opt_parser = OptionParser.new do |opt|
		  opt.banner = "Usage: gh_repo_stats [--after DATETIME] [--before DATETIME] [--event EVENT_NAME] [-n COUNT]"
		  opt.separator  ""
		  opt.separator  "Options"
		  opt.separator  ""
		  opt.on("--after DATETIME","The starting DATETIME") do |datetime|
		    @options[:after] = datetime
		  end
		  opt.on("--before DATETIME","The ending DATETIME") do |datetime|
		    @options[:before] = datetime
		  end
		  opt.on("--event EVENT_NAME","The event name") do |event_name|
		    @options[:event_name] = event_name
		  end
		  opt.on("-n","--count COUNT","The number of records to retrieve") do |count|
		    @options[:count] = count
		  end
		  opt.on("-h","--help","help") do
		    puts opt_parser
		  end
		end
		opt_parser.parse!
	end

	def connect
		puts "Retrieving archives for the selected range. This could take a while..."
		index = 0

	#TODO: Clean up this section.
		before_date = Date._strptime(@options[:before], '%FT%T%z')
		before = DateTime.strptime(before_date[:year].to_s+','+before_date[:mon].to_s+','+before_date[:mday].to_s+','+before_date[:hour].to_s, '%Y,%m,%e,%k')
		after_date = Date._strptime(@options[:after], '%FT%T%z')
		after = DateTime.strptime(after_date[:year].to_s+','+after_date[:mon].to_s+','+after_date[:mday].to_s+','+after_date[:hour].to_s, '%Y,%m,%e,%k')
		hour_step = (1.to_f/24)
		before.step(after,hour_step).each{|e| @combos.push [e.year, e.month, e.day, e.hour]}

		total = @combos.count
		@combos.each do | year, month, day, hour |
			percent = index.to_f/total.to_f*100
			puts "["+percent.to_i.to_s+"%]..."
			index = index + 1
			url = 'http://data.githubarchive.org/'+year.to_s+'-'+'%02d' % month.to_s+'-'+'%02d' % day.to_s+'-'+'%02d' % hour.to_s+'.json.gz'
			@gz = open url
			js = Zlib::GzipReader.new(@gz).read

			Yajl::Parser.parse(js) do |event|
				exists = 0
		  	if event['type'] == @options[:event_name] then
				  @events.each do |stored_event|
				  	if stored_event.name == event['repo']['name'] then 
				  		exists = 1
			  		#increment 
			  			stored_event.event_count = 1
				  	end
				  end
				  if exists == 0 then
			  	#create new unique project-event
				  	new_e = EventObject.new
				  	new_e.name = event['repo']['name']
				  	new_e.event_count = 1
				  	@events.push new_e
				  end
				end
			end
		end
	end

#sort values least to greatest
	def sort
		@events.sort_by! { |k| k.event_count }
		@events.reverse!
	end

	def display
		@events.each do |event|
			if @counter < @options[:count].to_i then
				puts event.name + " - " + event.event_count.to_s + " events"
				@counter = @counter + 1
			end
		end
	end
end

##
## 	 The Project's Event Type Object
##################################

class EventObject
	attr_accessor :name
	attr_reader :event_count

	def initialize
		@name = ""
		@event_count = 0
	end

  def event_count=(additional_count)
    @event_count = @event_count + additional_count
  end
end

##
## 	 Main application section
##################################

if !ARGV.include? File.basename(__FILE__)
	gh = Gh_archive.new
	if gh.options[:before] then
		gh.connect
		gh.sort
		gh.display 
	end
end

gh_repo_stats_spec.rb

require_relative 'gh_repo_stats'

describe Gh_archive do

    before do
      ARGV.push("--event","PushEvent","--count","30","--before","2015-01-11T16:00:00Z","--after","2015-01-11T16:00:00Z")
      @gh = Gh_archive.new
      @gh.connect
      @gh.sort
      @gh.display
    end

    describe "parameters" do
      it "are correctly set for event type" do
        expect(@gh.options).to include(:event_name)
        expect(@gh.options[:event_name]).to eq("PushEvent")
      end
      it "are correctly set for count" do
        expect(@gh.options).to include(:count)
        expect(@gh.options[:count]).to eq("30")
      end
      it "are correctly set for before_date" do
        expect(@gh.options).to include(:before)
        expect(@gh.options[:before]).to eq("2015-01-11T16:00:00Z")
      end
      it "are correctly set for after_date" do
        expect(@gh.options).to include(:after)
        expect(@gh.options[:after]).to eq("2015-01-11T16:00:00Z")
      end
    end

    describe "combinations" do
      it "should contain at least one datetime reference" do
        expect(@gh.instance_variable_get(:@combos).count).to be > 0
      end
    end

    describe "output" do
      it "quantity of output should be equal to the count parameter." do
        expect(@gh.counter).to be >= @gh.options[:count].to_i
      end
    end
end