Octostats

octostats.rb

require 'octokit'
require 'csv'
require 'hashie'
require 'active_support/all'

class PR < Hashie::Dash
  property "pr"
  property "created_at"
  property "comment_count"
  property "commit_count"
  property "collaborative"
  property "merged"

  def merged?
    merged == "1"
  end

  def collaborative?
    collaborative == "1"
  end
end

#repos = [ { :name => 'spine/spine', :alias => 'spine' },
#          { :name => 'octokit/octokit.rb', :alias => 'octokit' },
#          { :name => 'github/linguist', :alias => 'linguist'}]
repos = [ { :name => 'zestedesavoir/zds-site', :alias => 'zds-site' },
	  { :name => 'ARMmbed/mbed-os', :alias => 'mbed-os' }]

Octokit.auto_paginate = true

client = Octokit::Client.new(access_token: "xxxxx")

repos.each do |repo|
  closed = client.pulls(repo[:name], :state => 'closed')
  puts "closed"
  puts closed.count
  
  pull_requests = []
  
  closed.each do |pr|
    pull_requests << { :number => pr.number, :created_at => pr.created_at }
  end
  
  open = client.pulls(repo[:name], :state => 'open')
  puts "open"
  puts open.count

  open.each do |pr|
    pull_requests << { :number => pr.number, :created_at => pr.created_at }
  end
  
  puts "Working with #{pull_requests.count} pull requests"
  
  results = File.new("#{repo[:alias]}_raw_prs.csv", 'w')
  results.puts "pr,created_at,comment_count,commit_count,collaborative,merged"
  
  pull_requests.each do |pr|
    pull_request = client.pull_request(repo[:name], pr[:number])
    pull_request_commits = client.pull_request_commits(repo[:name], pr[:number])
  
    collaborative = 0
    merged = 0
  
    # Skip if the PR has zero commits
    next unless pull_request_commits.any?
  
    # Were more commits added to the PR over time
    if pull_request_commits.last.commit.committer.date > pull_request.created_at
      collaborative = 1
    end
  
    if pull_request.merged
      merged = 1
    end
  
    result = "#{pr[:number]},#{pr[:created_at]},#{pull_request.comments},#{pull_request.commits},#{collaborative},#{merged}"
    puts result
    results.puts result
  end
  
  results.close

  prs = []

  puts "== go =="

  CSV.foreach("#{repo[:alias]}_raw_prs.csv", :headers => true) do |row|
    puts row
    puts row.to_hash
    prs << PR.new(row.to_hash)
  end

  prs_grouped_by_month = prs.group_by { |m| m.created_at.to_date.beginning_of_month }

  # Now we want the merge fraction over time for pull requests. This means looping
  # through PRs grouped by week and calculating the fraction that are closed.

  merged_fraction_results = File.new("#{repo[:alias]}_merged_fraction.tsv", 'w')
  
  merged_fraction_results.puts("date\tunix_date\tmerged_count\tnot_merged_count\tcount\tmerge_fraction")
  
  total_prs = prs.count
  total_merged = 0
  
  prs_grouped_by_month.sort.each do |month, prs|
  
    merged_count = 0
    prs.each do |pr|
      merged_count += 1 if pr.merged?
      total_merged += 1 if pr.merged?
    end
  
    f_merged = merged_count.to_f / prs.size.to_f
  
    merged_fraction_results.puts("#{month.to_s}\t#{month.to_datetime.to_i*1000}\t#{merged_count}\t#{prs.size-merged_count}\t#{prs.size}\t#{f_merged}")
  end
  
  # Overall stats (for debug)
  puts "#{repo[:alias]} is #{(total_merged.to_f / total_prs) * 100}% merged"

  merged_fraction_results.close

  collaborative_fraction_results = File.new("#{repo[:alias]}_collaborative_fraction.tsv", 'w')
  
  collaborative_fraction_results.puts("date\tunix_date\tcollab_count\tnot_collab_count\tcount\tcollab_fraction")

  total_collab = 0

  prs_grouped_by_month.sort.each do |month, prs|

    collab_count = 0
    prs.each do |pr|
      collab_count += 1 if pr.collaborative?
      total_collab += 1 if pr.collaborative?
    end
  
    f_colab = collab_count.to_f / prs.size.to_f
  
    collaborative_fraction_results.puts("#{month.to_s}\t#{month.to_datetime.to_i*1000}\t#{collab_count}\t#{prs.size-collab_count}\t#{prs.size}\t#{f_colab}")
  end
  
  # Overall stats (for debug)
  puts "#{repo[:alias]} is #{(total_collab.to_f / total_prs) * 100}% collaborative"

  collaborative_fraction_results.close
end