wheredid.rb

require "rubygems"
require "bundler"

Bundler.require(:default)

def find_ref(body)
  refs = body.split("\n").map do |line|
    next unless line =~ /https:/
    line.gsub(%r{.*(https://[^\s\)">]+).*}, '\1')
  end
    .compact
    .select { |url| !%w(.jpg .jpeg .png .gif).include?(File.extname(url)) }
    .join("\n")
  return nil if refs == body
  refs
end

def search_query(last_cursor:, predicates:[])
  {
    variables: {
      query: predicates.compact.join(" "),
      cursor: last_cursor,
    },

    query: <<~GQL
      query search($query: String!, $cursor: String) {
        search(query: $query, type: ISSUE, first: 100, after: $cursor) {
          nodes {
            ...pullRequestFragment
          }
          pageInfo {
            hasNextPage
            endCursor
          }
        }
      }

      fragment pullRequestFragment on PullRequest {
        author { login }
        title
        body
        createdAt
        url
        state
      }
    GQL
  }
end

def search_by_author(cursor:, since:, org:, repo:, login:)
  query = search_query(
    last_cursor: cursor,
    predicates: [
      "is:pr",
      "created:>=#{since}",
      "author:#{login}",
      "org:#{org}",
      "repo:#{repo}"
    ]
  ).to_json
  gh_client.post("/graphql", query)
end

def author(since:, org:, repo:, member_logins:)
  nodes = []
  member_logins.each do |login|
    has_next = true
    cursor = nil
    while has_next
      puts "querying PRs created by: #{login}"
      response = search_by_author(cursor: cursor, since: since, org: org, repo: repo, login: login)
      response = response[:data][:search]

      nodes += response[:nodes]
      has_next = response[:pageInfo][:hasNextPage]
      cursor = response[:pageInfo][:endCursor]
    end
    sleep(2)
  end
  nodes
end

def scan_author(since:, org:, repo:, member_logins:)
  nodes = author(since: since, org: org, repo: repo, member_logins: member_logins)
  CSV.open(ENV.fetch("CSV_FILENAME", "author_org-#{org}_since-#{since}.csv"), "wb") do |csv|
    csv << %w(created status author title url ref)
    nodes.each do |node|
      date = Date.parse(node[:createdAt]).iso8601 rescue nil
      login = node[:author][:login] rescue node[:author]
      title = node[:title]
      ref = find_ref(node[:body] || '')
      url = node[:url]
      status = node[:state]
      csv << [date, status, login, title, url, ref]
    end
  end
end

def search_by_reviewer(cursor:, since:, org:, repo:, login:)
  query = search_query(
    last_cursor: cursor,
    predicates: [
      "is:pr",
      "created:>=#{since}",
      "reviewed-by:#{login}",
      "org:#{org}",
      "repo:#{repo}"
    ]
  ).to_json
  gh_client.post("/graphql", query)
end

def reviewer(since:, org:, repo:, member_logins:)
  nodes = []
  member_logins.each do |login|
    has_next = true
    cursor = nil
    while has_next
      puts "querying PRs reviewed by: #{login}"
      response = search_by_reviewer(cursor: cursor, since: since, org: org, repo: repo, login: login)
      response = response[:data][:search]

      nodes += response[:nodes].map { |node| node[:reviewer] = login; node }
      has_next = response[:pageInfo][:hasNextPage]
      cursor = response[:pageInfo][:endCursor]
    end
    sleep(2)
  end
  nodes
end

def scan_reviewer(since:, org:, repo:, member_logins:)
  nodes = reviewer(since: since, org: org, repo: repo, member_logins: member_logins)
  CSV.open(ENV.fetch("CSV_FILENAME", "reviewer_org-#{org}_since-#{since}.csv"), "wb") do |csv|
    csv << %w(created status reviewer title url ref)
    nodes.each do |node|
      date = Date.parse(node[:createdAt]).iso8601 rescue nil
      login = node[:reviewer]
      title = node[:title]
      ref = find_ref(node[:body] || '')
      url = node[:url]
      status = node[:state]
      csv << [date, status, login, title, url, ref]
    end
  end
end

def gh_client
  @gh_client ||= Octokit::Client.new(access_token: ENV["GITHUB_TOKEN"])
end

logins = ENV["GITHUB_LOGINS"].split(",")

if ENV["USE_GITHUB_TEAM"] == "true"
  team_id = ENV.fetch("GITHUB_TEAM_ID", 2265354)
  logins = gh_client.team_members(team_id).map(&:login)
end

scan_author(since: "2020-03-16", org: "quipper", repo: "quipper/monorepo-global", member_logins: logins)
scan_reviewer(since: "2020-03-16", org: "quipper", repo: "quipper/monorepo-global", member_logins: logins)