konmari.rb

# The origin of this script is https://gist.github.com/iambchan/273219729eb88155e11672c4ccac72b3 
# as linked in the blog post https://building.kickstarter.com/how-to-konmari-your-git-repo-474907a29d2e
# I applied two changes: a fix for cursor loop and add multiple dates for the PR for better reporting

require 'http'
require 'date'

REPOSITORY_OWNER = ""
REPOSITORY_NAME = ""
REFS_PREFIX = "refs/heads/"

GITHUB_AUTH_TOKEN = ""

# Simple request to the Github Graphql api
def self.perform_request(auth_token:, graphql:)
  response = HTTP
    .auth("Bearer #{auth_token}")
    .post(
      "https://api.github.com/graphql",
      body: { query: graphql }.to_json
    )

  unless response.code == 200
    raise response.inspect
  end

  JSON.parse(response.body)
end

# Gets first 100 branches and associated pull requests
# Takes a cursor for paging
def self.pull_requests(token:, repository:, owner:, cursor: nil)
  query = <<~GRAPHQL
    query {
      repository(name: "#{repository}", owner: "#{owner}") {
        refs(refPrefix: "refs/heads/", first: 100 #{", after: \"#{cursor}\"" if cursor}) {
          edges {
            cursor
            node {
              name
              target {
                oid
              }
              associatedPullRequests(last: 100) {
                edges {
                  node {
                    number
                    publishedAt
                    updatedAt
                    closedAt
                    state
                    isDraft
                    changedFiles
                    mergeable
                    author{
                      login
                    }
                  }
                }
              }
            }
          }
        }
      }
    }
  GRAPHQL

  data = perform_request(auth_token: token, graphql: query)
  edges = data.dig("data", "repository", "refs", "edges")
  raise data.inspect if edges.nil?

  # For each branch output: date, branch name, author, pull request state, todo
  # todo is set to 'DELETE' if state of the PR is merged, deleted,
  # or older than 3 months ago (hard coded to before Sept 1, 2018)
  edges.map do |edge|
    branch = edge.fetch("node")
    date = Time.at(`git show #{branch.dig("target", "oid")} --format="format:%at" | head -n 1`.to_i).to_date
    pr = (branch.dig("associatedPullRequests", "edges").first || {})["node"] || {}

    publishedAt = pr.dig("publishedAt") ? DateTime.iso8601(pr.dig("publishedAt")).to_date : ""
    closedAt = pr.dig("closedAt") ? DateTime.iso8601(pr.dig("publishedAt")).to_date : ""
    updatedAt = pr.dig("updatedAt") ? DateTime.iso8601(pr.dig("publishedAt")).to_date : ""

    # Very naive implementation for setting the `todo` column
    todo = ""
    todo = "DELETE" if pr["state"] == "MERGED"
    todo = "DELETE" if pr["state"] == "CLOSED"
    todo = "DELETE" if date < Date.new(2018, 9, 1)
    todo = "" if pr["state"] == "OPEN"
    todo = "" if branch["name"].start_with?("hd") # hackday
    todo = "" if branch["name"] == "master"

    puts [
      date,
      branch["name"],
      pr.dig("number"),
      publishedAt,
      updatedAt,
      closedAt,
      pr.dig("state"),
      pr.dig("isDraft"),
      pr.dig("changedFiles"),
      pr.dig("mergeable"),
      pr.dig("author", "login"),
      pr.dig("state"),
      todo
    ].join(",")
  end

  exit if edges.size < 100
  cursor = edges.last["cursor"]
end

# Process refs in batches of 100.
cursor = nil
loop do
  cursor = pull_requests(token: GITHUB_AUTH_TOKEN, repository: REPOSITORY_NAME, owner: REPOSITORY_OWNER, cursor: cursor)
end