github-prs.js

/*
Gets all your PRs from GitHub. A simple-ish wrapper around the
GitHub cli which doesn't seem to paginate (?) or offer output
formats besides JSON (and personally I like to review these
things in Excel)
* Example usage:
  node github-prs --from 2022-01-01 --to 2022-12-31 > prs.csv 
* I can't be bothered including a package.json :P
  You need to do:
  > npm install lodash papaparse yargs moment
  or equivalent.
  Because who doesn't love a JS dependency.
* You also need to have installed the GitHub cli tool:
  > brew install gh
  And have authenticated it:
  > gh auth login
  (then follow the prompts).
* N.B. moment is probably not needed as a dependency as in theory
  we're dealing only with ISO date strings that can be compared
  directly for ordering. But that feels slightly icky.
* The script prints the CSV data directly to stdout in the UNIX
  spirit of things, redirect or pipe it as you will.
  
* I hit rate limits in an earlier version of the script that made
  requests in parallel but I don't think it should be an issue now
  everything is serial.
*/

const util = require('util');
const exec = util.promisify(require('child_process').exec);
const _ = require('lodash')
const papa = require('papaparse')
const yargs = require('yargs')
const moment = require('moment')

function sleep(ms) {
  return new Promise((resolve) => {
    setTimeout(resolve, ms)
  })
}

async function getArgv() {
  return await yargs
    .option('from', {
      type: 'date',
      description: 'PR created from this date (as an ISO string)',
      demandOption: true,
    })
    .option('to', {
      type: 'date',
      description: 'PR created to this date (as an ISO string) - defaults to now',
      demandOption: false,
    })
    .option('json', {
      type: 'boolean',
      description: 'Output JSON instead of CSV',
      demandOption: false
    })
    .parse()
}

async function prsBefore(createdBefore) {
  const delay = sleep(8500)
  const cmd = [`gh search prs`, 
    `--sort=created`,
    `--author=@me`,
    `--created "<=${createdBefore.format()}"`,
    `--json=createdAt,repository,url,title,body`,
  ].join(' ')
  const { stdout } = await exec(cmd);
  await delay
  return JSON.parse(stdout)
}

async function getPrDump(from, to) {
  let before = to
  const allPrs = []
  while (before.isAfter(from)) {
    const prs = await prsBefore(before)
    allPrs.push(...prs)
    if (prs.length > 0) {
      const { createdAt } = prs[prs.length - 1]
      before = moment(createdAt, moment.ISO_8601)
    } else {
      break
    }
  }
  return allPrs  
}

function processPrs(prs, from, to) {
  return _(prs)
    .uniqBy(pr => `${pr.title}::${pr.createdAt}`)
    .filter(pr => 
      moment(pr.createdAt, moment.ISO_8601).isBetween(from, to)
    )
    .sortBy('createdAt')
    .map(pr => Object.assign(pr, { repository: pr.repository.name }))
    .value() 
}

async function main() {
  const { from: fromString, to: toString, json } = await getArgv()
  const from = moment(fromString, moment.ISO_8601)
  const to = toString ? moment(toString, moment.ISO_8601) : moment()
  const allPrs = await getPrDump(from, to)
  const processed = processPrs(allPrs, from, to)
  const output = json ?
    JSON.stringify(processed) :
    papa.unparse(processed)
  console.log(output)
}

main()