github scraper. get user info for the users watching a particular repo

github-scraper.php

#!/usr/bin/env php
<?php

function get_watchers($login, $repo) {
  $per_page = 100; // this is the max set by github http://developer.github.com/v3/#pagination
  $page_number = 1;
  $watchers = array();

  while (1) {
    $ch = curl_init(sprintf('https://api.github.com/repos/%s/%s/watchers?page=%d&per_page=%d',
      $login, $repo, $page_number, $per_page)); 
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
    $new_watchers = json_decode(curl_exec($ch), true);
    curl_close($ch);
    $watchers = array_merge($watchers, $new_watchers);

    if (count($new_watchers) < $per_page || $page_number > 15) {
      return $watchers;
    }
    
    $page_number++;
  }
}

function get_user($login) {
  $ch = curl_init('https://api.github.com/users/'.$login);
  curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
  $json = json_decode(curl_exec($ch), true);
  fprintf(STDERR, "$login processed!\n");
  curl_close($ch);
  return $json;
}

function get_users($users) {
  $user_data = array();

  foreach ($users as $user) {
    $login = $user['login'];
    $company = $user['company'];
    $user_data["$company:$login"] = get_user($login);
  }
  
  ksort($user_data);
  return $user_data;
}

function output_data($users) {
  $buf = array();

  foreach ($users as $user) {
    $user['company'] = str_replace(',', ' ', $user['company']);
    $buf[] = "{$user['company']}, {$user['login']}, {$user['followers']}, {$user['email']}, {$user['blog']}";
  }

  echo implode("\n", $buf);
}


global $argc;
global $argv;
$options = getopt('h');

if (isset($options['h']) || $argc < 2) {
  echo <<<HELP
Usage:
You need PHP 5.3 or above with curl enabled.

$ php github-scraper.php <login name>:<repo name>

e.g.
$ php github-scraper fluent:fluentd

HELP;
  exit(1);
}

list($login, $repo) = explode(':', $argv[1]);
output_data(get_users(get_watchers($login, $repo)));