Quick and dirty import into CiviCRM from a webcrawl

import_events.php

<?php

define('SOURCE', '[redacted]');
define('DRUPAL_ROOT', '.');
define('IMPORT_LOCATION_MAIN', 3);
define('IMPORT_WEBSITE_MAIN', 6);
define('IMPORT_PHONE_MOBILE', 2);
define('IMPORT_PHONE_FAX', 3);

$stdout = fopen('php://stdout', 'w');
fwrite($stdout, "Import Script\n");

// Site specific variables
$username = "admin";
$drupal_base_url = '[redacted]'

$_SERVER['HTTP_HOST'] = $drupal_base_url['host'];
// $_SERVER['PHP_SELF'] = $drupal_base_url['path'] . '/index.php';
$_SERVER['REQUEST_URI'] = $_SERVER['SCRIPT_NAME'] = $_SERVER['PHP_SELF'];
$_SERVER['REMOTE_ADDR'] = NULL;
$_SERVER['REQUEST_METHOD'] = NULL;

require_once 'includes/bootstrap.inc';
require_once 'lib/simple_html_dom.php';
drupal_bootstrap(DRUPAL_BOOTSTRAP_FULL);

global $user;

init();
crawl_events(SOURCE);

function create_event($params)
{
  $results = civicrm_event_create($params);
  if ( civicrm_error( $result ))
    return $result['error_message'];
  else
    print_r($result);
}

function crawl_events($url)
{
  $html       = file_get_html($url);
  $crawl_urls = array(); //list of event URLs to crawl

  foreach($html->find('a') as $element) {
    $link_url = $element->href;
    if(preg_match("/calendar_day/i", $link_url)){
      $crawl_urls[] = $link_url;
    }
  }
  
  foreach ($crawl_urls as $crawl_url) {    
    if(parse_event($crawl_url))
      print "Created event from $crawl_url\n";
    else
      print "Could not create event from $crawl_url\n";
  }

}

function parse_event($crawl_url)
{
    $data = file_get_html($crawl_url);

    /* Date parse */
    preg_match("/date\(s\)\:(.*)/i", $data->plaintext, $matches);
    if(empty($matches[1])) {
      print "Could not import $crawl_url .. skipping\n";
      return false;      
    }
    $date = trim(html_entity_decode($matches[1], ENT_QUOTES, "UTF-8"));
    $date = date_parse_from_format("F j, Y", $date);
    $date = sprintf('%04d%02d%02d', $date["year"], $date["month"], $date["day"]);

    /* Time parse, except no time support for civicrm event creation with time */
    preg_match("/time\:(.*)/i", $data->plaintext, $matches);
    $time = trim(html_entity_decode($matches[1], ENT_QUOTES, "UTF-8"));
    $time = date_parse_from_format("g:i A T", $time);
    $time = $time['hour'] . ":" . $time["minute"] . ":00";
    
    /*get Content */
    $content = $data->find(".content", 2)->outertext;

    $title = $data->find(".calendar_header2", 0);
    $title = trim(html_entity_decode($title->innertext, ENT_QUOTES, "UTF-8"));

    $event = array(
      'title' => $title,
      'description' => $content,
      'is_public' => true,
      'event_type_id' => 1,
      'start_date' => "$date",
      'is_active' => true,
    );
    create_event($event);
    return true;
}

function init()
{
  require_once('sites/default/civicrm.settings.php');

  require_once 'CRM/Core/Config.php';
  $config = & CRM_Core_Config::singleton();

  require_once 'CRM/Core/Error.php';
  require_once 'api/v2/Event.php';


  mb_language('uni');
  mb_internal_encoding('UTF-8');  
}


function import_test()
{
  $result = civicrm_api('contact', 'get', array('email' => '[email protected]', 'version' => 3));
  print_r($result);
}