Example Drupal Migration Sub Class for OI HTML Content

<?php

/*
 * https://gist.github.com/marktheunissen/2596787
 */

function oi_migrate_migrate_api() {
	$api = array(
    'api' => 2,
    'groups' => array(
      'oi' => array(
        'title' => t('OI Migrations'),
		'default_format' => 'filtered_html',
	),
	),
    'migrations' => array(
      'OIPage' => array(
        'class_name' => 'OIPageMigration',
        'group_name' => 'oi',
		),
	  ),
	);
	return $api;
}

/**
 *
 * Dynamic migration for html files from server
 * @author tommyt
 *
 */
class OIPageMigration extends Migration {
	public $base_dir;
	public $testMenuCount = 0;

	/**
	 * Constructor.
	 */
	public function __construct() {
		parent::__construct();

		// dercription
		$this->description = t('POC for migrating html content into Drupal.');

		// A map of source HTML filename -> destination node id.
		$this->map = new MigrateSQLMap($this->machineName,
		array(
          'sourceid' => array(
            'type' => 'varchar',
            'length' => 255,
            'not null' => TRUE,
		)
		),
		MigrateDestinationNode::getKeySchema()
		);

		// The source fields.
		$fields = array(
	      'title' => t('Title'),
	      'body' => t('Body'),
		  'taxonomy_reference' => t('Section taxonomy reference'),
	      'source_path' => t('Source Path'),
		  'path' => t('Node path alias'),
		  'menu' => t('Menu item'),
	      'uid' => t('User id'),
		);

		// Since the base directory of the HTML files can change depending on the
		// environment, we keep it in a variable. There is no interface for this,
		// set it using drush vset.
		//$this->base_dir = variable_get(my_html_source_path, '');
		//$xml_folder = DRUPAL_ROOT . '/' . drupal_get_path('module', 'migrate_example') . '/xml/';
		//$this->base_dir = '/html_source/research/projects';
		$this->base_dir = DRUPAL_ROOT . '/' . drupal_get_path('module', 'oi_migrate') . '/html_source/';
		// Match HTML files.
		// $regex = '/.*\.html/';
		$regex = '/(.*\.htm$|.*\.html$|.*\.shtml$)/i';

		// The source of the migration is HTML files from the old site.
		$list_files = new MigrateListFiles(array($this->base_dir), $this->base_dir, $regex);
		$item_file = new MigrateItemFile($this->base_dir);
		$this->source = new MigrateSourceList($list_files, $item_file, $fields);

		// The destination is the basic page content type.
		$this->destination = new MigrateDestinationNode('page');

		// Map the fields, pretty straightforward in this case.
		$this->addFieldMapping('uid', 'uid');
		$this->addFieldMapping('title', 'title');
		$this->addFieldMapping('field_section_taxonomy_id', 'taxonomy_reference');
		$this->addFieldMapping('field_source_path', 'source_path');
		$this->addFieldMapping('menu', 'menu');
		$this->addFieldMapping('body', 'body')
		->arguments(array('format' => 'full_html'));
		$this->addFieldMapping('pathauto')->defaultValue(0);
	}

	/**
	 * Prepare a row.
	 */
	public function prepareRow($row) {
		
		// Set to admin for now.
		$row->uid = 1;

		// Create a new SourceParser to handle HTML content.
		$source_parser = new SourceParser(substr($row->sourceid, 1), $row->filedata);
		$row->body = $source_parser->getContent();

		// The title is the html page title.
		$row->title = $source_parser->getPageTitle();

		// The source_path is the path/filename
		$row->source_path = $row->sourceid;
		
		// experimenting with some string processing
		$path_array = explode("/", $row->sourceid);
		$raw_page = count($path_array) >= 0 ?  $path_array[count($path_array)-1] : $path_array[0];
		$page_array = explode(".",$raw_page);
		$link_title = $page_array[0] == "index" ?   ucfirst($path_array[count($path_array)-2]) : ucfirst($page_array[0]);

		// path alias
		$row->path = substr($row->sourceid, 0, strripos($row->sourceid, "/")+1)."/".strtolower($link_title);
		
		// add $row to menu
		$row->menu = array(
		  'link_title' => $link_title,
		  'menu_name' => 'menu-research',
		  'plid' => 0,
		  'enabled' => 1,
		);
	}
	
	// Post processing of node($page) after import and creation entity
  	public function complete(stdClass $page, stdClass $row) {
        // dpm($page->nid);
        $node = node_load( $page->nid );
        $path_array = explode("/", $row->sourceid);
        $taxonomy_id = in_array($path_array[count($path_array)-2], array('research','projects')) ? 4 : 5;
         $node->field_section_taxonomy_id[$node->language][0]['tid'] = $taxonomy_id; 
       // dpm($node);
       
        node_save($node);
       
        // $page->field_testpostimportfield[LANGUAGE_NONE][0]['value'] = $page->nid;
  	}
}