State Decoded parser for Philadelphia for American Legal data

class.Philadelphia.inc.php

<?php

/**
 * San Francisco parser for State Decoded.
 * Extends AmericanLegal base classes.
 *
 * PHP version 5
 *
 * @license		http://www.gnu.org/licenses/gpl.html GPL 3
 * @version		0.8
 * @link		http://www.statedecoded.com/
 * @since		0.3
*/

/**
 * This class may be populated with custom functions.
 */

require 'class.AmericanLegal.inc.php';

class State extends AmericanLegalState {}

class Parser extends AmericanLegalParser
{
	/*
	 * Most codes have a Table of Contents as the first LEVEL.
	 */
	public function pre_parse_chapter(&$chapter)
	{

		// Get all of the titles.
		$titles = $chapter->REFERENCE->TITLE;

		$title_count = count($titles);

		// We skip the last title, thus " - 1"
		for($i = 0; $i < $title_count - 1; $i++)
		{
			unset($name, $identifier, $order_by);

			$title = $titles[$i];

			// Clean up the title, strip some html.
			$title = trim(str_replace('&quot;', '', $title));

			if(preg_match($this->structure_regex, $title, $matches))
			{
				$name = $matches['name'];
				$identifier = $matches['number'];
			}
			else
			{
				switch($title)
				{
					case 'THE PHILADELPHIA CODE' :
						$name = 'The Philadelphia Code';
						$identifier = 'Philadelphia';
						$order_by = '0';
						break;

					case 'PHILADELPHIA HOME RULE CHARTER' :
						$name = 'Philadelphia Home Rule Charter';
						$identifier = 'Charter';
						$order_by = '1';
						break;
				}
			}


			if(isset($name) && isset($identifier))
			{
				$structure = new stdClass();
				$structure->name = $this->clean_title($name);
				$structure->label = 'Code';
				$structure->identifier = $this->clean_identifier($identifier);

				if(!isset($order_by))
				{
					$order_by = $this->get_structure_order_by($structure);
				}
				$structure->order_by = $order_by;

				if(isset($last_structure))
				{
					$structure->parent_id = $last_structure->id;
				}

				$structure->level = count($this->structures) + 1;

				// if(isset($matches[3]) && strlen(trim($matches[3])))
				// {
				// 	$structure->metadata = new stdClass();
				// 	$structure->metadata->text = trim($matches[3]);
				// }

				$this->create_structure($structure);

				$last_structure = $structure;

				$this->structures[] = $structure;
			}
			else {
				$this->logger->message('Cannot get title from structure: "' . $title . '"', 5);
			}

		}

		// If there's two levels,this has a table of contents.
		if(count($chapter->LEVEL->LEVEL) == 2 && isset($chapter->LEVEL->LEVEL[1]->LEVEL) && isset($chapter->LEVEL->LEVEL[1]->LEVEL->LEVEL))
		{
			$this->logger->message('Skipping first level.', 2);
			unset($chapter->LEVEL->LEVEL[0]);
		}
	}

	public function get_structure_order_by($structure)
	{
		// Make room for the charter by adding one.
		return 1 + (int) parent::get_structure_order_by($structure);
	}

	public function get_section_parts($section)
	{
		/*
		 * Parse the catch line and section number.
		 */
		$section_title = $this->clean_title((string) $section->RECORD->HEADING);

		$this->logger->message('Title: ' . $section_title, 2);

		switch($section_title)
		{
			case 'PREFACE TO THE TENTH EDITION' :
				$section_parts = array(
					'catch_line' => 'Preface to the Tenth Edition',
					'number' => 'Preface'
				);
				break;

			case 'PREAMBLE' :
				$section_parts = array(
					'catch_line' => 'Preamble',
					'number' => 'Preamble'
				);
				break;
		}

		if(!isset($section_parts))
		{
			$section_parts = parent::get_section_parts($section_title);
		}

		return $section_parts;
	}

}