Parse PTT data

parse.php

<?php

define(URL_BASE, 'https://www.ptt.cc');
define(URL_INDEX, URL_BASE . '/bbs/Badminton/index.html');
define(MAX_LIMIT, 300);
define(KEY_WROD, '[揪人]');

$result = array();

function get_html($url) {
	return file_get_contents($url);
}

function parse($url) {
	global $result;

	if ($url === URL_BASE || count($result) > MAX_LIMIT) {
		return;
	} else {
		printf("Parse url: %s\n", $url);
		$html = get_html($url);
		get_list($html);


		$pattern = '/<a class="btn wide" href="(.+)">&lsaquo; 上頁<\/a>/';
		preg_match_all($pattern, $html, $matches);

		sleep(0.5);
		parse(URL_BASE . $matches[1][0]);
	}
}

function get_list($content) {
	global $result;

	$pattern = '/<div class="title">\s*<a href="(.+)">(.+)<\/a>\s*<\/div>/i';	
	
	if (preg_match_all($pattern, $content, $matches) !== FALSE) {
		if ($matches[1] !== NULL && $matches[2] !== NULL) {
			$size = count($matches[1]);
			for ($i = 0; $i < $size; $i++) {
				if (filter_key_word($matches[2][$i])) {
					$data = array(
						'title' => $matches[2][$i],
						'url' => URL_BASE . $matches[1][$i]
					);
					array_push($result, $data);
				}
			}
		}
	}
}

function filter_key_word($content) {
	$result = strpos($content, KEY_WROD);
	return ($result !== FALSE) ? TRUE : FALSE;
}

function get_content() {
	//	TODO:
}



//============ start =============
parse(URL_INDEX);
var_dump($result);