spider/lib/doc/Core.php

<?php
namespace Spider\Lib\Doc;
use Dever;
use Spider\Lib\Doc as Doc;

class Core
{
	public static function match($pattern, $string)
	{
		$temp = explode('||', $pattern);
		$index = isset($temp[1]) ? $temp[1] : 1;
		preg_match_all('/' . $temp[0] . '/i', $string, $match);
		$result = '';
		if (isset($match[$index][0])) {
			$result = $match[$index][0];
		}
		
		return $result;
	}

	public static function rule($doc, $data, $col, $rule, $key)
	{
		$rule = explode("\n", str_replace("\r", '', $rule));
		if (isset($rule[0]) && $rule[0]) {
			if (isset($col[$rule[0]]) && $rule[0] != $key) {
				# 此处开task
				$result = Doc::getInstance($doc->getUrl($data, $col, $col[$rule[0]]), $rule[1])->get();
				array_shift($rule);
			} elseif (isset($data[$rule[0]])) {
				$result = $data[$rule[0]];
			} else {
				$result = Dom::find($data, $rule[0]);
			}
		}
		if (isset($rule[1]) && $rule[1]) $result = self::match($rule[1], $result);
		
		return $result;
	}
}