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)
	{
		$result = '';
		$rule = Dever::split($rule);
		if (isset($rule[0]) && $rule[0]) {
			if (isset($col[$rule[0]]) && $rule[0] != $key) {
				if (isset($rule[1]) && $rule[1]) {
					$result = Doc::getInstance($doc->getUrl($data, $col, $col[$rule[0]]), $rule[1])->get();
					array_shift($rule);
				} else {
					$result = self::rule($doc, $data, $col, $col[$rule[0]]['collect_rule'], $key);
				}
			} else {
				if (is_string($data)) {
					$data = json_decode($data, true);
				}
				if (is_object($data)) {
					$result = Dom::find($data, $rule[0]);
				} elseif (is_array($data)) {
					$temp = explode('.', $rule[0]);

					$result = $data;
					foreach ($temp as $k => $v) {
						if (isset($result[$v])) {
							$result = $result[$v];
						}
					}
				}
			}
		}

		/*
		if (isset($rule[0]) && $rule[0]) {
			if (isset($col[$rule[0]]) && $rule[0] != $key) {
				# 此处开task
				if (isset($rule[1]) && $rule[1]) {
					$result = Doc::getInstance($doc->getUrl($data, $col, $col[$rule[0]]), $rule[1])->get();
					array_shift($rule);
				} else {
					$result = self::rule($doc, $data, $col, $col[$rule[0]]['collect_rule'], $key);
				}
			} elseif (isset($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;
	}
}