jiazhi/ku/spider/Handle.php

<?php
namespace Ku\Spider;

use Dever;

class Handle
{
	public function createProduct($id = false)
    {
    	$id = Dever::input('id', $id);
        $product = Dever::db('ku/product')->one($id);
        $spider = $this;
        $param = array();
        if ($product && $product['link']) {
            $update = false;
            if (!$product['name'] || !$product['pic'] || !$product['price']) {
                $update = true;
            }
            if ($update) {
                $data = $spider->get($product['link']);
                if ($data) {
                    $param['name'] = $product['name'] ? $product['name'] : $data['name'];
                    $param['pic'] = $product['pic'] ? $product['pic'] : $this->pic($data);
                    $param['price'] = $product['price'] ? $product['price'] : $data['price'];
                }
            }

            Dever::config('base')->hook = true;
            if ($param) {
                $param['where_id'] = $id;
                Dever::db('ku/product')->update($param);
            }
        }
    }

    private function pic($result)
    {
		if (isset($result['pic']) && $result['pic']) {
			if (!strstr($result['pic'], 'http')) {
				$result['pic'] = 'http:' . $result['pic'];
			}
			$pic = Dever::load('upload/save')->copy($result['pic']);
			if (isset($pic['url']) && $pic['url']) {
				$result['pic'] = $pic['url'];
			}
		}
		return $result['pic'];
    }

	public function get($link)
	{
        $type = $this->getType($link);

        if ($type) {
            $link = trim($link, "\r\n");
        	$link = $type->link($link);

	        $data = $this->filter(Dever::curl($link));

	        $result = $this->rule($type->rule(), $data, $type);

	        return $result;
        }

        return false;
	}

	private function getType($link)
	{
		$type = '';
		if (strpos($link, 'taobao.com')) {
        	$type = 'taobao';
        } elseif (strpos($link, 'tmall.com')) {
            $type = '';
        } elseif (strpos($link, 'youzan.com')) {
            $type = 'youzan';
        }

        if (!$type) {
        	return false;
        }
        return Dever::load('ku/spider/' . $type);
	}

	private function rule($rule, $data, $type)
    {
    	$result = array();
    	foreach ($rule as $key => $value) {
    		$m = 1;
    		if (is_array($value)) {
    			$m = $value[1];
    			$value = $value[0];
    		}
    		if (strstr($value, '/')) {
	            $value = str_replace('/', '\/', $value);
	        }
	        preg_match_all('/'.$value.'/i', $data, $matches);
	        $result[$key] = '';
	        if (isset($matches[$m][0])) {
                $result[$key] = trim($matches[$m][0]);
                $result[$key] = $type->replace($key, $result[$key]);
                $result[$key] = str_replace(array(' ', "\t"), '', $result[$key]);
	        }
    	}
        
        return $result;
    }

    private function filter($string)
    {
        $encode = mb_detect_encoding($string, array('GB2312','GBK','UTF-8'));
        $config = array('GB2312', 'GBK', 'EUC-CN', 'CP936');
        if (in_array($encode, $config)) {
            $string = mb_convert_encoding($string, 'UTF-8', 'GBK');
        }
        $string = str_replace(PHP_EOL, '', $string);
        return $string;
    }
}