本文介绍了一个用于抓取hao123网站上多个城市团购信息的函数。该函数通过定义一系列URL来获取不同城市的团购数据，并解析这些数据以提取包括网站名称、商品标题等在内的关键信息。
/**
* function 抓取hao123的团购信息
* @access public
* @return integer
*/
public function hao123()
{
$arrUrl = array(
array(
'province' => 'beijing',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_1_0_jtpage.js',
'replace' => '/u5317/u4eac',
),
array(
'province' => 'shanghai',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_2_0_jtpage.js',
'replace' => '/u4e0a/u6d77',
),
array(
'province' => 'guangzhou',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_3_0_jtpage.js',
'replace' => '/u5e7f/u5dde',
),
array(
'province' => 'shenzhen',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_4_0_jtpage.js',
'replace' => '/u6df1/u5733',
),
array(
'province' => 'wuhan',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_5_0_jtpage.js',
'replace' => '/u6b66/u6c49',
),
array(
'province' => 'hangzhou',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_6_0_jtpage.js',
'replace' => '/u676d/u5dde',
),
array(
'province' => 'tianjin',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_7_0_jtpage.js',
'replace' => '/u5929/u6d25',
),
array(
'province' => 'nanjing',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_8_0_jtpage.js',
'replace' => '/u5357/u4eac',
),
array(
'province' => 'chengdu',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_9_0_jtpage.js',
'replace' => '/u6210/u90fd',
),
array(
'province' => 'chongqing',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_10_0_jtpage.js',
'replace' => '/u91cd/u5e86',
),
array(
'province' => 'shenyang',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_11_0_jtpage.js',
'replace' => '/u6c88/u9633',
),
array(
'province' => 'suzhou',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_12_0_jtpage.js',
'replace' => '/u82cf/u5dde',
),
array(
'province' => 'wuxi',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_13_0_jtpage.js',
'replace' => '/u65e0/u9521',
),
array(
'province' => 'wuxi',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_13_0_jtpage.js',
'replace' => '/u65e0/u9521',
),
array(
'province' => 'xian',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_14_0_jtpage.js',
'replace' => '/u897f/u5b89',
),
array(
'province' => 'zhengzhou',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_15_0_jtpage.js',
'replace' => '/u90d1/u5dde',
),
array(
'province' => 'dalian',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_16_0_jtpage.js',
'replace' => '/u5927/u8fde',
),
array(
'province' => 'qingdao',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_17_0_jtpage.js',
'replace' => '/u9752/u5c9b',
),
array(
'province' => 'changsha',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_18_0_jtpage.js',
'replace' => '/u957f/u6c99',
),
array(
'province' => 'hefei',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_19_0_jtpage.js',
'replace' => '/u5408/u80a5',
),
array(
'province' => 'taiyuan',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_20_0_jtpage.js',
'replace' => '/u592a/u539f',
),
array(
'province' => 'fuzhou',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_21_0_jtpage.js',
'replace' => '/u798f/u5dde',
),
array(
'province' => 'kunming',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_22_0_jtpage.js',
'replace' => '/u6606/u660e',
),
array(
'province' => 'shijiazhuang',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_23_0_jtpage.js',
'replace' => '/u77f3/u5bb6/u5e84',
),
array(
'province' => 'jinan',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_24_0_jtpage.js',
'replace' => '/u6d4e/u5357',
),
array(
'province' => 'xiamen',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_25_0_jtpage.js',
'replace' => '/u53a6/u95e8',
),
array(
'province' => 'langfang',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_26_0_jtpage.js',
'replace' => '/u5eca/u574a',
),
array(
'province' => 'nanchang',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_27_0_jtpage.js',
'replace' => '/u5357/u660c',
),
array(
'province' => 'ningbo',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_29_0_jtpage.js',
'replace' => '/u5b81/u6ce2',
),
array(
'province' => 'changzhou',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_30_0_jtpage.js',
'replace' => '/u5b81/u6ce2',
),
array(
'province' => 'haerbin',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_31_0_jtpage.js',
'replace' => '/u54c8/u5c14/u6ee8',
),
array(
'province' => 'shaoxing',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_33_0_jtpage.js',
'replace' => '/u7ecd/u5174',
),
array(
'province' => 'changchun',
'url' => 'http://app.hao123.com/tuangou/data/tgdata_34_0_jtpage.js',
'replace' => '/u957f/u6625',
),
);
$file_process = "hao123.txt";
clearstatcache();
$process = file_exists($file_process) ? file_get_contents($file_process) : 0 ;
$processAdd = count($arrUrl) <= $process ? 0 : $process +1 ;
file_put_contents($file_process,$processAdd);
for($i = 1; $i <= 3 ; $i++)
{
$url = str_replace('jtpage',$i,$arrUrl[$process]['url']);
$content = file_get_contents($url);
if(stripos($content,'var strTgData=') === false)
{
break;
}
$content = str_replace('var strTgData=','',$content);
$content = str_replace('};','}',$content);
$content = str_replace('{"'.$arrUrl[$process]['replace'],'{"items',$content);
$content = json_decode($content);
foreach($content->items as $item)
{
$data['loc'] = $item->item_url;
$data['website'] = iconv('UTF-8','GBK',$item->website);
$data['siteurl'] = $item->siteurl;
$data['city'] = iconv('UTF-8','GBK',$item->city);
$data['city_id'] = $this->getCityId($data['city']);
$data['category'] = $item->category;
$data['major'] = $item->major;
$data['title'] = iconv('UTF-8','GBK',$item->title);
$data['image'] = $item->image;
$data['startTime'] = $item->startTime;
$data['endTime'] = $item->endTime;
$data['value'] = $item->value;
$data['price'] = $item->price;
$data['rebate'] = $item->rebate;
$data['endTime'] = $item->endTime;
$data['bought'] = $item->bought;
$data['range'] = iconv('UTF-8','GBK',$item->shopProp[0]->range);
$data['address'] = iconv('UTF-8','GBK',$item->shopProp[0]->address);
$data['isapi'] = 1;
$this->dealInfo($data);
}
}
}
扫一扫
评论
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
查看更多评论
添加红包
