Share 蟧.gif＂ alt=＂Script Home＂ />Home Page Web Production Script Column Network Programming Database Script Download CMS Tutorial Electronic Books Graphic Design Media_PHP Tutorial

复制代码 代码如下:

/**
 * HOST: www.icbase.com
 */
//set_time_limit(0);
// base function
function curl_get($url, $data = array(), $header = array(), $timeout = 15, $port = 80, $reffer = '', $proxy = '')
{
 $ch = curl_init();
 if (!empty($data)) {
 $data = is_array($data)?http_build_query($data): $data;
 $url .= (strpos($url,'?')? '&': "?") . $data;
 }
 curl_setopt($ch, CURLOPT_URL, $url);
 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, $timeout);
 curl_setopt($ch, CURLOPT_POST, 0);
 curl_setopt($ch, CURLOPT_PORT, $port);
 curl_setopt($ch, CURLOPT_HTTPHEADER, $header);
 curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1); //是否抓取跳转后的页面
 $reffer && curl_setopt($ch, CURLOPT_REFERER, $reffer);
 if($proxy) {
 curl_setopt($ch, CURLOPT_PROXY, $proxy);
 curl_setopt($ch, CURLOPT_PROXYPORT, 1723);
 curl_setopt($ch, CURLOPT_PROXYUSERPWD,"andhm001:andhm123");
 }

$result = array();
 $result['result'] = curl_exec($ch);
 if (0 != curl_errno($ch)) {
 $result['error'] = "Error:n" . curl_error($ch);

}
 curl_close($ch);
 return $result;
}

复制代码 代码如下:

function curl_post($url, $data = array(), $header = array(), $timeout = 5, $port = 80)
{
 $ch = curl_init();
 curl_setopt($ch, CURLOPT_URL, $url);
 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, $timeout);
 //curl_setopt($ch, CURLOPT_PORT, $port);
 !empty ($header) && curl_setopt($ch, CURLOPT_HTTPHEADER, $header);
 curl_setopt($ch, CURLOPT_POST, 1);
 curl_setopt($ch, CURLOPT_POSTFIELDS, $data);

$result = array();
 $result['result'] = curl_exec($ch);
 if (0 != curl_errno($ch)) {
 $result['error'] = "Error:n" . curl_error($ch);

}
 curl_close($ch);

return $result;
}

/**
* Get the html source code of the list page
* @param string $keywords Search keywords
* @param int $page page number
* @return boolean|array
*/
function getListHtml($keywords, $page=1)
{
 if ($page < 0)
 {
 return false;
 }
 $page = $page == 0 ? 1 : intval($page);
 if ($page == 1)
 {
 $result = curl_get('http://www.icbase.com/ProResult.aspx', array('ProKey' => $keywords));
 if ( isset($result['error']) )
 {
 return false;
 //exit($result['error']);
 }
 $result = $result['result'];

 // asp.net post提交数据
 if(! defined('__VIEWSTATE') && preg_match('/ {
 define('__VIEWSTATE', $matches[1]);
 } else {
 return false;
 }

 if(! defined('__PREVIOUSPAGE') && preg_match('/ {
 define('__PREVIOUSPAGE', $matches[1]);
 } else {
 return false;
 }

 if(! defined('__EVENTVALIDATION') && preg_match('/ {
 define('__EVENTVALIDATION', $matches[1]);
 } else {
 return false;
 }

 return $result;
 }
 $data = array(
 '__EVENTTARGET' => 'pager',
 '__EVENTARGUMENT' => $page,
 '__VIEWSTATE' => __VIEWSTATE,
 '__PREVIOUSPAGE' => __PREVIOUSPAGE,
 '__EVENTVALIDATION' => __EVENTVALIDATION,
 );
 $result = curl_post('http://www.icbase.com/ProResult.aspx?ProKey=' . $keywords, $data);
 if ( isset($result['error']) )
 {
 return false;
 //exit($result['error']);
 }
 $result = $result['result'];
 return $result;
}

/**
* Get the url of the list page a link
* @param string $html html source code
* @return array
*/
function getListHref($html)
{
 $pattern = '/[sn]*]/>/isU';
 if (preg_match_all($pattern, $html, $matches))
 {
 return $matches[1];
 } else {
 // 没有匹配项
 return array();
 }
}

/**
* Get the next page number
* @param string $html html source code
* @return number
*/
function getListNextPage($html)
{
 $pattern = '/]>.+>/isU';
 if (preg_match($pattern, $html, $matches))
 {
 return intval($matches[1]);
 } else {
 return -1;
 }
}

/**
* Get all hrefs in the list
* @param string $keywords Search keywords
* @return boolean|array
*/
function getListHrefAll($keywords)
{
 if (empty($keywords))
 {
 return false;
 }

 $html = getListHtml($keywords);
 $hrefList = getListHref($html);
 if (empty($hrefList))
 {
 // 没有结果
 return array();
 }
 $nextPage = getListNextPage($html);
 while ($nextPage > 0)
 {
 $html = getListHtml($keywords, $nextPage);
 $tmpHrefList = getListHref($html);
 $hrefList = array_merge($hrefList, $tmpHrefList);
 $nextPage = getListNextPage($html);
 }
 return $hrefList;
}

/**
* Get details page information
* @param string $url The url address or the captured html source code is distinguished according to @see $is_url
* @param int $is_url 1 uses url address 0 Directly process html source code
* @return boolean|multitype:|multitype:string
*/
function getDetail($url, $is_url = 1)
{
if ( empty($url) )
{
return false;
}
$host = 'www.icbase.com';
$html = $url;
if ($is_url) {
$url = '/' .ltrim ($url, '/');
$result = curl_get($host . $url);
if ( isset($result['error']) )
{
exit($ result['error']);
}
$html = $result['result'];
}

$result = array(
'sup_part' => '', // Supplier model
'sup_id' => '', // Supplier ID
'mfg_part' => '', // Manufacturer model
'mfg_name' => ; '', //Manufacturer name
'cat_name' => '', // Category name
'para' => '', // Attribute
'desc' => ' ', // Description
'pdf_url' => '', // PDF address
'sup_stock' => '', // Stock
'min_purch' => '', // Minimum order quantity
'price' => '', // Price
'img_url' => '', // Image address
'createtime' => '', // Creation time
'datacode' => '', // Batch number
'package' => '', // Package
'page_url' => '', // Page address
) ;

// mfg_part
$pattern = '/Product Model(.[^<]+) if (preg_match($pattern, $html, $matches))
{
$result['mfg_part'] = trim($matches[1]);
} else {
// This item does not exist, and the instructions are nowhere to be found
return array();
}

// mfg_name
$pattern = '/Manufacturer[ sn]*(.+)/isU';
if (preg_match($pattern, $html, $matches))
{
$result['mfg_name' ] = trim($matches[1]);
}

// para
$pattern = '/(.+)/isU';
if (preg_match($pattern, $html, $matches))
{
if (preg_match_all('/(.+)/isU', $matches[1], $matches))
{
$count = count($matches[1 ]);
$count = intval($count / 2 );
foreach ($matches[1] as $k=>$v)
{
if ($k >= $count)
{
break;
}
if (trim($v) == 'Description')
{
// desc
$result['desc '] = trim($matches[1][$count + $k]);
continue;
}
$v = trim($v);
$result['para'] [$v] = trim($matches[1][$count + $k]);
}
}
}

// pdf_url
$pattern = '/ Details if (preg_match($pattern, $html, $matches ))
{
$result['pdf_url'] = trim($matches[1]);
}

// sup_stock
$pattern = '/[sn]*(d+)/isU';
if (preg_match($pattern, $html, $matches))
{
$result['sup_stock'] = trim($matches[1]);
}

// price
$pattern = '/]+>(d+)+]+>.[^d]*([d.]+) /isU';
if (preg_match_all($pattern, $html, $matches))
{
foreach ($matches[1] as $k=>$v)
{
$result['price'][$v] = '￥' . $matches[2][$k];
}
}

//img_url
$pattern = '/Pictures if (preg_match($pattern, $html, $ matches))
{
$result['img_url'] = trim($matches[1]);
}

// page_url
if ($is_url)
{
$result['page_url'] = $host . $url;
}

return $result;
}

/**
* Final call function
* @param string $keywords Search keywords
* @return array
* /
function getData($keywords)
{
$hrefList = getListHrefAll($keywords);
$result = array();

foreach ($hrefList as $k= >$v)
{
$result[] = getDetail($v);
}

return $result;
}

// Test Script
$keywords = trim($_GET['keywords']);
$result = getData($keywords);

print_r($result);

http://www.bkjia.com/PHPjc/728095.htmlwww.bkjia.comtruehttp://www.bkjia.com/PHPjc/728095.htmlTechArticle复制代码 代码如下: ?php /*** HOST: www.icbase.com*/ //set_time_limit(0); // base function function curl_get($url, $data = array(), $header = array(), $timeout = 15, $port...