复制代码 代码如下:
/**
* HOST: components.arrow.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 = 15, $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;
}
/**
* 获取列表页的html源码
* @param string $keywords 搜索关键字
* @param int $start 开始记录数
* @return boolean|array
*/
function getListHtml($keywords, $start = 0)
{
if ($start {
return false;
}
$postData = array(
'search_token' => $keywords,
'start' => $start,
'limit' => 100,
);
$result = curl_post('http://components.arrow.com/part/search/' . $keywords, http_build_query($postData));
if ( isset($result['error']) )
{
return false;
//exit($result['error']);
}
$result = $result['result'];
return $result;
}
/**
* 获取列表页 连接href
* @param string $html html源码
* @return array
*/
function getListHref($html)
{
$pattern = '/
if (preg_match_all($pattern, $html, $matches))
{
return $matches[1];
} else {
// 没有匹配项
return array();
}
}
/**
* 获取下一页数字start
* @param string $html html源码
* @return number
*/
function getListNextPage($html)
{
$pattern = '/<script>buildPagination\(\'\d+\',\'\d+\',\'(\d+)\',\d+\);<\/script>/isU';<br> if (preg_match($pattern, $html, $matches))<br> {<br> return intval($matches[1]);<br> } else {<br> return -1;<br> }<br> }<br><br>/**<br> * 获取列表也所有的详细列表<br> * @param string $keywords 搜索关键字<br> * @return boolean|array<br> */<br> function getListHrefAll($keywords)<br> {<br> if (empty($keywords))<br> {<br> return false;<br> }<br><br>$html = getListHtml($keywords);<br> $hrefList = getListHref($html);<br> if (empty($hrefList))<br> {<br> // 没有结果<br> return array();<br> }<br> $nextPage = getListNextPage($html);<br> $loop =0;<br> while ($nextPage > 0)<br> {<br> $html = getListHtml($keywords, $nextPage);<br> $tmpHrefList = getListHref($html);<br> $hrefList = array_merge($hrefList, $tmpHrefList);<br> $nextPage = getListNextPage($html);<br> $loop ++;<br> }<br> return $hrefList;<br> }<br><br>/**<br> * 获取详情页信息<br> * @param string $url url地址<br> * @return array()<br> */<br> function getDetail($url)<br> {<br> if ( empty($url) )<br> {<br> return false;<br> }<br> $host = 'http://components.arrow.com';<br><br>$url = $host . $url;<br> $result = curl_get($url);<br> if ( isset($result['error']) )<br> {<br> return array();<br> //exit($result['error']);<br> }<br> $html = $result['result'];<br><br>$result = array(<br> 'sup_part' => '', // 供应商型<br> 'sup_id' => '', // 供应商ID<br> 'mfg_part' => '', // 制造商型号<br> 'mfg_name' => '', // 制造商名称<br> 'cat_name' => '', // 分类名称<br> 'para' => '', // 属性<br> 'desc' => '', // 描述<br> 'pdf_url' => '', // PDF地址<br> 'sup_stock' => '', // 库存<br> 'min_purch' => '', // 最小订购量<br> 'price' => '', // 价格<br> 'img_url' => '', // 图片地址<br> 'createtime' => '', // 创建时间<br> 'datacode' => '', // 批号<br> 'package' => '', // 封装<br> 'page_url' => '', // 页面地址<br> );<br><br>// mfg_part<br> $pattern = '/<li>[\s\n]*<strong>Part No:\s*<\/strong>(.+)<\/li>/isU';<br> if (preg_match($pattern, $html, $matches))<br> {<br> $result['mfg_part'] = trim($matches[1]);<br> } else {file_put_contents('page.txt', $html);die('xxx');<br> return array();<br> }<br><br>// mfg_name<br> $pattern = '/<li>[\s\n]*<strong>Manufacturer: <\/strong>(.+)<\/li>/isU';<br> if (preg_match($pattern, $html, $matches))<br> {<br> $result['mfg_name'] = trim($matches[1]);<br> }<br><br>// cat_name<br> $pattern = '/displayCategory\(\'(.[^\']+)\'\);/isU';<br> if (preg_match($pattern, $html, $matches))<br> {<br> $result['cat_name'] = trim($matches[1]);<br> $result['cat_name'] = str_replace('|', '>', $result['cat_name']);<br> }<br><br>// para<br> $tablepattern = '/<table\s+id="part_specs".[^>]*>(.+)<\/table>/isU';<br> if (preg_match($tablepattern, $html, $matches))<br> {<br> $pattern = '/<tr>[\s\n]*<td><strong>(.+)<\/strong><\/td><td>(.+)<\/td>[\s\n]*<\/tr>/isU';<br> if (preg_match_all($pattern, $matches[1], $matches))<br> {<br> foreach($matches[1] as $k=>$v)<br> {<br> $v = trim($v);<br> if ('Package Type' == $v)<br> {<br> $result['package'] = trim($matches[2][$k]);<br> continue;<br> }<br> $result['para'][$v] = trim($matches[2][$k]);<br> }<br> }<br> }<br><br>// desc<br> $pattern = '/<div\s+id="part_title">.+<h4>(.+)<\/h4>[\s\n]*<\/div>/isU';<br> if (preg_match($pattern, $html, $matches))<br> {<br> $result['desc'] = trim($matches[1]);<br> }<br><br>// pdf_url<br> $pattern = '/<li\s+class="datasheet">[\s\n]*<strong>Datasheet:<\/strong><a\s+href="(.[^"]+)"/isU';<br> if (preg_match($pattern, $html, $matches))<br> {<br> $result['pdf_url'] = $host . trim($matches[1]);<br> }<br><br>// sup_stock<br> $pattern = '/<td\s+id="inv_1"\s+class="li_inv">([\d,]+)<\/td>/isU';<br> if (preg_match($pattern, $html, $matches))<br> {<br> $result['sup_stock'] = trim($matches[1]);<br> $result['sup_stock'] = str_replace(',', '', $result['sup_stock']);<br> }<br><br>// min_purch<br> $pattern = '/<span\s+id="multiples">[\s\n]*<strong>Multiple:\s*<\/strong>(.+)<\/span>/isU';<br> if (preg_match($pattern, $html, $matches))<br> {<br> $result['min_purch'] = trim($matches[1]);<br> }<br><br>// price<br> $pattern = '/<div\s+id="price_1"\s+class="li_price">(.[^<]+)<\/div>/isU';<br> if (preg_match($pattern, $html, $matches))<br> {<br> $result['price'][1] = trim($matches[1]);<br> }<br> $pattern = '/<div\s+id="price_1"\s+class="li_price">[\s\n]*<span.[^>]+title="(.[^"]+)">/isU';<br> if (preg_match($pattern, $html, $matches))<br> {<br> $priceurl = str_replace('&', '&', $matches[1]);<br> $json = curl_get($priceurl);<br> $json = $json['result'];<br> if (! empty($json))<br> {<br> $jsonresult = json_decode($json, true);<br> foreach ($jsonresult['parts'][0]['webprice']['resale'] as $k=>$v)<br> {<br> $result['price'][$v['minqty']] = $v['price'];<br> }<br> }<br> }<br><br>// img_url<br> $pattern = '/<div\s+id="part_image">[\s\n]*<img \s+src="(.[^"]+)"/isU';<br alt="分享下页面关键字抓取components.arrow.com站点代码_PHP教程" > if (preg_match($pattern, $html, $matches))<br> {<br> $result['img_url'] = trim($matches[1]);<br> }<br><br>// page_url<br> $result['page_url'] = $url;<br><br>return $result;<br> }<br><br>/**<br> * 最终调用函数<br> * @param string $keywords 搜索关键字<br> * @return array<br> */<br> function getData($keywords)<br> {<br> $hrefList = getListHrefAll($keywords);<br> $result = array();<br><br>foreach ($hrefList as $k=>$v)<br> {<br> $result[] = getDetail($v);<br> }<br><br>return $result;<br> }<br><br>// Test Script<br> $keywords = trim($_GET['keywords']);<br> $result = getData($keywords);<br><br>print_r($result);<br><p align="left"><div style="max-width:90%"><span id="url" itemprop="url">http://www.bkjia.com/PHPjc/728094.html<span id="indexUrl" itemprop="indexUrl">www.bkjia.com<span id="isOriginal" itemprop="isOriginal">true<span id="isBasedOnUrl" itemprop="isBasedOnUrl">http://www.bkjia.com/PHPjc/728094.html<span id="genre" itemprop="genre">TechArticle<span id="description" itemprop="description">复制代码 代码如下: ?php /** * HOST: components.arrow.com */ //set_time_limit(0); // base function function curl_get($url, $data = array(), $header = array(), $timeout = 15,...<div class="art_confoot"> </script>
