php 正则采集获取文章,PHP正则过滤采集文章

$news_info['cont'] = preg_replace('/|]+?)>/','',$news_info['cont']);

if(isset($_GET['iphone'])  &&  $_GET['iphone'] ) {

$news_info['cont'] = preg_replace('/\d+px/',"50px", $news_info['cont'] );

$news_info['cont'] = preg_replace('/font-size[^;]+/i','',$news_info['cont']);

$news_info['cont'] = ''.$news_info['cont'];

}else {

$news_info['cont'] = preg_replace('/\d+px/',"24px", $news_info['cont'] );

$news_info['cont'] = preg_replace('/font-size[^;]+/i','',$news_info['cont']);

$news_info['cont'] = ''.$news_info['cont'];

}

$news_info['cont'] = preg_replace('/|

$news_info['cont'] = preg_replace('/<\/iframe>|]+?)>/','',$news_info['cont']);

$news_info['cont'] = preg_replace('/<\/meta>|]+?)>/','',$news_info['cont']);

//替换掉图片，仅仅保留图片src

             $news_info['cont'] = preg_replace('/]+?)src="([^>]+?)"([^>]+?)>/','',$news_info['cont']);

//$news_info['cont'] = preg_replace('/<\/meta>|]+?)>/','',$news_info['cont']);

//$news_info['cont'] = str_ireplace('brush:js;toolbar:false','', $news_info['cont']);

$news_info['cont'] .= "

window.onload = function() {

var w = window.innerWidth || document.documentElement.clientWidth || document.body.clientWidth;

var imgs = document.querySelectorAll('img');if(imgs.length > 0) {for(var i in imgs) {if(imgs[i].width > w) {imgs[i].width = w-20;}}}

var frame = document.querySelectorAll('iframe');if(frame.length > 0) {for(var i in frame) {if(frame[i].width > w) {frame[i].width = w-20;}}}}

";

$news_info['cont'] = preg_replace('/(\\t){2,}/',"\t",$news_info['cont']);

$news_info['cont'] = preg_replace('/(\\n){2,}/',"\n",$news_info['cont']);

$news_info['cont'] = str_ireplace('

$news_info['cont'] = str_ireplace('data-lazyload="','src="',$news_info['cont']);