WEBAPP新闻客户端

<?php header("Content-Type:text/html;charset=utf-8"); $cb = $_GET['cb']; $site = $_GET['site']; $page = $_GET['page']; $net = $_GET['net']; if ($site == 'sina') { $url = 'http://slide.news.sina.com.cn/w/'.$page.'.html'; } $con = file_get_contents($url); $con = iconv('GB2312', 'UTF-8', $con); if ($site == 'sina') { //匹配数据 $re = '/div class="content" id="picList">(.*)title="last page">最末页<!-- --><\/a>/is'; preg_match($re, $con, $matches); $str = $matches[0]; //获取最大页数 preg_match('/<a[^<>]+href="([^<>]+).html" title="last page">/', $str, $maxpage); $maxpage = $maxpage[1]; if ($page <= $maxpage) { $re = '/div class="content" id="picList">(.*)<div class="page">/is'; preg_match($re, $con, $matches); $str = $matches[0]; //去除标签 //去除开头多余标签 $str = preg_replace('/div class="content" id="picList">/', '', $str); //去除结尾多余标签 $str = preg_replace('/<div class="clearit"><\/div>(\b|\s)*<\/div>(\b|\s)*<div class="page">$/', '', $str); //去除注释 $str = preg_replace('/<!--(.*)-->/i', '', $str); //格式化字符串 //$str = preg_replace('/<div class="picBox([^<>]*)>(\b|\s)*<table[^<>]*>(\b|\s)*<tr[^<>]*>(\b|\s)*<td[^<>]*>(\b|\s)*<a href="([^<>\s\b]*)" (target="_blank"><img src=")([^<>\s\b]*)" class="b1"><\/a><\/td>(\b|\s)*<\/tr>(\b|\s)*<\/table>(\b|\s)*<h2[^<>]*>(\b|\s)*<a([^<>]*)subtitle="([^<>]*)">([^<>]*)<\/a><\/h2>(\b|\s)*<span>(\b|\s)*\((\d{2})(\/)(\d{2})(\b|\s)*(\d+:\d+)\)(\b|\s)*<\/span>(\b|\s)*<\/div>(\b|\s)*/is', '{\'url\': \'\\6\', \'pic\': \'\\8\', \'subtitle\': \'\\14\', \'title\': \'\\15\', \'date\': \'\\18月\\20日 \\22\'}, ', $str); $str = preg_replace('/<div class="picBox([^<>]*)>(\b|\s)*<table[^<>]*>(\b|\s)*<tr[^<>]*>(\b|\s)*<td[^<>]*>(\b|\s)*<a href="([^<>\s\b]*)" (target="_blank"><img src=")([^<>\s\b]*)" class="b1"><\/a><\/td>(\b|\s)*<\/tr>(\b|\s)*<\/table>(\b|\s)*<h2[^<>]*>(\b|\s)*<a([^<>]*)subtitle="([^<>]*)">([^<>]*)<\/a><\/h2>(\b|\s)*<span>(\b|\s)*\((\d{2})(\/)(\d{2})(\b|\s)*(\d+:\d+)\)(\b|\s)*<\/span>(\b|\s)*<\/div>(\b|\s)*/is', '{\'url\': \'\\6\', \'pic\': \'\\8\', \'subtitle\': \'\\14\', \'title\': \'\\15\', \'date\': \'\\18月\\20日 \\22\'}, ', $str); $str = preg_replace('/^(\b|\s)*/i', '{\'code\': 0, \'page\': '.$page.', \'maxpage\': '.$maxpage.', \'data\': [', $str); $str = preg_replace('/(, )$/i', ']}', $str); $str = $cb.'('.$str.')'; } else { $str = $cb.'({\'code\': 1});'; } } echo $str; ?>