function findArticlePosById($articles, $id)
{
foreach ($articles as $pos => $article) {
$thisId = get_article_id($article);
if ($thisId == $id) {
return $pos;
}
}
}
$nf_text[] = "<strong>Category:</strong> <a href=\"" . WW_WEB_ROOT . '/' . $_GET['category_url'] . "/\">" . $_GET['category_url'] . "</a> was found!";
}
}
if (isset($_GET['author_url'])) {
if (!isset($_GET['author_id'])) {
$nf_text[] = "<strong>Author:</strong> " . $_GET['author_url'] . " - not found";
} else {
$nf_text[] = "<strong>Author:</strong> <a href=\"" . WW_WEB_ROOT . '/author/' . $_GET['author_url'] . "/\">" . $_GET['author_url'] . "</a> was found!";
}
}
if (isset($_GET['tag_url'])) {
if (!isset($_GET['tag_id'])) {
$nf_text[] = "<strong>Tag:</strong> " . $_GET['tag_url'] . " - not found";
} else {
$nf_text[] = "<strong>Tag:</strong> <a href=\"" . WW_WEB_ROOT . '/tag/' . $_GET['tag_url'] . "/\">" . $_GET['tag_url'] . "</a> was found!";
}
}
if (isset($_GET['article_url'])) {
$article_id = get_article_id();
if (empty($article_id)) {
$nf_text[] = "<strong>Article:</strong> " . $_GET['article_url'] . " - not found";
} else {
$nf_text[] = "<strong>Article:</strong> <a href=\"" . WW_WEB_ROOT . '/id/' . $article_id . "/\">" . $_GET['article_url'] . "</a> was found!";
}
}
// build content
echo "\n\t<h1>404 error</h1>\n\t<h4>Based on the url you entered we tried to find the following:</h4>\n\t<ul>";
foreach ($nf_text as $nf) {
echo "\n\t\t<li>" . $nf . "</li>";
}
echo "\n\t</ul>";
/**
* 采集互联网上的文章，用于学习和交流
* @author suhanyu<*****@*****.**>
*/
define('PHP_QUERY', '/www/8hao/html/tool/QueryList/');
include PHP_QUERY . 'QueryList.class.php';
require PHP_QUERY . '../include/connector.php';
//获取左岸的文章列表标题
$url = 'http://www.zreading.cn/';
$hj = QueryList::Query($url, array("title" => array('#content .entry-name a', 'text'), "url" => array('#content .entry-name', 'text', 'a', 'getUrl'), 'tag' => array('#content .entry-meta', 'html', 'a'), 'category' => array('#content .entry-meta', 'html', 'a', 'getCategory')));
$nowDate = time();
if ($hj->jsonArr) {
foreach ($hj->jsonArr as $key => $value) {
$data = array();
$data['sourceId'] = get_article_id($value['url']);
# 如果存在，则不执行添加操作
$res1 = $arti->select('select "X" from uctoo_article_article where sourceId=' . $data['sourceId']);
if ($res1) {
continue;
}
$data['title'] = $value['title'];
$data['source'] = $value['url'];
$data['sourceType'] = 2;
// 2表示来源是互联网
$data['classId'] = 3;
// 3表示左岸频道
$data['category'] = 1;
// 1表示创造之路
$data['isPublish'] = 1;
$data['publishTime'] = $nowDate;
alert('出错的URL是：' . $list_url . $i);
} else {
curl_close($list_handle);
alert("第 {$i} 列表页html内容获取成功！");
}
$html = str_get_html($pageContent);
$thread_lists = $html->find('#container');
$urlPrev = 'http://yu.pk1024.com/bbs/';
foreach ($thread_lists as $threadTable) {
$dataArr = array();
// 标题和链接地址 find('tr .f_title a[href^="thread"]')
foreach ($threadTable->find('#content header h2 a') as $url) {
$title = trim($url->innertext);
#$data_url = "http://yu.pk1024.com/bbs/" . $url->href;
$detailUrl = trim($url->href);
$sourceId = get_article_id($detailUrl);
$addTime = date('Y-m-d H:i:s');
if (check_exists_data($sourceId)) {
echo "{$sourceId} 存在 \r\n";
continue;
}
$dataArr[] = '("' . $detailUrl . '","' . $sourceId . '","' . $addTime . '")';
}
$partSql = implode(',', $dataArr);
# 没有数据 则跳过
if (!$dataArr) {
continue;
}
try {
// $stmt = $pdo->prepare("INSERT INTO list(
// detail_url,source_id,add_time)
<title>Blog</title>
<link href="/style.css" rel="stylesheet" type="text/css">
</head>
<body>
<h1>My Blog</h1><?php
if (count($articles) > 0) {
?>
<h2>List of Articles</h2>
<div>
<ul>
<?php
foreach ($articles as $article) {
?>
<li><a href="#<?php
echo get_article_id($article);
?>
"><?php
echo $article['title'];
?>
</a></li>
<?php
}
?>
</ul>
</div>
<div><?php
}
foreach ($articles as $article1) {
echoArticle($article1);
}
/**
* process request
*
* this function is called at the top of the main index page
* it processes the GET parameters created by process_url()
* and determines which data should be retrieved, creating additional
* parameters if necessary for specific datasets
*/
function process_request()
{
// search request - redirect to keep tidy url
if (isset($_POST['search'])) {
$search_redirect = WW_WEB_ROOT . '/search/' . $_POST['search'] . '/';
header('HTTP/1.1 302 Moved Temporarily');
header('Location: ' . $search_redirect);
}
// redirect jump links from select boxes
if (isset($_POST['select_link'])) {
$select_redirect = $_POST['select_link'];
header('HTTP/1.1 302 Moved Temporarily');
header('Location: ' . $select_redirect);
}
// now get urldata
$urldata = process_url();
// author url
if (isset($_GET['author_url']) && !empty($_GET['author_url'])) {
$author_id = get_author_id($_GET['author_url']);
if (!empty($author_id)) {
$_GET['author_id'] = $author_id;
} else {
$_GET['page_name'] = '404';
}
}
// category url
if (isset($_GET['category_url']) && !empty($_GET['category_url'])) {
$category_id = get_category_id($_GET['category_url']);
if (!empty($category_id)) {
$_GET['category_id'] = $category_id;
} else {
$_GET['page_name'] = '404';
}
}
// tag url
if (isset($_GET['tag_url']) && !empty($_GET['tag_url'])) {
$tag_id = get_tag_id($_GET['tag_url']);
if (!empty($tag_id)) {
$_GET['tag_id'] = $tag_id;
} else {
$_GET['page_name'] = '404';
}
}
// have we already 404ed?
if (isset($_GET['page_name']) && $_GET['page_name'] == '404') {
show_404($urldata);
return false;
}
// start getting data
// was a single article requested?
if (isset($_GET['article_id']) && !empty($_GET['article_id']) || isset($_GET['article_url']) && !empty($_GET['article_url'])) {
$article_id = get_article_id();
if (!empty($article_id)) {
// n.b. page_name could be 'feed' if looking for article comments rss
$_GET['page_name'] = isset($_GET['page_name']) ? $_GET['page_name'] : 'article';
$_GET['article_id'] = $article_id;
return true;
} else {
show_404($urldata);
return false;
}
// or a search term or feed request
} elseif (isset($_GET['feed']) || isset($_GET['feed_listing'])) {
return true;
// or did we get a valid author, category, or tag id - or a search, or date request?
} elseif (!empty($author_id) || !empty($category_id) || !empty($tag_id) || isset($_GET['year']) || isset($_GET['search'])) {
$_GET['page_name'] = 'listing';
return true;
// if anything else has been requested we sure can't find it
} elseif (!empty($urldata)) {
show_404($urldata);
return false;
}
// default
$_GET['page_name'] = 'front';
return false;
}
