function cleanse_urls($str, $article_url)
{
$html = SimpleHtmlDom\str_get_html($str);
foreach ($html->find('a') as $element) {
// Fix cross posts, especially those from LMMSChallenge facebook
if (!$element->find('img') && is_feed_image($element)) {
$element->innertext = '<img class="img-thumbnail fb-thumb" src="' . $element->href . '"/>';
$element->href = $article_url;
}
// Fix unnecessary facebook redirects
$pos = strpos(strtolower($element->href), '.php?u=http%3a%2f%2f') + strpos(strtolower($element->href), '.php?u=https%3a%2f%2f');
if ($pos) {
// Remove js callbacks
$element->onmouseover = '';
$element->onclick = '';
// Strip out the facebook params
$href = explode('&', $element->href)[0];
// Isolate and decode the proper url
$element->href = urldecode(substr($href, $pos + 7));
}
// Fix relative facebook URLs
if (str_startswith($element->href, '/')) {
$element->href = 'https://www.facebook.com' . $element->href;
}
}
foreach ($html->find('img') as $img) {
$img->class = ($img->class ? $img->class . ' ' : '') . 'img-thumbnail fb-thumb';
}
return $html->save();
}
function cleanse_html($atom, $href)
{
$html = SimpleHtmlDom\str_get_html($atom);
foreach ($html->find('img') as $element) {
// Skip smilies
if (str_contains($element->src, '/smilies/')) {
continue;
}
$class = 'img img-thumbnail forum-thumb';
$element->outertext = '<a href="' . $href . '"><img class="' . $class . '" src="' . scale_image($element->src, 200) . '"></a>';
}
foreach ($html->find('p') as $element) {
if ($element->class == 'dummy') {
$element->outertext = '<small class="feed-small">' . $element->innertext . '</small>';
}
}
$html->save;
return $html;
}
public function sendAction($request)
{
$account = $this->repo('Account')->find($request->get('account'));
if (!$account) {
return $this->responseAJAX(['status' => 'error', 'message' => 'Brak skonfigurowanego konta pocztowego, z którego można wysłać wiadomość.']);
}
$this->repo('Account')->decryptPasswords($account);
$this->completeAccountWithPassword($account);
try {
$to = new EmailAddressParser($request->get('to'));
$cc = new EmailAddressParser($request->get('cc'));
$bcc = new EmailAddressParser($request->get('bcc'));
$message = \Swift_Message::newInstance();
$content = $request->get('content');
$contentObj = \SimpleHtmlDom\str_get_html($content);
if ($request->get('originalMessageId')) {
$originalMessage = $this->openBoxFromRequest($request)->readMessage($request->get('originalMessageId'), MailReader::READ_BODY_ATTACHMENTS);
if ($originalMessage) {
foreach ($originalMessage->attachments as $attachment) {
/**
* Add attachment to new message, and retrive it's ID.
*/
$newAttId = $message->embed(\Swift_Image::fromPath($attachment->filePath));
/**
* Search for elements with given attribute and given ID of attachment.
*/
$elements = $contentObj->find('*[data-attid="cid:' . $attachment->id . '"]');
foreach ($elements as $element) {
/**
* Few lines abowe we add attachment to new mail, so we save it's id
* in searched src of image (inline attachment). Old SRC we replace by
* new attachment's ID.
*/
$element->src = $newAttId;
/**
* We only add this attribute to tage the ID of attachment in original message.
* Now we only remove it, couse we don't need it anymore.
*/
$element->removeAttribute('data-attid');
}
}
}
}
$content = $contentObj->save();
$message->setSubject($request->get('subject'))->setFrom(array($account->getSmtpUsername() => $account->getSenderName()))->setTo($to->getAsPairs())->setBody($content, 'text/html');
if ($cc->hasAny()) {
$message->setCc($cc->getAsPairs());
}
if ($bcc->hasAny()) {
$message->setBcc($bcc->getAsPairs());
}
$attachmentsDir = BASEPATH . '/app/Cache/outgoing-attachments/' . $this->request()->getSession()->getId() . '/' . $request->request->get('windowId');
if (is_dir($attachmentsDir)) {
foreach (new \DirectoryIterator($attachmentsDir) as $fileInfo) {
if ($fileInfo->isDot() || $fileInfo->isDir()) {
continue;
}
$message->attach(\Swift_Attachment::fromPath($fileInfo->getPathname()));
}
}
$transport = \Swift_SmtpTransport::newInstance($account->getSmtpHost(), $account->getSmtpPort(), $account->getSmtpSecurity())->setUsername($account->getSmtpUsername())->setPassword($account->getSmtpPassword());
$mailer = \Swift_Mailer::newInstance($transport);
// Send the message
$result = $mailer->send($message);
$this->openBoxFromRequest($request)->appendMessage($message->toString(), $this->imap($account)->getSentBoxName(), '\\Seen');
// Save stat
if ($this->openSettings('app')->get('mod.mail.stat.savesentmailsinfo') == 1) {
$stat = new SentMail();
$stat->setUserId($this->user()->getId());
$stat->setDate(time());
$this->repo('SentMail')->save($stat);
}
} catch (\Exception $e) {
return $this->responseAJAX(['status' => 'error', 'message' => $e->getMessage()]);
}
return $this->responseAJAX(['status' => $result ? 'success' : 'error', 'message' => 'Wiadomość została wysłana.']);
}
// 初始化curl
$list_handle = curl_init();
// curl配置参数
$options = [CURLOPT_URL => $list_url . $i, CURLOPT_RETURNTRANSFER => true, CURLOPT_USERAGENT => $cmd->rand_user_agent()];
curl_setopt_array($list_handle, $options);
// 执行curl
$pageContent = curl_exec($list_handle);
if (curl_errno($list_handle)) {
$cmd->beep();
$cmd->alert('curl出错:' . curl_error($list_handle));
$cmd->alert('出错的URL是：' . $list_url . $i);
} else {
curl_close($list_handle);
$cmd->alert("第 {$i} 列表页html内容获取成功！");
}
$html = SimpleHtmlDom\str_get_html($pageContent);
$thread_lists = $html->find('tr[class=tr3 t_one]');
foreach ($thread_lists as $thread_tr) {
// 标题和链接地址
foreach ($thread_tr->find('td h3 a') as $url) {
$title = iconv('gb2312', 'utf-8//IGNORE', $url->innertext);
$url = "http://t66y.com/" . $url->href;
}
// 回复数量
foreach ($thread_tr->find('td[class=tal f10 y-style]') as $comment) {
$comments = $comment->innertext;
}
// 作者
foreach ($thread_tr->find('a[class=bl]') as $author) {
$author = iconv('gb2312', 'utf-8', $author->innertext);
}
for ($j = $key_start; $j < $key_end; $j++) {
// 初始化curl
$thread_handle = curl_init();
// curl配置参数
$options = [CURLOPT_URL => $threads[$j]['url'], CURLOPT_RETURNTRANSFER => true, CURLOPT_USERAGENT => $cmd->rand_user_agent()];
curl_setopt_array($thread_handle, $options);
// 执行curl
$thread_content = curl_exec($thread_handle);
if (curl_errno($thread_handle)) {
$cmd->beep();
$cmd->alert('curl出错:' . curl_error($thread_handle));
$cmd->alert('出错的URL是：' . $threads[$j]['url']);
} else {
curl_close($thread_handle);
}
$html = SimpleHtmlDom\str_get_html($thread_content);
// 查找正文中所有的图片url并写入数据库
if (is_object($html)) {
foreach ($html->find('input') as $element) {
if ($element->src != '') {
$cmd->alert($element->src);
// 写入数据库
try {
$pdo = new PDO('mysql:host=127.0.0.1;dbname=caoliu;charset=utf8', 'root', '881224');
$pdo->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
$stmt = $pdo->prepare("insert into test_img_url (thread_id, url) VALUES (:thread_id, :url)");
$stmt->bindParam(':thread_id', $threads[$j]['id']);
$stmt->bindParam(':url', $element->src);
$stmt->execute();
} catch (PDOException $error) {
$cmd->alert($error->getMessage());
}
?>
<h1>标题关键词排名前20</h1>
<?php
try {
$stmt = $pdo->query("select title from thread");
$result = $stmt->fetchAll();
} catch (PDOException $error) {
$cmd->alert($error->getMessage());
}
// 将标题组装成一个回车符分隔的字符串
foreach ($result as $value) {
// 先去掉标题中类似 [10p] 的部分，每行都有，代表帖子的图片数量
$title = preg_replace("/[0-9][0-9]P/", ' ', $value['title']);
$text .= SimpleHtmlDom\str_get_html($title)->plaintext . "\n";
}
$cws = new PSCWS4();
$cws->set_charset('utf8');
// 编码
$cws->set_dict(ROOT_PATH . '/vendor/scws/pscws4/dict/dict.utf8.xdb');
// 加载字典文件
$cws->set_rule(ROOT_PATH . '/vendor/scws/pscws4/etc/rules.ini');
// 人名地名规则
//$cws->set_multi(3);
$cws->set_ignore(true);
// 忽略标点
//$cws->set_debug(true);
$cws->set_duality(true);
// 对 单字 格外进行二元法匹配
$cws->send_text($text);