Loads the dom from a document file/url
| public loadFromFile ( string $file, array $options = [] ) | ||
| $file | string | |
| $options | array | |
private function getAndParseSubteInfo()
{
if ($cachedData = $this->isCached()) {
$this->lineas = $cachedData;
return true;
}
$dom = new Dom();
$dom->loadFromFile($this->sourceURL);
foreach ($this->lineas as $linea => $info) {
$lineInfo = $dom->find("#status-line-{$linea}-container")[0];
$lineStatusClass = $lineInfo->getAttribute('class');
if (strpos($lineStatusClass, 'suspendido') !== false) {
$this->lineas[$linea]->status = 'CANCELLED';
}
if (strpos($lineStatusClass, 'demorado') !== false) {
$this->lineas[$linea]->status = 'DELAYED';
}
if ($this->isSleepingTime()) {
$this->lineas[$linea]->status = 'SLEEPING';
}
// get raw status msg
$status_msg = $lineInfo->find("#status-line-{$linea}")->text;
$this->lineas[$linea]->statusMessage = html_entity_decode($status_msg, ENT_QUOTES, 'ISO-8859-1');
}
$this->updateStatusInfo();
$this->cacheLines();
}
/**
* Parses the given file and returns a Game object
*
* @param string $filename
*
* @return Game
*/
protected function processFile($filename)
{
$this->command->out("Processing " . $filename);
// Create a game object with home/away teams and other info
$game = $this->createGameWithInfo($filename);
$dom = new Dom();
$dom->loadFromFile($filename);
$lines = [];
/** @var AbstractNode $tr */
foreach ($dom->find('tr.evenColor') as $tr) {
$lineContent = [];
$lineCount = 0;
/** @var AbstractNode $td */
foreach ($tr->getChildren() as $td) {
$value = $this->cleanUpLine($td->text);
if ($value) {
$lineCount++;
// Each event is actually 6 lines
$lineContent[] = $value;
if ($lineCount % 6 == 0) {
$lines[] = $lineContent;
$lineContent = [];
}
}
}
}
// Add each event line to the game log
foreach ($lines as $line) {
if ($event = $this->createParsedEvent($line)) {
$game->addEvent($event);
}
}
return $game;
}
public function addPhoto($session, File $file, $number)
{
$path = $session . '.html';
$update = $this->client->api('repo')->contents()->exists($this->user, $this->repo, $path, $this->branch);
if ($update) {
$set = new \PHPHtmlParser\Dom();
$set->load($this->client->api('repo')->contents()->download($this->user, $this->repo, $path, $this->branch));
$info = $this->client->api('repo')->contents()->show($this->user, $this->repo, $path, $this->branch);
} else {
$set = new \PHPHtmlParser\Dom();
$set->loadFromFile($this->page);
}
$div = $set->find('#photos')[0];
$photo = new \PHPHtmlParser\Dom();
$photo->loadFromFile($this->photo);
$img = $photo->find('img')[0];
$img->setAttribute('src', $this->getPhotoFilename($file));
$div->addChild($photo->root);
$images = $div->find('img');
$count = count($images);
if ($count <= 4) {
$meta = new Dom();
$meta->load('<meta name="twitter:image' . --$count . '" content="' . $this->web . $this->getPhotoFilename($file, 'th') . '">');
$set->find('head', 0)->addChild($meta->root);
}
$content = \Mihaeu\HtmlFormatter::format((string) $set);
$content = preg_replace("#\n\\s*\n#", "\n", $content);
if ($update) {
$response = $this->client->api('repo')->contents()->update($this->user, $this->repo, $path, $content, 'Adding Photo ' . PHP_EOL . $file, $info['sha'], $this->branch);
} else {
$response = $this->client->api('repo')->contents()->create($this->user, $this->repo, $path, $content, 'Adding Page and Photo ' . PHP_EOL . $file, $this->branch);
$this->addSession($session, $number, $file);
}
}
public function testRemoveScriptsFalse()
{
$dom = new Dom();
$dom->setOptions(['removeScripts' => false]);
$dom->loadFromFile('tests/files/horrible.html');
$this->assertEquals(1, count($dom->find('script')));
$this->assertEquals('text/JavaScript', $dom->find('script')->getAttribute('type'));
}
/**
* Creates a new dom object and calls loadFromFile() on the
* new object.
* @param string $file
* @return $this
*/
public static function loadFromFile($file)
{
$dom = new Dom();
self::$dom = $dom;
return $dom->loadFromFile($file);
}
public function testLoadFileBigTwicePreserveOption()
{
$dom = new Dom();
$dom->loadFromFile('tests/files/big.html', ['preserveLineBreaks' => true]);
$post = $dom->find('.post-row', 0);
$this->assertEquals('<p>Журчанье воды<br />
Черно-белые тени<br />
Вновь на фонтане</p>', trim($post->find('.post-message', 0)->innerHtml));
}
public function testLoadFileBigTwice()
{
$dom = new Dom();
$dom->loadFromFile('tests/files/big.html');
$post = $dom->find('.post-row', 0);
$this->assertEquals(' <p>Журчанье воды<br /> Черно-белые тени<br /> Вновь на фонтане</p> ', $post->find('.post-message', 0)->innerHtml);
}
include "vendor/autoload.php";
use PHPHtmlParser\Dom;
?>
<!doctype html>
<html lang="ru">
<head>
<meta charset="UTF-8">
<title>Таблица</title>
</head>
<body>
<table>
<?php
$dom = new Dom();
$dom->loadFromFile('index.html');
for ($i = 0; $i < 100; $i++) {
$contents = $dom->find("tr td")[$i];
if ($contents->text == "Адрес") {
$i += 5;
continue;
}
if (preg_match("|^[\\d]+\$|", $contents->text)) {
echo "<tr><td>" . $contents->text . "</td></tr>";
}
echo "<br/>";
echo $contents->text;
}
?>
</table>
</body>
private function collectSites()
{
$old_data = json_decode(file_get_contents(Yii::getAlias('@runtime/data/chatapp.json')), true);
$dom = new Dom();
$dom->loadFromFile(Yii::getAlias('@runtime/sites.html'));
$clinks = $blinks = [];
$exclude = array_key_exists('exclude', $old_data) ? $old_data['exclude'] : ['auburn.craigslist.org', 'bham.craigslist.org', 'dothan.craigslist.org', 'shoals.craigslist.org', 'gadsden.craigslist.org', 'huntsville.craigslist.org'];
/* @var Dom\AbstractNode $link */
foreach ($dom->find('a') as $link) {
$href = $link->getAttribute('href');
if (0 !== strpos($href, '//')) {
continue;
}
$href = str_replace('/', '', $href);
if (in_array($href, $exclude, true)) {
continue;
}
$clinks[] = $href;
}
shuffle($clinks);
$dom->loadFromFile(Yii::getAlias('@runtime/backpage.html'));
/* @var Dom\AbstractNode $link */
foreach ($dom->find('a') as $link) {
$href = $link->getAttribute('href');
$href = str_replace(['http:', '/'], '', $href);
if (in_array($href, $exclude, true)) {
continue;
}
$blinks[] = $href;
}
shuffle($blinks);
$data = ['total_count' => (int) $old_data['total_count'], 'current_site' => '', 'sites' => array_merge($clinks, $blinks), 'exclude' => $exclude];
file_put_contents(Yii::getAlias('@runtime/data/chatapp.json'), json_encode($data));
}
