/**
* @inheritdoc
*/
protected function loadDom()
{
if (!$this->reply_url) {
preg_match('/(\\w+)\\/(\\d+)\\.html/', $this->url, $matches);
if (!isset($matches[1])) {
throw new ObjectException('Can\'t match reply URL.');
}
$this->object_id = (int) $matches[2];
$this->reply_url = 'http://' . parse_url($this->url, PHP_URL_HOST) . '/reply/lax/' . $matches[1] . '/' . $this->object_id;
}
$curl = GlabsController::$curl;
$curl::$referer = $this->url;
try {
self::$dom->loadFromUrl($this->reply_url, [], $curl);
} catch (CurlException $e) {
if (false !== strpos($e->getMessage(), 'timed out')) {
GlabsController::showMessage(' ...trying again', false);
return $this->loadDom();
}
throw new ObjectException($e->getMessage());
} catch (EmptyCollectionException $e) {
throw new ObjectException($e->getMessage());
}
return true;
}
/**
* @inheritdoc
*/
protected function collectObjects($url)
{
if (!array_key_exists($url, $this->collectedCount)) {
$this->collectedCount[$url] = 0;
}
$host = 'http://' . parse_url($url, PHP_URL_HOST);
$dom = new Dom();
try {
$dom->loadFromUrl($url, [], GlabsController::$curl);
} catch (CurlException $e) {
if (false === strpos($e->getMessage(), 'timed out')) {
throw new CurlException($e->getMessage());
}
GlabsController::showMessage(' ...trying again', false);
return $this->collectObjects($url);
}
if (false !== strpos($dom, 'This IP has been automatically blocked.')) {
throw new CurlException('IP has been blocked.');
}
// end collect. no results
if ($dom->find('#moon')[0]) {
return true;
}
$this->checkTotalObjects($dom);
/* @var \PHPHtmlParser\Dom\AbstractNode $span */
foreach ($dom->find('.txt') as $span) {
if ($this->isEnoughCollect()) {
break;
}
/* @var \PHPHtmlParser\Dom\AbstractNode $link */
if ($link = $span->find('a')[0]) {
$href = $this->checkObjectLink($host, $link->getAttribute('href'));
if (false === $href) {
continue;
}
$title = $link->text() ?: strip_tags($link->innerHtml());
try {
$object = $this->getObjectModel($url, $href, $title, $this->categoryId, $this->type);
$object->setPrice($span);
} catch (ObjectException $e) {
continue;
}
$this->collected[] = $href;
$this->objects[] = $object;
$this->collectedCount[$url]++;
BaseSite::$doneObjects++;
BaseSite::progress();
}
}
if (!$this->isEnoughCollect()) {
$curl = GlabsController::$curl;
$curl::$referer = $url;
$url = str_replace([self::$pageParam . self::$page, '#list'], '', $url);
self::$page += 100;
return $this->collectObjects($this->getPagedUrl($url));
}
return true;
}
/**
* @inheritdoc
*/
protected function collectObjects($url)
{
if (!array_key_exists($url, $this->collectedCount)) {
$this->collectedCount[$url] = 0;
}
$dom = new Dom();
try {
$dom->loadFromUrl($url, [], GlabsController::$curl);
} catch (CurlException $e) {
if (false === strpos($e->getMessage(), 'timed out')) {
throw new CurlException($e->getMessage());
}
if (false === strpos($e->getMessage(), '525')) {
throw new CurlException($e->getMessage());
}
GlabsController::showMessage(' ...trying again', false);
return $this->collectObjects($url);
}
// end collect. no results
if (false !== strpos($dom, 'No matches found.')) {
return true;
}
$this->checkTotalObjects($dom);
/* @var \PHPHtmlParser\Dom\AbstractNode $span */
foreach ($dom->find('.summaryHeader') as $span) {
if ($this->isEnoughCollect()) {
break;
}
/* @var \PHPHtmlParser\Dom\AbstractNode $link */
if ($link = $span->find('a', 0)) {
$href = $link->getAttribute('href');
if (in_array($href, $this->collected, true)) {
continue;
}
$object = new Object($url, $href, $link->text(), $this->categoryId, $this->type);
try {
$object->setPrice();
} catch (ObjectException $e) {
continue;
}
$this->collected[] = $href;
$this->objects[] = $object;
$this->collectedCount[$url]++;
BaseSite::$doneObjects++;
BaseSite::progress();
}
}
if (!$this->isEnoughCollect()) {
$curl = GlabsController::$curl;
$curl::$referer = $url;
$url = str_replace(self::$pageParam . self::$page, '', $url);
self::$page += self::$page ? 1 : 2;
return $this->collectObjects($this->getPagedUrl($url));
}
return true;
}
/**
* @inheritdoc
*/
protected function collectObjects($url)
{
if (!array_key_exists($url, $this->collectedCount)) {
$this->collectedCount[$url] = 0;
}
$dom = new Dom();
try {
$dom->loadFromUrl($url, [], GlabsController::$curl);
} catch (CurlException $e) {
if (false === strpos($e->getMessage(), 'timed out')) {
throw new CurlException($e->getMessage());
}
if (false === strpos($e->getMessage(), '525')) {
throw new CurlException($e->getMessage());
}
GlabsController::showMessage(' ...trying again', false);
return $this->collectObjects($url);
}
// end collect. no results
if (false !== strpos($dom, 'No matches found') || false !== strpos($dom, 'Keine Entsprechungen gefunden') || false !== strpos($dom, 'No hay resultados') || false !== strpos($dom, 'Nessuna corrispondenza trovata') || false !== strpos($dom, 'Aucune correspondance n’a été trouvée') || false !== strpos($dom, 'Nenhuma correspondência encontrada') || false !== strpos($dom, 'Совпадений нет') || false !== strpos($dom, 'Ingen match fundet') || false !== strpos($dom, 'Nebyly nalezeny žádné shody') || false !== strpos($dom, 'Ingen match funnet') || false !== strpos($dom, 'Nie znaleziono') || false !== strpos($dom, 'Eşleşme bulunamadı') || false !== strpos($dom, 'Eredmény nem található') || false !== strpos($dom, 'Δεν βρέθηκαν εγγραφές') || false !== strpos($dom, 'Aucune correspondance n’a été trouvée')) {
return true;
}
$this->checkTotalObjects($dom);
/* @var \PHPHtmlParser\Dom\AbstractNode $span */
foreach ($dom->find('.cat') as $span) {
if ($this->isEnoughCollect()) {
break;
}
/* @var \PHPHtmlParser\Dom\AbstractNode $link */
if ($link = $span->find('a', 0)) {
$href = $link->getAttribute('href');
if (in_array($href, $this->collected, true)) {
continue;
}
try {
$object = new Object($url, $href, $link->text(), $this->categoryId, $this->type);
} catch (ObjectException $e) {
continue;
}
$this->collected[] = $href;
$this->objects[] = $object;
$this->collectedCount[$url]++;
BaseSite::$doneObjects++;
BaseSite::progress();
}
}
if (!$this->isEnoughCollect()) {
$curl = GlabsController::$curl;
$curl::$referer = $url;
$url = str_replace(self::$pageParam . self::$page, '', $url);
self::$page += self::$page ? 1 : 2;
return $this->collectObjects($this->getPagedUrl($url));
}
return true;
}
/**
* Parse category page.
*
* @throws CurlException
* @throws InvalidParamException
* @throws ObjectException
*/
public function parse()
{
GlabsController::showMessage("\n" . 'Parsing category "' . $this->title . '"');
/** @var \app\models\glabs\objects\massmail\Craigslist $object */
foreach ($this->objects as $object) {
if (in_array($object->getUrl(), $this->doneObjects, true)) {
continue;
}
$this->i++;
GlabsController::showMessage("\t" . $this->i . ') Parsing object "' . $object->getTitle() . '" (' . $object->getUrl() . ')');
try {
$object->parse();
$this->doneObjects[] = $object->getUrl();
} catch (ObjectException $e) {
GlabsController::showMessage("\t\t" . 'Object skipped because of reason: ' . $e->getMessage());
continue;
} catch (EmptyCollectionException $e) {
GlabsController::showMessage("\t\t" . 'Object skipped because of reason: ' . $e->getMessage());
continue;
}
GlabsController::showMessage("\t\t" . 'Sending object... ', false);
try {
//$object->send();
GlabsController::$sentObjects++;
GlabsController::showMessage('Success.');
} catch (TransportException $e) {
$object->removeFiles();
GlabsController::showMessage('Fail with message: "' . $e->getMessage() . '"');
}
/* @var \app\models\glabs\objects\massmail\Craigslist $object */
GlabsController::saveMassmailLinks($object);
}
$done_count = count($this->doneObjects);
if ($done_count < $this->needCount && count($this->objects)) {
$this->count = $this->needCount - $done_count;
$this->objects = [];
$this->collectObjects($this->getPagedUrl(reset($this->url)));
$this->parse();
}
}
/**
* Show progress bar.
*/
public static function progress()
{
GlabsController::showMessage(sprintf(self::$progressFormat, self::$doneCategories, self::$doneObjects), false);
}
/**
* Send object to Chatapp.mobi
*
* @param bool $isTest
*
* @return bool
*
* @throws TransportException
* @throws InvalidParamException
* @throws ImageException
*/
public function send($isTest = false)
{
try {
(new TransportChatapp($this))->send($isTest);
} catch (TransportException $e) {
$message = $e->getMessage();
if ($message === 'Error retrieving: Username ' . $this->username . ' already taken') {
GlabsController::showMessage('re-generate username and trying again... ', false);
$this->setUsername();
return $this->send($isTest);
} elseif ($message === 'Error retrieving: Email "' . $this->email . '" already exist.') {
GlabsController::showMessage('re-generate email and trying again... ', false);
$this->setEmail();
return $this->send($isTest);
} else {
throw new TransportException($message);
}
}
return true;
}
/**
* Load DOM.
*
* @return bool
*
* @throws ObjectException
*/
protected function loadDom()
{
try {
$curl = GlabsController::$curl;
$curl::$referer = $this->categoryUrl;
self::$dom->loadFromUrl($this->url, [], GlabsController::$curl);
} catch (CurlException $e) {
if (false !== strpos($e->getMessage(), 'timed out')) {
GlabsController::showMessage(' ...trying again', false);
return $this->loadDom();
}
throw new ObjectException($e->getMessage());
} catch (EmptyCollectionException $e) {
throw new ObjectException($e->getMessage());
}
return true;
}
