getHostPagesByIndexed() as $hostPage) {
if (false !== stripos(Filter::mime($hostPage->mime), 'text/html')) {
if ($hostPageDescription = $db->getLastPageDescription($hostPage->hostPageId)) {
$hostPagesProcessedTotal++;
if (!empty($hostPageDescription->data)) {
$html = str_get_html(base64_decode($hostPageDescription->data));
foreach ((array) explode(',', CRAWL_HOST_PAGE_DOM_SELECTORS) as $selector) {
foreach($html->find($selector) as $element) {
if (!empty($element->innertext)) {
$hostPageDOMAddedTotal++;
$db->addHostPageDom($hostPage->hostPageId,
time(),
$selector,
trim(CRAWL_HOST_PAGE_DOM_STRIP_TAGS ? strip_tags(
preg_replace('/[\s]+/',
' ',
str_replace(['
', '
', '
', 'innertext))) : $element->innertext));
}
}
}
}
}
}
}
echo sprintf(_('Host pages processed: %s'), $hostPagesProcessedTotal) . PHP_EOL;
echo sprintf(_('Host page DOM elements added: %s'), $hostPageDOMAddedTotal) . PHP_EOL;
exit;
}
echo PHP_EOL . _('CRAWL_HOST_PAGE_DOM_SELECTORS not provided in the configuration file') . PHP_EOL;
exit;
break;
case 'truncate':
$db->truncateHostPageDom();
echo _('hostPageDom table successfully truncated') . PHP_EOL;
exit;
break;
default:
echo PHP_EOL . _('undefined action argument') . PHP_EOL;
}
break;
}
// Default message
echo '__ ______________ __' . PHP_EOL;
echo '\ \/ / ____/ ____/___ / /' . PHP_EOL;
echo ' \ / / __/ / __/ __ \/ /' . PHP_EOL;
echo ' / / /_/ / /_/ / /_/ /_/' . PHP_EOL;
echo '/_/\____/\____/\____(_)' . PHP_EOL;
echo PHP_EOL . _('available options:') . PHP_EOL . PHP_EOL;
echo _(' help - this message') . PHP_EOL;
echo _(' hostPageDom generate - make hostPageDom index based on related hostPage.data field') . PHP_EOL;
echo _(' hostPageDom truncate - flush hostPageDom table') . PHP_EOL . PHP_EOL;
echo _('get support: https://github.com/YGGverse/YGGo/issues') . PHP_EOL . PHP_EOL;