}
$item["body"] = HTML::toBBCode($body, $basepath);
+ // Remove tracking pixels
+ $item["body"] = preg_replace("/\[img=1x1\]([^\[\]]*)\[\/img\]/Usi", '', $item["body"]);
+
if (($item["body"] == '') && ($item["title"] != '')) {
$item["body"] = $item["title"];
$item["title"] = '';
$replace = true;
}
+ $saved_body = $item["body"];
+ $saved_title = $item["title"];
+
if ($replace) {
$item["body"] = trim($item["title"]);
}
}
}
+ $data = PageInfo::queryUrl($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_denylist"] ?? '');
+
+ // Take the data that was provided by the feed if the query is empty
+ if (($data['type'] == 'link') && empty($data['title']) && empty($data['text'])) {
+ $data['title'] = $saved_title;
+ $item["body"] = $saved_body;
+ }
+
+ $data_text = strip_tags(trim($data['text'] ?? ''));
+ $item_body = strip_tags(trim($item['body'] ?? ''));
+
+ if (!empty($data_text) && (($data_text == $item_body) || strstr($item_body, $data_text))) {
+ $data['text'] = '';
+ }
+
// We always strip the title since it will be added in the page information
$item["title"] = "";
- $item["body"] = $item["body"] . "\n" . PageInfo::getFooterFromUrl($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_denylist"] ?? '');
+ $item["body"] = $item["body"] . "\n" . PageInfo::getFooterFromData($data, false);
$taglist = $contact["fetch_further_information"] == 2 ? PageInfo::getTagsFromUrl($item["plink"], $preview, $contact["ffi_keyword_denylist"] ?? '') : [];
$item["object-type"] = Activity\ObjectType::BOOKMARK;
unset($item["attach"]);
*/
private static function adjustPollFrequency(array $contact, array $creation_dates)
{
- if (($contact['priority'] > 3) || ($contact['network'] != Protocol::FEED)) {
- Logger::info('Contact is no feed or has a low priority, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'priority' => $contact['priority'], 'network' => $contact['network']]);
+ if ($contact['network'] != Protocol::FEED) {
+ Logger::info('Contact is no feed, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'network' => $contact['network']]);
return;
}
}
}
- if (($newest == $oldest) && count($creation_dates) > 1) {
- Logger::info('Feed has no different creation dates, quitting', ['date' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
- return;
+ if (count($creation_dates) == 1) {
+ Logger::info('Feed had posted a single time, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
+ $priority = 8; // Poll once a day
}
- if (empty($frequency)) {
- Logger::info('Feed had not posted for at least a week, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
- $priority = 3; // Poll once a day
+ if (empty($priority) && (((time() / 86400) - $newest) > 730)) {
+ Logger::info('Feed had not posted for two years, switching to monthly polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
+ $priority = 10; // Poll every month
}
- if (empty($priority) && (count($creation_dates) == 1)) {
- Logger::info('Feed had posted a single time, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
- $priority = 3; // Poll once a day
+ if (empty($priority) && (((time() / 86400) - $newest) > 365)) {
+ Logger::info('Feed had not posted for a year, switching to weekly polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
+ $priority = 9; // Poll every week
+ }
+
+ if (empty($priority) && empty($frequency)) {
+ Logger::info('Feed had not posted for at least a week, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
+ $priority = 8; // Poll once a day
}
if (empty($priority)) {
// Calculate the highest "posts per day" value
- $max = 1;
+ $max = 0;
foreach ($frequency as $entry) {
if (($entry['count'] == 1) || ($entry['high'] == $entry['low'])) {
continue;
$max = $ppd;
}
}
-
- if ($max >= 24) {
- $priority = 0; // Poll with the minimum poll intervall
- } elseif ($max >= 12) {
- $priority = 1; // Poll hourly
- } elseif ($max > 1) {
- $priority = 2; // Poll twice a day
+ if ($max > 48) {
+ $priority = 1; // Poll every quarter hour
+ } elseif ($max > 24) {
+ $priority = 2; // Poll half an hour
+ } elseif ($max > 12) {
+ $priority = 3; // Poll hourly
+ } elseif ($max > 8) {
+ $priority = 4; // Poll every two hours
+ } elseif ($max > 4) {
+ $priority = 5; // Poll every three hours
+ } elseif ($max > 2) {
+ $priority = 6; // Poll every six hours
} else {
- /// @todo In the future we could calculate the days between the posts to set even lower priorities
- $priority = 3; // Poll once a day
+ $priority = 7; // Poll twice a day
}
Logger::info('Calculated priority by the posts per day', ['priority' => $priority, 'max' => round($max, 2), 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
}
} else {
Logger::info('No posts, switching to daily polling', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
- $priority = 3; // Poll once a day
+ $priority = 8; // Poll once a day
}
if ($contact['rating'] != $priority) {
projects / friendica.git / blobdiff