Merge pull request #9022 from annando/adjust-frequency

Automatically adjust feed poll frequencies
This commit is contained in:
Hypolite Petovan 2020-08-17 16:09:44 -04:00 committed by GitHub
commit 11701e2cd3
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 164 additions and 57 deletions

View File

@ -558,7 +558,7 @@ class Contact extends BaseModule
}
$poll_interval = null;
if (in_array($contact['network'], [Protocol::FEED, Protocol::MAIL])) {
if ((($contact['network'] == Protocol::FEED) && !DI::config()->get('system', 'adjust_poll_frequency')) || ($contact['network']== Protocol::MAIL)) {
$poll_interval = ContactSelector::pollInterval($contact['priority'], !$poll_enabled);
}

View File

@ -300,6 +300,7 @@ class Feed
}
$items = [];
$creation_dates = [];
// Limit the number of items that are about to be fetched
$total_items = ($entries->length - 1);
@ -354,16 +355,6 @@ class Feed
$item["parent-uri"] = $item["uri"];
if (!$dryRun) {
$condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
$importer["uid"], $item["uri"], Protocol::FEED, Protocol::DFRN];
$previous = Item::selectFirst(['id'], $condition);
if (DBA::isResult($previous)) {
Logger::info("Item with uri " . $item["uri"] . " for user " . $importer["uid"] . " already existed under id " . $previous["id"]);
continue;
}
}
$item["title"] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry);
if (empty($item["title"])) {
@ -403,6 +394,19 @@ class Feed
$item["edited"] = $updated;
}
if (!$dryRun) {
$condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
$importer["uid"], $item["uri"], Protocol::FEED, Protocol::DFRN];
$previous = Item::selectFirst(['id', 'created'], $condition);
if (DBA::isResult($previous)) {
// Use the creation date when the post had been stored. It can happen this date changes in the feed.
$creation_dates[] = $previous['created'];
Logger::info("Item with uri " . $item["uri"] . " for user " . $importer["uid"] . " already existed under id " . $previous["id"]);
continue;
}
$creation_dates[] = DateTimeFormat::utc($item['created']);
}
$creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
if (empty($creator)) {
@ -598,9 +602,130 @@ class Feed
}
}
if (!$dryRun && DI::config()->get('system', 'adjust_poll_frequency')) {
self::adjustPollFrequency($contact, $creation_dates);
}
return ["header" => $author, "items" => $items];
}
/**
* Automatically adjust the poll frequency according to the post frequency
*
* @param array $contact
* @param array $creation_dates
* @return void
*/
private static function adjustPollFrequency(array $contact, array $creation_dates)
{
if ($contact['network'] != Protocol::FEED) {
Logger::info('Contact is no feed, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'network' => $contact['network']]);
return;
}
if (!empty($creation_dates)) {
// Count the post frequency and the earliest and latest post date
$frequency = [];
$oldest = time();
$newest = 0;
$oldest_date = $newest_date = '';
foreach ($creation_dates as $date) {
$timestamp = strtotime($date);
$day = intdiv($timestamp, 86400);
$hour = $timestamp % 86400;
// Only have a look at values from the last seven days
if (((time() / 86400) - $day) < 7) {
if (empty($frequency[$day])) {
$frequency[$day] = ['count' => 1, 'low' => $hour, 'high' => $hour];
} else {
++$frequency[$day]['count'];
if ($frequency[$day]['low'] > $hour) {
$frequency[$day]['low'] = $hour;
}
if ($frequency[$day]['high'] < $hour) {
$frequency[$day]['high'] = $hour;
}
}
}
if ($oldest > $day) {
$oldest = $day;
$oldest_date = $date;
}
if ($newest < $day) {
$newest = $day;
$newest_date = $date;
}
}
if (count($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']]);
$priority = 8; // Poll once a day
}
if (empty($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) && (((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
$max = 0;
foreach ($frequency as $entry) {
if (($entry['count'] == 1) || ($entry['high'] == $entry['low'])) {
continue;
}
// We take the earliest and latest post day and interpolate the number of post per day
// that would had been created with this post frequency
// Assume at least four hours between oldest and newest post per day - should be okay for news outlets
$duration = max($entry['high'] - $entry['low'], 14400);
$ppd = (86400 / $duration) * $entry['count'];
if ($ppd > $max) {
$max = $ppd;
}
}
if ($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 {
$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']]);
$priority = 8; // Poll once a day
}
if ($contact['rating'] != $priority) {
Logger::notice('Adjusting priority', ['old' => $contact['rating'], 'new' => $priority, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
DBA::update('contact', ['rating' => $priority], ['id' => $contact['id']]);
}
}
/**
* Convert a tag array to a tag string
*

View File

@ -186,7 +186,7 @@ class Cron
Addon::reload();
$sql = "SELECT `contact`.`id`, `contact`.`nick`, `contact`.`name`, `contact`.`network`, `contact`.`archive`,
`contact`.`last-update`, `contact`.`priority`, `contact`.`rel`, `contact`.`subhub`
`contact`.`last-update`, `contact`.`priority`, `contact`.`rating`, `contact`.`rel`, `contact`.`subhub`
FROM `user`
STRAIGHT_JOIN `contact`
ON `contact`.`uid` = `user`.`uid` AND `contact`.`poll` != ''
@ -217,65 +217,43 @@ class Cron
}
while ($contact = DBA::fetch($contacts)) {
$ratings = [0, 3, 7, 8, 9, 10];
if (DI::config()->get('system', 'adjust_poll_frequency') && ($contact['network'] == Protocol::FEED)) {
$rating = $contact['rating'];
} elseif (array_key_exists($contact['priority'], $ratings)) {
$rating = $ratings[$contact['priority']];
} else {
$rating = -1;
}
// Friendica and OStatus are checked once a day
if (in_array($contact['network'], [Protocol::DFRN, Protocol::OSTATUS])) {
$contact['priority'] = 3;
$rating = 8;
}
// ActivityPub is checked once a week
if ($contact['network'] == Protocol::ACTIVITYPUB) {
$contact['priority'] = 4;
$rating = 9;
}
// Check archived contacts once a month
if ($contact['archive']) {
$contact['priority'] = 5;
$rating = 10;
}
if ($contact['priority'] >= 0) {
$update = false;
if ($rating < 0) {
continue;
}
/*
* Based on $contact['priority'], should we poll this site now? Or later?
*/
$t = $contact['last-update'];
$t = $contact['last-update'];
$poll_intervals = [$min_poll_interval . ' minute', '15 minute', '30 minute',
'1 hour', '2 hour', '3 hour', '6 hour', '12 hour' ,'1 day', '1 week', '1 month'];
/*
* Based on $contact['priority'], should we poll this site now? Or later?
*/
switch ($contact['priority']) {
case 5:
if (DateTimeFormat::utcNow() > DateTimeFormat::utc($t . " + 1 month")) {
$update = true;
}
break;
case 4:
if (DateTimeFormat::utcNow() > DateTimeFormat::utc($t . " + 1 week")) {
$update = true;
}
break;
case 3:
if (DateTimeFormat::utcNow() > DateTimeFormat::utc($t . " + 1 day")) {
$update = true;
}
break;
case 2:
if (DateTimeFormat::utcNow() > DateTimeFormat::utc($t . " + 12 hour")) {
$update = true;
}
break;
case 1:
if (DateTimeFormat::utcNow() > DateTimeFormat::utc($t . " + 1 hour")) {
$update = true;
}
break;
case 0:
default:
if (DateTimeFormat::utcNow() > DateTimeFormat::utc($t . " + " . $min_poll_interval . " minute")) {
$update = true;
}
break;
}
if (!$update) {
continue;
}
if (empty($poll_intervals[$rating]) || (DateTimeFormat::utcNow() > DateTimeFormat::utc($t . ' + ' . $poll_intervals[$rating]))) {
continue;
}
if ((($contact['network'] == Protocol::FEED) && ($contact['priority'] <= 3)) || ($contact['network'] == Protocol::MAIL)) {

View File

@ -82,6 +82,10 @@ return [
'php_path' => 'php',
],
'system' => [
// adjust_poll_frequency (Boolean)
// Automatically detect and set the best feed poll frequency.
'adjust_poll_frequency' => false,
// allowed_link_protocols (Array)
// Allowed protocols in links URLs, add at your own risk. http(s) is always allowed.
'allowed_link_protocols' => ['ftp://', 'ftps://', 'mailto:', 'cid:', 'gopher://'],