soselo/cron.php

<?php chdir(__DIR__) ?>
<?php require 'base.php' ?>
<?php

if (!isset($argv) || $argv === NULL) die;

if (file_exists('.cron.lock'))
	die('cron.php is locked by ".cron.lock"');
touch('.cron.lock');

function logi($thing) { echo date('Y/m/d H:i:s').' INFO  | '.strval($thing)."\n"; }
function loge($thing) { echo date('Y/m/d H:i:s').' ERROR | '.strval($thing)."\n"; }

function end_cronjobs() {
	unlink('.cron.lock');
	die;
}


////////////////
// INITIAL SETUP
//
$GLOBALS['_crondir'] = $GLOBALS['appconf']['data_dir'].'/cron';
if (!file_exists($GLOBALS['_crondir'])) {
	mkdir($GLOBALS['_crondir']);
	end_cronjobs();
}
$time_start = time();
//
////////////////


////////////////////////////////
// CRONJOBS for database caching
//
function run__cronjobs_db() {
	$dir = $GLOBALS['_crondir'].'/db';
	if (!file_exists($dir)) {
		mkdir($dir);
		return false;
	}
	$jobs = [];
	foreach (scandir($dir) as $jf) {
		if (in_array($jf,['.','..']))
			continue;
		$ps = explode(',',$jf);
		$jobs []= (object)[
			'id' => $ps[2],
			'file' => $dir.'/'.$jf,
			'software' => $ps[0],
			'instance' => $ps[1],
			'time' => intval($ps[3]),
		];
	}
	foreach ($jobs as $job) {
		$ms = time() - filemtime($job->file);
		$data = null;
		$data = @unserialize(file_get_contents($job->file));
		if ($data === false) {
			$data = ['last_accessed' => time(), 'sql' => $data];
			file_put_contents($job->file, serialize($data));
		}
		if ($ms < $job->time) continue;

		logi("Processing job: {$job->id}, software={$job->software}, instance={$job->instance}");
		$ok = false;
		switch ($job->software) {
			case 'mastodon':
				$pg = new PgDatabase($job->software, $job->instance);
				if ($pg->is_ok()) {
					$result = $pg->fetch_all($data['sql']);
					content_cache__put($job->software.$job->instance.$data['sql'], 'always,'.$job->time, $result);
					$pg->close();
					$ok = true;
				}
				break;
		}
		if ($ok) {
			logi("Job done: {$job->id}");
			touch($job->file);
		}
	}
}
run__cronjobs_db();
//
////////////////////////////////


///////////////////////////////
// CRONJOBS for caching cleanup
//
function run__cronjobs_cache_cleanup() {
	$cache_dir = $GLOBALS['appconf']['data_dir'].'/cache';
	if (!file_exists($cache_dir))
		return;
	$others = [];
	foreach (scandir($cache_dir) as $fl) {
		if (in_array($fl, ['.','..']))
			continue;
		$ps = explode(',',$fl);
		if ($ps[1] !== 'ondemand') {
			$others []= $fl;
			continue;
		}
		$time = intval($ps[2]);
		$mtime = filemtime($cache_dir.'/'.$fl);
		if (time() - $mtime >= $time) {
			logi('CACHE-CLEANUP remove old file: '.$fl);
			unlink($cache_dir.'/'.$fl);
		}
	}
	$dir = $GLOBALS['appconf']['data_dir'].'/cron/db';
	foreach ($others as $fl) {
		$ps = explode(',',$fl);
		if (count($ps) === 2 && $ps[1] === 'always') {
			$id = $ps[0];
			$contains = false;
			foreach (scandir($dir) as $f) {
				if (in_array($f, ['.','..']))
					continue;
				if (strpos($f, ','.$id.',') !== false) {
					$contains = true;
					break;
				}
			}
			if (!$contains) {
				logi('CACHE-CLEANUP remove unused "always" cache: '.$fl);
				unlink($cache_dir.'/'.$fl);
			}
		}
	}
}
run__cronjobs_cache_cleanup();
//
///////////////////////////////


//////////////////////////////////////////
// CRONJOBS for creating jobs from filters
//
function run__cronjobs_filter_job($software, $instance, $filter, $action) {
	$_GET = [
		'filter_id' => $filter['id'],
		'instance' => $instance
	];
	logi('Processing filter: '.serialize($_GET));
	$users = mod_php('api/v1/database/'.$software.'/accounts/search');
	if (gettype($users) !== 'array' || isset($users['error'])) {
		loge('Unexpected value for users: '.strval($users));
		return false;
	}
	if (!isset($users['data'])) {
		loge('Unexpected value for users: '.strval($users));
		return false;
	}
	foreach ($users['data'] as $user) {
		$jobname = $action['action']['name'].'_'.$user['id'];
		$jobmod = 'api/v1/http/'.$software.'/accounts/'.$action['action']['name'];
		$report = '[automatic_report] User has been reported because it matches filter "'.$filter['preset_name'].'"';
		$explain = $report;
		if (isset($action['action']['explain']))
			$explain = $action['action']['explain'];
		$args = base64_encode(json_encode([
			'_get' => [
				'id' => $user['id'],
				'instance' => $instance,
			],
			'_post' => [
				'report' => $report,
				'explain' => $explain,
			],
		]));
		$_POST = [
			'name' => $jobname,
			'module' => $jobmod,
			'args' => $args,
		];
		$result = mod_php('api/v1/cron/job/put');
		if (isset($result['ok']))
			logi($result['ok']);
		else if (isset($result['error']))
			loge($result['error']);
	}

	touch($GLOBALS['appconf']['data_dir'].'/filter_actions/users,'.$instance.','.$action['id']);
	return true;
}
function run__cronjobs_filter_jobs() {
	$_GET = ['type' => 'users'];
	$filters = mod_php('api/v1/config/filters/get');
	$config = mod_php('api/v1/config/get');
	$filter_get = function($filters, $id) {
		foreach ($filters as $f)
			if ($f['id'] === $id)
				return $f;
		return null;
	};
	foreach ($GLOBALS['supported_ap_software'] as $software) {
		foreach ($config['hosts'][$software] as $hostcfg) {
			$_GET['instance'] = $hostcfg['instance'];
			$actions = mod_php('api/v1/config/filters/actions/get');
			if ($actions === null) continue;
			foreach ($actions as $action) {
				$filter = $filter_get($filters, $action['id']);
				if ($filter === null) continue;
				$ts_diff = time() - $action['filemtime'];
				if ($ts_diff >= $action['action']['interval'])
					run__cronjobs_filter_job($software,
						$hostcfg['instance'], $filter, $action);
			}
		}
	}
}
run__cronjobs_filter_jobs();
//
//////////////////////////////////////////


///////////////////////////////////////
// CRONJOBS for running the actual jobs
//
function run_cronjobs() {
	$max = 1000;
	if (isset($GLOBALS['appconf']['cron__max_jobs']))
		$max = intval($GLOBALS['appconf']['cron__max_jobs']);
	if (getenv('MAX_JOBS') !== false)
		$max = intval(getenv('MAX_JOBS'));
	$_GET = ['max' => $max];
	$jobs = mod_php('api/v1/cron/jobs');
	foreach ($jobs as $job) {
		$_GET = ['id' => $job];
		$job_data = mod_php('api/v1/cron/job/get');
		if (isset($job_data['args']['_get']))
			$_GET = $job_data['args']['_get'];
		if (isset($job_data['args']['_post']))
			$_POST = $job_data['args']['_post'];
		logi('Running job with data: '.serialize($job_data));
		$r = mod_php($job_data['module']);
		if ($r === 'does_not_exist') {
			loge('Could not find module (mod.php file) for: '.$job_data['module']);
			continue;
		}
		$_GET = ['id' => $job];
		$result = mod_php('api/v1/cron/job/del');
		if (isset($result['ok']))
			logi($result['ok']);
		else if (isset($result['error']))
			loge($result['error']);
	}
}
run_cronjobs();
//
///////////////////////////////////////


$ms_total = time() - $time_start;
logi("Cron has finished in {$ms_total} seconds");

end_cronjobs();