grabber.module

<?php

require_once(__DIR__ . '/grabber.menu.php');
require_once(__DIR__ . '/include/Registry.php');
require_once(__DIR__ . '/include/ProcessHelper.php');
require_once(__DIR__ . '/include/Logger.php');
require_once(__DIR__ . '/include/Tracker.php');
require_once(__DIR__ . '/include/ProxyList.php');

function grabber_cron() {
    //$scheduler = \Grabber\Registry::get_scheduler();
    //$scheduler->execute();
}

function grabber_execute() {
    \Grabber\ProxyList::update_file();

    $scheduler = \Grabber\Registry::get_scheduler();
    $scheduler->execute();
}

function grabber_execute_processor($processor_name) {
    $processor_name = filter_var($processor_name, FILTER_SANITIZE_STRING);
    $semaphore = \Grabber\Registry::get_semaphore();

    // wait for called process finished
    sleep(3); // TODO check request flag (header|var)

    if ($semaphore->is_set("Processor-{$processor_name}")) {
        \Grabber\Logger::log("Execute processor: " . $processor_name . " - Already executed.");
        print "Already executed: {$processor_name}.";
        exit();
    }

    $semaphore->set("Processor-{$processor_name}");

    \Grabber\Logger::log("Execute processor: " . $processor_name . " ...");
    try {
        $processor = \Grabber\Registry::get_processor($processor_name);
        $processor->execute();
        \Grabber\Logger::log("Finish processor: " . get_class($processor) . " - [ OK ]");
        print 'OK';

    } catch (\Grabber\TrackerTimeoutException $e) {
        \Grabber\Logger::log("Processor timeout: " . $processor_name . ': ' . $e->getMessage() . " - [TIME]");
        \Grabber\ProcessHelper::fork();
        print 'TIME';

    } catch (\Exception $e) {
        \Grabber\Logger::error("Processor error: " . $processor_name . ': ' . $e->getMessage());
        print 'FAIL';
    }

    exit();
}


function grabber_execute_scanner($scanner_name, $thread_id) {
    $scanner_name = filter_var($scanner_name, FILTER_SANITIZE_STRING);
    $thread_id = filter_var($thread_id, FILTER_SANITIZE_NUMBER_INT);
    $semaphore = \Grabber\Registry::get_semaphore();

    if ($thread_id > \Grabber\ProcessHelper::MAX_SCANNER_THREADS) {
        print "Maximum: " . \Grabber\ProcessHelper::MAX_SCANNER_THREADS . " thread/scanner. (MAX_SCANNER_THREADS)";
        exit();
    }

    // prevent overload
    sleep(3);

    if ($semaphore->is_set("Scanner-{$scanner_name}-{$thread_id}")) {
        \Grabber\Logger::log("Execute scanner: {$scanner_name}/{$thread_id} - Already executed.");
        print "Already executed: {$scanner_name}/{$thread_id}.";
        exit();
    }

    $semaphore->set("Scanner-{$scanner_name}-{$thread_id}");

    \Grabber\Logger::log("Execute scanner: {$scanner_name}/{$thread_id}...");
    try {
        $scanner = \Grabber\Registry::get_scanner($scanner_name);
        $scanner->execute();
        \Grabber\Logger::log("Finish scanner: " . get_class($scanner) . "/{$thread_id} - [ OK ]");
        print 'OK';

    } catch (\Grabber\TrackerTimeoutException $e) {
        \Grabber\Logger::log("Scanner timeout: {$scanner_name}/{$thread_id}: " . $e->getMessage() . " - [TIME]");
        \Grabber\ProcessHelper::fork();
        print 'TIME';

    } catch (\Exception $e) {
        \Grabber\Logger::error("Scanner error: {$scanner_name}/{$thread_id}: " . $e->getMessage());
        print 'FAIL';
    }

    exit();
}

function grabber_execute_scanner_all($scanner_name) {
    $scanner_name = filter_var($scanner_name, FILTER_SANITIZE_STRING);
    \Grabber\ProcessHelper::execute_scanner($scanner_name);
}