GG.deals clone: CodeIgniter 4 + MongoDB game price tracker

- Full gg.deals-style UI: deal feed with filters, game pages with price history charts, rankings, vouchers, news, prepaids, subscriptions, bundles, franchises, wishlist, API docs - Real data crawlers: * gg:crawl (Steam Store API + GOG Catalog + Epic free games) * gg:catalog (252K game slugs from gg.deals sitemaps) * gg:ggdeals (per-game prices via FlareSolverr session, JSON-LD parsing) * gg:seed / gg:crawl-covers (demo seed + Steam CDN cover art) - 24/7 self-updating scheduler (scripts/scheduler_loop.php, lockfile-protected) - Tailwind v4 dark/light theme, Alpine.js, Chart.js
2026-08-23 07:06:02 +00:00
commit fff7a4b9b9
146 changed files with 15691 additions and 0 deletions
@@ -0,0 +1,91 @@
+#!/usr/bin/env php
+<?php
+
+declare(strict_types=1);
+
+/**
+ * Crash-safe recurring crawler scheduler with a single-instance lockfile.
+ *
+ * Loop:
+ *   1. Acquire lockfile (fail fast if another instance is running)
+ *   2. Refresh Steam + GOG + Epic prices (gg:crawl)
+ *   3. Finalize aggregates
+ *   4. Expand catalog: crawl next batch of N games from gg.deals via FlareSolverr
+ *   5. Release lockfile, sleep interval
+ *
+ * Usage: php scripts/scheduler_loop.php [interval_minutes] [batch_size]
+ * Example: php scripts/scheduler_loop.php 60 200
+ */
+
+$intervalMin = isset($argv[1]) ? max(5, (int) $argv[1]) : 60;
+$batchSize   = isset($argv[2]) ? max(10, (int) $argv[2]) : 150;
+$root        = dirname(__DIR__);
+$lockFile    = $root . '/writable/logs/scheduler.lock';
+$logFile     = $root . '/writable/logs/scheduler.log';
+
+function logMsg(string $file, string $msg): void
+{
+    $line = '[' . date('Y-m-d H:i:s') . '] ' . $msg . PHP_EOL;
+    file_put_contents($file, $line, FILE_APPEND);
+    echo $line;
+}
+
+function acquireLock(string $lockFile): bool
+{
+    if (file_exists($lockFile)) {
+        $pid = (int) trim((string) file_get_contents($lockFile));
+        if ($pid > 0 && posix_kill($pid, 0)) {
+            return false; // another instance alive
+        }
+        @unlink($lockFile); // stale lock
+    }
+    file_put_contents($lockFile, (string) getmypid());
+
+    return true;
+}
+
+function releaseLock(string $lockFile): void
+{
+    @unlink($lockFile);
+}
+
+function run(string $root, string $cmd, string $logFile, string $tag): int
+{
+    logMsg($logFile, ">>> [$tag] $cmd");
+    $output = [];
+    $code = 0;
+    exec('cd ' . escapeshellarg($root) . ' && ' . $cmd . ' 2>&1', $output, $code);
+    foreach (array_slice($output, -6) as $line) {
+        logMsg($logFile, "    $line");
+    }
+    logMsg($logFile, "<<< [$tag] exit $code");
+
+    return $code;
+}
+
+logMsg($logFile, "Scheduler loop started — interval {$intervalMin}min, batch {$batchSize}");
+
+while (true) {
+    if (! acquireLock($lockFile)) {
+        logMsg($logFile, 'Another crawler instance is running — exiting this loop.');
+        exit(1);
+    }
+
+    try {
+        // 1. refresh prices from direct sources
+        run($root, 'php spark gg:crawl steam', $logFile, 'steam');
+        run($root, 'php spark gg:crawl gog', $logFile, 'gog');
+        run($root, 'php spark gg:crawl epic', $logFile, 'epic');
+        run($root, 'php spark gg:crawl finalize', $logFile, 'finalize');
+
+        // 2. expand catalog via FlareSolverr session
+        run($root, "php spark gg:ggdeals --batch={$batchSize}", $logFile, 'ggdeals-expand');
+    } catch (\Throwable $e) {
+        logMsg($logFile, 'ERROR in loop: ' . $e->getMessage());
+    } finally {
+        releaseLock($lockFile);
+    }
+
+    logMsg($logFile, "Sleeping {$intervalMin}min until next cycle…");
+    sleep($intervalMin * 60);
+}