Automated EPG Download

2026-07-10 14:14:41 +02:00
parent 8b68ba2e9b
commit 3ae2d14e95
3 changed files with 375 additions and 60 deletions
@@ -1,5 +1,13 @@
 FROM php:8.4-apache

-RUN echo "memory_limit=512M" > /usr/local/etc/php/conf.d/tvguide.ini
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+        xz-utils \
+        gzip \
+        ca-certificates \
+    && rm -rf /var/lib/apt/lists/*
+
+RUN echo "memory_limit=512M" \
+    > /usr/local/etc/php/conf.d/tvguide.ini

 COPY --from=composer:latest /usr/bin/composer /usr/bin/composer
@@ -2,84 +2,361 @@

 require_once '/var/www/src/bootstrap.php';

-$config = new Config(
-    '/var/www/config/config.json',
-    '/var/www/config/channels.json'
-);
+const EPG_WORK_DIR = '/var/www/epg/work';
+const EPG_TARGET = '/var/www/epg/guide.xml';
+const EPG_CACHE = '/var/www/cache/guide.json';
+const EPG_LOCK_FILE = '/var/www/epg/update.lock';
+const EPG_LOG_FILE = '/var/www/logs/epg-update.log';

-$epg = $config->get('epg', []);
-$urls = $epg['urls'] ?? [];
+function logMessage(string $message, bool $error = false): void
+{
+    $line = sprintf(
+        "[%s] %s\n",
+        date('Y-m-d H:i:s'),
+        $message
+    );

-if (!$urls && !empty($epg['url'])) {
-    $urls = [$epg['url']];
+    echo $line;
+
+    if ($error) {
+        fwrite(STDERR, $line);
+    }
+
+    $logDirectory = dirname(EPG_LOG_FILE);
+
+    if (!is_dir($logDirectory)) {
+        mkdir($logDirectory, 0775, true);
+    }
+
+    file_put_contents(
+        EPG_LOG_FILE,
+        $line,
+        FILE_APPEND | LOCK_EX
+    );
 }

-if (!$urls) {
-    fwrite(STDERR, "Keine epg.urls in config.json gesetzt.\n");
+function downloadFile(string $url, string $target): bool
+{
+    $context = stream_context_create([
+        'http' => [
+            'timeout' => 120,
+            'follow_location' => 1,
+            'max_redirects' => 5,
+            'user_agent' => 'TVGuide EPG Updater/1.0',
+        ],
+        'https' => [
+            'timeout' => 120,
+            'follow_location' => 1,
+            'max_redirects' => 5,
+            'user_agent' => 'TVGuide EPG Updater/1.0',
+        ],
+    ]);
+
+    $input = @fopen($url, 'rb', false, $context);
+
+    if ($input === false) {
+        return false;
+    }
+
+    $output = @fopen($target, 'wb');
+
+    if ($output === false) {
+        fclose($input);
+        return false;
+    }
+
+    $bytes = stream_copy_to_stream($input, $output);
+
+    fclose($input);
+    fclose($output);
+
+    return $bytes !== false && $bytes > 0;
+}
+
+function detectCompression(string $url, string $file): string
+{
+    $path = strtolower(
+        parse_url($url, PHP_URL_PATH) ?? ''
+    );
+
+    if (str_ends_with($path, '.xz')) {
+        return 'xz';
+    }
+
+    if (
+        str_ends_with($path, '.gz') ||
+        str_ends_with($path, '.gzip')
+    ) {
+        return 'gz';
+    }
+
+    $handle = fopen($file, 'rb');
+    $header = $handle ? fread($handle, 6) : '';
+    if ($handle) {
+        fclose($handle);
+    }
+
+    if (str_starts_with($header, "\xFD\x37\x7A\x58\x5A\x00")) {
+        return 'xz';
+    }
+
+    if (str_starts_with($header, "\x1F\x8B")) {
+        return 'gz';
+    }
+
+    return 'xml';
+}
+
+function unpackFile(
+    string $source,
+    string $target,
+    string $compression
+): bool {
+    if ($compression === 'xml') {
+        return copy($source, $target);
+    }
+
+    $command = match ($compression) {
+        'xz' => sprintf(
+            'xz -dc %s > %s 2>&1',
+            escapeshellarg($source),
+            escapeshellarg($target)
+        ),
+
+        'gz' => sprintf(
+            'gzip -dc %s > %s 2>&1',
+            escapeshellarg($source),
+            escapeshellarg($target)
+        ),
+
+        default => null,
+    };
+
+    if ($command === null) {
+        return false;
+    }
+
+    exec($command, $output, $status);
+
+    return
+        $status === 0 &&
+        file_exists($target) &&
+        filesize($target) > 0;
+}
+
+function isValidXmlTv(string $file): bool
+{
+    if (!file_exists($file) || filesize($file) === 0) {
+        return false;
+    }
+
+    libxml_use_internal_errors(true);
+
+    $xml = simplexml_load_file($file);
+
+    libxml_clear_errors();
+
+    if (!$xml) {
+        return false;
+    }
+
+    return $xml->getName() === 'tv';
+}
+
+function removeOldWorkFiles(): void
+{
+    foreach (glob(EPG_WORK_DIR . '/*') ?: [] as $file) {
+        if (is_file($file)) {
+            @unlink($file);
+        }
+    }
+}
+
+$lockHandle = fopen(EPG_LOCK_FILE, 'c');
+
+if ($lockHandle === false) {
+    logMessage('Lock-Datei konnte nicht geöffnet werden.', true);
    exit(1);
 }

-$workDir = '/var/www/epg/work';
-$target = '/var/www/epg/guide.xml';
-
-if (!is_dir($workDir)) {
-    mkdir($workDir, 0775, true);
+if (!flock($lockHandle, LOCK_EX | LOCK_NB)) {
+    logMessage(
+        'Ein EPG-Update läuft bereits. Dieser Lauf wird beendet.'
+    );
+    fclose($lockHandle);
+    exit(0);
 }

-$xmlFiles = [];
+$startedAt = microtime(true);

-foreach ($urls as $index => $url) {
-    $base = $workDir . '/source-' . $index;
-    $xzFile = $base . '.xz';
-    $xmlFile = $base . '.xml';
+try {
+    logMessage('EPG-Update gestartet.');

-    echo "Lade EPG: {$url}\n";
+    $config = new Config(
+        '/var/www/config/config.json',
+        '/var/www/config/channels.json'
+    );

-    $data = file_get_contents($url);
+    $epg = $config->get('epg', []);
+    $urls = $epg['urls'] ?? [];

-    if ($data === false || trim($data) === '') {
-        fwrite(STDERR, "Download fehlgeschlagen oder leer: {$url}\n");
-        continue;
+    if (!$urls && !empty($epg['url'])) {
+        $urls = [$epg['url']];
    }

-    file_put_contents($xzFile, $data);
-
-    echo "Entpacke XZ...\n";
-
-    $command = 'xz -dc ' . escapeshellarg($xzFile) . ' > ' . escapeshellarg($xmlFile);
-    exec($command, $output, $code);
-
-    if ($code !== 0 || !file_exists($xmlFile) || filesize($xmlFile) === 0) {
-        fwrite(STDERR, "XZ konnte nicht entpackt werden: {$url}\n");
-        continue;
+    if (!$urls) {
+        throw new RuntimeException(
+            'Keine epg.urls in config.json gesetzt.'
+        );
    }

-    if (!@simplexml_load_file($xmlFile)) {
-        fwrite(STDERR, "XML ungültig: {$url}\n");
-        continue;
+    if (!is_dir(EPG_WORK_DIR)) {
+        mkdir(EPG_WORK_DIR, 0775, true);
    }

-    $xmlFiles[] = $xmlFile;
-}
+    removeOldWorkFiles();
+
+    $xmlFiles = [];
+
+    foreach (array_values($urls) as $index => $url) {
+        $downloadFile =
+            EPG_WORK_DIR . "/source-{$index}.download";
+
+        $xmlFile =
+            EPG_WORK_DIR . "/source-{$index}.xml";
+
+        logMessage("Lade Quelle " . ($index + 1) . ": {$url}");
+
+        if (!downloadFile($url, $downloadFile)) {
+            logMessage(
+                "Download fehlgeschlagen oder leer: {$url}",
+                true
+            );
+            continue;
+        }
+
+        $compression = detectCompression(
+            $url,
+            $downloadFile
+        );
+
+        logMessage(
+            sprintf(
+                'Download abgeschlossen: %.2f MB, Format: %s',
+                filesize($downloadFile) / 1024 / 1024,
+                strtoupper($compression)
+            )
+        );
+
+        if (!unpackFile(
+            $downloadFile,
+            $xmlFile,
+            $compression
+        )) {
+            logMessage(
+                "Datei konnte nicht entpackt werden: {$url}",
+                true
+            );
+            continue;
+        }
+
+        if (!isValidXmlTv($xmlFile)) {
+            logMessage(
+                "XMLTV-Datei ist ungültig: {$url}",
+                true
+            );
+            continue;
+        }
+
+        $xmlFiles[] = $xmlFile;
+    }
+
+    if (!$xmlFiles) {
+        throw new RuntimeException(
+            'Keine gültigen XMLTV-Dateien geladen.'
+        );
+    }
+
+    /*
+     * Zuerst in eine temporäre Datei mergen.
+     * Die bestehende guide.xml bleibt bei einem Fehler unangetastet.
+     */
+    $temporaryTarget =
+        EPG_WORK_DIR . '/guide-' . getmypid() . '.xml';
+
+    logMessage(
+        count($xmlFiles) . ' gültige Quelle(n), Merge gestartet.'
+    );
+
+    $merge = new XmlTvMergeService();
+    $merge->merge($xmlFiles, $temporaryTarget);
+
+    if (!isValidXmlTv($temporaryTarget)) {
+        throw new RuntimeException(
+            'Die zusammengeführte XMLTV-Datei ist ungültig.'
+        );
+    }
+
+    /*
+     * Atomarer Austausch innerhalb desselben Dateisystems.
+     */
+    if (!rename($temporaryTarget, EPG_TARGET)) {
+        throw new RuntimeException(
+            'guide.xml konnte nicht atomar ersetzt werden.'
+        );
+    }
+
+    logMessage(
+        sprintf(
+            'guide.xml aktualisiert: %.2f MB',
+            filesize(EPG_TARGET) / 1024 / 1024
+        )
+    );
+
+    $cacheService = new XmlTvCacheService(
+        EPG_TARGET,
+        $config->get('timezone', 'Europe/Vienna'),
+        EPG_CACHE
+    );
+
+    $cacheData = $cacheService->rebuildCache();
+
+    $channelCount =
+        count($cacheData['channels'] ?? []);
+
+    $programmeCount = 0;
+
+    foreach ($cacheData['channels'] ?? [] as $channel) {
+        $programmeCount +=
+            count($channel['programmes'] ?? []);
+    }
+
+    $duration = microtime(true) - $startedAt;
+
+    logMessage(
+        sprintf(
+            'EPG-Update erfolgreich: %d Sender, %d Sendungen, %.1f Sekunden.',
+            $channelCount,
+            $programmeCount,
+            $duration
+        )
+    );
+
+    exit(0);
+} catch (Throwable $exception) {
+    $duration = microtime(true) - $startedAt;
+
+    logMessage(
+        sprintf(
+            'EPG-Update fehlgeschlagen nach %.1f Sekunden: %s',
+            $duration,
+            $exception->getMessage()
+        ),
+        true
+    );

-if (!$xmlFiles) {
-    fwrite(STDERR, "Keine gültigen XMLTV-Dateien geladen.\n");
    exit(1);
+} finally {
+    flock($lockHandle, LOCK_UN);
+    fclose($lockHandle);
 }
-
-echo "Merge XMLTV...\n";
-
-$merge = new XmlTvMergeService();
-$merge->merge($xmlFiles, $target);
-
-echo "guide.xml aktualisiert.\n";
-
-$cache = new XmlTvCacheService(
-    $target,
-    $config->get('timezone', 'Europe/Vienna'),
-    '/var/www/cache/guide.json'
-);
-
-$cache->rebuildCache();
-
-echo "guide.json und guide.version aktualisiert.\n";
@@ -16,3 +16,33 @@ services:
      - ./epg:/var/www/epg
      - ./logos:/var/www/html/logos
      - ./posters:/var/www/html/posters
+      - ./logs:/var/www/logs
+
+  epg-scheduler:
+    build: .
+    container_name: tvguide-epg-scheduler
+    restart: unless-stopped
+
+    environment:
+      EPG_UPDATE_INTERVAL: "21600"
+
+    volumes:
+      - ./app/src:/var/www/src
+      - ./app/config:/var/www/config
+      - ./app/cache:/var/www/cache
+      - ./app/bin:/var/www/bin
+      - ./vendor:/var/www/vendor
+      - ./epg:/var/www/epg
+      - ./logs:/var/www/logs
+
+    command:
+      - /bin/sh
+      - -c
+      - |
+        echo "TVGuide EPG-Scheduler gestartet."
+        echo "Intervall: $${EPG_UPDATE_INTERVAL:-21600} Sekunden"
+
+        while true; do
+          php /var/www/bin/update-epg.php
+          sleep "$${EPG_UPDATE_INTERVAL:-21600}"
+        done