This commit is contained in:
Jürgen Mummert
2026-01-09 22:04:52 +01:00
parent 99ef883da5
commit 2bd52f77e0
6 changed files with 147 additions and 139 deletions
+14 -20
View File
@@ -3,7 +3,7 @@
namespace MummertMedia\ContaoMeilisearchBundle\Command;
use Contao\CoreBundle\Framework\ContaoFramework;
use Contao\Database;
use Doctrine\DBAL\Connection;
use Symfony\Component\Console\Command\Command;
use Symfony\Component\Console\Input\InputInterface;
use Symfony\Component\Console\Input\InputOption;
@@ -13,6 +13,7 @@ class MeilisearchFilesCleanupCommand extends Command
{
public function __construct(
private readonly ContaoFramework $framework,
private readonly Connection $connection,
) {
parent::__construct();
}
@@ -21,13 +22,13 @@ class MeilisearchFilesCleanupCommand extends Command
{
$this
->setName('meilisearch:files:cleanup')
->setDescription('Remove stale indexed files (PDF, DOCX, XLSX, PPTX) from tl_search_pdf')
->setDescription('Remove stale indexed files from tl_search_files')
->addOption(
'grace',
null,
InputOption::VALUE_OPTIONAL,
'Grace period in seconds (files newer than now-grace are kept)',
86400 // 24 Stunden
86400
)
->addOption(
'dry-run',
@@ -49,10 +50,10 @@ class MeilisearchFilesCleanupCommand extends Command
$cutoff = time() - $grace;
if ($dryRun) {
$count = Database::getInstance()
->prepare('SELECT COUNT(*) AS cnt FROM tl_search_pdf WHERE last_seen < ?')
->execute($cutoff)
->cnt;
$count = $this->connection->fetchOne(
'SELECT COUNT(*) FROM tl_search_files WHERE last_seen < ?',
[$cutoff]
);
$message = sprintf(
'[DRY-RUN] %d stale file(s) would be removed (last_seen < %s)',
@@ -63,14 +64,14 @@ class MeilisearchFilesCleanupCommand extends Command
$output->writeln('<comment>' . $message . '</comment>');
$this->log($message);
$this->log('Cleaner successfully stopped');
$this->log('Cleaner stopped (dry-run)');
return Command::SUCCESS;
}
$affected = Database::getInstance()
->prepare('DELETE FROM tl_search_pdf WHERE last_seen < ?')
->execute($cutoff)
->affectedRows;
$affected = $this->connection->executeStatement(
'DELETE FROM tl_search_files WHERE last_seen < ?',
[$cutoff]
);
$message = sprintf(
'Removed %d stale file(s) (last_seen < %s)',
@@ -92,15 +93,8 @@ class MeilisearchFilesCleanupCommand extends Command
}
}
/**
* Einheitliches Logging mit Zeitstempel
*/
private function log(string $message): void
{
error_log(sprintf(
'[%s] %s',
date('Y-m-d H:i:s'),
$message
));
error_log(sprintf('[%s] %s', date('Y-m-d H:i:s'), $message));
}
}
+72 -75
View File
@@ -3,15 +3,13 @@
namespace MummertMedia\ContaoMeilisearchBundle\EventListener;
use Contao\Config;
use MummertMedia\ContaoMeilisearchBundle\Service\PdfIndexService;
use MummertMedia\ContaoMeilisearchBundle\Service\OfficeIndexService;
use Contao\System;
class IndexPageListener
{
public function __construct(
private readonly PdfIndexService $pdfIndexService,
private readonly OfficeIndexService $officeIndexService,
) {}
public function __construct()
{
}
private function debug(string $message, array $context = []): void
{
@@ -103,7 +101,6 @@ class IndexPageListener
$this->debug('Meta: searchimage candidate', ['searchimage' => $searchImage]);
if (!empty($searchImage)) {
// >>> HINWEIS: falls dein tl_search-Feld "image" heißt, hier auf $set['image'] ändern!
$set['imagepath'] = trim((string) $searchImage);
}
@@ -139,20 +136,12 @@ class IndexPageListener
'class' => $e::class,
]);
}
$this->debug('Meta: final set snapshot', [
'priority' => $set['priority'] ?? null,
'keywords' => $set['keywords'] ?? null,
'imagepath' => $set['imagepath'] ?? null,
'startDate' => $set['startDate'] ?? null,
'checksum' => $set['checksum'] ?? null,
]);
}
}
/*
* =====================
* DATEI-INDEXIERUNG (PDF / OFFICE)
* DATEI-ERKENNUNG + UPSERT
* =====================
*/
if ((int) ($data['protected'] ?? 0) !== 0) {
@@ -160,15 +149,13 @@ class IndexPageListener
return;
}
$indexPdfs = (bool) Config::get('meilisearch_index_pdfs');
$indexOffice = (bool) Config::get('meilisearch_index_office');
$indexFiles = (bool) Config::get('meilisearch_index_files');
$this->debug('File indexing settings', [
'meilisearch_index_pdfs' => $indexPdfs,
'meilisearch_index_office' => $indexOffice,
$this->debug('File indexing setting', [
'meilisearch_index_files' => $indexFiles,
]);
if (!$indexPdfs && !$indexOffice) {
if (!$indexFiles) {
$this->debug('Abort: file indexing disabled');
return;
}
@@ -176,61 +163,85 @@ class IndexPageListener
$links = $this->findAllLinks($content);
$this->debug('Links found', ['count' => count($links)]);
$pdfLinks = [];
$officeLinks = [];
$fileLinks = [];
foreach ($links as $link) {
$type = $this->detectIndexableFileType($link['url']);
if ($type === 'pdf' && $indexPdfs) {
$pdfLinks[] = $link;
continue;
}
if (in_array($type, ['docx', 'xlsx', 'pptx'], true) && $indexOffice) {
$officeLinks[] = $link;
if ($type !== null) {
$fileLinks[] = $link + ['type' => $type];
}
}
$this->debug('Indexable file links', [
'pdf' => count($pdfLinks),
'office' => count($officeLinks),
$this->debug('Indexable file links found', [
'count' => count($fileLinks),
'types' => array_count_values(array_column($fileLinks, 'type')),
]);
try {
if ($pdfLinks !== []) {
$this->debug('PDF handlePdfLinks(): call', ['count' => count($pdfLinks)]);
$this->pdfIndexService->handlePdfLinks($pdfLinks);
$this->debug('PDF handlePdfLinks(): ok');
}
if ($fileLinks) {
$db = System::getContainer()->get('database_connection');
$time = time();
if ($officeLinks !== []) {
$this->debug('Office handleOfficeLinks(): call', ['count' => count($officeLinks)]);
$this->officeIndexService->handleOfficeLinks($officeLinks);
$this->debug('Office handleOfficeLinks(): ok');
}
} catch (\Throwable $e) {
$this->debug('File indexing failed', [
'error' => $e->getMessage(),
'class' => $e::class,
foreach ($fileLinks as $file) {
$url = strtok($file['url'], '#');
$path = parse_url($url, PHP_URL_PATH);
$abs = $path ? TL_ROOT . '/' . ltrim($path, '/') : null;
$mtime = ($abs && is_file($abs)) ? filemtime($abs) : 0;
$checksum = md5($url . '|' . $mtime);
$existing = $db->fetchAssociative(
'SELECT id, checksum FROM tl_search_files WHERE url = ?',
[$url]
);
if ($existing) {
$db->update(
'tl_search_files',
[
'tstamp' => $time,
'last_seen' => $time,
'page_id' => (int) ($data['pid'] ?? 0),
'file_mtime' => $mtime,
'checksum' => $checksum,
],
['id' => $existing['id']]
);
$this->debug('File updated', [
'url' => $url,
'checksum' => $checksum,
]);
} else {
$db->insert(
'tl_search_files',
[
'tstamp' => $time,
'last_seen' => $time,
'type' => $file['type'],
'url' => $url,
'title' => $file['linkText'] ?? basename($url),
'page_id' => (int) ($data['pid'] ?? 0),
'file_mtime' => $mtime,
'checksum' => $checksum,
]
);
$this->debug('File inserted', [
'url' => $url,
'checksum' => $checksum,
]);
}
}
}
$this->debug('Hook end', [
'final_set_keys' => array_keys($set),
'final_set' => [
'priority' => $set['priority'] ?? null,
'keywords' => $set['keywords'] ?? null,
'imagepath' => $set['imagepath'] ?? null,
'startDate' => $set['startDate'] ?? null,
'checksum' => $set['checksum'] ?? null,
],
]);
}
/**
* Extrahiert MEILISEARCH_JSON aus HTML-Kommentar
*/
/* === Hilfsmethoden unverändert === */
private function extractMeilisearchJson(string $content): ?array
{
if (!preg_match('/<!--\s*MEILISEARCH_JSON\s*(\{.*?\})\s*-->/s', $content, $m)) {
@@ -245,9 +256,6 @@ class IndexPageListener
: null;
}
/**
* Sammle alle <a href="…"> Links
*/
private function findAllLinks(string $content): array
{
if (!preg_match_all(
@@ -270,12 +278,8 @@ class IndexPageListener
return $result;
}
/**
* Ermittelt indexierbaren Dateityp (pdf|docx|xlsx|pptx) oder null
*/
private function detectIndexableFileType(string $url): ?string
{
// Hash entfernen
$url = strtok($url, '#');
$parts = parse_url($url);
@@ -283,7 +287,6 @@ class IndexPageListener
return null;
}
// direkter Pfad (/files/…)
if (!empty($parts['path'])) {
$ext = strtolower(pathinfo($parts['path'], PATHINFO_EXTENSION));
if (in_array($ext, ['pdf', 'docx', 'xlsx', 'pptx'], true)) {
@@ -291,18 +294,12 @@ class IndexPageListener
}
}
// Query-Parameter (Contao 4 + 5)
if (!empty($parts['query'])) {
parse_str($parts['query'], $query);
foreach (['file', 'p', 'f'] as $param) {
if (!empty($query[$param])) {
$candidate = (string) $query[$param];
// sicher decodieren (Contao 4 + 5)
$candidate = html_entity_decode($candidate, ENT_QUOTES);
$candidate = rawurldecode($candidate);
$candidate = rawurldecode(html_entity_decode((string) $query[$param], ENT_QUOTES));
$ext = strtolower(pathinfo($candidate, PATHINFO_EXTENSION));
if (in_array($ext, ['pdf', 'docx', 'xlsx', 'pptx'], true)) {
+1 -7
View File
@@ -3,7 +3,7 @@ services:
Psr\Container\ContainerInterface: '@service_container'
MummertMedia\ContaoMeilisearchBundle\:
resource: '../../{Command,Cron,EventListener,Service}'
resource: '../../{Command,EventListener,Service}'
autowire: true
autoconfigure: true
@@ -19,12 +19,6 @@ services:
tags:
- { name: contao.hook, hook: indexPage, method: onIndexPage }
MummertMedia\ContaoMeilisearchBundle\Cron\MeilisearchIndexCron:
autowire: true
autoconfigure: false
tags:
- { name: contao.cron, interval: daily, method: __invoke }
MummertMedia\ContaoMeilisearchBundle\Controller\FrontendModule\MeilisearchSearchController:
autowire: true
autoconfigure: false
+1 -1
View File
@@ -2,7 +2,7 @@
use Contao\DC_Table;
$GLOBALS['TL_DCA']['tl_search_pdf'] = [
$GLOBALS['TL_DCA']['tl_search_files'] = [
'config' => [
'dataContainer' => DC_Table::class,
'sql' => [
+45 -22
View File
@@ -4,8 +4,11 @@ use Contao\CoreBundle\DataContainer\PaletteManipulator;
use Contao\System;
/**
* -------------------------------------------------
* Fields
* -------------------------------------------------
*/
$GLOBALS['TL_DCA']['tl_settings']['fields']['meilisearch_host'] = [
'inputType' => 'text',
'eval' => [
@@ -59,17 +62,9 @@ $GLOBALS['TL_DCA']['tl_settings']['fields']['meilisearch_imagesize'] = [
'chosen' => true,
'includeBlankOption' => true,
],
// 🔥 DAS HAT GEFEHLT
'sql' => "int(10) unsigned NOT NULL default 0",
];
$GLOBALS['TL_DCA']['tl_settings']['fields']['meilisearch_index_past_events'] = [
'inputType' => 'checkbox',
'eval' => [
'tl_class' => 'w50 clr',
],
];
$GLOBALS['TL_DCA']['tl_settings']['fields']['meilisearch_fallback_image'] = [
'inputType' => 'fileTree',
'eval' => [
@@ -80,25 +75,54 @@ $GLOBALS['TL_DCA']['tl_settings']['fields']['meilisearch_fallback_image'] = [
'sql' => "varbinary(16) NULL",
];
$GLOBALS['TL_DCA']['tl_settings']['fields']['meilisearch_index_pdfs'] = [
'label' => &$GLOBALS['TL_LANG']['tl_settings']['meilisearch_index_pdfs'],
$GLOBALS['TL_DCA']['tl_settings']['fields']['meilisearch_index_past_events'] = [
'inputType' => 'checkbox',
'eval' => [
'tl_class' => 'w50',
'tl_class' => 'w50 clr',
],
'sql' => "char(1) NOT NULL default '1'",
];
$GLOBALS['TL_DCA']['tl_settings']['fields']['meilisearch_index_office'] = [
'label' => &$GLOBALS['TL_LANG']['tl_settings']['meilisearch_index_office'],
'inputType' => 'checkbox',
'eval' => ['tl_class' => 'w50'],
'sql' => "char(1) NOT NULL default '0'",
];
/**
* Palette
* -------------------------------------------------
* Datei-Indexierung (Tika)
* -------------------------------------------------
*/
$GLOBALS['TL_DCA']['tl_settings']['fields']['meilisearch_index_files'] = [
'inputType' => 'checkbox',
'eval' => [
'tl_class' => 'w50',
'submitOnChange' => true,
],
'sql' => "char(1) NOT NULL default '0'",
];
$GLOBALS['TL_DCA']['tl_settings']['fields']['meilisearch_tika_url'] = [
'inputType' => 'text',
'eval' => [
'rgxp' => 'url',
'mandatory' => true,
'tl_class' => 'w50 clr',
],
];
/**
* -------------------------------------------------
* Selector / Subpalette
* -------------------------------------------------
*/
$GLOBALS['TL_DCA']['tl_settings']['palettes']['__selector__'][] = 'meilisearch_index_files';
$GLOBALS['TL_DCA']['tl_settings']['subpalettes']['meilisearch_index_files']
= 'meilisearch_tika_url';
/**
* -------------------------------------------------
* Palette
* -------------------------------------------------
*/
PaletteManipulator::create()
->addLegend('meilisearch_legend', null, PaletteManipulator::POSITION_AFTER, true)
->addField('meilisearch_host', 'meilisearch_legend')
@@ -108,6 +132,5 @@ PaletteManipulator::create()
->addField('meilisearch_imagesize', 'meilisearch_legend')
->addField('meilisearch_fallback_image', 'meilisearch_legend')
->addField('meilisearch_index_past_events', 'meilisearch_legend')
->addField('meilisearch_index_pdfs', 'meilisearch_legend')
->addField('meilisearch_index_office', 'meilisearch_legend')
->addField('meilisearch_index_files', 'meilisearch_legend')
->applyToPalette('default', 'tl_settings');
@@ -28,10 +28,10 @@ $GLOBALS['TL_LANG']['tl_settings']['meilisearch_index_past_events'][0]
$GLOBALS['TL_LANG']['tl_settings']['meilisearch_index_past_events'][1]
= 'Vergangene Kalender-Events werden ebenfalls in Meilisearch indexiert.';
$GLOBALS['TL_LANG']['tl_settings']['meilisearch_index_pdfs'] = [
'PDFs indexieren',
'Aktiviert die Indexierung von PDF-Dateien für die Suche.',
$GLOBALS['TL_LANG']['tl_settings']['meilisearch_index_files'] = [
'Dateien indexieren',
'Aktiviert die Indexierung von PDF-Dateien sowie DOCX, XLSX und PPTX.',
];
$GLOBALS['TL_LANG']['tl_settings']['meilisearch_index_office']
= ['Office-Dateien indexieren', 'DOCX, XLSX und PPTX in die Suche aufnehmen.'];
$GLOBALS['TL_LANG']['tl_settings']['meilisearch_tika_url']
= ['Apache Tika URL', 'URL der Apache Tika Instanz (z. B. https://tika.domain.tld).'];