From 44be40a24de93a2b136849014dc3e051ba7ebcc2 Mon Sep 17 00:00:00 2001
From: team2 <team2@mitho-media.de>
Date: Fri, 27 Feb 2026 15:49:01 +0100
Subject: [PATCH] cleanup code

---
 src/Ingest/ChunkWriteService.php       | 28 --------------------------
 src/Knowledge/ChunkManager.php         |  5 -----
 src/Knowledge/Ingest/SimpleChunker.php | 26 +++++++++---------------
 src/Service/IngestOrchestrator.php     | 10 ++++-----
 4 files changed, 14 insertions(+), 55 deletions(-)

diff --git a/src/Ingest/ChunkWriteService.php b/src/Ingest/ChunkWriteService.php
index 4d445da..2967418 100644
--- a/src/Ingest/ChunkWriteService.php
+++ b/src/Ingest/ChunkWriteService.php
@@ -14,11 +14,6 @@ final readonly class ChunkWriteService
         private ChunkManager $chunkManager,
     ) {}
 
-    public function getIndexPath(): string
-    {
-        return $this->chunkManager->getIndexPath();
-    }
-
     public function countAllChunks(): int
     {
         return $this->chunkManager->countAllChunks();
@@ -37,29 +32,6 @@ final readonly class ChunkWriteService
         $this->chunkManager->appendChunks($chunks);
     }
 
-    /**
-     * Lokaler Ingest für eine einzelne DocumentVersion.
-     *
-     * Ablauf:
-     * 1. Entfernt bestehende Chunks dieses Dokuments
-     * 2. Appendet neue Chunks
-     *
-     * @param iterable<array<string,mixed>> $chunks
-     */
-    public function writeForDocumentVersion(
-        DocumentVersion $version,
-        iterable $chunks
-    ): void {
-        $documentId = $version->getDocument()->getId();
-
-        if (!$documentId instanceof Uuid) {
-            throw new \RuntimeException('Document ID must be a Uuid instance');
-        }
-
-        $this->chunkManager->compactByDocument($documentId);
-        $this->chunkManager->appendChunks($chunks);
-    }
-
     /**
      * Vollständiger Rewrite des NDJSON-Index (Global Reindex).
      *
diff --git a/src/Knowledge/ChunkManager.php b/src/Knowledge/ChunkManager.php
index 5409b57..cc3712f 100644
--- a/src/Knowledge/ChunkManager.php
+++ b/src/Knowledge/ChunkManager.php
@@ -17,11 +17,6 @@ final class ChunkManager
         $this->indexPath = rtrim($projectDir, '/') . $relativeIndexPath;
     }
 
-    public function getIndexPath(): string
-    {
-        return $this->indexPath;
-    }
-
     // ============================================================
     // COUNT (Streaming, robust)
     // ============================================================
diff --git a/src/Knowledge/Ingest/SimpleChunker.php b/src/Knowledge/Ingest/SimpleChunker.php
index 3445537..043e6f5 100644
--- a/src/Knowledge/Ingest/SimpleChunker.php
+++ b/src/Knowledge/Ingest/SimpleChunker.php
@@ -6,15 +6,16 @@ declare(strict_types=1);
 namespace App\Knowledge\Ingest;
 
 use App\Index\IndexConfigurationProvider;
+use App\Knowledge\Text\TextNormalizer;
 
-final class SimpleChunker
+final readonly class SimpleChunker
 {
-    private IndexConfigurationProvider $configurationProvider;
 
     public function __construct(
-        IndexConfigurationProvider $configurationProvider
-    ) {
-        $this->configurationProvider = $configurationProvider;
+        private IndexConfigurationProvider $configurationProvider,
+        private TextNormalizer             $textNormalizer
+    )
+    {
     }
 
     /** @return string[] */
@@ -25,7 +26,7 @@ final class SimpleChunker
         $maxWords = $config->getChunkSize();
         $overlapWords = $config->getChunkOverlap();
 
-        $text = $this->normalize($text);
+        $text = $this->textNormalizer->normalize($text);
         if ($text === '') {
             return [];
         }
@@ -60,7 +61,7 @@ final class SimpleChunker
             $wordEnd = min($wordPos + $maxWords, $totalWords);
 
             $tokenStart = $wordTokenIndexes[$wordPos];
-            $tokenEnd   = $wordTokenIndexes[$wordEnd - 1] + 1;
+            $tokenEnd = $wordTokenIndexes[$wordEnd - 1] + 1;
 
             $tokenEnd = $this->adjustCutToBoundary($tokens, $tokenStart, $tokenEnd);
 
@@ -84,15 +85,6 @@ final class SimpleChunker
         return $this->dedupe($chunks);
     }
 
-    private function normalize(string $text): string
-    {
-        $text = str_replace(["\r\n", "\r"], "\n", $text);
-        $text = preg_replace("/[ \t]+/u", " ", $text);
-        $text = preg_replace("/\n{3,}/u", "\n\n", $text);
-
-        return trim((string) $text);
-    }
-
     private function adjustCutToBoundary(array $tokens, int $start, int $end): int
     {
         $startToken = $tokens[$start] ?? '';
@@ -122,7 +114,7 @@ final class SimpleChunker
     private function dedupe(array $chunks): array
     {
         $seen = [];
-        $out  = [];
+        $out = [];
 
         foreach ($chunks as $chunk) {
             $key = mb_strtolower(
diff --git a/src/Service/IngestOrchestrator.php b/src/Service/IngestOrchestrator.php
index e4a914d..3fa73d1 100644
--- a/src/Service/IngestOrchestrator.php
+++ b/src/Service/IngestOrchestrator.php
@@ -9,13 +9,13 @@ use App\Ingest\IngestFlow;
 use Doctrine\ORM\EntityManagerInterface;
 use Symfony\Component\Uid\Uuid;
 
-final class IngestOrchestrator
+final readonly class IngestOrchestrator
 {
     public function __construct(
-        private readonly LockService            $lockService,
-        private readonly IngestJobService       $jobService,
-        private readonly EntityManagerInterface $em,
-        private readonly IngestFlow             $ingestFlow,
+        private LockService            $lockService,
+        private IngestJobService       $jobService,
+        private EntityManagerInterface $em,
+        private IngestFlow             $ingestFlow,
     )
     {
     }