first commit

src/Knowledge/Ingest/KnowledgeIngestService.php

@@ -0,0 +1,39 @@
+<?php
+// src/Knowledge/Ingest/KnowledgeIngestService.php
+
+declare(strict_types=1);
+
+namespace App\Knowledge\Ingest;
+
+final class KnowledgeIngestService
+{
+    public function __construct(
+        private DocumentLoader   $loader,
+        private SimpleChunker    $chunker,
+        private ChunkWriter      $writer,
+        private ChunkIndexWriter $indexWriter,
+    )
+    {
+    }
+
+    /** @return string[] written chunk filenames */
+    public function ingestFile(string $path, bool $optimize = false): array
+    {
+        $text = $this->loader->load($path);
+
+        if ($optimize) {
+            $text = preg_replace("/\n{3,}/", "\n\n", $text);
+            $text = preg_replace("/[ \t]+$/m", "", $text);
+        }
+
+        $sourceHash = sha1($text);
+        $sourceName = basename($path);
+
+        if ($this->indexWriter->hasSourceHash($sourceName, $sourceHash)) {
+            return [];
+        }
+
+        $chunks = $this->chunker->chunk($text);
+        return $this->writer->write($sourceName, $chunks, $sourceHash);
+    }
+}