Vectors: Built content vector indexing system

diff --git a/app/Config/services.php b/app/Config/services.php
index d73458231506cb56cc16ec81d4329709b14dbf95..a34b243f07d20606ad79a6b46cf80e286ac8c9b8 100644 (file)
--- a/app/Config/services.php
+++ b/app/Config/services.php
@@ -22,6 +22,16 @@ return [
     // Callback URL for social authentication methods
     'callback_url' => env('APP_URL', false),
 
     // Callback URL for social authentication methods
     'callback_url' => env('APP_URL', false),
 

+    // LLM Service
+    // Options: openai
+    'llm' => env('LLM_SERVICE', ''),
+
+    // OpenAI API-compatible service details
+    'openai' => [
+        'endpoint' => env('OPENAI_ENDPOINT', 'https://api.openai.com'),
+        'key' => env('OPENAI_KEY', ''),
+    ],
+

     'github'   => [
         'client_id'     => env('GITHUB_APP_ID', false),
         'client_secret' => env('GITHUB_APP_SECRET', false),
     'github'   => [
         'client_id'     => env('GITHUB_APP_ID', false),
         'client_secret' => env('GITHUB_APP_SECRET', false),

diff --git a/app/Search/SearchIndex.php b/app/Search/SearchIndex.php
index 36f71f6ccc7759cd09d3de29e8464a187c1507a6..9b34fa04e28a18babf77826e742015ef74b65117 100644 (file)
--- a/app/Search/SearchIndex.php
+++ b/app/Search/SearchIndex.php
@@ -6,6 +6,8 @@ use BookStack\Activity\Models\Tag;
 use BookStack\Entities\EntityProvider;
 use BookStack\Entities\Models\Entity;
 use BookStack\Entities\Models\Page;
 use BookStack\Entities\EntityProvider;
 use BookStack\Entities\Models\Entity;
 use BookStack\Entities\Models\Page;

+use BookStack\Search\Vectors\StoreEntityVectorsJob;
+use BookStack\Search\Vectors\VectorQueryServiceProvider;

 use BookStack\Util\HtmlDocument;
 use DOMNode;
 use Illuminate\Database\Eloquent\Builder;
 use BookStack\Util\HtmlDocument;
 use DOMNode;
 use Illuminate\Database\Eloquent\Builder;


@@ -25,7 +27,7 @@ class SearchIndex
     public static string $softDelimiters = ".-";
 
     public function __construct(
     public static string $softDelimiters = ".-";
 
     public function __construct(

-        protected EntityProvider $entityProvider
+        protected EntityProvider $entityProvider,

     ) {
     }
 
     ) {
     }
 


@@ -37,6 +39,10 @@ class SearchIndex
         $this->deleteEntityTerms($entity);
         $terms = $this->entityToTermDataArray($entity);
         $this->insertTerms($terms);
         $this->deleteEntityTerms($entity);
         $terms = $this->entityToTermDataArray($entity);
         $this->insertTerms($terms);

+
+        if (VectorQueryServiceProvider::isEnabled()) {
+            dispatch(new StoreEntityVectorsJob($entity));
+        }

     }
 
     /**
     }
 
     /**


@@ -47,9 +53,15 @@ class SearchIndex
     public function indexEntities(array $entities): void
     {
         $terms = [];
     public function indexEntities(array $entities): void
     {
         $terms = [];

+        $vectorQueryEnabled = VectorQueryServiceProvider::isEnabled();
+

         foreach ($entities as $entity) {
             $entityTerms = $this->entityToTermDataArray($entity);
             array_push($terms, ...$entityTerms);
         foreach ($entities as $entity) {
             $entityTerms = $this->entityToTermDataArray($entity);
             array_push($terms, ...$entityTerms);

+
+            if ($vectorQueryEnabled) {
+                dispatch(new StoreEntityVectorsJob($entity));
+            }

         }
 
         $this->insertTerms($terms);
         }
 
         $this->insertTerms($terms);

diff --git a/app/Search/Vectors/EntityVectorGenerator.php b/app/Search/Vectors/EntityVectorGenerator.php
new file mode 100644 (file)
index 0000000..8a49187
--- /dev/null
+++ b/app/Search/Vectors/EntityVectorGenerator.php
@@ -0,0 +1,84 @@
+<?php
+
+namespace BookStack\Search\Vectors;
+
+use BookStack\Entities\Models\Entity;
+use BookStack\Search\Vectors\Services\VectorQueryService;
+use Illuminate\Support\Facades\DB;
+
+class EntityVectorGenerator
+{
+    public function __construct(
+        protected VectorQueryServiceProvider $vectorQueryServiceProvider
+    ) {
+    }
+
+    public function generateAndStore(Entity $entity): void
+    {
+        $vectorService = $this->vectorQueryServiceProvider->get();
+
+        $text = $this->entityToPlainText($entity);
+        $chunks = $this->chunkText($text);
+        $embeddings = $this->chunksToEmbeddings($chunks, $vectorService);
+
+        $this->deleteExistingEmbeddingsForEntity($entity);
+        $this->storeEmbeddings($embeddings, $chunks, $entity);
+    }
+
+    protected function deleteExistingEmbeddingsForEntity(Entity $entity): void
+    {
+        SearchVector::query()
+            ->where('entity_type', '=', $entity->getMorphClass())
+            ->where('entity_id', '=', $entity->id)
+            ->delete();
+    }
+
+    protected function storeEmbeddings(array $embeddings, array $textChunks, Entity $entity): void
+    {
+        $toInsert = [];
+
+        foreach ($embeddings as $index => $embedding) {
+            $text = $textChunks[$index];
+            $toInsert[] = [
+                'entity_id' => $entity->id,
+                'entity_type' => $entity->getMorphClass(),
+                'embedding' => DB::raw('STRING_TO_VECTOR("[' . implode(',', $embedding) . ']")'),
+                'text' => $text,
+            ];
+        }
+
+        // TODO - Chunk inserts
+        SearchVector::query()->insert($toInsert);
+    }
+
+    /**
+     * @param string[] $chunks
+     * @return float[] array
+     */
+    protected function chunksToEmbeddings(array $chunks, VectorQueryService $vectorQueryService): array
+    {
+        $embeddings = [];
+        foreach ($chunks as $index => $chunk) {
+            $embeddings[$index] = $vectorQueryService->generateEmbeddings($chunk);
+        }
+        return $embeddings;
+    }
+
+    /**
+     * @return string[]
+     */
+    protected function chunkText(string $text): array
+    {
+        // TODO - Join adjacent smaller chunks up
+        return array_filter(array_map(function (string $section): string {
+            return trim($section);
+        }, explode("\n", $text)));
+    }
+
+    protected function entityToPlainText(Entity $entity): string
+    {
+        $text = $entity->name . "\n\n" . $entity->{$entity->textField};
+        // TODO - Add tags
+        return $text;
+    }
+}

diff --git a/app/Search/Vectors/SearchVector.php b/app/Search/Vectors/SearchVector.php
new file mode 100644 (file)
index 0000000..4a5555f
--- /dev/null
+++ b/app/Search/Vectors/SearchVector.php
@@ -0,0 +1,16 @@
+<?php
+
+namespace BookStack\Search\Vectors;
+
+use Illuminate\Database\Eloquent\Model;
+
+/**
+ * @property string $entity_type
+ * @property int $entity_id
+ * @property string $text
+ * @property string $embedding
+ */
+class SearchVector extends Model
+{
+    public $timestamps = false;
+}

diff --git a/app/Search/Vectors/Services/OpenAiVectorQueryService.php b/app/Search/Vectors/Services/OpenAiVectorQueryService.php
new file mode 100644 (file)
index 0000000..8d29109
--- /dev/null
+++ b/app/Search/Vectors/Services/OpenAiVectorQueryService.php
@@ -0,0 +1,36 @@
+<?php
+
+namespace BookStack\Search\Vectors\Services;
+
+use BookStack\Http\HttpRequestService;
+
+class OpenAiVectorQueryService implements VectorQueryService
+{
+    public function __construct(
+        protected string $endpoint,
+        protected string $key,
+        protected HttpRequestService $http,
+    ) {
+    }
+
+    protected function jsonRequest(string $method, string $uri, array $data): array
+    {
+        $fullUrl = rtrim($this->endpoint, '/') . '/' . ltrim($uri, '/');
+        $client = $this->http->buildClient(10);
+        $request = $this->http->jsonRequest($method, $fullUrl, $data)
+            ->withHeader('Authorization', 'Bearer ' . $this->key);
+
+        $response = $client->sendRequest($request);
+        return json_decode($response->getBody()->getContents(), true);
+    }
+
+    public function generateEmbeddings(string $text): array
+    {
+        $response = $this->jsonRequest('POST', 'v1/embeddings', [
+            'input' => $text,
+            'model' => 'text-embedding-3-small',
+        ]);
+
+        return $response['data'][0]['embedding'];
+    }
+}

diff --git a/app/Search/Vectors/Services/VectorQueryService.php b/app/Search/Vectors/Services/VectorQueryService.php
new file mode 100644 (file)
index 0000000..2cc4ed0
--- /dev/null
+++ b/app/Search/Vectors/Services/VectorQueryService.php
@@ -0,0 +1,12 @@
+<?php
+
+namespace BookStack\Search\Vectors\Services;
+
+interface VectorQueryService
+{
+    /**
+     * Generate embedding vectors from the given chunk of text.
+     * @return float[]
+     */
+    public function generateEmbeddings(string $text): array;
+}

diff --git a/app/Search/Vectors/StoreEntityVectorsJob.php b/app/Search/Vectors/StoreEntityVectorsJob.php
new file mode 100644 (file)
index 0000000..e96fc34
--- /dev/null
+++ b/app/Search/Vectors/StoreEntityVectorsJob.php
@@ -0,0 +1,28 @@
+<?php
+
+namespace BookStack\Search\Vectors;
+
+use BookStack\Entities\Models\Entity;
+use Illuminate\Contracts\Queue\ShouldQueue;
+use Illuminate\Foundation\Queue\Queueable;
+
+class StoreEntityVectorsJob implements ShouldQueue
+{
+    use Queueable;
+
+    /**
+     * Create a new job instance.
+     */
+    public function __construct(
+        protected Entity $entity
+    ) {
+    }
+
+    /**
+     * Execute the job.
+     */
+    public function handle(EntityVectorGenerator $generator): void
+    {
+        $generator->generateAndStore($this->entity);
+    }
+}

diff --git a/app/Search/Vectors/VectorQueryServiceProvider.php b/app/Search/Vectors/VectorQueryServiceProvider.php
new file mode 100644 (file)
index 0000000..c700307
--- /dev/null
+++ b/app/Search/Vectors/VectorQueryServiceProvider.php
@@ -0,0 +1,38 @@
+<?php
+
+namespace BookStack\Search\Vectors;
+
+use BookStack\Http\HttpRequestService;
+use BookStack\Search\Vectors\Services\OpenAiVectorQueryService;
+use BookStack\Search\Vectors\Services\VectorQueryService;
+
+class VectorQueryServiceProvider
+{
+    public function __construct(
+        protected HttpRequestService $http,
+    ) {
+    }
+
+    public function get(): VectorQueryService
+    {
+        $service = $this->getServiceName();
+
+        if ($service === 'openai') {
+            $key = config('services.openai.key');
+            $endpoint = config('services.openai.endpoint');
+            return new OpenAiVectorQueryService($endpoint, $key, $this->http);
+        }
+
+        throw new \Exception("No '{$service}' LLM service found");
+    }
+
+    protected static function getServiceName(): string
+    {
+        return strtolower(config('services.llm'));
+    }
+
+    public static function isEnabled(): bool
+    {
+        return !empty(static::getServiceName());
+    }
+}

diff --git a/database/migrations/2025_03_24_155748_create_search_vectors_table.php b/database/migrations/2025_03_24_155748_create_search_vectors_table.php
new file mode 100644 (file)
index 0000000..d7fb011
--- /dev/null
+++ b/database/migrations/2025_03_24_155748_create_search_vectors_table.php
@@ -0,0 +1,32 @@
+<?php
+
+use Illuminate\Database\Migrations\Migration;
+use Illuminate\Database\Schema\Blueprint;
+use Illuminate\Support\Facades\Schema;
+
+return new class extends Migration
+{
+    /**
+     * Run the migrations.
+     */
+    public function up(): void
+    {
+        // TODO - Handle compatibility with older databases that don't support vectors
+        Schema::create('search_vectors', function (Blueprint $table) {
+            $table->string('entity_type', 100);
+            $table->integer('entity_id');
+            $table->text('text');
+            $table->vector('embedding');
+
+            $table->index(['entity_type', 'entity_id']);
+        });
+    }
+
+    /**
+     * Reverse the migrations.
+     */
+    public function down(): void
+    {
+        Schema::dropIfExists('search_vectors');
+    }
+};