Adjust to assistants v2

sitegeist · Jul 15, 2024 · df5f824 · df5f824
1 parent 8aa83e4
commit df5f824
Show file tree

Hide file tree

Showing 7 changed files with 128 additions and 40 deletions.
diff --git a/Classes/Domain/AssistantDepartment.php b/Classes/Domain/AssistantDepartment.php
@@ -12,10 +12,10 @@
 use Sitegeist\Chatterbox\Domain\Instruction\InstructionCollection;
 use Sitegeist\Chatterbox\Domain\Instruction\InstructionContract;
 use Sitegeist\Chatterbox\Domain\Instruction\Manual;
-use Sitegeist\Chatterbox\Domain\Knowledge\KnowledgeFilename;
 use Sitegeist\Chatterbox\Domain\Knowledge\KnowledgeSourceDiscriminator;
 use Sitegeist\Chatterbox\Domain\Knowledge\KnowledgeSourceName;
 use Sitegeist\Chatterbox\Domain\Knowledge\Library;
+use Sitegeist\Chatterbox\Domain\Knowledge\VectorStoreName;
 use Sitegeist\Chatterbox\Domain\Tools\Toolbox;
 use Sitegeist\Chatterbox\Domain\Tools\ToolCollection;
 use Sitegeist\Chatterbox\Domain\Tools\ToolContract;
@@ -113,7 +113,7 @@ public function updateAssistant(AssistantRecord $assistantRecord): void
                 'description' => $assistantRecord->description ?: '',
                 'instructions' => $assistantRecord->instructions ?: '',
                 'tools' => $this->createToolConfiguration($assistantRecord),
-                'file_ids' => $this->createFileIdConfiguration($assistantRecord),
+                'tool_resources' => $this->createToolResourcesConfiguration($assistantRecord),
                 'metadata' => $this->createMetadataConfiguration($assistantRecord),
             ]
         );
@@ -140,7 +140,7 @@ private function createToolConfiguration(AssistantRecord $assistantRecord): arra
         $tools = [];
         if (!empty($assistantRecord->selectedSourcesOfKnowledge)) {
             $tools[] = [
-                'type' => 'retrieval'
+                'type' => 'file_search'
             ];
         }
         foreach ($assistantRecord->selectedTools as $toolId) {
@@ -166,36 +166,33 @@ private function createToolConfiguration(AssistantRecord $assistantRecord): arra
     /**
      * @return string[]
      */
-    private function createFileIdConfiguration(AssistantRecord $assistantRecord): array
+    private function createToolResourcesConfiguration(AssistantRecord $assistantRecord): array
     {
-        $fileListResponse = $this->client->files()->list();
-        $fileIds = [];
+        $vectorStoreListResponse = $this->client->vectorStores()->list();
+        $vectorStoreIds = [];
         foreach ($assistantRecord->selectedSourcesOfKnowledge as $knowledgeSourceName) {
-            $knowledgeSourceNameObject = new KnowledgeSourceName($knowledgeSourceName);
             $knowledgeSourceDiscriminator = new KnowledgeSourceDiscriminator(
                 $this->organizationDiscriminator,
-                $knowledgeSourceNameObject
+                new KnowledgeSourceName($knowledgeSourceName)
             );
-            $latestFilename = null;
-            foreach ($fileListResponse->data as $fileResponse) {
-                $knowledgeFilename = KnowledgeFilename::tryFromSystemFileName($fileResponse->filename);
-                if ($knowledgeFilename === null) {
+            $latestVectorStoreName = null;
+            foreach ($vectorStoreListResponse->data as $vectorStoreResponse) {
+                $vectorStoreName = VectorStoreName::tryFromNullableString($vectorStoreResponse->name);
+                if (!$vectorStoreName?->knowledgeSourceDiscriminator->equals($knowledgeSourceDiscriminator)) {
                     continue;
                 }
 
-                if ($latestFilename instanceof KnowledgeFilename) {
-                    if ($knowledgeFilename->takesPrecedenceOver($latestFilename)) {
-                        $latestFilename = $knowledgeFilename;
-                        $fileIds[$knowledgeSourceName] = $fileResponse->id;
-                    }
-                } else {
-                    if ($knowledgeFilename->knowledgeSourceDiscriminator->equals($knowledgeSourceDiscriminator)) {
-                        $latestFilename = $knowledgeFilename;
-                        $fileIds[$knowledgeSourceName] = $fileResponse->id;
-                    }
+                if (!$latestVectorStoreName || $vectorStoreName->takesPrecedenceOver($latestVectorStoreName)) {
+                    $latestVectorStoreName = $vectorStoreName;
+                    $vectorStoreIds[$knowledgeSourceName] = $vectorStoreResponse->id;
                 }
             }
         }
-        return array_values($fileIds);
+
+        return [
+            'file_search' => [
+                'vector_store_ids' => array_values($vectorStoreIds)
+            ],
+        ];
     }
 }
diff --git a/Classes/Domain/AssistantRecord.php b/Classes/Domain/AssistantRecord.php
@@ -7,8 +7,8 @@
 use OpenAI\Responses\Assistants\AssistantResponse;
 use Neos\Flow\Annotations as Flow;
 use OpenAI\Responses\Assistants\AssistantResponseToolCodeInterpreter;
+use OpenAI\Responses\Assistants\AssistantResponseToolFileSearch;
 use OpenAI\Responses\Assistants\AssistantResponseToolFunction;
-use OpenAI\Responses\Assistants\AssistantResponseToolRetrieval;
 
 #[Flow\Proxy(false)]
 final class AssistantRecord
@@ -19,7 +19,7 @@ final class AssistantRecord
      * @param string[] $selectedTools
      * @param string[] $selectedSourcesOfKnowledge
      * @param string[] $selectedInstructions
-     * @param string[] $fileIds
+     * @param mixed[] $toolResources
      */
     public function __construct(
         public readonly string $id,
@@ -32,7 +32,7 @@ public function __construct(
         public readonly array $selectedTools = [],
         public readonly array $selectedSourcesOfKnowledge = [],
         public readonly array $selectedInstructions = [],
-        public readonly array $fileIds = [],
+        public readonly array $toolResources = [],
     ) {
     }
 
@@ -48,12 +48,12 @@ public static function fromAssistantResponse(AssistantResponse $response): self
             $response->name,
             $response->description,
             $response->instructions,
-            array_map(fn(AssistantResponseToolCodeInterpreter|AssistantResponseToolRetrieval|AssistantResponseToolFunction $item) => $item->toArray(), $response->tools),
+            array_map(fn(AssistantResponseToolCodeInterpreter|AssistantResponseToolFileSearch|AssistantResponseToolFunction $item) => $item->toArray(), $response->tools),
             $response->metadata,
             $selectedTools,
             $selectedSourcesOfKnowledge,
             $selectedInstructions,
-            $response->fileIds,
+            $response->toolResources?->toArray() ?: [],
         );
     }
 }
diff --git a/Classes/Domain/Knowledge/Library.php b/Classes/Domain/Knowledge/Library.php
@@ -67,6 +67,7 @@ public function updateSourceOfKnowledge(SourceOfKnowledgeContract $sourceOfKnowl
             $sourceOfKnowledge->getName()
         );
         $filename = KnowledgeFilename::forKnowledgeSource($knowledgeSourceDiscriminator);
+        $vectorStoreName = VectorStoreName::forKnowledgeSource($knowledgeSourceDiscriminator);
         $this->databaseConnection->transactional(function () use ($content, $knowledgeSourceDiscriminator) {
             $this->databaseConnection->delete(
                 self::TABLE_NAME,
@@ -89,20 +90,34 @@ public function updateSourceOfKnowledge(SourceOfKnowledgeContract $sourceOfKnowl
         $path = $this->environment->getPathToTemporaryDirectory() . '/' . $filename->toSystemFilename();
         \file_put_contents($path, (string)$content);
 
-        $this->client->files()->upload([
+        $createFileResponse = $this->client->files()->upload([
             'file' => fopen($path, 'r'),
             'purpose' => 'assistants'
         ]);
         \unlink($path);
+
+        $this->client->vectorStores()->create([
+            'file_ids' => [
+                $createFileResponse->id
+            ],
+            'name' => $vectorStoreName->toString(),
+        ]);
     }
 
     public function cleanKnowledgePool(AssistantDepartment $assistantDepartment): void
     {
         $filesListResponse = $this->client->files()->list();
+        $vectorStoreListResponse = $this->client->vectorStores()->list();
 
+        $usedVectorStoreIds = [];
         $usedFileIds = [];
         foreach ($assistantDepartment->findAllRecords() as $assistant) {
-            $usedFileIds = array_merge($usedFileIds, $assistant->fileIds);
+            foreach ($assistant->toolResources['file_search']['vector_store_ids'] ?? [] as $vectorStoreId) {
+                $usedVectorStoreIds[] = $vectorStoreId;
+                foreach ($this->client->vectorStores()->files()->list($vectorStoreId)->data as $vectorStoreFileResponse) {
+                    $usedFileIds[] = $vectorStoreFileResponse->id;
+                }
+            }
         }
 
         foreach ($filesListResponse->data as $fileResponse) {
@@ -117,6 +132,19 @@ public function cleanKnowledgePool(AssistantDepartment $assistantDepartment): vo
                 $this->client->files()->delete($fileResponse->id);
             }
         }
+
+        foreach ($vectorStoreListResponse->data as $vectorStoreResponse) {
+            $vectorStoreName = VectorStoreName::tryFromNullableString($vectorStoreResponse->name);
+            if (
+                $vectorStoreName === null
+                || $this->organizationDiscriminator->equals($vectorStoreName->knowledgeSourceDiscriminator->organizationDiscriminator) === false
+            ) {
+                continue;
+            }
+            if (!in_array($vectorStoreResponse->id, $usedVectorStoreIds)) {
+                $this->client->vectorStores()->delete($vectorStoreResponse->id);
+            }
+        }
     }
 
     private function instantiateSourceOfKnowledge(string $name): SourceOfKnowledgeContract

diff --git a/Classes/Domain/Knowledge/QuoteString.php b/Classes/Domain/Knowledge/QuoteString.php
@@ -18,9 +18,12 @@ private function __construct(
     ) {
     }
 
-    public static function fromFileCitationObject(
+    public static function tryFromFileCitationObject(
         ThreadMessageResponseContentTextAnnotationFileCitationObject $fileCitationObject
-    ): self {
+    ): ?self {
+        if ($fileCitationObject->fileCitation->quote === null || $fileCitationObject->fileCitation->quote === '') {
+            return null;
+        }
         return new self($fileCitationObject->fileCitation->quote);
     }
 
@@ -40,9 +43,4 @@ public function wtf8Encode(): string
 
         return '%' . $value . '%';
     }
-
-    public function isEmpty(): bool
-    {
-        return $this->value === '';
-    }
 }
diff --git a/Classes/Domain/Knowledge/SourceOfKnowledgeCollection.php b/Classes/Domain/Knowledge/SourceOfKnowledgeCollection.php
@@ -56,8 +56,8 @@ public function resolveQuotations(
         $unresolvedQuotations = [];
         foreach ($annotations as $annotation) {
             if ($annotation instanceof ThreadMessageResponseContentTextAnnotationFileCitationObject) {
-                $quoteString = QuoteString::fromFileCitationObject($annotation);
-                if ($quoteString->isEmpty()) {
+                $quoteString = QuoteString::tryFromFileCitationObject($annotation);
+                if (!$quoteString) {
                     $unresolvedQuotations[] = new UnresolvedQuotation($annotation->text);
                     continue;
                 }

diff --git a/Classes/Domain/Knowledge/VectorStoreName.php b/Classes/Domain/Knowledge/VectorStoreName.php
@@ -0,0 +1,64 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Sitegeist\Chatterbox\Domain\Knowledge;
+
+use Neos\Flow\Annotations as Flow;
+use Sitegeist\Chatterbox\Domain\OrganizationDiscriminator;
+
+#[Flow\Proxy(false)]
+final readonly class VectorStoreName
+{
+    public function __construct(
+        public KnowledgeSourceDiscriminator $knowledgeSourceDiscriminator,
+        public int $timestamp,
+    ) {
+    }
+
+    public static function forKnowledgeSource(KnowledgeSourceDiscriminator $knowledgeSourceDiscriminator): self
+    {
+        return new self(
+            $knowledgeSourceDiscriminator,
+            time()
+        );
+    }
+
+    public static function tryFromNullableString(?string $value): ?self
+    {
+        if ($value === null) {
+            return null;
+        }
+
+        if (substr_count($value, '-') !== 2) {
+            return null;
+        }
+        list($discriminator, $sourceName, $timestamp) = explode('-', $value);
+
+        if (!is_numeric($timestamp)) {
+            return null;
+        }
+
+        return new self(
+            new KnowledgeSourceDiscriminator(
+                new OrganizationDiscriminator($discriminator),
+                new KnowledgeSourceName($sourceName)
+            ),
+            (int)$timestamp
+        );
+    }
+
+    public function takesPrecedenceOver(self $other): bool
+    {
+        if ($this->knowledgeSourceDiscriminator->equals($other->knowledgeSourceDiscriminator) === false) {
+            return false;
+        }
+
+        return $this->timestamp > $other->timestamp;
+    }
+
+    public function toString(): string
+    {
+        return $this->knowledgeSourceDiscriminator->toString() . '-' . $this->timestamp;
+    }
+}
diff --git a/composer.json b/composer.json
@@ -4,8 +4,9 @@
     "name": "sitegeist/chatterbox",
     "license": "GPL-3.0-or-later",
     "require": {
+        "php": "^8.2",
         "neos/neos": "*",
-        "sitegeist/flow-openaiclientfactory": "dev-main",
+        "sitegeist/flow-openaiclientfactory": "~0.1.0",
         "symfony/http-client": "^6.4",
         "league/commonmark": "~2.4",
         "league/html-to-markdown": "^5.0"