Setup some embedding functions and fix error handling

- Setup the OpenAIEmbeddingFunction, JinaEmbeddingFunction and HuggingFaceEmbeddingServerFunction

- Fix the way exceptions are being handled in the ApiCLient

- Added more tests to expect exceptions

- Added an examples folder for examples on usage

Author: CodeWithKyrian

.github/workflows/test.yml

@@ -25,7 +25,7 @@ jobs:
         uses: actions/checkout@v3
 
       - name: Cache dependencies
-        uses: actions/cache@v2
+        uses: actions/cache@v3
         with:
           path: ~/.composer/cache/files
           key: dependencies-php-${{ matrix.php }}-composer-${{ hashFiles('composer.json') }}

.gitignore

@@ -6,5 +6,4 @@
 *.swp
 *.swo
 playground/*
-.idea
-index.php
+.idea

examples/.gitignore

@@ -0,0 +1,2 @@
+vendor
+composer.lock

examples/composer.json

@@ -0,0 +1,27 @@
+{
+    "name": "kyrian/examples",
+    "type": "project",
+    "autoload": {
+        "psr-4": {
+        }
+    },
+    "authors": [
+        {
+            "name": "Kyrian Obikwelu",
+            "email": "koshnawaza@gmail.com"
+        }
+    ],
+    "repositories": [
+        {
+            "type": "path",
+            "url": "../",
+            "options": {
+                "symlink": true
+            }
+        }
+    ],
+    "require": {
+        "codewithkyrian/chromadb-php": "@dev",
+        "symfony/var-dumper": "^6.3"
+    }
+}

examples/index.php

@@ -0,0 +1,38 @@
+<?php
+
+declare(strict_types=1);
+
+require './vendor/autoload.php';
+
+use Codewithkyrian\ChromaDB\ChromaDB;
+use Codewithkyrian\ChromaDB\Embeddings\JinaEmbeddingFunction;
+
+$chroma = ChromaDB::factory()
+    ->withDatabase('test_database')
+    ->withTenant('test_tenant')
+    ->connect();
+
+$chroma->deleteAllCollections();
+
+$embeddingFunction = new JinaEmbeddingFunction(
+    'jina_8cbaafb9543e42f1a2fc7430d456c3faKKPc93W8Eur5T2XjAkryfwQ9TOv8'
+);
+
+$collection = $chroma->createCollection(
+    name: 'test_collection',
+    embeddingFunction: $embeddingFunction
+);
+
+$collection->add(
+    ids: ['hello', 'world'],
+    documents: ['This is a test document', 'The man is happy']
+);
+
+$queryResponse = $collection->query(
+    queryTexts: ['The man is excited'],
+    include: ['documents', 'distances']
+);
+
+dd($queryResponse->documents[0], $queryResponse->distances[0]);
+
+

src/Client.php

@@ -21,9 +21,6 @@ public function __construct(
     }
 
 
-    /**
-     * @throws Generated\Exceptions\ChromaApiExceptionInterface
-     */
     public function initDatabaseAndTenant(): void
     {
 
@@ -118,21 +115,18 @@ public function getOrCreateCollection(string $name, ?array $metadata = null, ?Em
     }
 
     /**
-     * Gets a collection with the specified name.
+     * Gets a collection with the specified name. Will raise an exception if the
+     * collection does not exist.
      *
      * @param string $name The name of the collection.
      * @param ?EmbeddingFunction $embeddingFunction Optional custom embedding function for the collection.
      *
-     * @return ?CollectionResource
+     * @return CollectionResource
      */
-    public function getCollection(string $name, ?EmbeddingFunction $embeddingFunction = null): ?CollectionResource
+    public function getCollection(string $name, ?EmbeddingFunction $embeddingFunction = null): CollectionResource
     {
         $collection = $this->apiClient->getCollection($name, $this->database, $this->tenant);
 
-        if ($collection === null) {
-            return null;
-        }
-
         return CollectionResource::make(
             $collection,
             $this->database,

…eddings/HuggingFaceEmbeddingFunction.php …s/HuggingFaceEmbeddingServerFunction.phpsrc/Embeddings/HuggingFaceEmbeddingFunction.php renamed to src/Embeddings/HuggingFaceEmbeddingServerFunction.php src/Embeddings/HuggingFaceEmbeddingFunction.php renamed to src/Embeddings/HuggingFaceEmbeddingServerFunction.php

@@ -8,25 +8,26 @@
 use GuzzleHttp\Client;
 use GuzzleHttp\Exception\GuzzleException;
 
-class HuggingFaceEmbeddingFunction implements EmbeddingFunction
+class HuggingFaceEmbeddingServerFunction implements EmbeddingFunction
 {
 
     public function __construct(
-        public readonly string $url,
+        public readonly string $baseUrl = 'http://localhost:8080',
     )
     {
     }
 
     public function generate(array $texts): array
     {
         $client = new Client([
+            'base_uri' => $this->baseUrl,
             'headers' => [
                 'Content-Type' => 'application/json',
             ]
         ]);
 
         try {
-            $response = $client->post($this->url, [
+            $response = $client->post('embed', [
                 'body' => json_encode([
                     'inputs' => $texts,
                 ])

src/Embeddings/JinaEmbeddingFunction.php

@@ -8,47 +8,47 @@
 use Codewithkyrian\ChromaDB\Embeddings\EmbeddingFunction;
 use GuzzleHttp\Client;
 use GuzzleHttp\Exception\GuzzleException;
+use Psr\Http\Client\ClientExceptionInterface;
 
 class JinaEmbeddingFunction implements EmbeddingFunction
 {
-
-    private const API_URL = 'https://api.jina.ai/v1/embeddings';
+    private Client $client;
 
     public function __construct(
         public readonly string $apiKey,
         public readonly string $model = 'jina-embeddings-v2-base-en',
     )
     {
-    }
-
-    /**
-     * @inheritDoc
-     */
-    public function generate(array $texts): array
-    {
-        $client = new Client([
+        $this->client = new Client([
+            'base_uri' => 'https://api.jina.ai/v1/',
             'headers' => [
                 'Authorization' => "Bearer $this->apiKey",
                 'Content-Type' => 'application/json',
                 'Accept-Encoding' => 'identity',
             ]
         ]);
+    }
 
+    /**
+     * @inheritDoc
+     */
+    public function generate(array $texts): array
+    {
         try {
-            $response = $client->post(self::API_URL, [
-                'body' => json_encode([
+            $response = $this->client->post('embeddings', [
+                'json' => [
                     'model' => $this->model,
                     'input' => $texts,
-                ])
+                ]
             ]);
 
             $result = json_decode($response->getBody()->getContents(), true);
             $embeddings = $result['data'];
             usort($embeddings, fn($a, $b) => $a['index'] <=> $b['index']);
 
             return array_map(fn($embedding) => $embedding['embedding'], $embeddings);
-        } catch (GuzzleException $e) {
-            throw new \RuntimeException('Failed to generate embeddings', 0, $e);
+        } catch (ClientExceptionInterface $e) {
+            throw new \RuntimeException("Error calling Jina AI API: {$e->getMessage()}", 0, $e);
         }
     }
 }

src/Embeddings/OpenAIEmbeddingFunction.php

@@ -5,37 +5,54 @@
 
 namespace Codewithkyrian\ChromaDB\Embeddings;
 
-use Codewithkyrian\ChromaDB\Embeddings\EmbeddingFunction;
-
-interface OpenAIAPIInterface {
-    /**
-     * Creates a new embedding function
-     * @param string $model
-     * @param string[] $input
-     * @param string $user
-     *
-     * @return int[][]
-     */
-    public function createEmbedding(string $model, array $input, string $user ): array;
-}
+use GuzzleHttp\Client;
+use Psr\Http\Client\ClientExceptionInterface;
 
 class OpenAIEmbeddingFunction implements EmbeddingFunction
 {
+    private Client $client;
 
     public function __construct(
         public readonly string $apiKey,
-        public readonly string $organizationId,
-        public readonly string $model,
-
+        public readonly string $organization = '',
+        public readonly string $model = 'text-embedding-ada-002',
     )
     {
+        $headers = [
+            'Authorization' => "Bearer $this->apiKey",
+            'Content-Type' => 'application/json',
+        ];
+
+        if (!empty($this->organization)) {
+            $headers['OpenAI-Organization'] = $this->organization;
+        }
+
+        $this->client = new Client([
+            'base_uri' => 'https://api.openai.com/v1/',
+            'headers' => $headers
+        ]);
     }
 
     /**
      * @inheritDoc
      */
     public function generate(array $texts): array
     {
-        return [[1,2,3,4,5,6,7,8,9,10], [1,2,3,4,5,6,7,8,9,10]];
+        try {
+            $response = $this->client->post('embeddings', [
+                'json' => [
+                    'model' => $this->model,
+                    'input' => $texts,
+                ]
+            ]);
+
+            $result = json_decode($response->getBody()->getContents(), true);
+            $embeddings = $result['data'];
+            usort($embeddings, fn($a, $b) => $a['index'] <=> $b['index']);
+
+            return array_map(fn($embedding) => $embedding['embedding'], $embeddings);
+        } catch (ClientExceptionInterface $e) {
+            throw new \RuntimeException("Error calling OpenAI API: {$e->getMessage()}", 0, $e);
+        }
     }
 }