Improved configurability, code cleanup

cognesy · Aug 28, 2024 · f9d7d9e · f9d7d9e
1 parent 78c263a
commit f9d7d9e
Show file tree

Hide file tree

Showing 104 changed files with 865 additions and 2,287 deletions.
diff --git a/README.md b/README.md
@@ -615,10 +615,10 @@ $yourApiKey = Env::get('OPENAI_API_KEY'); // use your own API key
 // Create instance of OpenAI client initialized with custom parameters
 $client = new OpenAIClient(
     $yourApiKey,
-    baseUri: 'https://api.openai.com', // you can change base URI
-    organization: '',
+    baseUri: 'https://api.openai.com/v1', // you can change base URI
     connectTimeout: 3,
     requestTimeout: 30,
+    metadata: ['organization' => ''],
 );
 
 /// Get Instructor with the default client component overridden with your own

diff --git a/composer.json b/composer.json
@@ -46,7 +46,8 @@
         "symfony/css-selector": "^7.1",
         "symfony/dom-crawler": "^7.1",
         "duzun/hquery": "^3.1",
-        "ext-dom": "*"
+        "ext-dom": "*",
+        "psy/psysh": "@stable"
     },
     "config": {
         "allow-plugins": {

diff --git a/config/instructor.php b/config/instructor.php
@@ -0,0 +1,123 @@
+<?php
+use Cognesy\Instructor\ApiClient\Enums\ClientType;
+use Cognesy\Instructor\Utils\Env;
+
+return [
+    'defaultConnection' => 'openai',
+    'cache' => [
+        'enabled' => Env::get('INSTRUCTOR_CACHE_ENABLED', false),
+        'expiry' => Env::get('INSTRUCTOR_CACHE_EXPIRY', 3600),
+        'path' => Env::get('INSTRUCTOR_CACHE_PATH', '/tmp/instructor/cache'),
+    ],
+    'debug' => [
+        'enabled' => Env::get('INSTRUCTOR_DEBUG', false),
+        'stopOnDebug' => Env::get('INSTRUCTOR_STOP_ON_DEBUG', false),
+        'forceDebug' => Env::get('INSTRUCTOR_FORCE_DEBUG', false),
+    ],
+    'useObjectReferences' => Env::get('INSTRUCTOR_USE_OBJECT_REFERENCES', false),
+
+    'connections' => [
+        'anthropic' => [
+            'clientType' => ClientType::Anthropic->value,
+            'apiUrl' => Env::get('ANTHROPIC_API_URL', 'https://api.anthropic.com/v1'),
+            'apiKey' => Env::get('ANTHROPIC_API_KEY', ''),
+            'metadata' => [
+                'apiVersion' => Env::get('ANTHROPIC_API_VERSION', '2023-06-01'),
+                'beta' => Env::get('ANTHROPIC_BETA', 'prompt-caching-2024-07-31'),
+            ],
+            'defaultModel' => Env::get('ANTHROPIC_DEFAULT_MODEL', 'claude-3-haiku-20240307'),
+            'defaultMaxTokens' => Env::get('ANTHROPIC_DEFAULT_MAX_TOKENS', 1024),
+            'connectTimeout' => Env::get('ANTHROPIC_CONNECT_TIMEOUT', 3),
+            'requestTimeout' => Env::get('ANTHROPIC_REQUEST_TIMEOUT', 30),
+        ],
+        'azure' => [
+            'clientType' => ClientType::Azure->value,
+            'apiUrl' => Env::get('AZURE_OPENAI_BASE_URI', 'https://api.openai.com/v1'),
+            'apiKey' => Env::get('AZURE_OPENAI_API_KEY', ''),
+            'metadata' => [
+                'apiVersion' => Env::get('AZURE_OPENAI_API_VERSION', '2023-03-15-preview'),
+                'resourceName' => Env::get('AZURE_OPENAI_RESOURCE_NAME', 'instructor-dev'),
+                'deploymentName' => Env::get('AZURE_OPENAI_DEPLOYMENT_NAME', 'gpt-4o-mini'),
+            ],
+            'defaultModel' => Env::get('AZURE_OPENAI_DEFAULT_MODEL', 'gpt-4o-mini'),
+            'defaultMaxTokens' => Env::get('AZURE_OPENAI_DEFAULT_MAX_TOKENS', 1024),
+            'connectTimeout' => Env::get('AZURE_OPENAI_CONNECT_TIMEOUT', 3),
+            'requestTimeout' => Env::get('AZURE_OPENAI_REQUEST_TIMEOUT', 30),
+        ],
+        'cohere' => [
+            'clientType' => ClientType::Cohere->value,
+            'apiUrl' => Env::get('COHERE_API_URL', 'https://api.cohere.ai/v1'),
+            'apiKey' => Env::get('COHERE_API_KEY', ''),
+            'defaultModel' => Env::get('COHERE_DEFAULT_MODEL', 'cohere-r-plus'),
+            'defaultMaxTokens' => Env::get('COHERE_DEFAULT_MAX_TOKENS', 1024),
+            'connectTimeout' => Env::get('COHERE_CONNECT_TIMEOUT', 3),
+            'requestTimeout' => Env::get('COHERE_REQUEST_TIMEOUT', 30),
+        ],
+        'fireworks' => [
+            'clientType' => ClientType::Fireworks->value,
+            'apiUrl' => Env::get('FIREWORKS_API_URL', 'https://api.fireworks.ai/inference/v1'),
+            'apiKey' => Env::get('FIREWORKS_API_KEY', ''),
+            'defaultModel' => Env::get('FIREWORKS_DEFAULT_MODEL', 'accounts/fireworks/models/mixtral-8x7b-instruct'),
+            'defaultMaxTokens' => Env::get('FIREWORKS_DEFAULT_MAX_TOKENS', 1024),
+            'connectTimeout' => Env::get('FIREWORKS_CONNECT_TIMEOUT', 3),
+            'requestTimeout' => Env::get('FIREWORKS_REQUEST_TIMEOUT', 30),
+        ],
+        'gemini' => [
+            'clientType' => ClientType::Gemini->value,
+            'apiUrl' => Env::get('GEMINI_API_URL', 'https://generativelanguage.googleapis.com/v1beta'),
+            'apiKey' => Env::get('GEMINI_API_KEY', ''),
+            'defaultModel' => Env::get('GEMINI_DEFAULT_MODEL', 'gemini-1.5-flash'),
+            'defaultMaxTokens' => Env::get('GEMINI_DEFAULT_MAX_TOKENS', 1024),
+            'connectTimeout' => Env::get('GEMINI_CONNECT_TIMEOUT', 3),
+            'requestTimeout' => Env::get('GEMINI_REQUEST_TIMEOUT', 30),
+        ],
+        'groq' => [
+            'clientType' => ClientType::Groq->value,
+            'apiUrl' => Env::get('GROQ_API_URL', 'https://api.groq.com/openai/v1'),
+            'apiKey' => Env::get('GROQ_API_KEY', ''),
+            'defaultModel' => Env::get('GROQ_DEFAULT_MODEL', 'llama3-8b-8192'),
+            'defaultMaxTokens' => Env::get('GROQ_DEFAULT_MAX_TOKENS', 1024),
+            'connectTimeout' => Env::get('GROQ_CONNECT_TIMEOUT', 3),
+            'requestTimeout' => Env::get('GROQ_REQUEST_TIMEOUT', 30),
+        ],
+        'mistral' => [
+            'clientType' => ClientType::Mistral->value,
+            'apiUrl' => Env::get('MISTRAL_API_URL', 'https://api.mistral.ai/v1'),
+            'apiKey' => Env::get('MISTRAL_API_KEY', ''),
+            'defaultModel' => Env::get('MISTRAL_DEFAULT_MODEL', 'mistral-small-latest'),
+            'defaultMaxTokens' => Env::get('MISTRAL_DEFAULT_MAX_TOKENS', 1024),
+            'connectTimeout' => Env::get('MISTRAL_CONNECT_TIMEOUT', 3),
+            'requestTimeout' => Env::get('MISTRAL_REQUEST_TIMEOUT', 30),
+        ],
+        'openai' => [
+            'clientType' => ClientType::OpenAI->value,
+            'apiUrl' => Env::get('OPENAI_API_URL', 'https://api.openai.com/v1'),
+            'apiKey' => Env::get('OPENAI_API_KEY', ''),
+            'metadata' => [
+                'organization' => ''
+            ],
+            'defaultModel' => Env::get('OPENAI_DEFAULT_MODEL', 'gpt-4o-mini'),
+            'defaultMaxTokens' => Env::get('OPENAI_DEFAULT_MAX_TOKENS', 1024),
+            'connectTimeout' => Env::get('OPENAI_CONNECT_TIMEOUT', 3),
+            'requestTimeout' => Env::get('OPENAI_REQUEST_TIMEOUT', 30),
+        ],
+        'openrouter' => [
+            'clientType' => ClientType::OpenRouter->value,
+            'apiUrl' => Env::get('OPENROUTER_API_URL', 'https://api.openrouter.io/v1'),
+            'apiKey' => Env::get('OPENROUTER_API_KEY', ''),
+            'defaultModel' => Env::get('OPENROUTER_DEFAULT_MODEL', 'gpt-4o-mini'),
+            'defaultMaxTokens' => Env::get('OPENROUTER_DEFAULT_MAX_TOKENS', 1024),
+            'connectTimeout' => Env::get('OPENROUTER_CONNECT_TIMEOUT', 3),
+            'requestTimeout' => Env::get('OPENROUTER_REQUEST_TIMEOUT', 30),
+        ],
+        'together' => [
+            'clientType' => ClientType::Together->value,
+            'apiUrl' => Env::get('TOGETHER_API_URL', 'https://api.together.xyz/v1'),
+            'apiKey' => Env::get('TOGETHER_API_KEY', ''),
+            'defaultModel' => Env::get('TOGETHER_DEFAULT_MODEL', 'mistralai/Mixtral-8x7B-Instruct-v0.1'),
+            'defaultMaxTokens' => Env::get('TOGETHER_DEFAULT_MAX_TOKENS', 1024),
+            'connectTimeout' => Env::get('TOGETHER_CONNECT_TIMEOUT', 3),
+            'requestTimeout' => Env::get('TOGETHER_REQUEST_TIMEOUT', 30),
+        ],
+    ],
+];
diff --git a/docs/advanced/model_options.mdx b/docs/advanced/model_options.mdx
@@ -31,9 +31,9 @@ use Cognesy\Instructor\Clients\OpenAI\OpenAIClient;
 $client = new OpenAIClient(
     apiKey: $yourApiKey,
     baseUri: 'https://api.openai.com/v1',
-    organization: '',
     connectTimeout: 3,
     requestTimeout: 30,
+    metadata: ['organization' => ''],
 );
 
 /// Get Instructor with the default client component overridden with your own

diff --git a/docs/cookbook/examples/advanced/partials.mdx b/docs/cookbook/examples/advanced/partials.mdx
@@ -43,15 +43,12 @@ function partialUpdate($partial) {
     // Clear the screen and move the cursor to the top
     echo chr(27).chr(91).'H'.chr(27).chr(91).'J';
 
-    // Print explanation
-    echo "Waiting 250ms on every update received to make changes easier to observe...\n";
-
     // Display the partial object
     dump($partial);
 
     // Wait a bit before clearing the screen to make partial changes slower.
     // Don't use this in your application :)
-    usleep(250000);
+    //usleep(250000);
 }
 ?>
 ```

diff --git a/docs/cookbook/examples/advanced/streaming.mdx b/docs/cookbook/examples/advanced/streaming.mdx
@@ -41,11 +41,7 @@ function partialUpdate($partial) {
     // Clear the screen and move the cursor to the top
     echo chr(27).chr(91).'H'.chr(27).chr(91).'J';
 
-    // Print explanation
-    echo "Waiting 250ms on every update received to make changes easier to observe...\n";
-
     // Display the partial object
-
     dump($partial);
 
     // Wait a bit before clearing the screen to make partial changes slower.

diff --git a/docs/cookbook/examples/api_support/azure_openai.mdx b/docs/cookbook/examples/api_support/azure_openai.mdx
@@ -40,9 +40,11 @@ class User {
 /// Custom client parameters: base URI
 $client = (new AzureClient(
     apiKey: Env::get('AZURE_OPENAI_API_KEY'), // set your own value/source
-    resourceName: Env::get('AZURE_OPENAI_RESOURCE_NAME'), // set your own value/source
-    deploymentId: Env::get('AZURE_OPENAI_DEPLOYMENT_NAME'), // set your own value/source
-    apiVersion: Env::get('AZURE_OPENAI_API_VERSION'), // set your own value/source
+    metadata: [
+        'resourceName' => Env::get('AZURE_OPENAI_RESOURCE_NAME'), // set your own value/source
+        'deploymentId' => Env::get('AZURE_OPENAI_DEPLOYMENT_NAME'), // set your own value/source
+        'apiVersion' => Env::get('AZURE_OPENAI_API_VERSION'), // set your own value/source
+    ],
 ));
 
 /// Get Instructor with the default client component overridden with your own

diff --git a/docs/cookbook/examples/api_support/openai.mdx b/docs/cookbook/examples/api_support/openai.mdx
@@ -47,9 +47,9 @@ $yourApiKey = Env::get('OPENAI_API_KEY'); // use your own API key
 $client = new OpenAIClient(
     apiKey: $yourApiKey,
     baseUri: 'https://api.openai.com/v1',
-    organization: '',
     connectTimeout: 3,
     requestTimeout: 30,
+    metadata: ['organization' => ''],
 );
 
 /// Get Instructor with the default client component overridden with your own

diff --git a/examples/02_Advanced/Caching/run.php b/examples/02_Advanced/Caching/run.php
@@ -59,6 +59,8 @@ class User {
     responseModel: User::class,
 )->withRequestCache()->get();
 
+eval(\Psy\sh());
+
 $delta = Profiler::mark('cache 1st call')->mili();
 echo "Time elapsed (cache on, 1st call): $delta msec\n\n";
 

diff --git a/examples/02_Advanced/PartialUpdates/run.php b/examples/02_Advanced/PartialUpdates/run.php
@@ -43,15 +43,12 @@ function partialUpdate($partial) {
     // Clear the screen and move the cursor to the top
     echo chr(27).chr(91).'H'.chr(27).chr(91).'J';
 
-    // Print explanation
-    echo "Waiting 250ms on every update received to make changes easier to observe...\n";
-
     // Display the partial object
     dump($partial);
 
     // Wait a bit before clearing the screen to make partial changes slower.
     // Don't use this in your application :)
-    usleep(250000);
+    //usleep(250000);
 }
 ?>
 ```

diff --git a/examples/02_Advanced/Streaming/run.php b/examples/02_Advanced/Streaming/run.php
@@ -41,11 +41,7 @@ function partialUpdate($partial) {
     // Clear the screen and move the cursor to the top
     echo chr(27).chr(91).'H'.chr(27).chr(91).'J';
 
-    // Print explanation
-    echo "Waiting 250ms on every update received to make changes easier to observe...\n";
-
     // Display the partial object
-
     dump($partial);
 
     // Wait a bit before clearing the screen to make partial changes slower.

diff --git a/examples/05_APISupport/LLMSupportAzureOAI/run.php b/examples/05_APISupport/LLMSupportAzureOAI/run.php
@@ -40,9 +40,11 @@ class User {
 /// Custom client parameters: base URI
 $client = (new AzureClient(
     apiKey: Env::get('AZURE_OPENAI_API_KEY'), // set your own value/source
-    resourceName: Env::get('AZURE_OPENAI_RESOURCE_NAME'), // set your own value/source
-    deploymentId: Env::get('AZURE_OPENAI_DEPLOYMENT_NAME'), // set your own value/source
-    apiVersion: Env::get('AZURE_OPENAI_API_VERSION'), // set your own value/source
+    metadata: [
+        'resourceName' => Env::get('AZURE_OPENAI_RESOURCE_NAME'), // set your own value/source
+        'deploymentId' => Env::get('AZURE_OPENAI_DEPLOYMENT_NAME'), // set your own value/source
+        'apiVersion' => Env::get('AZURE_OPENAI_API_VERSION'), // set your own value/source
+    ]
 ));
 
 /// Get Instructor with the default client component overridden with your own

diff --git a/examples/05_APISupport/LLMSupportFireworksAI/run.php b/examples/05_APISupport/LLMSupportFireworksAI/run.php
@@ -42,7 +42,7 @@ class User {
 }
 
 // Mistral instance params
-$yourApiKey = Env::get('FIREWORKSAI_API_KEY'); // set your own API key
+$yourApiKey = Env::get('FIREWORKS_API_KEY'); // set your own API key
 
 // Create instance of client initialized with custom parameters
 $client = new FireworksAIClient(

diff --git a/examples/05_APISupport/LLMSupportOpenAI/run.php b/examples/05_APISupport/LLMSupportOpenAI/run.php
@@ -47,9 +47,9 @@ class User {
 $client = new OpenAIClient(
     apiKey: $yourApiKey,
     baseUri: 'https://api.openai.com/v1',
-    organization: '',
     connectTimeout: 3,
     requestTimeout: 30,
+    metadata: ['organization' => ''],
 );
 
 /// Get Instructor with the default client component overridden with your own

diff --git a/examples/05_APISupport/LLMSupportTogetherAI/run.php b/examples/05_APISupport/LLMSupportTogetherAI/run.php
@@ -47,7 +47,7 @@ class User {
 
 // Create instance of client initialized with custom parameters
 $client = new TogetherAIClient(
-    apiKey: Env::get('TOGETHERAI_API_KEY'),
+    apiKey: Env::get('TOGETHER_API_KEY'),
 );
 
 /// Get Instructor with the default client component overridden with your own

diff --git a/notes/NOTES.md b/notes/NOTES.md
@@ -17,15 +17,18 @@ Catch up with the latest additions.
 
 ## TODOs
 
-- Parallel tool calls
 - Multiple tools with tool selection
-- Modules: Add module observability via events - currently no access to this info
+- Parallel tool calls
 - Generate unstructured, then format to structured - to improve reasoning
 
 ### API Client
 
 - Clean up predefined models, prices, etc.
 
+### Addon: Modules
+
+- Modules: Add module observability via events - currently no access to this info
+
 ### Configuration
 
 - Export configuration to user folder / use external configuration

diff --git a/src/ApiClient/ApiClient.php b/src/ApiClient/ApiClient.php
@@ -1,6 +1,7 @@
 <?php
 namespace Cognesy\Instructor\ApiClient;
 
+use Cognesy\Instructor\ApiClient\Enums\ClientType;
 use Cognesy\Instructor\ApiClient\RequestConfig\ApiRequestConfig;
 use Cognesy\Instructor\ApiClient\Contracts\CanCallApi;
 use Cognesy\Instructor\Enums\Mode;
@@ -26,10 +27,13 @@ abstract class ApiClient implements CanCallApi
     use Traits\HandlesStreamApiResponse;
     use Traits\ReadsStreamResponse;
 
+    protected ClientType $clientType;
+
     public function __construct(
         EventDispatcher $events = null,
     ) {
         $this->withEventDispatcher($events ?? new EventDispatcher('api-client'));
+        $this->clientType = ClientType::fromClientClass(static::class);
     }
 
     /// PUBLIC API //////////////////////////////////////////////////////////////////////////////////////////

diff --git a/src/ApiClient/ApiConnector.php b/src/ApiClient/ApiConnector.php
@@ -1,6 +1,8 @@
 <?php
 namespace Cognesy\Instructor\ApiClient;
 
+use Saloon\Contracts\Authenticator;
+use Saloon\Http\Auth\TokenAuthenticator;
 use Saloon\Http\Connector;
 use Saloon\Traits\Plugins\AlwaysThrowOnErrors;
 use Saloon\Traits\Plugins\HasTimeout;
@@ -46,4 +48,8 @@ protected function defaultHeaders(): array {
     public function defaultConfig(): array {
         return ['stream' => true];
     }
+
+    protected function defaultAuth() : Authenticator {
+        return new TokenAuthenticator($this->apiKey);
+    }
 }
diff --git a/src/ApiClient/Contracts/CanCallApi.php b/src/ApiClient/Contracts/CanCallApi.php
@@ -2,6 +2,7 @@
 
 namespace Cognesy\Instructor\ApiClient\Contracts;
 
+use Cognesy\Instructor\ApiClient\Requests\ApiRequest;
 use Cognesy\Instructor\ApiClient\Responses\ApiResponse;
 use Cognesy\Instructor\ApiClient\Responses\PartialApiResponse;
 use Cognesy\Instructor\Enums\Mode;
@@ -24,4 +25,6 @@ public function defaultMaxTokens() : int;
     public function defaultModel() : string;
 
     public function getModeRequestClass(Mode $mode) : string;
+
+    public function withApiRequest(ApiRequest $request) : static;
 }
diff --git a/src/ApiClient/Enums/ClientType.php b/src/ApiClient/Enums/ClientType.php
@@ -3,9 +3,11 @@
 
 enum ClientType : string
 {
-    use Traits\HandlesCreation;
     use Traits\HandlesAccess;
+    use Traits\HandlesCreation;
     use Traits\HandlesMapping;
+    use Traits\HandlesResponse;
+    use Traits\HandlesStreamData;
 
     case Anthropic = 'anthropic';
     //case Anyscale = 'anyscale';