Cleanup (consistent naming) + PartialJsonParser bugfix

cognesy · Oct 3, 2024 · 5411bf8 · 5411bf8
1 parent 4b73881
commit 5411bf8
Show file tree

Hide file tree

Showing 45 changed files with 296 additions and 205 deletions.
diff --git a/docs/cookbook/examples/advanced/context_cache_llm.mdx b/docs/cookbook/examples/advanced/context_cache_llm.mdx
@@ -45,7 +45,7 @@ $inference = (new Inference)->withConnection('anthropic')->withCachedContext(
 $response = $inference->create(
     messages: [['role' => 'user', 'content' => 'CTO of lead gen software vendor']],
     options: ['max_tokens' => 256],
-)->toApiResponse();
+)->toLLMResponse();
 
 print("----------------------------------------\n");
 print("\n# Summary for CTO of lead gen vendor\n");
@@ -60,7 +60,7 @@ assert(Str::contains($response->content, 'lead', false));
 $response2 = $inference->create(
     messages: [['role' => 'user', 'content' => 'CIO of insurance company']],
     options: ['max_tokens' => 256],
-)->toApiResponse();
+)->toLLMResponse();
 
 print("----------------------------------------\n");
 print("\n# Summary for CIO of insurance company\n");

diff --git a/docs/cookbook/examples/extras/web_to_objects.mdx b/docs/cookbook/examples/extras/web_to_objects.mdx
@@ -54,7 +54,11 @@ $companyGen = Webpage::withScraper('none')
     ->get('https://themanifest.com/pl/software-development/laravel/companies?page=1')
     ->cleanup()
     ->select('.directory-providers__list')
-    ->selectMany(selector: '.provider-card', callback: fn($item) => $item->asMarkdown(), limit: 3);
+    ->selectMany(
+        selector: '.provider-card',
+        callback: fn($item) => $item->asMarkdown(),
+        limit: 3
+    );
 
 $companies = [];
 foreach($companyGen as $companyDiv) {

diff --git a/docs/cookbook/examples/troubleshooting/token_usage_events.mdx b/docs/cookbook/examples/troubleshooting/token_usage_events.mdx
@@ -6,8 +6,8 @@ docname: 'token_usage_events'
 ## Overview
 
 Some use cases require tracking the token usage of the API responses.
-Currently, this can be done by listening to the `ApiResponseReceived`
-and `PartialApiResponseReceived` events and summing the token usage
+Currently, this can be done by listening to the `LLMResponseReceived`
+and `PartialLLMResponseReceived` events and summing the token usage
 of the responses.
 
 Code below demonstrates how it can be implemented using Instructor
@@ -24,10 +24,10 @@ event listeners.
 $loader = require 'vendor/autoload.php';
 $loader->add('Cognesy\\Instructor\\', __DIR__ . '../../src/');
 
-use Cognesy\Instructor\Events\ApiClient\ApiResponseReceived;
-use Cognesy\Instructor\Events\ApiClient\PartialApiResponseReceived;
-use Cognesy\Instructor\Extras\LLM\Data\LLMApiResponse;
-use Cognesy\Instructor\Extras\LLM\Data\PartialLLMApiResponse;
+use Cognesy\Instructor\Events\ApiClient\LLMResponseReceived;
+use Cognesy\Instructor\Events\ApiClient\PartialLLMResponseReceived;
+use Cognesy\Instructor\Extras\LLM\Data\LLMResponse;
+use Cognesy\Instructor\Extras\LLM\Data\PartialLLMResponse;
 use Cognesy\Instructor\Instructor;
 
 class User {
@@ -41,7 +41,7 @@ class TokenCounter {
     public int $cacheCreation = 0;
     public int $cacheRead = 0;
 
-    public function add(LLMApiResponse|PartialLLMApiResponse $response) {
+    public function add(LLMResponse|PartialLLMResponse $response) {
         $this->input += $response->inputTokens;
         $this->output += $response->outputTokens;
         $this->cacheCreation += $response->cacheCreationTokens;
@@ -68,7 +68,7 @@ $counter = new TokenCounter();
 echo "COUNTING TOKENS FOR SYNC RESPONSE\n";
 $text = "Jason is 25 years old and works as an engineer.";
 $instructor = (new Instructor)
-    ->onEvent(ApiResponseReceived::class, fn($e) => $counter->add($e->apiResponse))
+    ->onEvent(LLMResponseReceived::class, fn(LLMResponseReceived $e) => $counter->add($e->llmResponse))
     ->respond(
         messages: $text,
         responseModel: User::class,
@@ -84,7 +84,7 @@ $counter->reset();
 echo "\n\nCOUNTING TOKENS FOR STREAMED RESPONSE\n";
 $text = "Anna is 19 years old.";
 $instructor = (new Instructor)
-    ->onEvent(PartialApiResponseReceived::class, fn($e) => $counter->add($e->partialApiResponse))
+    ->onEvent(PartialLLMResponseReceived::class, fn(PartialLLMResponseReceived $e) => $counter->add($e->partialLLMResponse))
     ->respond(
         messages: $text,
         responseModel: User::class,

diff --git a/evals/LLMModes/CompareModes.php b/evals/LLMModes/CompareModes.php
@@ -48,8 +48,8 @@ private function execute(string $connection, Mode $mode, bool $isStreamed) : Eva
         $key = $this->makeKey($connection, $mode, $isStreamed);
         try {
             $time = microtime(true);
-            $apiResponse = $this->modes->callInferenceFor($this->query, $mode, $connection, $this->modes->schema(), $isStreamed);
-            $answer = $apiResponse->content;
+            $llmResponse = $this->modes->callInferenceFor($this->query, $mode, $connection, $this->modes->schema(), $isStreamed);
+            $answer = $llmResponse->content;
             $timeElapsed = microtime(true) - $time;
             $evalRequest = new EvalRequest(
                 answer: $answer,
@@ -58,16 +58,16 @@ private function execute(string $connection, Mode $mode, bool $isStreamed) : Eva
                 mode: $mode,
                 connection: $connection,
                 isStreamed: $isStreamed,
-                response: $apiResponse,
+                response: $llmResponse,
             );
             $isCorrect = ($this->evalFn)($evalRequest);
             $evalResponse = new EvalResponse(
                 id: $key,
                 answer: $answer,
                 isCorrect: $isCorrect,
                 timeElapsed: $timeElapsed,
-                inputTokens: $apiResponse->inputTokens,
-                outputTokens: $apiResponse->outputTokens,
+                inputTokens: $llmResponse->inputTokens,
+                outputTokens: $llmResponse->outputTokens,
             );
         } catch(Exception $e) {
             $timeElapsed = microtime(true) - $time;

diff --git a/evals/LLMModes/EvalRequest.php b/evals/LLMModes/EvalRequest.php
@@ -3,17 +3,17 @@
 namespace Cognesy\Evals\LLMModes;
 
 use Cognesy\Instructor\Enums\Mode;
-use Cognesy\Instructor\Extras\LLM\Data\LLMApiResponse;
+use Cognesy\Instructor\Extras\LLM\Data\LLMResponse;
 
 class EvalRequest
 {
     public function __construct(
-        public string          $answer = '',
-        public string|array    $query = '',
-        public array           $schema = [],
-        public Mode            $mode = Mode::Text,
-        public string          $connection = '',
-        public bool            $isStreamed = false,
-        public ?LLMApiResponse $response = null,
+        public string       $answer = '',
+        public string|array $query = '',
+        public array        $schema = [],
+        public Mode         $mode = Mode::Text,
+        public string       $connection = '',
+        public bool         $isStreamed = false,
+        public ?LLMResponse $response = null,
     ) {}
 }
diff --git a/evals/LLMModes/Modes.php b/evals/LLMModes/Modes.php
@@ -3,7 +3,7 @@
 namespace Cognesy\Evals\LLMModes;
 
 use Cognesy\Instructor\Enums\Mode;
-use Cognesy\Instructor\Extras\LLM\Data\LLMApiResponse;
+use Cognesy\Instructor\Extras\LLM\Data\LLMResponse;
 use Cognesy\Instructor\Extras\LLM\Inference;
 use Cognesy\Instructor\Extras\LLM\InferenceResponse;
 
@@ -25,7 +25,7 @@ public function schema() : array {
         return $this->model->schema();
     }
 
-    public function callInferenceFor(string|array $query, Mode $mode, string $connection, array $schema, bool $isStreamed) : LLMApiResponse {
+    public function callInferenceFor(string|array $query, Mode $mode, string $connection, array $schema, bool $isStreamed) : LLMResponse {
         $query = is_array($query) ? $query : [['role' => 'user', 'content' => $query]];
         $inferenceResponse = match($mode) {
             Mode::Tools => $this->forModeTools($query, $connection, $schema, $isStreamed),
@@ -34,7 +34,7 @@ public function callInferenceFor(string|array $query, Mode $mode, string $connec
             Mode::MdJson => $this->forModeMdJson($query, $connection, $schema, $isStreamed),
             Mode::Text => $this->forModeText($query, $connection, $isStreamed),
         };
-        return $inferenceResponse->toApiResponse();
+        return $inferenceResponse->toLLMResponse();
     }
 
     public function forModeTools(string|array $query, string $connection, array $schema, bool $isStreamed) : InferenceResponse {

diff --git a/evals/LLMModes/run.php b/evals/LLMModes/run.php
@@ -6,7 +6,6 @@
 use Cognesy\Evals\LLMModes\CompareModes;
 use Cognesy\Evals\LLMModes\EvalRequest;
 use Cognesy\Instructor\Enums\Mode;
-use Cognesy\Instructor\Extras\Debug\Debug;
 use Cognesy\Instructor\Utils\Json\Json;
 use Cognesy\Instructor\Utils\Str;
 
@@ -24,8 +23,8 @@
 ];
 
 $streamingModes = [
-    false,
     true,
+    false,
 ];
 
 $modes = [
@@ -59,7 +58,7 @@ function evalFn(EvalRequest $er) {
 function validateToolsData(array $data) : bool {
     return 'store_company' === ($data[0]['name'] ?? '')
         && 'ACME' === ($data[0]['arguments']['name'] ?? '')
-        && 2020 === ($data[0]['arguments']['year'] ?? 0);
+        && 2020 === (int) ($data[0]['arguments']['year'] ?? 0);
 }
 
 (new CompareModes(

diff --git a/examples/A02_Advanced/ContextCacheLLM/run.php b/examples/A02_Advanced/ContextCacheLLM/run.php
@@ -45,7 +45,7 @@
 $response = $inference->create(
     messages: [['role' => 'user', 'content' => 'CTO of lead gen software vendor']],
     options: ['max_tokens' => 256],
-)->toApiResponse();
+)->toLLMResponse();
 
 print("----------------------------------------\n");
 print("\n# Summary for CTO of lead gen vendor\n");
@@ -60,7 +60,7 @@
 $response2 = $inference->create(
     messages: [['role' => 'user', 'content' => 'CIO of insurance company']],
     options: ['max_tokens' => 256],
-)->toApiResponse();
+)->toLLMResponse();
 
 print("----------------------------------------\n");
 print("\n# Summary for CIO of insurance company\n");

diff --git a/examples/A03_Troubleshooting/TokenUsage/run.php b/examples/A03_Troubleshooting/TokenUsage/run.php
@@ -6,8 +6,8 @@
 ## Overview
 
 Some use cases require tracking the token usage of the API responses.
-Currently, this can be done by listening to the `ApiResponseReceived`
-and `PartialApiResponseReceived` events and summing the token usage
+Currently, this can be done by listening to the `LLMResponseReceived`
+and `PartialLLMResponseReceived` events and summing the token usage
 of the responses.
 
 Code below demonstrates how it can be implemented using Instructor
@@ -24,10 +24,10 @@
 $loader = require 'vendor/autoload.php';
 $loader->add('Cognesy\\Instructor\\', __DIR__ . '../../src/');
 
-use Cognesy\Instructor\Events\ApiClient\ApiResponseReceived;
-use Cognesy\Instructor\Events\ApiClient\PartialApiResponseReceived;
-use Cognesy\Instructor\Extras\LLM\Data\LLMApiResponse;
-use Cognesy\Instructor\Extras\LLM\Data\PartialLLMApiResponse;
+use Cognesy\Instructor\Events\ApiClient\LLMResponseReceived;
+use Cognesy\Instructor\Events\ApiClient\PartialLLMResponseReceived;
+use Cognesy\Instructor\Extras\LLM\Data\LLMResponse;
+use Cognesy\Instructor\Extras\LLM\Data\PartialLLMResponse;
 use Cognesy\Instructor\Instructor;
 
 class User {
@@ -41,7 +41,7 @@ class TokenCounter {
     public int $cacheCreation = 0;
     public int $cacheRead = 0;
 
-    public function add(LLMApiResponse|PartialLLMApiResponse $response) {
+    public function add(LLMResponse|PartialLLMResponse $response) {
         $this->input += $response->inputTokens;
         $this->output += $response->outputTokens;
         $this->cacheCreation += $response->cacheCreationTokens;
@@ -68,11 +68,11 @@ public function print() {
 echo "COUNTING TOKENS FOR SYNC RESPONSE\n";
 $text = "Jason is 25 years old and works as an engineer.";
 $instructor = (new Instructor)
-    ->onEvent(ApiResponseReceived::class, fn($e) => $counter->add($e->apiResponse))
+    ->onEvent(LLMResponseReceived::class, fn(LLMResponseReceived $e) => $counter->add($e->llmResponse))
     ->respond(
         messages: $text,
         responseModel: User::class,
-);
+    );
 echo "\nTEXT: $text\n";
 assert($counter->input > 0);
 assert($counter->output > 0);
@@ -84,7 +84,7 @@ public function print() {
 echo "\n\nCOUNTING TOKENS FOR STREAMED RESPONSE\n";
 $text = "Anna is 19 years old.";
 $instructor = (new Instructor)
-    ->onEvent(PartialApiResponseReceived::class, fn($e) => $counter->add($e->partialApiResponse))
+    ->onEvent(PartialLLMResponseReceived::class, fn(PartialLLMResponseReceived $e) => $counter->add($e->partialLLMResponse))
     ->respond(
         messages: $text,
         responseModel: User::class,

diff --git a/examples/A05_Extras/ComplexExtractionClaude/run.php b/examples/A05_Extras/ComplexExtractionClaude/run.php
@@ -113,7 +113,7 @@ enum StakeholderRole: string {
     ->request(
         messages: $report,
         responseModel: Sequence::of(ProjectEvent::class),
-        model: 'claude-3-haiku-20240307', //'claude-3-sonnet-20240229',
+        model: 'claude-3-5-sonnet-20240620', // 'claude-3-haiku-20240307'
         prompt: 'Extract a list of project events with all the details from the provided input in JSON format using schema: <|json_schema|>',
         mode: Mode::Json,
         examples: [['input' => 'Acme Insurance project to implement SalesTech CRM solution is currently in RED status due to delayed delivery of document production system, led by 3rd party vendor - Alfatech. Customer (Acme) is discussing the resolution with the vendor. Production deployment plan has been finalized on Aug 15th and awaiting customer approval.', 'output' => [["type" => "object", "title" => "sequenceOfProjectEvent", "description" => "A sequence of ProjectEvent", "properties" => ["list" => [["title" => "Absorbing delay by deploying extra resources", "description" => "System integrator (SysCorp) are working to absorb some of the delay by deploying extra resources to speed up development when the doc production is done.", "type" => "action", "status" => "open", "stakeholders" => [["name" => "SysCorp", "role" => "system integrator", "details" => "System integrator",],], "date" => "2021-09-01",], ["title" => "Finalization of production deployment plan", "description" => "Production deployment plan has been finalized on Aug 15th and awaiting customer approval.", "type" => "progress", "status" => "open", "stakeholders" => [["name" => "Acme", "role" => "customer", "details" => "Customer",],], "date" => "2021-08-15",],],]]]]],

diff --git a/examples/A05_Extras/ImageToDataAnthropic/run.php b/examples/A05_Extras/ImageToDataAnthropic/run.php
@@ -57,6 +57,7 @@ class Receipt {
     input: Image::fromFile(__DIR__ . '/receipt.png'),
     responseModel: Receipt::class,
     prompt: 'Extract structured data from the receipt. Return result as JSON following this schema: <|json_schema|>',
+    model: 'claude-3-5-sonnet-20240620',
     mode: Mode::Json,
     options: ['max_tokens' => 4096]
 );

diff --git a/examples/A05_Extras/WebToObjects/run.php b/examples/A05_Extras/WebToObjects/run.php
@@ -54,7 +54,11 @@ class Company {
     ->get('https://themanifest.com/pl/software-development/laravel/companies?page=1')
     ->cleanup()
     ->select('.directory-providers__list')
-    ->selectMany(selector: '.provider-card', callback: fn($item) => $item->asMarkdown(), limit: 3);
+    ->selectMany(
+        selector: '.provider-card',
+        callback: fn($item) => $item->asMarkdown(),
+        limit: 3
+    );
 
 $companies = [];
 foreach($companyGen as $companyDiv) {

diff --git a/src/Contracts/CanGeneratePartials.php b/src/Contracts/CanGeneratePartials.php
@@ -3,13 +3,13 @@
 namespace Cognesy\Instructor\Contracts;
 
 use Cognesy\Instructor\Data\ResponseModel;
-use Cognesy\Instructor\Extras\LLM\Data\PartialLLMApiResponse;
+use Cognesy\Instructor\Extras\LLM\Data\PartialLLMResponse;
 use Generator;
 
 interface CanGeneratePartials
 {
     /**
-     * @param Generator<PartialLLMApiResponse> $stream
+     * @param Generator<PartialLLMResponse> $stream
      * @param ResponseModel $responseModel
      * @return Generator<mixed>
      */

diff --git a/src/Contracts/CanGenerateResponse.php b/src/Contracts/CanGenerateResponse.php
@@ -3,10 +3,10 @@
 namespace Cognesy\Instructor\Contracts;
 
 use Cognesy\Instructor\Data\ResponseModel;
-use Cognesy\Instructor\Extras\LLM\Data\LLMApiResponse;
+use Cognesy\Instructor\Extras\LLM\Data\LLMResponse;
 use Cognesy\Instructor\Utils\Result\Result;
 
 interface CanGenerateResponse
 {
-    public function makeResponse(LLMApiResponse $response, ResponseModel $responseModel) : Result;
+    public function makeResponse(LLMResponse $response, ResponseModel $responseModel) : Result;
 }