anthropic.php
2 days ago
chatml.php
2 days ago
core.php
1 month ago
custom.php
1 month ago
factory.php
2 months ago
google-interactions.php
1 month ago
google.php
1 month ago
mistral.php
2 months ago
open-router.php
2 days ago
openai.php
2 weeks ago
ovh.php
1 month ago
perplexity.php
8 months ago
replicate.php
2 weeks ago
xai.php
3 months ago
chatml.php
2312 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * Base implementation of the ChatML API. |
| 5 | * This was first introduced by OpenAI and many providers keep compatibility with it. |
| 6 | * Engines relying on this original API can extend this class. |
| 7 | * |
| 8 | */ |
| 9 | |
| 10 | class Meow_MWAI_Engines_ChatML extends Meow_MWAI_Engines_Core { |
| 11 | // Base (OpenAI) |
| 12 | protected $apiKey = null; |
| 13 | protected $organizationId = null; |
| 14 | |
| 15 | // Azure |
| 16 | private $azureDeployments = null; |
| 17 | protected $azureApiVersion = 'api-version=2024-12-01-preview'; |
| 18 | |
| 19 | // Response |
| 20 | protected $inModel = null; |
| 21 | protected $inId = null; |
| 22 | protected $inThreadId = null; |
| 23 | |
| 24 | // Streaming |
| 25 | protected $streamFunctionCall = null; |
| 26 | protected $streamToolCalls = []; |
| 27 | protected $streamLastMessage = null; |
| 28 | protected $streamAnnotations = []; |
| 29 | protected $streamImageIds = []; |
| 30 | protected $streamThinking = null; // For reasoning/thinking content |
| 31 | protected $streamThinkTag = 0; // Inside an inline <think> block (Qwen-style models) |
| 32 | |
| 33 | protected $streamInTokens = null; |
| 34 | protected $streamOutTokens = null; |
| 35 | protected $streamCost = null; |
| 36 | protected $streamStartEmitted = false; |
| 37 | |
| 38 | public function __construct( $core, $env ) { |
| 39 | parent::__construct( $core, $env ); |
| 40 | $this->set_environment(); |
| 41 | } |
| 42 | |
| 43 | public function reset_stream() { |
| 44 | $this->streamContent = null; |
| 45 | $this->streamBuffer = null; |
| 46 | $this->streamFunctionCall = null; |
| 47 | $this->streamToolCalls = []; |
| 48 | $this->streamLastMessage = null; |
| 49 | $this->streamThinking = null; |
| 50 | $this->streamThinkTag = 0; |
| 51 | $this->streamInTokens = null; |
| 52 | $this->streamOutTokens = null; |
| 53 | $this->inModel = null; |
| 54 | $this->inId = null; |
| 55 | $this->emittedFunctionResults = []; |
| 56 | $this->streamStartEmitted = false; |
| 57 | } |
| 58 | |
| 59 | protected function set_environment() { |
| 60 | $env = $this->env; |
| 61 | $this->apiKey = $env['apikey']; |
| 62 | |
| 63 | if ( isset( $env['organizationId'] ) ) { |
| 64 | $this->organizationId = $env['organizationId']; |
| 65 | } |
| 66 | if ( $this->envType === 'azure' ) { |
| 67 | $this->azureDeployments = isset( $env['deployments'] ) ? $env['deployments'] : []; |
| 68 | } |
| 69 | } |
| 70 | |
| 71 | protected function get_azure_deployment_name( $model ) { |
| 72 | foreach ( $this->azureDeployments as $deployment ) { |
| 73 | if ( $deployment['model'] === $model && !empty( $deployment['name'] ) ) { |
| 74 | return $deployment['name']; |
| 75 | } |
| 76 | } |
| 77 | throw new Exception( 'Unknown deployment for model: ' . $model ); |
| 78 | } |
| 79 | |
| 80 | protected function get_service_name() { |
| 81 | return $this->envType === 'azure' ? 'Azure' : 'OpenAI'; |
| 82 | } |
| 83 | |
| 84 | private function is_o1_model( $model ) { |
| 85 | $modelDef = $this->retrieve_model_info( $model ); |
| 86 | return !empty( $modelDef['tags'] ) && in_array( 'o1-model', $modelDef['tags'] ); |
| 87 | } |
| 88 | |
| 89 | private function is_gpt5_model( $model ) { |
| 90 | // Check if the model is a GPT-5 variant |
| 91 | return strpos( $model, 'gpt-5' ) === 0; |
| 92 | } |
| 93 | |
| 94 | private function requires_developer_roles( $model ) { |
| 95 | return false; |
| 96 | } |
| 97 | |
| 98 | private function is_realtime_model( $model ) { |
| 99 | $modelDef = $this->retrieve_model_info( $model ); |
| 100 | return !empty( $modelDef['family'] ) && $modelDef['family'] === 'realtime'; |
| 101 | } |
| 102 | |
| 103 | protected function build_messages( $query ) { |
| 104 | $messages = []; |
| 105 | |
| 106 | // First, we need to add the first message (the instructions). |
| 107 | if ( !empty( $query->instructions ) ) { |
| 108 | //if ( !$this->is_o1_model( $query->model ) ) { |
| 109 | $messages[] = [ 'role' => 'system', 'content' => $query->instructions ]; |
| 110 | //} |
| 111 | } |
| 112 | |
| 113 | // Then, if any, we need to add the 'messages', they are already formatted. |
| 114 | foreach ( $query->messages as $message ) { |
| 115 | $messages[] = $message; |
| 116 | } |
| 117 | |
| 118 | // If there is a context, we need to add it with proper framing. |
| 119 | if ( !empty( $query->context ) ) { |
| 120 | $framedContext = $this->core->frame_context( $query->context ); |
| 121 | $messages[] = [ 'role' => 'system', 'content' => $framedContext ]; |
| 122 | } |
| 123 | |
| 124 | // Finally, we need to add the message, but if there is an image, we need to add it as a system message. |
| 125 | $attachments = method_exists( $query, 'getAttachments' ) ? $query->getAttachments() : []; |
| 126 | if ( !empty( $attachments ) ) { |
| 127 | // The message can be empty when the user only sends files; in that case the |
| 128 | // text part is skipped entirely, as some providers reject empty text parts. |
| 129 | $content = []; |
| 130 | $message = $query->get_message(); |
| 131 | if ( $message !== null && $message !== '' ) { |
| 132 | $content[] = [ |
| 133 | 'type' => 'text', |
| 134 | 'text' => $message |
| 135 | ]; |
| 136 | } |
| 137 | |
| 138 | // Handle all attachments (unified approach) |
| 139 | foreach ( $attachments as $file ) { |
| 140 | // Check file type BEFORE trying to access file data |
| 141 | // Images can be loaded via URL or base64, but PDFs might use OpenAI file_id references |
| 142 | $mimeType = $file->get_mimeType() ?? ''; |
| 143 | $isImage = strpos( $mimeType, 'image/' ) === 0; |
| 144 | |
| 145 | if ( $isImage ) { |
| 146 | $finalUrl = null; |
| 147 | if ( $query->image_remote_upload === 'url' ) { |
| 148 | $finalUrl = $file->get_url(); |
| 149 | } |
| 150 | else { |
| 151 | $finalUrl = $file->get_inline_base64_url(); |
| 152 | } |
| 153 | |
| 154 | $content[] = [ |
| 155 | 'type' => 'image_url', |
| 156 | 'image_url' => [ |
| 157 | 'url' => $finalUrl |
| 158 | ] |
| 159 | ]; |
| 160 | } |
| 161 | // Skip non-images for Chat Completions API |
| 162 | } |
| 163 | |
| 164 | // If every attachment was skipped and there was no message, fall back to an |
| 165 | // empty text part rather than sending an invalid empty content array. |
| 166 | if ( empty( $content ) ) { |
| 167 | $content[] = [ 'type' => 'text', 'text' => '' ]; |
| 168 | } |
| 169 | |
| 170 | $messages[] = [ |
| 171 | 'role' => 'user', |
| 172 | 'content' => $content |
| 173 | ]; |
| 174 | } |
| 175 | else { |
| 176 | $messages[] = [ 'role' => 'user', 'content' => $query->get_message() ]; |
| 177 | } |
| 178 | |
| 179 | // We need to convert all the 'system' role into 'developer' role. |
| 180 | if ( $this->requires_developer_roles( $query->model ) ) { |
| 181 | foreach ( $messages as &$message ) { |
| 182 | if ( $message['role'] === 'system' ) { |
| 183 | $message['role'] = 'developer'; |
| 184 | } |
| 185 | } |
| 186 | } |
| 187 | // But otherwise, if it's o1, we need to remove the message which are 'system' |
| 188 | else if ( $this->is_o1_model( $query->model ) ) { |
| 189 | $hasChanges = false; |
| 190 | foreach ( $messages as $index => $message ) { |
| 191 | if ( $message['role'] === 'system' ) { |
| 192 | unset( $messages[$index] ); |
| 193 | $hasChanges = true; |
| 194 | } |
| 195 | } |
| 196 | if ( $hasChanges ) { |
| 197 | $messages = array_values( $messages ); |
| 198 | Meow_MWAI_Logging::warn( 'The model ' . $query->model . ' doesn\'t support System nor Developer messages. They were removed.' ); |
| 199 | } |
| 200 | } |
| 201 | |
| 202 | return $messages; |
| 203 | } |
| 204 | |
| 205 | /** |
| 206 | * Convert functions to Responses API tools format. |
| 207 | * Shared by OpenAI and OpenRouter engines. |
| 208 | */ |
| 209 | protected function build_responses_tools( $functions ) { |
| 210 | $tools = []; |
| 211 | foreach ( $functions as $function ) { |
| 212 | $functionData = $function->serializeForOpenAI(); |
| 213 | if ( !isset( $functionData['name'] ) || empty( $functionData['name'] ) ) { |
| 214 | Meow_MWAI_Logging::warn( 'Function missing required name field' ); |
| 215 | continue; |
| 216 | } |
| 217 | $parameters = $functionData['parameters'] ?? null; |
| 218 | if ( !$parameters ) { |
| 219 | $parameters = [ |
| 220 | 'type' => 'object', |
| 221 | 'properties' => new stdClass(), |
| 222 | 'required' => [] |
| 223 | ]; |
| 224 | } |
| 225 | else { |
| 226 | if ( isset( $parameters['properties'] ) && |
| 227 | is_array( $parameters['properties'] ) && |
| 228 | empty( $parameters['properties'] ) ) { |
| 229 | $parameters['properties'] = new stdClass(); |
| 230 | } |
| 231 | } |
| 232 | $tool = [ |
| 233 | 'type' => 'function', |
| 234 | 'name' => $functionData['name'], |
| 235 | 'description' => $functionData['description'] ?? '', |
| 236 | 'parameters' => $parameters, |
| 237 | 'strict' => false |
| 238 | ]; |
| 239 | $tools[] = $tool; |
| 240 | } |
| 241 | return $tools; |
| 242 | } |
| 243 | |
| 244 | /** |
| 245 | * Built-in tools this engine can honor on the Chat Completions path. Empty here: the |
| 246 | * OpenAI engine only honors them on its Responses path, which never reaches this builder. |
| 247 | * Subclasses that map a tool onto something native (OpenRouter's web plugin) override it. |
| 248 | */ |
| 249 | protected function supported_builtin_tools(): array { |
| 250 | return []; |
| 251 | } |
| 252 | |
| 253 | /** |
| 254 | * The OpenAI-style built-in tools (web_search, image_generation, code_interpreter) only |
| 255 | * exist on the Responses API. A Chat Completions body has nowhere to put them, so any that |
| 256 | * reach this builder are dropped. Dropping them silently produced reports of "Web Search |
| 257 | * is enabled but the model says it cannot browse" with nothing in the logs, because the |
| 258 | * model never even saw the tool. Say what happened and why, once per request. |
| 259 | * |
| 260 | * Logged as an error on purpose: warn() only reaches the plugin's own log when Dev Tools |
| 261 | * and Server Debug Mode are both on, which nobody has on by default. error() always |
| 262 | * reaches PHP's error_log, which is the first place anyone looks. |
| 263 | */ |
| 264 | protected function warn_dropped_builtin_tools( $query ): void { |
| 265 | if ( empty( $query->tools ) || !is_array( $query->tools ) ) { |
| 266 | return; |
| 267 | } |
| 268 | $builtin = [ 'web_search', 'image_generation', 'code_interpreter' ]; |
| 269 | $dropped = array_values( array_diff( array_intersect( $query->tools, $builtin ), $this->supported_builtin_tools() ) ); |
| 270 | if ( empty( $dropped ) ) { |
| 271 | return; |
| 272 | } |
| 273 | Meow_MWAI_Logging::error( sprintf( |
| 274 | 'Built-in tool(s) %s were enabled for model "%s", but this request went through the Chat Completions API, which cannot carry them, so they were dropped. They need the native OpenAI environment with a Responses-capable model whose id matches a built-in entry exactly (for example gpt-5.6-luna), or OpenRouter for web search.', |
| 275 | implode( ', ', $dropped ), |
| 276 | $query->model |
| 277 | ) ); |
| 278 | } |
| 279 | |
| 280 | protected function build_body( $query, $streamCallback = null, $extra = null ) { |
| 281 | if ( $query instanceof Meow_MWAI_Query_Text ) { |
| 282 | $this->warn_dropped_builtin_tools( $query ); |
| 283 | $body = [ |
| 284 | 'model' => $query->model, |
| 285 | 'stream' => !is_null( $streamCallback ), |
| 286 | ]; |
| 287 | |
| 288 | if ( !empty( $query->maxTokens ) ) { |
| 289 | // max_tokens has been deprecated in favor of max_completion_tokens in 2025. |
| 290 | $body['max_completion_tokens'] = $query->maxTokens; |
| 291 | } |
| 292 | |
| 293 | // Null means "not set"; 0 is a valid temperature and must still be sent. |
| 294 | if ( isset( $query->temperature ) ) { |
| 295 | // GPT-5 and o1 models don't support temperature parameter |
| 296 | if ( !$this->is_o1_model( $query->model ) && !$this->is_gpt5_model( $query->model ) ) { |
| 297 | $body['temperature'] = $query->temperature; |
| 298 | } |
| 299 | else if ( $this->is_o1_model( $query->model ) ) { |
| 300 | // o1 models require temperature to be 1 if specified |
| 301 | $body['temperature'] = 1; |
| 302 | } |
| 303 | // For GPT-5 models, we simply don't include the temperature parameter |
| 304 | } |
| 305 | |
| 306 | if ( !empty( $query->maxResults ) ) { |
| 307 | $body['n'] = $query->maxResults; |
| 308 | } |
| 309 | |
| 310 | if ( !empty( $query->stop ) ) { |
| 311 | $body['stop'] = $query->stop; |
| 312 | } |
| 313 | |
| 314 | if ( !empty( $query->responseFormat ) ) { |
| 315 | if ( $query->responseFormat === 'json' ) { |
| 316 | $body['response_format'] = [ 'type' => 'json_object' ]; |
| 317 | } |
| 318 | } |
| 319 | |
| 320 | // Usage Data (only for OpenAI) |
| 321 | // https://cookbook.openai.com/examples/how_to_stream_completions#4-how-to-get-token-usage-data-for-streamed-chat-completion-response |
| 322 | if ( !empty( $streamCallback ) && $this->envType === 'openai' ) { |
| 323 | $body['stream_options'] = [ |
| 324 | 'include_usage' => true, |
| 325 | ]; |
| 326 | } |
| 327 | |
| 328 | if ( !empty( $query->functions ) ) { |
| 329 | $model = $this->retrieve_model_info( $query->model ); |
| 330 | if ( !empty( $model['tags'] ) && !in_array( 'functions', $model['tags'] ) ) { |
| 331 | Meow_MWAI_Logging::warn( 'The model ' . $query->model . ' doesn\'t support Function Calling.' ); |
| 332 | } |
| 333 | else if ( strpos( $query->model, 'ft:' ) === 0 ) { |
| 334 | Meow_MWAI_Logging::warn( 'OpenAI doesn\'t support Function Calling with fine-tuned models yet.' ); |
| 335 | } |
| 336 | else { |
| 337 | $body['tools'] = []; |
| 338 | // Dynamic function: they will interactively enhance the completion (tools). |
| 339 | foreach ( $query->functions as $function ) { |
| 340 | $body['tools'][] = [ |
| 341 | 'type' => 'function', |
| 342 | 'function' => $function->serializeForOpenAI() |
| 343 | ]; |
| 344 | } |
| 345 | // Static functions: they will be executed at the end of the completion. |
| 346 | //$body['function_call'] = $query->functionCall; |
| 347 | } |
| 348 | } |
| 349 | $body['messages'] = $this->build_messages( $query ); |
| 350 | |
| 351 | // Add the feedback if it's a feedback query. |
| 352 | if ( $query instanceof Meow_MWAI_Query_Feedback ) { |
| 353 | if ( !empty( $query->blocks ) ) { |
| 354 | foreach ( $query->blocks as $feedback_block ) { |
| 355 | $body['messages'][] = $feedback_block['rawMessage']; |
| 356 | foreach ( $feedback_block['feedbacks'] as $feedback ) { |
| 357 | // Ensure content is a string for the API |
| 358 | $content = $feedback['reply']['value']; |
| 359 | if ( !is_string( $content ) ) { |
| 360 | $content = json_encode( $content ); |
| 361 | } |
| 362 | |
| 363 | $body['messages'][] = [ |
| 364 | 'tool_call_id' => $feedback['request']['toolId'], |
| 365 | 'role' => 'tool', |
| 366 | 'name' => $feedback['request']['name'], |
| 367 | 'content' => $content |
| 368 | ]; |
| 369 | |
| 370 | // Note: Function result events are now emitted centrally in core.php |
| 371 | // when the function is actually executed |
| 372 | } |
| 373 | } |
| 374 | } |
| 375 | return $body; |
| 376 | } |
| 377 | |
| 378 | return $body; |
| 379 | } |
| 380 | else if ( $query instanceof Meow_MWAI_Query_Transcribe ) { |
| 381 | // Determine filename and MIME type |
| 382 | $filename = 'audio.mp3'; // default |
| 383 | $mimeType = 'audio/mpeg'; // default |
| 384 | $attachments = method_exists( $query, 'getAttachments' ) ? $query->getAttachments() : []; |
| 385 | if ( !empty( $attachments ) ) { |
| 386 | $firstAttachment = $attachments[0]; |
| 387 | if ( method_exists( $firstAttachment, 'get_filename' ) ) { |
| 388 | $filename = $firstAttachment->get_filename(); |
| 389 | } |
| 390 | if ( method_exists( $firstAttachment, 'get_mimeType' ) ) { |
| 391 | $detectedMime = $firstAttachment->get_mimeType(); |
| 392 | if ( !empty( $detectedMime ) ) { |
| 393 | $mimeType = $detectedMime; |
| 394 | } |
| 395 | } |
| 396 | } |
| 397 | else if ( !empty( $query->url ) ) { |
| 398 | $filename = basename( $query->url ); |
| 399 | } |
| 400 | |
| 401 | // Guard against malformed filenames coming from form prompts/placeholders |
| 402 | if ( empty( $filename ) ) { |
| 403 | $filename = 'audio.mp3'; |
| 404 | } |
| 405 | $filename = preg_replace( "/[\r\n]+/", '', $filename ); |
| 406 | $mimeType = preg_replace( "/[\r\n]+/", '', $mimeType ); |
| 407 | if ( empty( $mimeType ) ) { |
| 408 | $mimeType = 'audio/mpeg'; |
| 409 | } |
| 410 | |
| 411 | $body = [ |
| 412 | 'prompt' => $query->message, |
| 413 | 'model' => $query->model, |
| 414 | 'response_format' => 'text', |
| 415 | 'file' => $filename, |
| 416 | 'data' => $extra, |
| 417 | 'mime' => $mimeType |
| 418 | ]; |
| 419 | return $body; |
| 420 | } |
| 421 | else if ( $query instanceof Meow_MWAI_Query_Embed ) { |
| 422 | $body = [ 'input' => $query->message, 'model' => $query->model ]; |
| 423 | if ( $this->envType === 'azure' ) { |
| 424 | $body = [ 'input' => $query->message ]; |
| 425 | } |
| 426 | // Dimensions are only supported by v3 models |
| 427 | if ( !empty( $query->dimensions ) && strpos( $query->model, 'ada-002' ) === false ) { |
| 428 | $body['dimensions'] = $query->dimensions; |
| 429 | } |
| 430 | return $body; |
| 431 | } |
| 432 | else if ( $query instanceof Meow_MWAI_Query_EditImage ) { |
| 433 | $resolution = !empty( $query->resolution ) ? $query->resolution : '1024x1024'; |
| 434 | $attachments = method_exists( $query, 'getAttachments' ) ? $query->getAttachments() : []; |
| 435 | $firstFile = !empty( $attachments ) ? $attachments[0] : null; |
| 436 | $filename = $firstFile ? $firstFile->get_filename() : ''; |
| 437 | $mimeType = $firstFile ? $firstFile->get_mimeType() : null; |
| 438 | $body = [ |
| 439 | 'prompt' => $query->message, |
| 440 | 'n' => $query->maxResults, |
| 441 | 'size' => $resolution, |
| 442 | 'image' => $filename, |
| 443 | 'data' => $extra |
| 444 | ]; |
| 445 | if ( !empty( $mimeType ) ) { |
| 446 | $body['mime'] = $mimeType; |
| 447 | } |
| 448 | |
| 449 | // Add mask if provided |
| 450 | if ( !empty( $query->mask ) ) { |
| 451 | $maskData = $query->mask->get_data(); |
| 452 | $maskFilename = 'mask.png'; |
| 453 | $maskMimeType = $query->mask->get_mimeType(); |
| 454 | |
| 455 | $body['mask'] = $maskFilename; |
| 456 | $body['mask_data'] = $maskData; |
| 457 | if ( !empty( $maskMimeType ) ) { |
| 458 | $body['mask_mime'] = $maskMimeType; |
| 459 | } |
| 460 | } |
| 461 | // 'response_format' => 'b64_json', |
| 462 | if ( !empty( $query->model ) ) { |
| 463 | $body['model'] = $query->model; |
| 464 | } |
| 465 | return $body; |
| 466 | } |
| 467 | else if ( $query instanceof Meow_MWAI_Query_Image ) { |
| 468 | $model = $query->model; |
| 469 | $resolution = !empty( $query->resolution ) ? $query->resolution : '1024x1024'; |
| 470 | $body = [ |
| 471 | 'model' => $model, |
| 472 | 'prompt' => $query->message, |
| 473 | 'n' => $query->maxResults, |
| 474 | 'size' => $resolution, |
| 475 | 'moderation' => 'low', |
| 476 | ]; |
| 477 | // Only forward 'quality' when the caller (or final_checks) resolved one. |
| 478 | // Letting the API default decide ('auto' for gpt-image-*) keeps costs predictable |
| 479 | // and avoids forcing 'high' on every request. |
| 480 | if ( !empty( $query->quality ) ) { |
| 481 | $body['quality'] = $query->quality; |
| 482 | } |
| 483 | return $body; |
| 484 | } |
| 485 | } |
| 486 | |
| 487 | protected function build_url( $query, $endpoint = null ) { |
| 488 | $url = ''; |
| 489 | $env = $this->env; |
| 490 | // This endpoint is basically OpenAI or Azure, but in the case this class |
| 491 | // is overriden, we can pass the endpoint directly (for OpenRouter or HuggingFace, for example). |
| 492 | if ( empty( $endpoint ) ) { |
| 493 | if ( $this->envType === 'openai' ) { |
| 494 | $endpoint = apply_filters( 'mwai_openai_endpoint', 'https://api.openai.com/v1', $this->env ); |
| 495 | $this->organizationId = isset( $env['organizationId'] ) ? $env['organizationId'] : null; |
| 496 | } |
| 497 | else if ( $this->envType === 'azure' ) { |
| 498 | $endpoint = isset( $env['endpoint'] ) ? $env['endpoint'] : null; |
| 499 | // Ensure the endpoint has the proper protocol if it's just a domain |
| 500 | if ( $endpoint && strpos( $endpoint, 'http' ) !== 0 ) { |
| 501 | $endpoint = 'https://' . $endpoint; |
| 502 | } |
| 503 | } |
| 504 | else { |
| 505 | if ( empty( $this->envType ) ) { |
| 506 | throw new Exception( 'Endpoint is not defined, and this envType is not known.' ); |
| 507 | } |
| 508 | throw new Exception( 'Endpoint is not defined, and this envType is not known: ' . $this->envType ); |
| 509 | } |
| 510 | } |
| 511 | // Add the base API to the URL |
| 512 | if ( $query instanceof Meow_MWAI_Query_Text || $query instanceof Meow_MWAI_Query_Feedback ) { |
| 513 | if ( $this->envType === 'azure' ) { |
| 514 | $deployment_name = $this->get_azure_deployment_name( $query->model ); |
| 515 | $url = trailingslashit( $endpoint ) . 'openai/deployments/' . $deployment_name; |
| 516 | $url .= '/chat/completions?' . $this->azureApiVersion; |
| 517 | } |
| 518 | else { |
| 519 | $url .= trailingslashit( $endpoint ) . 'chat/completions'; |
| 520 | } |
| 521 | return $url; |
| 522 | } |
| 523 | else if ( $query instanceof Meow_MWAI_Query_Transcribe ) { |
| 524 | $modeEndpoint = $query->feature === 'translation' ? 'translations' : 'transcriptions'; |
| 525 | $url .= trailingslashit( $endpoint ) . 'audio/' . $modeEndpoint; |
| 526 | return $url; |
| 527 | } |
| 528 | else if ( $query instanceof Meow_MWAI_Query_Embed ) { |
| 529 | $url .= trailingslashit( $endpoint ) . 'embeddings'; |
| 530 | if ( $this->envType === 'azure' ) { |
| 531 | $deployment_name = $this->get_azure_deployment_name( $query->model ); |
| 532 | $url = trailingslashit( $endpoint ) . 'openai/deployments/' . |
| 533 | $deployment_name . '/embeddings?' . $this->azureApiVersion; |
| 534 | } |
| 535 | return $url; |
| 536 | } |
| 537 | else if ( $query instanceof Meow_MWAI_Query_EditImage ) { |
| 538 | $url .= trailingslashit( $endpoint ) . 'images/edits'; |
| 539 | if ( $this->envType === 'azure' ) { |
| 540 | $deployment_name = $this->get_azure_deployment_name( $query->model ); |
| 541 | $url = trailingslashit( $endpoint ) . 'openai/deployments/' . |
| 542 | $deployment_name . '/images/edits?' . $this->azureApiVersion; |
| 543 | } |
| 544 | return $url; |
| 545 | } |
| 546 | else if ( $query instanceof Meow_MWAI_Query_Image ) { |
| 547 | $url .= trailingslashit( $endpoint ) . 'images/generations'; |
| 548 | if ( $this->envType === 'azure' ) { |
| 549 | $deployment_name = $this->get_azure_deployment_name( $query->model ); |
| 550 | $url = trailingslashit( $endpoint ) . 'openai/deployments/' . |
| 551 | $deployment_name . '/images/generations?' . $this->azureApiVersion; |
| 552 | } |
| 553 | return $url; |
| 554 | } |
| 555 | throw new Exception( 'The query is not supported by build_url().' ); |
| 556 | } |
| 557 | |
| 558 | protected function build_headers( $query ) { |
| 559 | if ( $query->apiKey ) { |
| 560 | $this->apiKey = $query->apiKey; |
| 561 | } |
| 562 | if ( empty( $this->apiKey ) ) { |
| 563 | throw new Exception( 'No API Key provided. Please visit the Settings. (ChatML Engine)' ); |
| 564 | } |
| 565 | $headers = [ |
| 566 | 'Content-Type' => 'application/json', |
| 567 | 'Authorization' => 'Bearer ' . $this->apiKey, |
| 568 | ]; |
| 569 | if ( $this->organizationId ) { |
| 570 | $headers['OpenAI-Organization'] = $this->organizationId; |
| 571 | } |
| 572 | if ( $this->envType === 'azure' ) { |
| 573 | $headers = [ 'Content-Type' => 'application/json', 'api-key' => $this->apiKey ]; |
| 574 | } |
| 575 | return $headers; |
| 576 | } |
| 577 | |
| 578 | protected function build_options( $headers, $json = null, $forms = null, $method = 'POST' ) { |
| 579 | $body = null; |
| 580 | if ( !empty( $forms ) ) { |
| 581 | $boundary = wp_generate_password( 24, false ); |
| 582 | $headers['Content-Type'] = 'multipart/form-data; boundary=' . $boundary; |
| 583 | $body = $this->build_form_body( $forms, $boundary ); |
| 584 | } |
| 585 | else if ( !empty( $json ) ) { |
| 586 | $body = $this->safe_json_encode( $json, 'request body' ); |
| 587 | } |
| 588 | $options = [ |
| 589 | 'headers' => $headers, |
| 590 | 'method' => $method, |
| 591 | 'timeout' => MWAI_TIMEOUT, |
| 592 | 'body' => $body, |
| 593 | 'sslverify' => MWAI_SSL_VERIFY |
| 594 | ]; |
| 595 | return $options; |
| 596 | } |
| 597 | // object: "thread.message.delta" |
| 598 | protected function stream_data_handler( $json ) { |
| 599 | $content = null; |
| 600 | $handledCondition = false; // Track if we entered any condition |
| 601 | |
| 602 | // Get additional data from the JSON |
| 603 | if ( isset( $json['model'] ) ) { |
| 604 | $this->inModel = $json['model']; |
| 605 | } |
| 606 | if ( isset( $json['id'] ) ) { |
| 607 | $this->inId = $json['id']; |
| 608 | |
| 609 | // Send start event if debug mode is enabled and not already sent |
| 610 | if ( $this->currentDebugMode && $this->streamCallback && !$this->streamStartEmitted ) { |
| 611 | $this->streamStartEmitted = true; |
| 612 | $event = Meow_MWAI_Event::status( 'Starting stream...' ) |
| 613 | ->set_metadata( 'model', $this->inModel ) |
| 614 | ->set_metadata( 'id', $this->inId ); |
| 615 | call_user_func( $this->streamCallback, $event ); |
| 616 | } |
| 617 | } |
| 618 | |
| 619 | // Providers send no-op frames (keepalive, ping) purely to hold the SSE connection |
| 620 | // open through proxies. They carry no content and there is nothing to handle, so |
| 621 | // return before the "unhandled streaming JSON structure" log at the end of this |
| 622 | // method reports them as a parsing problem. They are expected traffic, not errors. |
| 623 | $eventType = $json['type'] ?? null; |
| 624 | if ( in_array( $eventType, [ 'keepalive', 'ping' ], true ) ) { |
| 625 | return null; |
| 626 | } |
| 627 | |
| 628 | $object = $json['object'] ?? null; |
| 629 | |
| 630 | if ( $object === 'thread.run' ) { |
| 631 | $handledCondition = true; |
| 632 | $this->inThreadId = $json['thread_id']; |
| 633 | if ( $json['status'] === 'failed' ) { |
| 634 | $error = $json['last_error']['message'] ?? 'The run failed.'; |
| 635 | throw new Exception( $error ); |
| 636 | } |
| 637 | } |
| 638 | else if ( $object === 'thread.run.step.delta' ) { |
| 639 | $handledCondition = true; |
| 640 | if ( $json['delta']['step_details']['type'] === 'tool_calls' ) { |
| 641 | foreach ( $json['delta']['step_details']['tool_calls'] as $tool_call ) { |
| 642 | $index = $tool_call['index'] ?? null; |
| 643 | // unset (not "= null") to break the reference from the previous loop |
| 644 | // iteration. Assigning null would null out the tool call we just stored |
| 645 | // in $this->streamToolCalls, leaving a null hole for parallel tool calls |
| 646 | // (Mistral rejects that with a 422; OpenAI silently tolerates it). |
| 647 | unset( $currentStreamToolCall ); |
| 648 | if ( $index !== null && isset( $this->streamToolCalls[$index] ) ) { |
| 649 | $currentStreamToolCall = &$this->streamToolCalls[$index]; |
| 650 | } |
| 651 | else { |
| 652 | $this->streamToolCalls[] = [ |
| 653 | 'id' => null, |
| 654 | 'type' => null, |
| 655 | 'function' => [ 'name' => '', 'arguments' => '' ], |
| 656 | 'code_interpreter' => [ 'input' => '', 'outputs' => [] ], |
| 657 | ]; |
| 658 | end( $this->streamToolCalls ); |
| 659 | $currentStreamToolCall = &$this->streamToolCalls[ key( $this->streamToolCalls ) ]; |
| 660 | |
| 661 | // Send tool call initiated event |
| 662 | if ( $this->currentDebugMode && $this->streamCallback ) { |
| 663 | $event = Meow_MWAI_Event::status( 'Initiating tool call...' ); |
| 664 | call_user_func( $this->streamCallback, $event ); |
| 665 | } |
| 666 | } |
| 667 | if ( !empty( $tool_call['id'] ) ) { |
| 668 | $currentStreamToolCall['id'] = $tool_call['id']; |
| 669 | } |
| 670 | if ( !empty( $tool_call['type'] ) ) { |
| 671 | $currentStreamToolCall['type'] = $tool_call['type']; |
| 672 | } |
| 673 | if ( isset( $tool_call['function'] ) ) { |
| 674 | $function = $tool_call['function']; |
| 675 | if ( isset( $function['name'] ) ) { |
| 676 | $currentStreamToolCall['function']['name'] .= $function['name']; |
| 677 | } |
| 678 | if ( isset( $function['arguments'] ) ) { |
| 679 | $currentStreamToolCall['function']['arguments'] .= $function['arguments']; |
| 680 | } |
| 681 | } |
| 682 | if ( isset( $tool_call['code_interpreter'] ) ) { |
| 683 | $code_interpreter = $tool_call['code_interpreter']; |
| 684 | if ( isset( $code_interpreter['input'] ) ) { |
| 685 | $currentStreamToolCall['code_interpreter']['input'] .= $code_interpreter['input']; |
| 686 | } |
| 687 | if ( isset( $code_interpreter['outputs'] ) ) { |
| 688 | $currentStreamToolCall['code_interpreter']['outputs'] = $code_interpreter['outputs']; |
| 689 | } |
| 690 | } |
| 691 | $this->streamLastMessage['tool_calls'] = $this->streamToolCalls; |
| 692 | } |
| 693 | } |
| 694 | } |
| 695 | else if ( $object === 'thread.message.delta' ) { |
| 696 | $handledCondition = true; |
| 697 | $delta = $json['delta']['content'][0] ?? null; |
| 698 | if ( $delta ) { |
| 699 | switch ( $delta['type'] ?? null ) { |
| 700 | case 'text': |
| 701 | if ( !empty( $delta['value'] ) && is_string( $delta['value'] ) ) { |
| 702 | $content = $delta['value']; |
| 703 | } |
| 704 | else if ( !empty( $delta['text'] ) && is_string( $delta['text'] ) ) { |
| 705 | $content = $delta['text']; |
| 706 | } |
| 707 | else if ( !empty( $delta['text'] ) && is_array( $delta['text'] ) ) { |
| 708 | $text = $delta['text']; |
| 709 | if ( !empty( $text['annotations'] ) ) { |
| 710 | $this->streamAnnotations = array_merge( $this->streamAnnotations, $text['annotations'] ); |
| 711 | } |
| 712 | if ( isset( $text['value'] ) ) { |
| 713 | $content = $text['value']; |
| 714 | } |
| 715 | } |
| 716 | else { |
| 717 | error_log( 'AI Engine: Unknown text format: ' . json_encode( $delta ) ); |
| 718 | } |
| 719 | break; |
| 720 | case 'image': |
| 721 | $content = $delta['url']; |
| 722 | break; |
| 723 | case 'image_file': |
| 724 | $fileId = $delta['image_file']['file_id']; |
| 725 | $content = '<!-- IMG #' . $fileId . ' -->'; |
| 726 | $this->streamImageIds[] = $fileId; |
| 727 | break; |
| 728 | case 'function_call': |
| 729 | // TODO: Re-evaluate after 2027-01-26. This is the deprecated OpenAI |
| 730 | // "function_call" streaming format, superseded by "tool_calls". It decodes |
| 731 | // each delta and replaces (instead of concatenating like the tool_calls |
| 732 | // path), so a provider that streams fragmented args here would lose them. |
| 733 | // Confirm no supported provider still uses this format, then remove it. |
| 734 | if ( empty( $this->streamFunctionCall ) ) { |
| 735 | $this->streamFunctionCall = [ 'name' => '', 'arguments' => [] ]; |
| 736 | } |
| 737 | $this->streamFunctionCall['name'] = $delta['function_call']['name'] ?? $this->streamFunctionCall['name']; |
| 738 | if ( isset( $delta['function_call']['arguments'] ) ) { |
| 739 | $args = json_decode( $delta['function_call']['arguments'], true ); |
| 740 | $this->streamFunctionCall['arguments'] = $args ?? []; |
| 741 | } |
| 742 | break; |
| 743 | case 'tool_call': |
| 744 | $tool_call = $delta['tool_call']; |
| 745 | $index = $tool_call['index'] ?? null; |
| 746 | // unset (not "= null") to break the reference from the previous loop |
| 747 | // iteration. Assigning null would null out the tool call we just stored |
| 748 | // in $this->streamToolCalls, leaving a null hole for parallel tool calls |
| 749 | // (Mistral rejects that with a 422; OpenAI silently tolerates it). |
| 750 | unset( $currentStreamToolCall ); |
| 751 | if ( $index !== null && isset( $this->streamToolCalls[$index] ) ) { |
| 752 | $currentStreamToolCall = &$this->streamToolCalls[$index]; |
| 753 | } |
| 754 | else { |
| 755 | $this->streamToolCalls[] = [ |
| 756 | 'id' => null, |
| 757 | 'type' => null, |
| 758 | 'function' => [ 'name' => '', 'arguments' => '' ] |
| 759 | ]; |
| 760 | end( $this->streamToolCalls ); |
| 761 | $currentStreamToolCall = &$this->streamToolCalls[ key( $this->streamToolCalls ) ]; |
| 762 | } |
| 763 | break; |
| 764 | } |
| 765 | } |
| 766 | } |
| 767 | else if ( $object === 'thread.run.step' ) { |
| 768 | $handledCondition = true; |
| 769 | //$type = $json['step']; |
| 770 | // Could be tool_calls, means an OpenAI Assistant is doing something. |
| 771 | } |
| 772 | else { |
| 773 | if ( isset( $json['choices'][0]['text'] ) ) { |
| 774 | $handledCondition = true; |
| 775 | $content = $json['choices'][0]['text']; |
| 776 | } |
| 777 | else if ( isset( $json['choices'][0]['delta']['content'] ) ) { |
| 778 | $handledCondition = true; |
| 779 | $content = $json['choices'][0]['delta']['content']; |
| 780 | } |
| 781 | else if ( isset( $json['choices'][0]['delta']['function_call'] ) ) { |
| 782 | // TODO: Re-evaluate after 2027-01-26. Deprecated "function_call" stream path |
| 783 | // (same deprecation as the object-based case above); remove once no provider uses it. |
| 784 | $handledCondition = true; |
| 785 | if ( empty( $this->streamFunctionCall ) ) { |
| 786 | $this->streamFunctionCall = [ 'name' => '', 'arguments' => [] ]; |
| 787 | } |
| 788 | $this->streamFunctionCall['name'] = $json['choices'][0]['delta']['function_call']['name'] ?? $this->streamFunctionCall['name']; |
| 789 | if ( isset( $json['choices'][0]['delta']['function_call']['arguments'] ) ) { |
| 790 | $args = json_decode( $json['choices'][0]['delta']['function_call']['arguments'], true ); |
| 791 | $this->streamFunctionCall['arguments'] = $args ?? []; |
| 792 | } |
| 793 | } |
| 794 | else if ( isset( $json['choices'][0]['delta']['tool_calls'] ) ) { |
| 795 | $handledCondition = true; |
| 796 | // New schema detected – drop any half-built legacy call to prevent duplicates |
| 797 | $this->streamFunctionCall = null; |
| 798 | |
| 799 | foreach ( $json['choices'][0]['delta']['tool_calls'] as $tool_call ) { |
| 800 | $index = $tool_call['index'] ?? null; |
| 801 | // unset (not "= null") to break the reference from the previous loop |
| 802 | // iteration. Assigning null would null out the tool call we just stored |
| 803 | // in $this->streamToolCalls, leaving a null hole for parallel tool calls |
| 804 | // (Mistral rejects that with a 422; OpenAI silently tolerates it). |
| 805 | unset( $currentStreamToolCall ); |
| 806 | if ( $index !== null && isset( $this->streamToolCalls[$index] ) ) { |
| 807 | $currentStreamToolCall = &$this->streamToolCalls[$index]; |
| 808 | } |
| 809 | else { |
| 810 | $this->streamToolCalls[] = [ |
| 811 | 'id' => null, |
| 812 | 'type' => null, |
| 813 | 'function' => [ 'name' => '', 'arguments' => '' ] |
| 814 | ]; |
| 815 | end( $this->streamToolCalls ); |
| 816 | $currentStreamToolCall = &$this->streamToolCalls[ key( $this->streamToolCalls ) ]; |
| 817 | } |
| 818 | if ( !empty( $tool_call['id'] ) ) { |
| 819 | $currentStreamToolCall['id'] = $tool_call['id']; |
| 820 | } |
| 821 | if ( !empty( $tool_call['type'] ) ) { |
| 822 | $currentStreamToolCall['type'] = $tool_call['type']; |
| 823 | } |
| 824 | if ( isset( $tool_call['function'] ) ) { |
| 825 | $function = $tool_call['function']; |
| 826 | if ( isset( $function['name'] ) ) { |
| 827 | $currentStreamToolCall['function']['name'] .= $function['name']; |
| 828 | } |
| 829 | if ( isset( $function['arguments'] ) ) { |
| 830 | $currentStreamToolCall['function']['arguments'] .= $function['arguments']; |
| 831 | } |
| 832 | } |
| 833 | $this->streamLastMessage['tool_calls'] = $this->streamToolCalls; |
| 834 | } |
| 835 | } |
| 836 | else if ( isset( $json['choices'][0]['delta']['role'] ) ) { |
| 837 | $handledCondition = true; |
| 838 | $this->streamLastMessage = [ |
| 839 | 'role' => $json['choices'][0]['delta']['role'], |
| 840 | 'content' => null |
| 841 | ]; |
| 842 | } |
| 843 | |
| 844 | // Handle thinking/reasoning content (from Ollama and potentially other models) |
| 845 | // This can appear alongside or instead of regular content |
| 846 | if ( isset( $json['choices'][0]['delta']['reasoning'] ) ) { |
| 847 | $handledCondition = true; |
| 848 | $thinking = $json['choices'][0]['delta']['reasoning']; |
| 849 | if ( !empty( $thinking ) ) { |
| 850 | $this->streamThinking = ( $this->streamThinking ?? '' ) . $thinking; |
| 851 | } |
| 852 | } |
| 853 | // Also check for 'thinking' field (OpenAI's o1 models use this) |
| 854 | else if ( isset( $json['choices'][0]['delta']['thinking'] ) ) { |
| 855 | $handledCondition = true; |
| 856 | $thinking = $json['choices'][0]['delta']['thinking']; |
| 857 | if ( !empty( $thinking ) ) { |
| 858 | $this->streamThinking = ( $this->streamThinking ?? '' ) . $thinking; |
| 859 | } |
| 860 | } |
| 861 | } |
| 862 | |
| 863 | $usage = $json['usage'] ?? []; |
| 864 | if ( isset( $usage['prompt_tokens'], $usage['completion_tokens'] ) ) { |
| 865 | $this->streamInTokens = (int) $usage['prompt_tokens']; |
| 866 | $this->streamOutTokens = (int) $usage['completion_tokens']; |
| 867 | |
| 868 | if ( isset( $usage['cost'] ) ) { |
| 869 | $this->streamCost = (float) $usage['cost']; |
| 870 | } |
| 871 | } |
| 872 | |
| 873 | // If content is an array, let's try to convert it into a string. Normally, there would be a 'value' key. |
| 874 | if ( is_array( $content ) ) { |
| 875 | if ( isset( $content['value'] ) ) { |
| 876 | $content = $content['value']; |
| 877 | } |
| 878 | else { |
| 879 | throw new Exception( 'Could not read this: ' . json_encode( $content ) ); |
| 880 | } |
| 881 | } |
| 882 | |
| 883 | // Log unhandled JSON in dev mode |
| 884 | if ( !$handledCondition && $this->core->get_option( 'dev_mode' ) ) { |
| 885 | error_log( '[AI Engine] Unhandled streaming JSON structure: ' . json_encode( $json ) ); |
| 886 | } |
| 887 | |
| 888 | // Avoid some endings |
| 889 | $endings = [ '', '</s>' ]; |
| 890 | if ( in_array( $content, $endings ) ) { |
| 891 | $content = null; |
| 892 | } |
| 893 | |
| 894 | if ( $content !== '0' && empty( $content ) ) { |
| 895 | return null; |
| 896 | } |
| 897 | if ( is_string( $content ) ) { |
| 898 | $content = $this->filter_inline_thinking( $content ); |
| 899 | if ( $content === '' ) { |
| 900 | return null; |
| 901 | } |
| 902 | } |
| 903 | return $content; |
| 904 | } |
| 905 | |
| 906 | /** |
| 907 | * Qwen-style open models emit their reasoning INLINE in the content as |
| 908 | * <think>...</think>. Route it to streamThinking (with a single thinking |
| 909 | * event per block) so the visible reply only contains the actual answer. |
| 910 | */ |
| 911 | protected function filter_inline_thinking( $content ) { |
| 912 | if ( $this->streamThinkTag === 0 && |
| 913 | strpos( $content, '<think>' ) === false && strpos( $content, '</think>' ) === false ) { |
| 914 | return $content; |
| 915 | } |
| 916 | $out = ''; |
| 917 | $thought = ''; |
| 918 | $remaining = $content; |
| 919 | while ( $remaining !== '' ) { |
| 920 | if ( $this->streamThinkTag === 0 ) { |
| 921 | $pos = strpos( $remaining, '<think>' ); |
| 922 | if ( $pos === false ) { |
| 923 | $out .= $remaining; |
| 924 | break; |
| 925 | } |
| 926 | $out .= substr( $remaining, 0, $pos ); |
| 927 | $remaining = substr( $remaining, $pos + 7 ); |
| 928 | $this->streamThinkTag = 1; |
| 929 | if ( $this->currentDebugMode && !empty( $this->streamCallback ) ) { |
| 930 | call_user_func( $this->streamCallback, Meow_MWAI_Event::thinking( 'Thinking...' ) ); |
| 931 | } |
| 932 | } |
| 933 | else { |
| 934 | $pos = strpos( $remaining, '</think>' ); |
| 935 | if ( $pos === false ) { |
| 936 | $thought .= $remaining; |
| 937 | break; |
| 938 | } |
| 939 | $thought .= substr( $remaining, 0, $pos ); |
| 940 | // The answer usually follows the block after blank lines. |
| 941 | $remaining = ltrim( substr( $remaining, $pos + 8 ), "\n" ); |
| 942 | $this->streamThinkTag = 0; |
| 943 | } |
| 944 | } |
| 945 | if ( $thought !== '' ) { |
| 946 | $this->streamThinking = ( $this->streamThinking ?? '' ) . $thought; |
| 947 | } |
| 948 | return $out; |
| 949 | } |
| 950 | |
| 951 | /** |
| 952 | * Same cleanup for non-streaming replies: closed blocks are removed, and a |
| 953 | * trailing unclosed block (the model spent its whole budget thinking) too. |
| 954 | */ |
| 955 | public static function strip_inline_thinking( $text ) { |
| 956 | if ( !is_string( $text ) || strpos( $text, '<think>' ) === false ) { |
| 957 | return $text; |
| 958 | } |
| 959 | $text = preg_replace( '/<think>.*?<\/think>/s', '', $text ); |
| 960 | $text = preg_replace( '/<think>.*$/s', '', $text ); |
| 961 | return ltrim( $text ); |
| 962 | } |
| 963 | |
| 964 | public function run( $query, $streamCallback = null, $maxDepth = 5 ) { |
| 965 | // Check if this is a realtime model being used with chat completions |
| 966 | if ( $this->is_realtime_model( $query->model ) ) { |
| 967 | throw new Exception( |
| 968 | 'Realtime models (like ' . $query->model . ') are designed for voice/audio interactions and cannot be used with this API.' |
| 969 | ); |
| 970 | } |
| 971 | |
| 972 | return parent::run( $query, $streamCallback, $maxDepth ); |
| 973 | } |
| 974 | |
| 975 | public function run_query( $url, $options, $isStream = false ) { |
| 976 | try { |
| 977 | $options['stream'] = $isStream; |
| 978 | if ( $isStream ) { |
| 979 | $options['filename'] = tempnam( sys_get_temp_dir(), 'mwai-stream-' ); |
| 980 | } |
| 981 | |
| 982 | // Check if queries debug is enabled |
| 983 | $queries_debug = $this->core->get_option( 'queries_debug_mode' ); |
| 984 | |
| 985 | // Log the request if queries debug is enabled |
| 986 | if ( $queries_debug ) { |
| 987 | error_log( '[AI Engine Queries] --> Request to: ' . $url ); |
| 988 | |
| 989 | if ( isset( $options['body'] ) ) { |
| 990 | // This is the actual body being sent to the AI service |
| 991 | $body_log = is_string( $options['body'] ) ? $options['body'] : json_encode( $options['body'] ); |
| 992 | |
| 993 | // Pretty print JSON if possible |
| 994 | $decoded = json_decode( $body_log, true ); |
| 995 | if ( json_last_error() === JSON_ERROR_NONE ) { |
| 996 | error_log( json_encode( $decoded, JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES ) ); |
| 997 | } |
| 998 | else { |
| 999 | error_log( $body_log ); |
| 1000 | } |
| 1001 | } |
| 1002 | } |
| 1003 | |
| 1004 | $res = wp_remote_get( $url, $options ); |
| 1005 | |
| 1006 | if ( is_wp_error( $res ) ) { |
| 1007 | throw new Exception( $res->get_error_message() ); |
| 1008 | } |
| 1009 | |
| 1010 | $responseCode = wp_remote_retrieve_response_code( $res ); |
| 1011 | |
| 1012 | // For streaming requests the response body is written to a temp file, so |
| 1013 | // wp_remote_retrieve_body() comes back empty. Read it back here so provider |
| 1014 | // error details (e.g. a 422 from Mistral) aren't lost on streaming requests. |
| 1015 | $errorBody = wp_remote_retrieve_body( $res ); |
| 1016 | if ( empty( $errorBody ) && $isStream && !empty( $options['filename'] ) && file_exists( $options['filename'] ) ) { |
| 1017 | $errorBody = file_get_contents( $options['filename'] ); |
| 1018 | } |
| 1019 | |
| 1020 | if ( $responseCode === 404 ) { |
| 1021 | // A 404 here is almost always an invalid model name for this provider |
| 1022 | // (the endpoint URL is correct), so surface the provider's own message. |
| 1023 | $detail = !empty( $errorBody ) ? ' Response: ' . $errorBody : ''; |
| 1024 | throw new Exception( 'The API returned 404 for ' . $url . '. This usually means the model is not valid for this provider.' . $detail ); |
| 1025 | } |
| 1026 | else if ( $responseCode === 400 ) { |
| 1027 | error_log( '[AI Engine] 400 Bad Request - Full response body: ' . $errorBody ); |
| 1028 | $message = !empty( $errorBody ) ? $errorBody : wp_remote_retrieve_response_message( $res ); |
| 1029 | if ( empty( $message ) ) { |
| 1030 | $message = 'Bad Request'; |
| 1031 | } |
| 1032 | throw new Exception( $message ); |
| 1033 | } |
| 1034 | else if ( $responseCode >= 400 ) { |
| 1035 | // Any other 4xx/5xx (401, 403, 422, 429, 500, 502...). The body is thrown |
| 1036 | // raw on purpose: the callers run it through try_decode_error() to pull out |
| 1037 | // the provider's message, whatever its shape. Before this branch was generic, |
| 1038 | // a Mistral 403 (tier_not_allowed) fell through to the JSON path and only |
| 1039 | // ever surfaced as "Invalid response (no model information)". |
| 1040 | $message = !empty( $errorBody ) ? $errorBody : wp_remote_retrieve_response_message( $res ); |
| 1041 | if ( empty( $message ) ) { |
| 1042 | $message = 'HTTP ' . $responseCode; |
| 1043 | } |
| 1044 | throw new Exception( $message ); |
| 1045 | } |
| 1046 | |
| 1047 | if ( $isStream ) { |
| 1048 | return [ 'stream' => true ]; |
| 1049 | } |
| 1050 | |
| 1051 | $response = wp_remote_retrieve_body( $res ); |
| 1052 | $headersRes = wp_remote_retrieve_headers( $res ); |
| 1053 | $headers = $headersRes->getAll(); |
| 1054 | |
| 1055 | // Check if Content-Type is 'multipart/form-data' or 'text/plain' |
| 1056 | // If so, we don't need to decode the response |
| 1057 | $normalizedHeaders = array_change_key_case( $headers, CASE_LOWER ); |
| 1058 | $resContentType = $normalizedHeaders['content-type'] ?? ''; |
| 1059 | if ( strpos( $resContentType, 'multipart/form-data' ) !== false || strpos( $resContentType, 'text/plain' ) !== false ) { |
| 1060 | // Log the response if queries debug is enabled |
| 1061 | if ( $queries_debug && !$isStream ) { |
| 1062 | error_log( '[AI Engine Queries] Response Headers: ' . json_encode( $headers ) ); |
| 1063 | error_log( '[AI Engine Queries] Response Body (raw): ' . substr( $response, 0, 1000 ) . '...' ); |
| 1064 | } |
| 1065 | return [ 'stream' => false, 'headers' => $headers, 'data' => $response ]; |
| 1066 | } |
| 1067 | |
| 1068 | $data = json_decode( $response, true ); |
| 1069 | $this->handle_response_errors( $data ); |
| 1070 | |
| 1071 | // Log the response if queries debug is enabled |
| 1072 | if ( $queries_debug && !$isStream ) { |
| 1073 | // Log the raw response as received from the AI service |
| 1074 | error_log( '[AI Engine Queries] <-- Response:' ); |
| 1075 | |
| 1076 | // Pretty print JSON if possible |
| 1077 | if ( json_last_error() === JSON_ERROR_NONE && is_array( $data ) ) { |
| 1078 | error_log( json_encode( $data, JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES ) ); |
| 1079 | } |
| 1080 | else { |
| 1081 | error_log( $response ); |
| 1082 | } |
| 1083 | } |
| 1084 | |
| 1085 | return [ 'headers' => $headers, 'data' => $data ]; |
| 1086 | } |
| 1087 | catch ( Exception $e ) { |
| 1088 | $service = $this->get_service_name(); |
| 1089 | Meow_MWAI_Logging::error( "$service: " . $e->getMessage() ); |
| 1090 | |
| 1091 | // Log error response if queries debug is enabled |
| 1092 | if ( $queries_debug ) { |
| 1093 | error_log( '[AI Engine Queries] Error occurred: ' . $e->getMessage() ); |
| 1094 | } |
| 1095 | |
| 1096 | throw $e; |
| 1097 | } |
| 1098 | finally { |
| 1099 | if ( $isStream && file_exists( $options['filename'] ) ) { |
| 1100 | unlink( $options['filename'] ); |
| 1101 | } |
| 1102 | } |
| 1103 | } |
| 1104 | |
| 1105 | private function get_audio( $url ) { |
| 1106 | require_once( ABSPATH . 'wp-admin/includes/media.php' ); |
| 1107 | |
| 1108 | // Validate URL scheme to prevent SSRF attacks |
| 1109 | $parts = wp_parse_url( $url ); |
| 1110 | if ( !isset( $parts['scheme'] ) || !in_array( $parts['scheme'], [ 'http', 'https' ], true ) ) { |
| 1111 | throw new Exception( 'Invalid URL scheme; only HTTP/HTTPS allowed.' ); |
| 1112 | } |
| 1113 | |
| 1114 | // Use wp_safe_remote_get to block localhost/private IPs (SSRF protection) |
| 1115 | $response = wp_safe_remote_get( $url, [ |
| 1116 | 'timeout' => 60, |
| 1117 | 'redirection' => 0 // Prevent redirect-based bypass |
| 1118 | ] ); |
| 1119 | if ( is_wp_error( $response ) ) { |
| 1120 | throw new Exception( 'Failed to download audio: ' . $response->get_error_message() ); |
| 1121 | } |
| 1122 | $audio_data = wp_remote_retrieve_body( $response ); |
| 1123 | if ( empty( $audio_data ) ) { |
| 1124 | throw new Exception( 'Failed to download audio: empty response.' ); |
| 1125 | } |
| 1126 | |
| 1127 | $tmpFile = tempnam( sys_get_temp_dir(), 'audio_' ); |
| 1128 | file_put_contents( $tmpFile, $audio_data ); |
| 1129 | $length = null; |
| 1130 | $metadata = wp_read_audio_metadata( $tmpFile ); |
| 1131 | if ( isset( $metadata['length'] ) ) { |
| 1132 | $length = $metadata['length']; |
| 1133 | } |
| 1134 | $data = file_get_contents( $tmpFile ); |
| 1135 | unlink( $tmpFile ); |
| 1136 | return [ 'data' => $data, 'length' => $length ]; |
| 1137 | } |
| 1138 | |
| 1139 | public function run_transcribe_query( $query ) { |
| 1140 | $audioData = null; |
| 1141 | |
| 1142 | // Priority 1: Direct audio data |
| 1143 | if ( !empty( $query->audioData ) ) { |
| 1144 | $audioData = [ |
| 1145 | 'data' => $query->audioData, |
| 1146 | 'length' => strlen( $query->audioData ) / 1024 // KB |
| 1147 | ]; |
| 1148 | } |
| 1149 | // Priority 2: File path |
| 1150 | else if ( !empty( $query->path ) ) { |
| 1151 | // Last line of defense: REST callers never get to set this (the routes strip the |
| 1152 | // key), but simpleTranscribeAudio() is public PHP API, so a path can legitimately |
| 1153 | // arrive from site code. Confinement is not applied here on purpose: real callers |
| 1154 | // do transcribe files from outside the uploads directory. Stream wrappers are the |
| 1155 | // part that is never legitimate. |
| 1156 | $path = Meow_MWAI_Core::sanitize_file_path( $query->path ); |
| 1157 | if ( !file_exists( $path ) ) { |
| 1158 | throw new Exception( 'Audio file not found: ' . $path ); |
| 1159 | } |
| 1160 | if ( !is_readable( $path ) ) { |
| 1161 | throw new Exception( 'Audio file is not readable: ' . $path ); |
| 1162 | } |
| 1163 | $audioData = [ |
| 1164 | 'data' => file_get_contents( $path ), |
| 1165 | 'length' => filesize( $path ) / 1024 // KB |
| 1166 | ]; |
| 1167 | } |
| 1168 | // Priority 3: Attached file object |
| 1169 | else if ( method_exists( $query, 'getAttachments' ) ) { |
| 1170 | $attachments = $query->getAttachments(); |
| 1171 | if ( !empty( $attachments ) ) { |
| 1172 | $file = $attachments[0]; |
| 1173 | $audioData = [ |
| 1174 | 'data' => $file->get_data(), |
| 1175 | 'length' => strlen( $file->get_data() ) / 1024 // KB |
| 1176 | ]; |
| 1177 | } |
| 1178 | } |
| 1179 | // Priority 4: URL (backward compatibility) |
| 1180 | else if ( !empty( $query->url ) ) { |
| 1181 | if ( !filter_var( $query->url, FILTER_VALIDATE_URL ) ) { |
| 1182 | throw new Exception( 'Invalid URL for transcription.' ); |
| 1183 | } |
| 1184 | $audioData = $this->get_audio( $query->url ); |
| 1185 | } |
| 1186 | else { |
| 1187 | throw new Exception( 'No audio source provided for transcription. Please provide either audioData, path, attachedFiles, or url.' ); |
| 1188 | } |
| 1189 | |
| 1190 | $body = $this->build_body( $query, null, $audioData['data'] ); |
| 1191 | $url = $this->build_url( $query ); |
| 1192 | $headers = $this->build_headers( $query ); |
| 1193 | $options = $this->build_options( $headers, null, $body ); |
| 1194 | |
| 1195 | // Perform the request |
| 1196 | try { |
| 1197 | $res = $this->run_query( $url, $options ); |
| 1198 | $data = $res['data']; |
| 1199 | if ( empty( $data ) ) { |
| 1200 | throw new Exception( 'Invalid data for transcription.' ); |
| 1201 | } |
| 1202 | $usage = $this->core->record_audio_usage( $query->model, $audioData['length'] ); |
| 1203 | $reply = new Meow_MWAI_Reply( $query ); |
| 1204 | $reply->set_usage( $usage ); |
| 1205 | $reply->set_choices( $data ); |
| 1206 | return $reply; |
| 1207 | } |
| 1208 | catch ( Exception $e ) { |
| 1209 | $service = $this->get_service_name(); |
| 1210 | Meow_MWAI_Logging::error( "$service: " . $e->getMessage() ); |
| 1211 | throw new Exception( "$service: " . $e->getMessage() ); |
| 1212 | } |
| 1213 | } |
| 1214 | |
| 1215 | public function run_embedding_query( $query ) { |
| 1216 | // Pre-validate token count against the embedding model's limit. |
| 1217 | // OpenAI embedding models (text-embedding-3-small/large, ada-002) all cap |
| 1218 | // at 8192 input tokens; sending a longer chunk yields an opaque |
| 1219 | // "Invalid 'input': maximum context length is 8192 tokens" error. Count |
| 1220 | // locally via tiktoken (bundled) so we fail with a clear message up front. |
| 1221 | $embedding_max_tokens = apply_filters( 'mwai_embedding_max_tokens', 8192, $query->model ); |
| 1222 | if ( !empty( $query->message ) && $embedding_max_tokens > 0 ) { |
| 1223 | $token_count = Meow_MWAI_Core::estimate_tokens( $query->message, $query->model ); |
| 1224 | if ( $token_count > $embedding_max_tokens ) { |
| 1225 | throw new Exception( sprintf( |
| 1226 | 'Embedding input is %d tokens, exceeds the %d-token limit for %s. Split this entry into smaller chunks and try again.', |
| 1227 | $token_count, |
| 1228 | $embedding_max_tokens, |
| 1229 | $query->model |
| 1230 | ) ); |
| 1231 | } |
| 1232 | } |
| 1233 | |
| 1234 | $body = $this->build_body( $query ); |
| 1235 | $url = $this->build_url( $query ); |
| 1236 | $headers = $this->build_headers( $query ); |
| 1237 | $options = $this->build_options( $headers, $body ); |
| 1238 | |
| 1239 | try { |
| 1240 | $res = $this->run_query( $url, $options ); |
| 1241 | $data = $res['data']; |
| 1242 | if ( empty( $data ) || !isset( $data['data'] ) ) { |
| 1243 | throw new Exception( 'Invalid data for embedding.' ); |
| 1244 | } |
| 1245 | $usage = $data['usage']; |
| 1246 | $this->core->record_tokens_usage( $query->model, $usage['prompt_tokens'] ); |
| 1247 | $reply = new Meow_MWAI_Reply( $query ); |
| 1248 | $reply->set_usage( $usage ); |
| 1249 | $reply->set_choices( $data['data'] ); |
| 1250 | return $reply; |
| 1251 | } |
| 1252 | catch ( Exception $e ) { |
| 1253 | $message = $e->getMessage(); |
| 1254 | $error = $this->try_decode_error( $message ); |
| 1255 | if ( !is_null( $error ) ) { |
| 1256 | $message = $error; |
| 1257 | } |
| 1258 | $service = $this->get_service_name(); |
| 1259 | Meow_MWAI_Logging::error( "$service: " . $message ); |
| 1260 | throw new Exception( "$service: " . $message ); |
| 1261 | } |
| 1262 | } |
| 1263 | |
| 1264 | public function try_decode_error( $data ) { |
| 1265 | $json = json_decode( $data, true ); |
| 1266 | return is_array( $json ) ? $this->extract_error_message( $json ) : null; |
| 1267 | } |
| 1268 | |
| 1269 | /** |
| 1270 | * Pulls the human-readable message out of a decoded provider error body, or null |
| 1271 | * when the body does not look like an error. Two shapes exist in the wild: |
| 1272 | * - OpenAI-style, nested: {"error": {"message": "...", "type": "..."}} (or "error" as a string) |
| 1273 | * - Mistral-style, flat: {"object": "error", "message": "...", "type": "...", "code": "..."} |
| 1274 | */ |
| 1275 | protected function extract_error_message( array $json ): ?string { |
| 1276 | if ( !empty( $json['error'] ) ) { |
| 1277 | $err = $json['error']; |
| 1278 | if ( is_array( $err ) ) { |
| 1279 | return isset( $err['message'] ) ? (string) $err['message'] : json_encode( $err ); |
| 1280 | } |
| 1281 | return (string) $err; |
| 1282 | } |
| 1283 | $isFlatError = ( $json['object'] ?? '' ) === 'error' || ( isset( $json['type'] ) && !isset( $json['choices'] ) ); |
| 1284 | if ( $isFlatError && !empty( $json['message'] ) && is_string( $json['message'] ) ) { |
| 1285 | return $json['message']; |
| 1286 | } |
| 1287 | return null; |
| 1288 | } |
| 1289 | |
| 1290 | protected function finalize_choices( $choices, $responseData, $query ) { |
| 1291 | // Clean up duplicate function calls: prefer tool_calls over legacy function_call |
| 1292 | foreach ( $choices as &$choice ) { |
| 1293 | if ( isset( $choice['message'] ) ) { |
| 1294 | // If we have both tool_calls and function_call, remove function_call |
| 1295 | if ( isset( $choice['message']['tool_calls'] ) && !empty( $choice['message']['tool_calls'] ) && |
| 1296 | isset( $choice['message']['function_call'] ) ) { |
| 1297 | unset( $choice['message']['function_call'] ); |
| 1298 | } |
| 1299 | } |
| 1300 | } |
| 1301 | return $choices; |
| 1302 | } |
| 1303 | |
| 1304 | public function run_completion_query( $query, $streamCallback = null ): Meow_MWAI_Reply { |
| 1305 | // Check if this is a GPT-5 model - they don't support Chat Completions API |
| 1306 | if ( $this->is_gpt5_model( $query->model ) ) { |
| 1307 | throw new Exception( 'GPT-5 models only support the Responses API. Please enable "Use Responses API" in AI Engine settings to use ' . $query->model . '.' ); |
| 1308 | } |
| 1309 | |
| 1310 | $isStreaming = !is_null( $streamCallback ); |
| 1311 | |
| 1312 | // Initialize debug mode |
| 1313 | $this->init_debug_mode( $query ); |
| 1314 | |
| 1315 | // Store the callback for event emission (both streaming and non-streaming debug mode) |
| 1316 | if ( !is_null( $streamCallback ) ) { |
| 1317 | $this->streamCallback = $streamCallback; |
| 1318 | } |
| 1319 | |
| 1320 | if ( $isStreaming ) { |
| 1321 | add_action( 'http_api_curl', [ $this, 'stream_handler' ], 10, 3 ); |
| 1322 | } |
| 1323 | |
| 1324 | $this->reset_stream(); |
| 1325 | $body = $this->build_body( $query, $streamCallback ); |
| 1326 | $url = $this->build_url( $query ); |
| 1327 | $headers = $this->build_headers( $query ); |
| 1328 | $options = $this->build_options( $headers, $body ); |
| 1329 | |
| 1330 | // Emit "Request sent" event for feedback queries |
| 1331 | if ( $this->currentDebugMode && !empty( $streamCallback ) && |
| 1332 | ( $query instanceof Meow_MWAI_Query_Feedback || $query instanceof Meow_MWAI_Query_AssistFeedback ) ) { |
| 1333 | $event = Meow_MWAI_Event::request_sent() |
| 1334 | ->set_metadata( 'is_feedback', true ) |
| 1335 | ->set_metadata( 'feedback_count', count( $query->blocks ) ); |
| 1336 | call_user_func( $streamCallback, $event ); |
| 1337 | } |
| 1338 | |
| 1339 | try { |
| 1340 | $res = $this->run_query( $url, $options, $streamCallback ); |
| 1341 | $reply = new Meow_MWAI_Reply( $query ); |
| 1342 | |
| 1343 | $returned_id = null; |
| 1344 | $returned_model = $this->inModel; |
| 1345 | $returned_in_tokens = null; |
| 1346 | $returned_out_tokens = null; |
| 1347 | $returned_price = null; |
| 1348 | $returned_choices = []; |
| 1349 | |
| 1350 | // Streaming Mode |
| 1351 | if ( $isStreaming ) { |
| 1352 | if ( empty( $this->streamContent ) ) { |
| 1353 | $error = $this->try_decode_error( $this->streamBuffer ); |
| 1354 | if ( !is_null( $error ) ) { |
| 1355 | throw new Exception( $error ); |
| 1356 | } |
| 1357 | // A stream that ends without a single chunk (no id, no content, no |
| 1358 | // tool call, no thinking) is a provider failure, not a reply. OVH |
| 1359 | // answers 200 with a lone [DONE] when it rejects a parameter, which |
| 1360 | // used to be saved as a silent empty message. |
| 1361 | if ( empty( $this->streamToolCalls ) && empty( $this->streamFunctionCall ) && |
| 1362 | empty( $this->streamThinking ) && is_null( $this->inId ) ) { |
| 1363 | throw new Exception( 'The AI service closed the stream without sending a reply. A parameter (often Max Tokens) may exceed what this model supports.' ); |
| 1364 | } |
| 1365 | } |
| 1366 | $returned_id = $this->inId; |
| 1367 | $returned_model = $this->inModel ? $this->inModel : $query->model; |
| 1368 | |
| 1369 | // Use regular content if available, otherwise fall back to thinking/reasoning |
| 1370 | $finalContent = $this->streamContent; |
| 1371 | if ( empty( $finalContent ) && !empty( $this->streamThinking ) ) { |
| 1372 | // Use thinking content as fallback when there's no regular content |
| 1373 | // This happens with Ollama when it returns only reasoning/thinking |
| 1374 | // Wrap in asterisks to show as italics in markdown |
| 1375 | $finalContent = '*' . $this->streamThinking . '*'; |
| 1376 | |
| 1377 | // Log this for debugging |
| 1378 | if ( $this->core->get_option( 'queries_debug_mode' ) ) { |
| 1379 | error_log( '[AI Engine] Using thinking/reasoning content as fallback (no regular content available)' ); |
| 1380 | } |
| 1381 | } |
| 1382 | |
| 1383 | $message = [ 'role' => 'assistant', 'content' => $finalContent ]; |
| 1384 | // Prefer tool_calls; fall back to legacy only if necessary |
| 1385 | if ( !empty( $this->streamToolCalls ) ) { |
| 1386 | $message['tool_calls'] = $this->streamToolCalls; |
| 1387 | } |
| 1388 | elseif ( !empty( $this->streamFunctionCall ) ) { |
| 1389 | $message['function_call'] = $this->streamFunctionCall; |
| 1390 | } |
| 1391 | |
| 1392 | // Optionally include thinking as metadata if both content and thinking exist |
| 1393 | if ( !empty( $this->streamContent ) && !empty( $this->streamThinking ) ) { |
| 1394 | $message['thinking'] = $this->streamThinking; |
| 1395 | } |
| 1396 | if ( !is_null( $this->streamInTokens ) ) { |
| 1397 | $returned_in_tokens = $this->streamInTokens; |
| 1398 | } |
| 1399 | if ( !is_null( $this->streamOutTokens ) ) { |
| 1400 | $returned_out_tokens = $this->streamOutTokens; |
| 1401 | } |
| 1402 | if ( !is_null( $this->streamCost ) ) { |
| 1403 | $returned_price = $this->streamCost; |
| 1404 | } |
| 1405 | $returned_choices = [ [ 'message' => $message ] ]; |
| 1406 | $returned_choices = $this->finalize_choices( $returned_choices, null, $query ); |
| 1407 | |
| 1408 | // Log streaming response data if queries debug is enabled |
| 1409 | $queries_debug = $this->core->get_option( 'queries_debug_mode' ); |
| 1410 | if ( $queries_debug ) { |
| 1411 | error_log( '[AI Engine Queries] Streaming Response Collected (ChatML):' ); |
| 1412 | $streaming_data = [ |
| 1413 | 'id' => $returned_id, |
| 1414 | 'model' => $returned_model, |
| 1415 | 'content_length' => strlen( $this->streamContent ), |
| 1416 | 'content_preview' => substr( $this->streamContent, 0, 200 ) . ( strlen( $this->streamContent ) > 200 ? '...' : '' ), |
| 1417 | 'function_calls' => !empty( $this->streamFunctionCall ) ? '1 function call' : 'none', |
| 1418 | 'tool_calls' => !empty( $this->streamToolCalls ) ? count( $this->streamToolCalls ) . ' tool calls' : 'none', |
| 1419 | 'usage' => [ |
| 1420 | 'input_tokens' => $returned_in_tokens, |
| 1421 | 'output_tokens' => $returned_out_tokens, |
| 1422 | 'cost' => $returned_price |
| 1423 | ] |
| 1424 | ]; |
| 1425 | |
| 1426 | // Log tool calls details if present |
| 1427 | if ( !empty( $this->streamToolCalls ) ) { |
| 1428 | $streaming_data['tool_calls_details'] = []; |
| 1429 | foreach ( $this->streamToolCalls as $tool_call ) { |
| 1430 | $streaming_data['tool_calls_details'][] = [ |
| 1431 | 'id' => $tool_call['id'] ?? 'unknown', |
| 1432 | 'name' => $tool_call['function']['name'] ?? 'unknown', |
| 1433 | 'arguments' => substr( $tool_call['function']['arguments'] ?? '{}', 0, 100 ) . '...' |
| 1434 | ]; |
| 1435 | } |
| 1436 | } |
| 1437 | |
| 1438 | // Log function call if present |
| 1439 | if ( !empty( $this->streamFunctionCall ) ) { |
| 1440 | $streaming_data['function_call'] = [ |
| 1441 | 'name' => $this->streamFunctionCall['name'] ?? 'unknown', |
| 1442 | 'arguments' => substr( $this->streamFunctionCall['arguments'] ?? '{}', 0, 100 ) . '...' |
| 1443 | ]; |
| 1444 | } |
| 1445 | |
| 1446 | error_log( json_encode( $streaming_data, JSON_PRETTY_PRINT ) ); |
| 1447 | } |
| 1448 | } |
| 1449 | // Standard Mode |
| 1450 | else { |
| 1451 | $data = $res['data']; |
| 1452 | if ( empty( $data ) ) { |
| 1453 | throw new Exception( 'No content received (res is null).' ); |
| 1454 | } |
| 1455 | |
| 1456 | // Comprehensive logging for non-streaming mode - capture FULL response |
| 1457 | $queries_debug = $this->core->get_option( 'queries_debug_mode' ); |
| 1458 | if ( $queries_debug ) { |
| 1459 | error_log( '[AI Engine Queries] ========================================' ); |
| 1460 | error_log( '[AI Engine Queries] FULL RESPONSE STRUCTURE (Non-streaming ChatML):' ); |
| 1461 | error_log( json_encode( $data, JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES ) ); |
| 1462 | error_log( '[AI Engine Queries] ========================================' ); |
| 1463 | |
| 1464 | // Look specifically for container_id |
| 1465 | $this->search_for_container_id_recursive( $data, '' ); |
| 1466 | } |
| 1467 | |
| 1468 | if ( !$data['model'] ) { |
| 1469 | $service = $this->get_service_name(); |
| 1470 | Meow_MWAI_Logging::error( "$service: Invalid response (no model information)." ); |
| 1471 | Meow_MWAI_Logging::error( print_r( $data, 1 ) ); |
| 1472 | throw new Exception( 'Invalid response (no model information).' ); |
| 1473 | } |
| 1474 | $returned_id = $data['id']; |
| 1475 | $returned_model = $data['model']; |
| 1476 | $usage = $data['usage'] ?? []; |
| 1477 | $returned_in_tokens = $usage['prompt_tokens'] ?? null; |
| 1478 | $returned_out_tokens = $usage['completion_tokens'] ?? null; |
| 1479 | $returned_price = $usage['total_cost'] ?? $usage['cost'] ?? null; |
| 1480 | $returned_choices = $data['choices']; |
| 1481 | $returned_choices = $this->finalize_choices( $returned_choices, $data, $query ); |
| 1482 | |
| 1483 | // Strip inline <think> reasoning (Qwen-style models) from the reply. |
| 1484 | foreach ( $returned_choices as &$choice ) { |
| 1485 | if ( isset( $choice['message']['content'] ) && is_string( $choice['message']['content'] ) ) { |
| 1486 | $choice['message']['content'] = self::strip_inline_thinking( $choice['message']['content'] ); |
| 1487 | } |
| 1488 | } |
| 1489 | unset( $choice ); |
| 1490 | } |
| 1491 | |
| 1492 | // Set the results. |
| 1493 | $reply->set_choices( $returned_choices ); |
| 1494 | if ( !empty( $returned_id ) ) { |
| 1495 | $reply->set_id( $returned_id ); |
| 1496 | } |
| 1497 | if ( !empty( $returned_id ) ) { |
| 1498 | $reply->set_id( $returned_id ); |
| 1499 | } |
| 1500 | |
| 1501 | // Handle tokens. |
| 1502 | $this->handle_tokens_usage( |
| 1503 | $reply, |
| 1504 | $query, |
| 1505 | $returned_model, |
| 1506 | $returned_in_tokens, |
| 1507 | $returned_out_tokens, |
| 1508 | $returned_price |
| 1509 | ); |
| 1510 | |
| 1511 | return $reply; |
| 1512 | } |
| 1513 | catch ( Exception $e ) { |
| 1514 | $service = $this->get_service_name(); |
| 1515 | Meow_MWAI_Logging::error( "$service: " . $e->getMessage() ); |
| 1516 | $message = "$service: " . $e->getMessage(); |
| 1517 | throw new Exception( $message ); |
| 1518 | } |
| 1519 | finally { |
| 1520 | if ( !is_null( $streamCallback ) ) { |
| 1521 | remove_action( 'http_api_curl', [ $this, 'stream_handler' ] ); |
| 1522 | } |
| 1523 | } |
| 1524 | } |
| 1525 | |
| 1526 | public function handle_tokens_usage( |
| 1527 | $reply, |
| 1528 | $query, |
| 1529 | $returned_model, |
| 1530 | $returned_in_tokens, |
| 1531 | $returned_out_tokens, |
| 1532 | $returned_price = null |
| 1533 | ) { |
| 1534 | $returned_in_tokens = !is_null( $returned_in_tokens ) ? $returned_in_tokens : |
| 1535 | $reply->get_in_tokens( $query ); |
| 1536 | $returned_out_tokens = !is_null( $returned_out_tokens ) ? $returned_out_tokens : |
| 1537 | $reply->get_out_tokens(); |
| 1538 | $returned_price = !is_null( $returned_price ) ? $returned_price : |
| 1539 | $reply->get_price(); |
| 1540 | $usage = $this->core->record_tokens_usage( |
| 1541 | $returned_model, |
| 1542 | $returned_in_tokens, |
| 1543 | $returned_out_tokens, |
| 1544 | $returned_price |
| 1545 | ); |
| 1546 | $reply->set_usage( $usage ); |
| 1547 | |
| 1548 | // Store the actual model returned by the API |
| 1549 | $reply->set_model( $returned_model ); |
| 1550 | |
| 1551 | // Set default accuracy to 'estimated' for engines that don't override |
| 1552 | // Most engines (Google, Anthropic, etc.) estimate tokens and calculate price |
| 1553 | $reply->set_usage_accuracy( 'estimated' ); |
| 1554 | } |
| 1555 | |
| 1556 | // Request to the OpenAI Images API (gpt-image models, token-based pricing) |
| 1557 | public function run_image_query( $query, $streamCallback = null ) { |
| 1558 | $body = $this->build_body( $query ); |
| 1559 | $url = $this->build_url( $query ); |
| 1560 | $headers = $this->build_headers( $query ); |
| 1561 | $options = $this->build_options( $headers, $body ); |
| 1562 | |
| 1563 | try { |
| 1564 | $res = $this->run_query( $url, $options ); |
| 1565 | $data = $res['data']; |
| 1566 | $choices = $data['data']; |
| 1567 | $reply = new Meow_MWAI_Reply( $query ); |
| 1568 | $model = $query->model; |
| 1569 | |
| 1570 | // gpt-image models always return usage with input/output tokens |
| 1571 | // (Images API uses 'input_tokens'/'output_tokens', not 'prompt_tokens'/'completion_tokens') |
| 1572 | $usage = $data['usage'] ?? []; |
| 1573 | $promptTokens = $usage['input_tokens'] ?? $usage['prompt_tokens'] ?? 0; |
| 1574 | $completionTokens = $usage['output_tokens'] ?? $usage['completion_tokens'] ?? 0; |
| 1575 | $this->core->record_tokens_usage( $model, $promptTokens, $completionTokens ); |
| 1576 | $usage['queries'] = 1; |
| 1577 | $usage['accuracy'] = 'tokens'; |
| 1578 | $reply->set_usage( $usage ); |
| 1579 | $reply->set_usage_accuracy( 'tokens' ); |
| 1580 | |
| 1581 | $reply->set_choices( $choices ); |
| 1582 | $reply->set_type( 'images' ); |
| 1583 | |
| 1584 | if ( $query->localDownload === 'uploads' || $query->localDownload === 'library' ) { |
| 1585 | foreach ( $reply->results as &$result ) { |
| 1586 | $fileId = $this->core->files->upload_file( $result, null, 'generated', [ |
| 1587 | 'query_envId' => $query->envId, |
| 1588 | 'query_session' => $query->session, |
| 1589 | 'query_model' => $query->model, |
| 1590 | ], $query->envId, $query->localDownload, $query->localDownloadExpiry ); |
| 1591 | $fileUrl = $this->core->files->get_url( $fileId ); |
| 1592 | $result = $fileUrl; |
| 1593 | } |
| 1594 | } |
| 1595 | $reply->result = $reply->results[0]; |
| 1596 | return $reply; |
| 1597 | } |
| 1598 | catch ( Exception $e ) { |
| 1599 | $service = $this->get_service_name(); |
| 1600 | Meow_MWAI_Logging::error( "$service: " . $e->getMessage() ); |
| 1601 | throw new Exception( "$service: " . $e->getMessage() ); |
| 1602 | } |
| 1603 | } |
| 1604 | |
| 1605 | public function run_editimage_query( $query ) { |
| 1606 | $attachments = method_exists( $query, 'getAttachments' ) ? $query->getAttachments() : []; |
| 1607 | if ( empty( $attachments ) ) { |
| 1608 | throw new Exception( 'No image provided for editing.' ); |
| 1609 | } |
| 1610 | // Ensure the model supports image editing |
| 1611 | $modelInfo = $this->retrieve_model_info( $query->model ); |
| 1612 | if ( empty( $modelInfo['tags'] ) || !in_array( 'image-edit', $modelInfo['tags'] ) ) { |
| 1613 | throw new Exception( 'The model ' . $query->model . ' does not support image editing.' ); |
| 1614 | } |
| 1615 | $file = $attachments[0]; |
| 1616 | $imageData = $file->get_data(); |
| 1617 | $body = $this->build_body( $query, null, $imageData ); |
| 1618 | $url = $this->build_url( $query ); |
| 1619 | $headers = $this->build_headers( $query ); |
| 1620 | $options = $this->build_options( $headers, null, $body ); |
| 1621 | |
| 1622 | try { |
| 1623 | $res = $this->run_query( $url, $options ); |
| 1624 | $data = $res['data']; |
| 1625 | $choices = $data['data']; |
| 1626 | $reply = new Meow_MWAI_Reply( $query ); |
| 1627 | $model = $query->model; |
| 1628 | |
| 1629 | // gpt-image models always return usage with input/output tokens |
| 1630 | // (Images API uses 'input_tokens'/'output_tokens', not 'prompt_tokens'/'completion_tokens') |
| 1631 | $usage = $data['usage'] ?? []; |
| 1632 | $promptTokens = $usage['input_tokens'] ?? $usage['prompt_tokens'] ?? 0; |
| 1633 | $completionTokens = $usage['output_tokens'] ?? $usage['completion_tokens'] ?? 0; |
| 1634 | $this->core->record_tokens_usage( $model, $promptTokens, $completionTokens ); |
| 1635 | $usage['queries'] = 1; |
| 1636 | $usage['accuracy'] = 'tokens'; |
| 1637 | $reply->set_usage( $usage ); |
| 1638 | $reply->set_usage_accuracy( 'tokens' ); |
| 1639 | |
| 1640 | $reply->set_choices( $choices ); |
| 1641 | $reply->set_type( 'images' ); |
| 1642 | |
| 1643 | if ( $query->localDownload === 'uploads' || $query->localDownload === 'library' ) { |
| 1644 | foreach ( $reply->results as &$result ) { |
| 1645 | $fileId = $this->core->files->upload_file( $result, null, 'generated', [ |
| 1646 | 'query_envId' => $query->envId, |
| 1647 | 'query_session' => $query->session, |
| 1648 | 'query_model' => $query->model, |
| 1649 | ], $query->envId, $query->localDownload, $query->localDownloadExpiry ); |
| 1650 | $fileUrl = $this->core->files->get_url( $fileId ); |
| 1651 | $result = $fileUrl; |
| 1652 | } |
| 1653 | } |
| 1654 | $reply->result = $reply->results[0]; |
| 1655 | return $reply; |
| 1656 | } |
| 1657 | catch ( Exception $e ) { |
| 1658 | $service = $this->get_service_name(); |
| 1659 | Meow_MWAI_Logging::error( "$service: " . $e->getMessage() ); |
| 1660 | throw new Exception( "$service: " . $e->getMessage() ); |
| 1661 | } |
| 1662 | } |
| 1663 | |
| 1664 | /* |
| 1665 | This is the rest of the OpenAI API support, not related to the models directly. |
| 1666 | */ |
| 1667 | |
| 1668 | // Check if there are errors in the response from OpenAI, and throw an exception if so. |
| 1669 | // OpenAI uses { error: { message: ... } }; some compatible providers (xAI, etc.) use a |
| 1670 | // plain string in the error field — handle both shapes. |
| 1671 | protected function handle_response_errors( $data ) { |
| 1672 | $message = is_array( $data ) ? $this->extract_error_message( $data ) : null; |
| 1673 | if ( is_null( $message ) ) { |
| 1674 | return; |
| 1675 | } |
| 1676 | if ( preg_match( '/API key provided(: .*)\./', $message, $matches ) ) { |
| 1677 | $message = str_replace( $matches[1], '', $message ); |
| 1678 | } |
| 1679 | throw new Exception( $message ); |
| 1680 | } |
| 1681 | |
| 1682 | public function list_files( $purposeFilter = null ) { |
| 1683 | if ( empty( $purposeFilter ) ) { |
| 1684 | return $this->execute( 'GET', '/files' ); |
| 1685 | } |
| 1686 | return $this->execute( 'GET', '/files', [ 'purpose' => $purposeFilter ] ); |
| 1687 | } |
| 1688 | |
| 1689 | public static function get_suffix_for_model( $model ) { |
| 1690 | // Legacy fine-tuned models |
| 1691 | preg_match( "/:([a-zA-Z0-9\-]{1,40})-([0-9]{4})-([0-9]{2})-([0-9]{2})/", $model, $matches ); |
| 1692 | if ( count( $matches ) > 0 ) { |
| 1693 | return $matches[1]; |
| 1694 | } |
| 1695 | |
| 1696 | // New fine-tuned models |
| 1697 | preg_match( '/:([^:]+)(?=:[^:]+$)/', $model, $matches ); |
| 1698 | if ( count( $matches ) > 0 ) { |
| 1699 | return $matches[1]; |
| 1700 | } |
| 1701 | |
| 1702 | return 'N/A'; |
| 1703 | } |
| 1704 | |
| 1705 | public static function get_model_without_release_date( $model ) { |
| 1706 | if ( empty( $model ) ) { |
| 1707 | return null; |
| 1708 | } |
| 1709 | return preg_replace( '/-\d{4}-\d{2}-\d{2}$/', '', $model ); |
| 1710 | } |
| 1711 | |
| 1712 | // TODO: Remove every list_finetunes / cancel_finetune / delete_finetune / run_finetune method below after 2027-02 (OpenAI ends fine-tune job creation on 2027-01-06). The `finetunes` env keys in init.php and the `finetune` key on every entry in constants/models.php go with them. |
| 1713 | public function list_deleted_finetunes( $envId = null, $legacy = false ) { |
| 1714 | $finetunes = $this->list_finetunes( $legacy ); |
| 1715 | $deleted = []; |
| 1716 | |
| 1717 | foreach ( $finetunes as $finetune ) { |
| 1718 | $name = $finetune['model']; |
| 1719 | $isSucceeded = $finetune['status'] === 'succeeded'; |
| 1720 | if ( $isSucceeded ) { |
| 1721 | try { |
| 1722 | $finetune = $this->get_model( $name ); |
| 1723 | } |
| 1724 | catch ( Exception $e ) { |
| 1725 | $deleted[] = $name; |
| 1726 | } |
| 1727 | } |
| 1728 | } |
| 1729 | if ( $legacy ) { |
| 1730 | $this->core->update_ai_env( $this->envId, 'legacy_finetunes_deleted', $deleted ); |
| 1731 | } |
| 1732 | else { |
| 1733 | $this->core->update_ai_env( $this->envId, 'finetunes_deleted', $deleted ); |
| 1734 | } |
| 1735 | return $deleted; |
| 1736 | } |
| 1737 | |
| 1738 | // TODO: This was used to retrieve the fine-tuned models, but not sure this is how we should |
| 1739 | // retrieve all the models since Summer 2023, let's see! WIP. |
| 1740 | public function list_finetunes( $legacy = false ) { |
| 1741 | if ( $legacy ) { |
| 1742 | $res = $this->execute( 'GET', '/fine-tunes' ); |
| 1743 | } |
| 1744 | else { |
| 1745 | $res = $this->execute( 'GET', '/fine_tuning/jobs' ); |
| 1746 | } |
| 1747 | $finetunes = $res['data']; |
| 1748 | |
| 1749 | // Add suffix |
| 1750 | $finetunes = array_map( function ( $finetune ) { |
| 1751 | if ( isset( $finetune['user_provided_suffix'] ) ) { |
| 1752 | $finetune['suffix'] = $finetune['user_provided_suffix']; |
| 1753 | } |
| 1754 | else { |
| 1755 | $finetune['suffix'] = self::get_suffix_for_model( $finetune['fine_tuned_model'] ); |
| 1756 | } |
| 1757 | $finetune['createdOn'] = date( 'Y-m-d H:i:s', $finetune['created_at'] ) . ' UTC'; |
| 1758 | if ( isset( $finetune['estimated_finish'] ) ) { |
| 1759 | $finetune['estimatedOn'] = date( 'Y-m-d H:i:s', $finetune['estimated_finish'] ) . ' UTC'; |
| 1760 | } |
| 1761 | else { |
| 1762 | $finetune['estimatedOn'] = null; |
| 1763 | } |
| 1764 | //$finetune['updatedOn'] = date( 'Y-m-d H:i:s', $finetune['updated_at'] ); |
| 1765 | $finetune['base_model'] = $finetune['model']; |
| 1766 | $finetune['model'] = $finetune['fine_tuned_model']; |
| 1767 | unset( $finetune['object'] ); |
| 1768 | unset( $finetune['hyperparams'] ); |
| 1769 | unset( $finetune['result_files'] ); |
| 1770 | unset( $finetune['training_files'] ); |
| 1771 | unset( $finetune['validation_files'] ); |
| 1772 | unset( $finetune['created_at'] ); |
| 1773 | unset( $finetune['updated_at'] ); |
| 1774 | unset( $finetune['fine_tuned_model'] ); |
| 1775 | return $finetune; |
| 1776 | }, $finetunes ); |
| 1777 | |
| 1778 | usort( $finetunes, function ( $a, $b ) { |
| 1779 | return strtotime( $b['createdOn'] ) - strtotime( $a['createdOn'] ); |
| 1780 | } ); |
| 1781 | |
| 1782 | if ( $legacy ) { |
| 1783 | $this->core->update_ai_env( $this->envId, 'legacy_finetunes', $finetunes ); |
| 1784 | } |
| 1785 | else { |
| 1786 | $this->core->update_ai_env( $this->envId, 'finetunes', $finetunes ); |
| 1787 | } |
| 1788 | |
| 1789 | return $finetunes; |
| 1790 | } |
| 1791 | |
| 1792 | public function moderate( $input ) { |
| 1793 | $result = $this->execute( 'POST', '/moderations', [ |
| 1794 | 'input' => $input |
| 1795 | ] ); |
| 1796 | return $result; |
| 1797 | } |
| 1798 | |
| 1799 | public function upload_file( $filename, $data, $purpose = 'fine-tune' ) { |
| 1800 | global $wp_filter; |
| 1801 | |
| 1802 | // Build multipart form data |
| 1803 | $boundary = wp_generate_password( 24, false ); |
| 1804 | $body = ''; |
| 1805 | $body .= "--$boundary\r\n"; |
| 1806 | $body .= "Content-Disposition: form-data; name=\"purpose\"\r\n\r\n"; |
| 1807 | $body .= $purpose . "\r\n"; |
| 1808 | $body .= "--$boundary\r\n"; |
| 1809 | $body .= "Content-Disposition: form-data; name=\"file\"; filename=\"{$filename}\"\r\n"; |
| 1810 | $body .= "Content-Type: application/octet-stream\r\n\r\n"; |
| 1811 | $body .= $data . "\r\n"; |
| 1812 | $body .= "--$boundary--\r\n"; |
| 1813 | |
| 1814 | // Temporarily remove ALL http_api_curl hooks to prevent streaming hook interference |
| 1815 | // Save current hooks |
| 1816 | $saved_hooks = null; |
| 1817 | if ( isset( $wp_filter['http_api_curl'] ) ) { |
| 1818 | $saved_hooks = $wp_filter['http_api_curl']; |
| 1819 | unset( $wp_filter['http_api_curl'] ); |
| 1820 | } |
| 1821 | |
| 1822 | // Upload using WordPress HTTP API |
| 1823 | $url = 'https://api.openai.com/v1/files'; |
| 1824 | $response = wp_remote_post( $url, [ |
| 1825 | 'headers' => [ |
| 1826 | 'Authorization' => 'Bearer ' . $this->apiKey, |
| 1827 | 'Content-Type' => 'multipart/form-data; boundary=' . $boundary |
| 1828 | ], |
| 1829 | 'body' => $body, |
| 1830 | 'timeout' => 60 |
| 1831 | ] ); |
| 1832 | |
| 1833 | // Restore hooks |
| 1834 | if ( $saved_hooks !== null ) { |
| 1835 | $wp_filter['http_api_curl'] = $saved_hooks; |
| 1836 | } |
| 1837 | |
| 1838 | if ( is_wp_error( $response ) ) { |
| 1839 | throw new Exception( 'File upload failed: ' . $response->get_error_message() ); |
| 1840 | } |
| 1841 | |
| 1842 | $response_body = wp_remote_retrieve_body( $response ); |
| 1843 | return json_decode( $response_body, true ); |
| 1844 | } |
| 1845 | |
| 1846 | public function create_vector_store( $name = null, $expiry = null, $metadata = null ) { |
| 1847 | $body = [ |
| 1848 | 'name' => !empty( $name ) ? $name : 'default', |
| 1849 | // Ensure metadata is an object (stdClass) for JSON encoding, not an array |
| 1850 | 'metadata' => !empty( $metadata ) ? $metadata : new stdClass() |
| 1851 | ]; |
| 1852 | if ( $expiry !== 'never' ) { |
| 1853 | if ( is_string( $expiry ) ) { |
| 1854 | error_log( 'AI Engine: Expiry is a string, setting it to 7 days.' ); |
| 1855 | $expiry = 7; |
| 1856 | } |
| 1857 | $expiryInDays = $expiry ? max( 1, ceil( (int) $expiry / 86400 ) ) : 7; |
| 1858 | if ( $expiry && is_numeric( $expiry ) ) { |
| 1859 | $body['expires_after'] = [ |
| 1860 | 'anchor' => 'last_active_at', |
| 1861 | 'days' => $expiryInDays |
| 1862 | ]; |
| 1863 | } |
| 1864 | } |
| 1865 | $result = $this->execute( 'POST', '/vector_stores', $body, null, true, [ 'OpenAI-Beta' => 'assistants=v2' ] ); |
| 1866 | return $result['id']; |
| 1867 | } |
| 1868 | |
| 1869 | public function get_vector_store( $vectorStoreId ) { |
| 1870 | return $this->execute( 'GET', '/vector_stores/' . $vectorStoreId, null, null, true, [ 'OpenAI-Beta' => 'assistants=v2' ] ); |
| 1871 | } |
| 1872 | |
| 1873 | public function add_vector_store_file( $vectorStoreId, $fileId ) { |
| 1874 | $result = $this->execute( 'POST', '/vector_stores/' . $vectorStoreId . '/files', [ |
| 1875 | 'file_id' => $fileId |
| 1876 | ], null, true, [ 'OpenAI-Beta' => 'assistants=v2' ] ); |
| 1877 | return $result['id']; |
| 1878 | |
| 1879 | } |
| 1880 | |
| 1881 | public function delete_file( $fileId ) { |
| 1882 | return $this->execute( 'DELETE', '/files/' . $fileId ); |
| 1883 | } |
| 1884 | |
| 1885 | public function get_model( $modelId ) { |
| 1886 | return $this->execute( 'GET', '/models/' . $modelId ); |
| 1887 | } |
| 1888 | |
| 1889 | public function cancel_finetune( $fineTuneId ) { |
| 1890 | return $this->execute( 'POST', '/fine-tunes/' . $fineTuneId . '/cancel' ); |
| 1891 | } |
| 1892 | |
| 1893 | public function delete_finetune( $modelId ) { |
| 1894 | return $this->execute( 'DELETE', '/models/' . $modelId ); |
| 1895 | } |
| 1896 | |
| 1897 | public function download_file( $fileId, $newFile = null ) { |
| 1898 | $fileInfo = $this->execute( 'GET', '/files/' . $fileId, null, null, false ); |
| 1899 | $fileInfo = json_decode( (string) $fileInfo, true ); |
| 1900 | if ( empty( $fileInfo ) ) { |
| 1901 | throw new Exception( 'File (' . ( $fileId ?? 'N/A' ) . ') not found.' ); |
| 1902 | } |
| 1903 | $filename = $fileInfo['filename']; |
| 1904 | $extension = pathinfo( $filename, PATHINFO_EXTENSION ); |
| 1905 | if ( empty( $newFile ) ) { |
| 1906 | include_once( ABSPATH . 'wp-admin/includes/file.php' ); |
| 1907 | $tempFile = wp_tempnam( $filename ); |
| 1908 | if ( !$tempFile ) { |
| 1909 | $tempFile = tempnam( sys_get_temp_dir(), 'download_' ); |
| 1910 | } |
| 1911 | if ( pathinfo( $tempFile, PATHINFO_EXTENSION ) != $extension ) { |
| 1912 | $newFile = $tempFile . '.' . $extension; |
| 1913 | } |
| 1914 | else { |
| 1915 | $newFile = $tempFile; |
| 1916 | } |
| 1917 | } |
| 1918 | $data = $this->execute( 'GET', '/files/' . $fileId . '/content', null, null, false ); |
| 1919 | file_put_contents( $newFile, $data ); |
| 1920 | return $newFile; |
| 1921 | } |
| 1922 | |
| 1923 | public function run_finetune( $fileId, $model, $suffix, $hyperparams = [], $legacy = false ) { |
| 1924 | $n_epochs = isset( $hyperparams['nEpochs'] ) ? (int) $hyperparams['nEpochs'] : null; |
| 1925 | $batch_size = isset( $hyperparams['batchSize'] ) ? (int) $hyperparams['batchSize'] : null; |
| 1926 | $learning_rate_multiplier = isset( $hyperparams['learningRateMultiplier'] ) ? |
| 1927 | (float) $hyperparams['learningRateMultiplier'] : null; |
| 1928 | $prompt_loss_weight = isset( $hyperparams['promptLossWeight'] ) ? |
| 1929 | (float) $hyperparams['promptLossWeight'] : null; |
| 1930 | $arguments = [ |
| 1931 | 'training_file' => $fileId, |
| 1932 | 'model' => $model, |
| 1933 | 'suffix' => $suffix |
| 1934 | ]; |
| 1935 | if ( $legacy ) { |
| 1936 | $result = $this->execute( 'POST', '/fine-tunes', $arguments ); |
| 1937 | } |
| 1938 | else { |
| 1939 | if ( $n_epochs ) { |
| 1940 | $arguments['hyperparams'] = []; |
| 1941 | $arguments['hyperparams']['n_epochs'] = $n_epochs; |
| 1942 | } |
| 1943 | if ( $batch_size ) { |
| 1944 | if ( empty( $arguments['hyperparams'] ) ) { |
| 1945 | $arguments['hyperparams'] = []; |
| 1946 | } |
| 1947 | $arguments['hyperparams']['batch_size'] = $batch_size; |
| 1948 | } |
| 1949 | if ( $learning_rate_multiplier ) { |
| 1950 | if ( empty( $arguments['hyperparams'] ) ) { |
| 1951 | $arguments['hyperparams'] = []; |
| 1952 | } |
| 1953 | $arguments['hyperparams']['learning_rate_multiplier'] = $learning_rate_multiplier; |
| 1954 | } |
| 1955 | if ( $prompt_loss_weight ) { |
| 1956 | if ( empty( $arguments['hyperparams'] ) ) { |
| 1957 | $arguments['hyperparams'] = []; |
| 1958 | } |
| 1959 | $arguments['hyperparams']['prompt_loss_weight'] = $prompt_loss_weight; |
| 1960 | } |
| 1961 | if ( $model === 'turbo' ) { |
| 1962 | $arguments['model'] = 'gpt-3.5-turbo'; |
| 1963 | } |
| 1964 | $result = $this->execute( 'POST', '/fine_tuning/jobs', $arguments ); |
| 1965 | } |
| 1966 | return $result; |
| 1967 | } |
| 1968 | |
| 1969 | /** |
| 1970 | * Build the body of a form request. |
| 1971 | * If the field name is 'file', then the field value is the filename of the file to upload. |
| 1972 | * The file contents are taken from the 'data' field. |
| 1973 | * |
| 1974 | * @param array $fields |
| 1975 | * @param string $boundary |
| 1976 | * @return string |
| 1977 | */ |
| 1978 | public function build_form_body( $fields, $boundary ) { |
| 1979 | $body = ''; |
| 1980 | foreach ( $fields as $name => $value ) { |
| 1981 | if ( in_array( $name, [ 'data', 'mime', 'mask_data', 'mask_mime' ] ) ) { |
| 1982 | continue; |
| 1983 | } |
| 1984 | $body .= "--$boundary\r\n"; |
| 1985 | $body .= "Content-Disposition: form-data; name=\"$name\""; |
| 1986 | if ( $name === 'image' || $name === 'file' ) { |
| 1987 | $body .= "; filename=\"{$value}\"\r\n"; |
| 1988 | $mime = !empty( $fields['mime'] ) ? $fields['mime'] : 'application/octet-stream'; |
| 1989 | $body .= "Content-Type: {$mime}\r\n\r\n"; |
| 1990 | $body .= $fields['data'] . "\r\n"; |
| 1991 | } |
| 1992 | else if ( $name === 'mask' ) { |
| 1993 | $body .= "; filename=\"{$value}\"\r\n"; |
| 1994 | $mime = !empty( $fields['mask_mime'] ) ? $fields['mask_mime'] : 'application/octet-stream'; |
| 1995 | $body .= "Content-Type: {$mime}\r\n\r\n"; |
| 1996 | $body .= $fields['mask_data'] . "\r\n"; |
| 1997 | } |
| 1998 | else { |
| 1999 | // JSON encode arrays and objects, keep scalars as-is |
| 2000 | $encodedValue = is_array( $value ) || is_object( $value ) ? json_encode( $value ) : $value; |
| 2001 | $body .= "\r\n\r\n$encodedValue\r\n"; |
| 2002 | } |
| 2003 | } |
| 2004 | $body .= "--$boundary--\r\n"; |
| 2005 | return $body; |
| 2006 | } |
| 2007 | |
| 2008 | /** |
| 2009 | * Run a request to the OpenAI API. |
| 2010 | * Fore more information about the $formFields, refer to the build_form_body method. |
| 2011 | * |
| 2012 | * @param string $method POST, PUT, GET, DELETE... |
| 2013 | * @param string $url The API endpoint |
| 2014 | * @param array $query The query parameters (json) |
| 2015 | * @param array $formFields The form fields (multipart/form-data) |
| 2016 | * @param bool $json Whether to return the response as json or not |
| 2017 | * @return array |
| 2018 | */ |
| 2019 | public function execute( |
| 2020 | $method, |
| 2021 | $url, |
| 2022 | $query = null, |
| 2023 | $formFields = null, |
| 2024 | $json = true, |
| 2025 | $extraHeaders = null, |
| 2026 | $streamCallback = null |
| 2027 | ) { |
| 2028 | $isAzure = $this->envType === 'azure'; |
| 2029 | $isOpenAI = !$isAzure; |
| 2030 | |
| 2031 | // Prepare the headers |
| 2032 | $headers = "Content-Type: application/json\r\n"; |
| 2033 | if ( $isOpenAI ) { |
| 2034 | $headers .= 'Authorization: Bearer ' . $this->apiKey . "\r\n"; |
| 2035 | if ( $this->organizationId ) { |
| 2036 | $headers .= 'OpenAI-Organization: ' . $this->organizationId . "\r\n"; |
| 2037 | } |
| 2038 | } |
| 2039 | else if ( $isAzure ) { |
| 2040 | $headers .= 'api-key: ' . $this->apiKey . "\r\n"; |
| 2041 | } |
| 2042 | |
| 2043 | // Prepare the body with json_encode, if it's not a string or null, otherwise we keep it as is. |
| 2044 | if ( !empty( $query ) && !is_string( $query ) ) { |
| 2045 | $body = $this->safe_json_encode( $query, 'query body' ); |
| 2046 | } |
| 2047 | else { |
| 2048 | $body = $query; |
| 2049 | } |
| 2050 | |
| 2051 | // If we have form fields, we need to change the headers and the body. |
| 2052 | if ( !empty( $formFields ) ) { |
| 2053 | $boundary = wp_generate_password( 24, false ); |
| 2054 | $headers = [ |
| 2055 | 'Content-Type' => 'multipart/form-data; boundary=' . $boundary |
| 2056 | ]; |
| 2057 | if ( $isOpenAI ) { |
| 2058 | $headers['Authorization'] = 'Bearer ' . $this->apiKey; |
| 2059 | if ( $this->organizationId ) { |
| 2060 | $headers['OpenAI-Organization'] = $this->organizationId; |
| 2061 | } |
| 2062 | } |
| 2063 | else if ( $isAzure ) { |
| 2064 | $headers['api-key'] = $this->apiKey; |
| 2065 | } |
| 2066 | $body = $this->build_form_body( $formFields, $boundary ); |
| 2067 | } |
| 2068 | |
| 2069 | // Maybe we should have headers always as an array... not sure why we have it as a string. |
| 2070 | if ( !empty( $extraHeaders ) ) { |
| 2071 | foreach ( $extraHeaders as $key => $value ) { |
| 2072 | if ( is_array( $headers ) ) { |
| 2073 | $headers[$key] = $value; |
| 2074 | } |
| 2075 | else { |
| 2076 | $headers .= "$key: $value\r\n"; |
| 2077 | } |
| 2078 | } |
| 2079 | } |
| 2080 | |
| 2081 | // Create the URL |
| 2082 | if ( $isOpenAI ) { |
| 2083 | $url = 'https://api.openai.com/v1' . $url; |
| 2084 | } |
| 2085 | else if ( $isAzure ) { |
| 2086 | $url = trailingslashit( $this->env['endpoint'] ) . 'openai' . $url; |
| 2087 | $hasQuery = strpos( $url, '?' ) !== false; |
| 2088 | $url = $url . ( $hasQuery ? '&' : '?' ) . $this->azureApiVersion; |
| 2089 | } |
| 2090 | |
| 2091 | // If it's a GET, body should be null, and we should append the query to the URL. |
| 2092 | if ( $method === 'GET' ) { |
| 2093 | if ( !empty( $query ) ) { |
| 2094 | $hasQuery = strpos( $url, '?' ) !== false; |
| 2095 | $url = $url . ( $hasQuery ? '&' : '?' ) . http_build_query( $query ); |
| 2096 | } |
| 2097 | $body = null; |
| 2098 | } |
| 2099 | |
| 2100 | $options = [ |
| 2101 | 'headers' => $headers, |
| 2102 | 'method' => $method, |
| 2103 | 'timeout' => MWAI_TIMEOUT, |
| 2104 | 'body' => $body, |
| 2105 | 'sslverify' => MWAI_SSL_VERIFY |
| 2106 | ]; |
| 2107 | |
| 2108 | // Check if queries debug is enabled |
| 2109 | $queries_debug = $this->core->get_option( 'queries_debug_mode' ); |
| 2110 | |
| 2111 | // Log the request if queries debug is enabled |
| 2112 | if ( $queries_debug ) { |
| 2113 | error_log( '[AI Engine Queries] HTTP Request to: ' . $url ); |
| 2114 | |
| 2115 | if ( !empty( $body ) ) { |
| 2116 | error_log( '[AI Engine Queries] Request Body:' ); |
| 2117 | error_log( $body ); |
| 2118 | } |
| 2119 | |
| 2120 | if ( !is_null( $streamCallback ) ) { |
| 2121 | error_log( '[AI Engine Queries] (Streaming mode - response will be streamed)' ); |
| 2122 | } |
| 2123 | } |
| 2124 | |
| 2125 | try { |
| 2126 | if ( !is_null( $streamCallback ) ) { |
| 2127 | $options['stream'] = true; |
| 2128 | $options['filename'] = tempnam( sys_get_temp_dir(), 'mwai-stream-' ); |
| 2129 | // The stream handler calls the streamCallback every time there is content |
| 2130 | // TODO: For assistants, we should probably have a different stream handler to |
| 2131 | // handle the assistant's specific reply and perform the necessary actions. |
| 2132 | $this->streamCallback = $streamCallback; |
| 2133 | add_action( 'http_api_curl', [ $this, 'stream_handler' ], 10, 3 ); |
| 2134 | } |
| 2135 | $res = wp_remote_request( $url, $options ); |
| 2136 | if ( is_wp_error( $res ) ) { |
| 2137 | throw new Exception( $res->get_error_message() ); |
| 2138 | } |
| 2139 | |
| 2140 | $res = wp_remote_retrieve_body( $res ); |
| 2141 | |
| 2142 | // Handle empty responses for container LIST API only (not for file content downloads) |
| 2143 | if ( strpos( $url, '/containers/' ) !== false && |
| 2144 | strpos( $url, '/files' ) !== false && |
| 2145 | strpos( $url, '/content' ) === false && // Don't apply this to content downloads |
| 2146 | empty( $res ) ) { |
| 2147 | // Return empty array for empty container files LIST response |
| 2148 | $data = $json ? [] : ''; |
| 2149 | error_log( '[AI Engine] Container LIST API returned empty response, treating as empty array' ); |
| 2150 | } |
| 2151 | else { |
| 2152 | $data = $json ? json_decode( $res, true ) : $res; |
| 2153 | } |
| 2154 | |
| 2155 | // Debug logging for decoded data (skip for content downloads) |
| 2156 | if ( strpos( $url, '/containers/' ) !== false && strpos( $url, '/files' ) !== false && strpos( $url, '/content' ) === false ) { |
| 2157 | error_log( '[AI Engine] After json_decode:' ); |
| 2158 | error_log( '[AI Engine] - Data type: ' . gettype( $data ) ); |
| 2159 | error_log( '[AI Engine] - Data is null: ' . ( $data === null ? 'YES' : 'NO' ) ); |
| 2160 | if ( $data !== null && is_array( $data ) ) { |
| 2161 | error_log( '[AI Engine] - Data keys: ' . implode( ', ', array_keys( $data ) ) ); |
| 2162 | error_log( '[AI Engine] - Data count: ' . count( $data ) ); |
| 2163 | } |
| 2164 | if ( $json && $data === null && !empty( $res ) ) { |
| 2165 | error_log( '[AI Engine] - JSON decode error: ' . json_last_error_msg() ); |
| 2166 | } |
| 2167 | } |
| 2168 | |
| 2169 | $this->handle_response_errors( $data ); |
| 2170 | |
| 2171 | // Log the response if queries debug is enabled |
| 2172 | if ( $queries_debug && is_null( $streamCallback ) ) { |
| 2173 | error_log( '[AI Engine Queries] Response Body:' ); |
| 2174 | error_log( $res ); |
| 2175 | } |
| 2176 | return $data; |
| 2177 | } |
| 2178 | catch ( Exception $e ) { |
| 2179 | $service = $this->get_service_name(); |
| 2180 | Meow_MWAI_Logging::error( "$service: " . $e->getMessage() ); |
| 2181 | throw new Exception( "$service: " . $e->getMessage() ); |
| 2182 | } |
| 2183 | finally { |
| 2184 | if ( !is_null( $streamCallback ) ) { |
| 2185 | remove_action( 'http_api_curl', [ $this, 'stream_handler' ] ); |
| 2186 | } |
| 2187 | if ( !empty( $options['stream'] ) && file_exists( $options['filename'] ) ) { |
| 2188 | unlink( $options['filename'] ); |
| 2189 | } |
| 2190 | } |
| 2191 | } |
| 2192 | |
| 2193 | public function get_models() { |
| 2194 | $models = apply_filters( 'mwai_openai_models', MWAI_OPENAI_MODELS ); |
| 2195 | // TODO: Drop the fine-tune injection loop below after 2027-02 (OpenAI ends fine-tune job creation on 2027-01-06). |
| 2196 | $finetunes = !empty( $this->env['finetunes'] ) ? $this->env['finetunes'] : []; |
| 2197 | foreach ( $finetunes as $finetune ) { |
| 2198 | if ( empty( $finetune['status'] ) || $finetune['status'] !== 'succeeded' ) { |
| 2199 | continue; |
| 2200 | } |
| 2201 | if ( empty( $finetune['base_model'] ) || empty( $finetune['model'] ) || empty( $finetune['suffix'] ) ) { |
| 2202 | continue; |
| 2203 | } |
| 2204 | $baseModel = self::get_model_without_release_date( $finetune['base_model'] ); |
| 2205 | if ( !empty( $baseModel ) ) { |
| 2206 | $model = null; |
| 2207 | foreach ( $models as $currentModel ) { |
| 2208 | if ( $currentModel['model'] === $baseModel ) { |
| 2209 | $model = $currentModel; |
| 2210 | break; |
| 2211 | } |
| 2212 | } |
| 2213 | if ( !empty( $model ) ) { |
| 2214 | $model['model'] = $finetune['model']; |
| 2215 | $model['name'] = $finetune['suffix']; |
| 2216 | $models[] = $model; |
| 2217 | } |
| 2218 | } |
| 2219 | } |
| 2220 | return $models; |
| 2221 | } |
| 2222 | |
| 2223 | public static function get_models_static() { |
| 2224 | return MWAI_OPENAI_MODELS; |
| 2225 | } |
| 2226 | |
| 2227 | /** |
| 2228 | * Recursively search for container_id in the response data |
| 2229 | */ |
| 2230 | protected function search_for_container_id_recursive( $data, $path = '' ) { |
| 2231 | if ( is_array( $data ) || is_object( $data ) ) { |
| 2232 | foreach ( $data as $key => $value ) { |
| 2233 | $currentPath = $path ? $path . '.' . $key : $key; |
| 2234 | |
| 2235 | // Check if this key is container_id |
| 2236 | if ( $key === 'container_id' ) { |
| 2237 | error_log( '[AI Engine Queries] *** FOUND container_id at path: ' . $currentPath . ' = ' . $value . ' ***' ); |
| 2238 | } |
| 2239 | |
| 2240 | // Recursively search in nested structures |
| 2241 | if ( is_array( $value ) || is_object( $value ) ) { |
| 2242 | $this->search_for_container_id_recursive( $value, $currentPath ); |
| 2243 | } |
| 2244 | } |
| 2245 | } |
| 2246 | } |
| 2247 | |
| 2248 | // TODO: After 2027-02 (OpenAI ends fine-tune job creation on 2027-01-06), drop the $finetune branch from calculate_price() and the finetune-detection block in get_price(). |
| 2249 | private function calculate_price( $modelFamily, $inUnits, $outUnits, $resolution = null, $finetune = false ) { |
| 2250 | $modelFamily = self::get_model_without_release_date( $modelFamily ); |
| 2251 | $models = $this->get_models(); |
| 2252 | foreach ( $models as $currentModel ) { |
| 2253 | if ( $currentModel['model'] === $modelFamily ) { |
| 2254 | if ( $finetune ) { |
| 2255 | if ( isset( $currentModel['finetune']['price'] ) ) { |
| 2256 | $currentModel['price'] = $currentModel['finetune']['price']; |
| 2257 | } |
| 2258 | else if ( isset( $currentModel['finetune']['in'] ) ) { |
| 2259 | $currentModel['price'] = [ |
| 2260 | 'in' => $currentModel['finetune']['in'], |
| 2261 | 'out' => $currentModel['finetune']['out'] |
| 2262 | ]; |
| 2263 | } |
| 2264 | } |
| 2265 | $inPrice = $currentModel['price']; |
| 2266 | $outPrice = $currentModel['price']; |
| 2267 | if ( is_array( $currentModel['price'] ) ) { |
| 2268 | $inPrice = $currentModel['price']['in']; |
| 2269 | $outPrice = $currentModel['price']['out']; |
| 2270 | } |
| 2271 | $inTotalPrice = $inPrice * $currentModel['unit'] * $inUnits; |
| 2272 | $outTotalPrice = $outPrice * $currentModel['unit'] * $outUnits; |
| 2273 | return $inTotalPrice + $outTotalPrice; |
| 2274 | } |
| 2275 | } |
| 2276 | Meow_MWAI_Logging::warn( "(OpenAI) Invalid model ($modelFamily)." ); |
| 2277 | return null; |
| 2278 | } |
| 2279 | |
| 2280 | public function get_price( Meow_MWAI_Query_Base $query, Meow_MWAI_Reply $reply ) { |
| 2281 | $model = $query->model; |
| 2282 | $units = 0; |
| 2283 | $finetune = false; |
| 2284 | if ( is_a( $query, 'Meow_MWAI_Query_Text' ) || is_a( $query, 'Meow_MWAI_Query_Assistant' ) ) { |
| 2285 | if ( preg_match( '/^([a-zA-Z]{0,32}):/', $model, $matches ) ) { |
| 2286 | $finetune = true; |
| 2287 | } |
| 2288 | $inUnits = $reply->get_in_tokens( $query ); |
| 2289 | $outUnits = $reply->get_out_tokens(); |
| 2290 | return $this->calculate_price( $model, $inUnits, $outUnits, null, $finetune ); |
| 2291 | } |
| 2292 | else if ( is_a( $query, 'Meow_MWAI_Query_Image' ) || is_a( $query, 'Meow_MWAI_Query_EditImage' ) ) { |
| 2293 | // gpt-image models are billed by input/output tokens like text models. |
| 2294 | // IMPORTANT: Images API uses 'input_tokens'/'output_tokens', not 'prompt_tokens'/'completion_tokens' |
| 2295 | $inUnits = $reply->usage['input_tokens'] ?? $reply->usage['prompt_tokens'] ?? 0; |
| 2296 | $outUnits = $reply->usage['output_tokens'] ?? $reply->usage['completion_tokens'] ?? 0; |
| 2297 | return $this->calculate_price( $model, $inUnits, $outUnits, null, $finetune ); |
| 2298 | } |
| 2299 | else if ( is_a( $query, 'Meow_MWAI_Query_Transcribe' ) ) { |
| 2300 | $model = 'whisper'; |
| 2301 | $units = $reply->get_units(); |
| 2302 | return $this->calculate_price( $model, 0, $units, null, $finetune ); |
| 2303 | } |
| 2304 | else if ( is_a( $query, 'Meow_MWAI_Query_Embed' ) ) { |
| 2305 | $units = $reply->get_total_tokens(); |
| 2306 | return $this->calculate_price( $model, 0, $units, null, $finetune ); |
| 2307 | } |
| 2308 | Meow_MWAI_Logging::warn( "(OpenAI) Cannot calculate price for $model." ); |
| 2309 | return null; |
| 2310 | } |
| 2311 | } |
| 2312 |