anthropic.php
2 months ago
chatml.php
1 month ago
core.php
1 month ago
custom.php
1 month ago
factory.php
1 month ago
google.php
3 months ago
mistral.php
5 months ago
open-router.php
5 months ago
openai.php
1 month ago
perplexity.php
6 months ago
replicate.php
5 months ago
xai.php
1 month ago
core.php
768 lines
| 1 | <?php |
| 2 | |
| 3 | class Meow_MWAI_Engines_Core { |
| 4 | protected $core = null; |
| 5 | public $env = null; |
| 6 | public $envId = null; |
| 7 | public $envType = null; |
| 8 | |
| 9 | // Streaming |
| 10 | protected $streamCallback = null; |
| 11 | protected $streamTemporaryBuffer = ''; |
| 12 | protected $streamBuffer = ''; |
| 13 | protected $streamHeaders = []; |
| 14 | protected $streamContent = ''; |
| 15 | |
| 16 | // Debug mode for stream events |
| 17 | protected $currentDebugMode = false; |
| 18 | protected $currentQuery = null; |
| 19 | protected $emittedFunctionResults = []; |
| 20 | |
| 21 | public function __construct( $core, $env ) { |
| 22 | $this->core = $core; |
| 23 | $this->env = $env; |
| 24 | $this->envId = isset( $env['id'] ) ? $env['id'] : null; |
| 25 | $this->envType = isset( $env['type'] ) ? $env['type'] : null; |
| 26 | } |
| 27 | |
| 28 | /** |
| 29 | * Reset all request-specific state variables. |
| 30 | * This should be called at the start of each new request to prevent |
| 31 | * state leakage between requests. |
| 32 | */ |
| 33 | protected function reset_request_state() { |
| 34 | // Reset streaming state |
| 35 | $this->streamCallback = null; |
| 36 | $this->streamTemporaryBuffer = ''; |
| 37 | $this->streamBuffer = ''; |
| 38 | $this->streamHeaders = []; |
| 39 | $this->streamContent = ''; |
| 40 | |
| 41 | // Reset debug/event state |
| 42 | $this->currentDebugMode = false; |
| 43 | $this->currentQuery = null; |
| 44 | $this->emittedFunctionResults = []; |
| 45 | } |
| 46 | |
| 47 | /** |
| 48 | * Safely encode data to JSON for API requests with UTF-8 error handling. |
| 49 | * |
| 50 | * WordPress content can contain malformed UTF-8 characters from various sources: |
| 51 | * - Copy-paste from Microsoft Word or other rich text editors |
| 52 | * - Database migrations from different character sets |
| 53 | * - Old content created before proper UTF-8 handling |
| 54 | * - User input with mixed encodings |
| 55 | * - WooCommerce product descriptions with special characters |
| 56 | * |
| 57 | * Without proper handling, json_encode() silently returns FALSE when encountering |
| 58 | * invalid UTF-8, causing API requests to fail cryptically with no error message. |
| 59 | * |
| 60 | * This method: |
| 61 | * 1. Uses JSON_INVALID_UTF8_SUBSTITUTE to replace invalid UTF-8 with � (U+FFFD) |
| 62 | * 2. Detects encoding failures and logs detailed debugging information |
| 63 | * 3. Throws descriptive exceptions instead of failing silently |
| 64 | * |
| 65 | * The replacement character (�) is handled correctly by all modern AI APIs and is |
| 66 | * far better than complete request failure. |
| 67 | * |
| 68 | * @param mixed $data The data to encode (array, object, string, etc.) |
| 69 | * @param string $context Optional context for error messages (e.g., 'request body', 'query') |
| 70 | * @return string The JSON-encoded string |
| 71 | * @throws Exception If JSON encoding fails even with UTF-8 substitution |
| 72 | */ |
| 73 | protected function safe_json_encode( $data, $context = 'data' ) { |
| 74 | // Use JSON_INVALID_UTF8_SUBSTITUTE to handle malformed UTF-8 gracefully |
| 75 | // This flag replaces invalid sequences with the Unicode replacement character (U+FFFD) |
| 76 | $json = json_encode( $data, JSON_INVALID_UTF8_SUBSTITUTE ); |
| 77 | |
| 78 | if ( $json === false ) { |
| 79 | // Encoding failed even with UTF-8 substitution - log detailed debug info |
| 80 | $error_msg = json_last_error_msg(); |
| 81 | error_log( "[AI Engine] JSON encode failed for {$context}: {$error_msg}" ); |
| 82 | error_log( '[AI Engine] Data type: ' . gettype( $data ) ); |
| 83 | |
| 84 | if ( is_array( $data ) || is_object( $data ) ) { |
| 85 | // Log structure (limited to prevent massive logs) |
| 86 | $structure = print_r( $data, true ); |
| 87 | $preview = substr( $structure, 0, 1000 ); |
| 88 | if ( strlen( $structure ) > 1000 ) { |
| 89 | $preview .= "\n... (truncated, total length: " . strlen( $structure ) . ' chars)'; |
| 90 | } |
| 91 | error_log( "[AI Engine] Data structure: {$preview}" ); |
| 92 | } |
| 93 | |
| 94 | throw new Exception( "Failed to encode {$context} as JSON: {$error_msg}" ); |
| 95 | } |
| 96 | |
| 97 | return $json; |
| 98 | } |
| 99 | |
| 100 | /** |
| 101 | * Prepare query before execution. |
| 102 | * This method is called BEFORE any streaming hooks are set up. |
| 103 | * Engines should override this to perform preliminary tasks like: |
| 104 | * - Uploading files to provider APIs |
| 105 | * - Preprocessing data |
| 106 | * - Validating query parameters |
| 107 | * |
| 108 | * @param Meow_MWAI_Query_Base $query The query to prepare |
| 109 | */ |
| 110 | protected function prepare_query( $query ) { |
| 111 | // Base implementation does nothing |
| 112 | // Child engines can override to add provider-specific preparation |
| 113 | } |
| 114 | |
| 115 | public function run( $query, $streamCallback = null, $maxDepth = 5 ) { |
| 116 | |
| 117 | // Apply filter to allow overriding maxDepth (only on first call) |
| 118 | if ( !isset( $query->_maxDepthConfigured ) ) { |
| 119 | $maxDepth = apply_filters( 'mwai_function_call_max_depth', $maxDepth, $query ); |
| 120 | $query->_maxDepthConfigured = $maxDepth; |
| 121 | } |
| 122 | |
| 123 | // Check if queries debug is enabled |
| 124 | $queries_debug = $this->core->get_option( 'queries_debug_mode' ); |
| 125 | |
| 126 | // Log query start if debug is enabled |
| 127 | if ( $queries_debug ) { |
| 128 | // We'll let the individual engines log the actual HTTP requests/responses |
| 129 | // Just log a simple start marker here |
| 130 | error_log( '[AI Engine Queries] ========================================' ); |
| 131 | $query_type = get_class( $query ); |
| 132 | error_log( '[AI Engine Queries] Starting ' . $query_type . ' to ' . ( $query->model ?? 'unknown model' ) ); |
| 133 | } |
| 134 | |
| 135 | // Check if the query is allowed. |
| 136 | $limits = $this->core->get_option( 'limits' ); |
| 137 | $allowed = apply_filters( 'mwai_ai_allowed', true, $query, $limits ); |
| 138 | if ( $allowed !== true ) { |
| 139 | $message = is_string( $allowed ) ? $allowed : 'Unauthorized query.'; |
| 140 | throw new Exception( $message ); |
| 141 | } |
| 142 | |
| 143 | // Important as it makes sure everything is consolidated in the query and the engine. |
| 144 | $this->final_checks( $query ); |
| 145 | |
| 146 | // Run the query |
| 147 | $reply = null; |
| 148 | if ( $query instanceof Meow_MWAI_Query_Text || $query instanceof Meow_MWAI_Query_Feedback ) { |
| 149 | $reply = $this->run_completion_query( $query, $streamCallback ); |
| 150 | } |
| 151 | else if ( $query instanceof Meow_MWAI_Query_Assistant || $query instanceof Meow_MWAI_Query_AssistFeedback ) { |
| 152 | $reply = $this->run_assistant_query( $query, $streamCallback ); |
| 153 | if ( $reply === null ) { |
| 154 | throw new Exception( 'Assistants are not supported in this version of AI Engine.' ); |
| 155 | } |
| 156 | } |
| 157 | else if ( $query instanceof Meow_MWAI_Query_Embed ) { |
| 158 | $reply = $this->run_embedding_query( $query ); |
| 159 | } |
| 160 | else if ( $query instanceof Meow_MWAI_Query_EditImage ) { |
| 161 | $reply = $this->run_editimage_query( $query ); |
| 162 | } |
| 163 | else if ( $query instanceof Meow_MWAI_Query_Image ) { |
| 164 | $reply = $this->run_image_query( $query, $streamCallback ); |
| 165 | } |
| 166 | else if ( $query instanceof Meow_MWAI_Query_Transcribe ) { |
| 167 | $reply = $this->run_transcribe_query( $query ); |
| 168 | } |
| 169 | else { |
| 170 | throw new Exception( 'Unknown query type.' ); |
| 171 | } |
| 172 | |
| 173 | // Allow to modify the reply before it is sent. |
| 174 | $reply = apply_filters( 'mwai_ai_reply', $reply, $query ); |
| 175 | |
| 176 | // Log query completion if debug is enabled |
| 177 | if ( $queries_debug && empty( $reply->needFeedbacks ) ) { |
| 178 | // For embedding queries, just log the dimensions count |
| 179 | if ( $query instanceof Meow_MWAI_Query_Embed && !empty( $reply->result ) && is_array( $reply->result ) ) { |
| 180 | error_log( '[AI Engine Queries] Embedding completed with ' . count( $reply->result ) . ' dimensions' ); |
| 181 | } |
| 182 | else { |
| 183 | error_log( '[AI Engine Queries] Query completed' ); |
| 184 | } |
| 185 | error_log( '[AI Engine Queries] ========================================' ); |
| 186 | } |
| 187 | |
| 188 | // Function Call Handling - This is where the magic happens! |
| 189 | // When the AI model requests function calls, we execute them and send results back |
| 190 | if ( !empty( $reply->needFeedbacks ) ) { |
| 191 | |
| 192 | // Debug: Log how many needFeedbacks we have |
| 193 | if ( $queries_debug ) { |
| 194 | error_log( '[AI Engine Queries] Core: Processing ' . count( $reply->needFeedbacks ) . ' needFeedbacks' ); |
| 195 | foreach ( $reply->needFeedbacks as $idx => $feedback ) { |
| 196 | error_log( '[AI Engine Queries] Core: needFeedback[' . $idx . ']: name=' . $feedback['name'] . ', toolId=' . ( $feedback['toolId'] ?? 'none' ) ); |
| 197 | } |
| 198 | } |
| 199 | |
| 200 | // Prevent infinite loops - each function call reduces maxDepth by 1 |
| 201 | if ( $maxDepth <= 0 ) { |
| 202 | // Build call stack for better debugging |
| 203 | $callStack = []; |
| 204 | foreach ( $reply->needFeedbacks as $feedback ) { |
| 205 | $callStack[] = $feedback['name'] ?? 'unknown'; |
| 206 | } |
| 207 | |
| 208 | throw Meow_MWAI_FunctionCallException::loop_detected( |
| 209 | $query->_maxDepthConfigured ?? 5, // Use configured max depth |
| 210 | $callStack |
| 211 | ); |
| 212 | } |
| 213 | |
| 214 | // Create a feedback query if we're not already in one |
| 215 | // This wraps the original query with function execution results |
| 216 | if ( !( $query instanceof Meow_MWAI_Query_AssistFeedback ) && !( $query instanceof Meow_MWAI_Query_Feedback ) ) { |
| 217 | $queryClass = $query instanceof Meow_MWAI_Query_Assistant ? |
| 218 | Meow_MWAI_Query_AssistFeedback::class : Meow_MWAI_Query_Feedback::class; |
| 219 | // Note: $reply->query contains the original query that produced this reply |
| 220 | $query = new $queryClass( $reply, $reply->query ); |
| 221 | } |
| 222 | else { |
| 223 | // Already inside a feedback chain (recursion depth >= 2). Each new reply produces a |
| 224 | // fresh response_id; the next request must reference THAT, not the original turn's id. |
| 225 | // Otherwise OpenAI's Responses API rejects with "No tool output found" because, from |
| 226 | // the original response's perspective, the new function_call_output answers a call |
| 227 | // that doesn't exist in its chain. Refresh previousResponseId from the latest reply |
| 228 | // so each recursion targets the immediately preceding turn. |
| 229 | if ( !empty( $reply->id ) ) { |
| 230 | $query->previousResponseId = $reply->id; |
| 231 | } |
| 232 | // Reset the per-turn feedback blocks; they'll be repopulated below from the new reply. |
| 233 | if ( method_exists( $query, 'clear_feedback_blocks' ) ) { |
| 234 | $query->clear_feedback_blocks(); |
| 235 | } |
| 236 | } |
| 237 | |
| 238 | // Determine whether every function call in this turn is "static" (no AI |
| 239 | // feedback wanted). If so, we still execute the functions below but |
| 240 | // skip the recursive AI round-trip — the AI's reply stands as-is. |
| 241 | // Mixed turns degrade to dynamic for all calls, since OpenAI/Anthropic |
| 242 | // tool-call protocols require a result for every requested call in the |
| 243 | // same message. |
| 244 | $all_static = true; |
| 245 | foreach ( $reply->needFeedbacks as $needFeedback ) { |
| 246 | $fn = $needFeedback['function'] ?? null; |
| 247 | if ( !$fn || !isset( $fn->behavior ) || $fn->behavior !== 'static' ) { |
| 248 | $all_static = false; |
| 249 | break; |
| 250 | } |
| 251 | } |
| 252 | |
| 253 | // OpenAI's Responses API is stateful: when the next turn references previous_response_id, |
| 254 | // every function_call from the prior response must have a matching function_call_output, |
| 255 | // or the API rejects with "No tool output found for function call call_xxx". Skipping the |
| 256 | // round-trip on all-static turns satisfies the local user-facing reply but leaves the |
| 257 | // server-side conversation in a half-answered state, breaking any follow-up. So we |
| 258 | // disable the static-skip optimization for Responses API replies — the round-trip stays |
| 259 | // mandatory there. Chat Completions and Anthropic are stateless on tool calls and remain |
| 260 | // safe to skip. |
| 261 | if ( $all_static && ! empty( $reply->id ) && $this->core->responseIdManager->is_responses_api_id( $reply->id ) ) { |
| 262 | $all_static = false; |
| 263 | } |
| 264 | |
| 265 | // Validate that all function calls have proper function definitions |
| 266 | foreach ( $reply->needFeedbacks as $needFeedback ) { |
| 267 | if ( !isset( $needFeedback['function'] ) ) { |
| 268 | $functionName = $needFeedback['name'] ?? 'unknown'; |
| 269 | $availableFunctions = array_map( function ( $f ) { return $f->name; }, $query->functions ); |
| 270 | |
| 271 | throw new Exception( sprintf( |
| 272 | "Function '%s' not found in query functions. Available functions: %s", |
| 273 | $functionName, |
| 274 | implode( ', ', $availableFunctions ) |
| 275 | ) ); |
| 276 | } |
| 277 | } |
| 278 | |
| 279 | // Group function calls by their source message to maintain proper context |
| 280 | // This ensures related function calls are processed together |
| 281 | $feedback_blocks = []; |
| 282 | |
| 283 | // Special handling for Responses API - group all function calls together |
| 284 | // Check if we're using Responses API by looking at the query's previous response ID or reply ID |
| 285 | $isResponsesApi = false; |
| 286 | |
| 287 | // Method 1: Check if query has a previous response ID from Responses API |
| 288 | if ( !empty( $query->previousResponseId ) && $this->core->responseIdManager->is_responses_api_id( $query->previousResponseId ) ) { |
| 289 | $isResponsesApi = true; |
| 290 | } |
| 291 | |
| 292 | // Method 2: Check if the reply has a Responses API response ID |
| 293 | if ( !$isResponsesApi && !empty( $reply->id ) && $this->core->responseIdManager->is_responses_api_id( $reply->id ) ) { |
| 294 | $isResponsesApi = true; |
| 295 | } |
| 296 | |
| 297 | // Method 3: Check the model tags for 'responses' tag |
| 298 | if ( !$isResponsesApi && !empty( $query->model ) ) { |
| 299 | $modelInfo = $this->retrieve_model_info( $query->model ); |
| 300 | if ( $modelInfo && !empty( $modelInfo['tags'] ) && in_array( 'responses', $modelInfo['tags'] ) ) { |
| 301 | $isResponsesApi = true; |
| 302 | } |
| 303 | } |
| 304 | |
| 305 | // Method 4: For OpenAI engine, check if we're already using Responses API |
| 306 | // This is important for models that use Responses API but don't have the tag |
| 307 | if ( !$isResponsesApi && method_exists( $this, 'should_use_responses_api' ) ) { |
| 308 | // This is an OpenAI engine, check if it should use Responses API |
| 309 | $isResponsesApi = $this->should_use_responses_api( $query->model ); |
| 310 | } |
| 311 | |
| 312 | // Debug: Log grouping information |
| 313 | if ( $queries_debug ) { |
| 314 | error_log( '[AI Engine Queries] Grouping ' . count( $reply->needFeedbacks ) . ' function calls' ); |
| 315 | error_log( '[AI Engine Queries] Is Responses API: ' . ( $isResponsesApi ? 'yes' : 'no' ) ); |
| 316 | error_log( '[AI Engine Queries] Detection methods:' ); |
| 317 | error_log( '[AI Engine Queries] - previousResponseId: ' . ( $query->previousResponseId ?? 'null' ) ); |
| 318 | error_log( '[AI Engine Queries] - reply->id: ' . ( $reply->id ?? 'null' ) ); |
| 319 | error_log( '[AI Engine Queries] - model: ' . ( $query->model ?? 'null' ) ); |
| 320 | error_log( '[AI Engine Queries] - method_exists should_use_responses_api: ' . ( method_exists( $this, 'should_use_responses_api' ) ? 'yes' : 'no' ) ); |
| 321 | error_log( '[AI Engine Queries] - engine class: ' . get_class( $this ) ); |
| 322 | if ( $isResponsesApi ) { |
| 323 | error_log( '[AI Engine Queries] All function calls will be grouped together for Responses API' ); |
| 324 | } |
| 325 | } |
| 326 | |
| 327 | foreach ( $reply->needFeedbacks as $idx => $needFeedback ) { |
| 328 | // For Responses API, use a single key to group all function calls together |
| 329 | $rawMessageKey = md5( serialize( $needFeedback['rawMessage'] ) ); |
| 330 | |
| 331 | if ( $queries_debug ) { |
| 332 | error_log( '[AI Engine Queries] Function call ' . $idx . ': ' . $needFeedback['name'] . ' (key: ' . substr( $rawMessageKey, 0, 8 ) . ')' ); |
| 333 | } |
| 334 | |
| 335 | // Initialize the feedback block for this rawMessage if it hasn't been initialized yet |
| 336 | if ( !isset( $feedback_blocks[$rawMessageKey] ) ) { |
| 337 | $feedback_blocks[$rawMessageKey] = [ |
| 338 | 'rawMessage' => $needFeedback['rawMessage'], |
| 339 | 'feedbacks' => [] |
| 340 | ]; |
| 341 | } |
| 342 | |
| 343 | // Allow modifying function call arguments before execution |
| 344 | $needFeedback['arguments'] = apply_filters( |
| 345 | 'mwai_function_call_params', |
| 346 | $needFeedback['arguments'], |
| 347 | $needFeedback, |
| 348 | $reply |
| 349 | ); |
| 350 | |
| 351 | // Get the value related to this feedback (usually, a function call) |
| 352 | $value = apply_filters( 'mwai_ai_feedback', null, $needFeedback, $reply ); |
| 353 | |
| 354 | if ( $value === null ) { |
| 355 | // Check if the function handler exists |
| 356 | if ( !has_filter( 'mwai_ai_feedback' ) ) { |
| 357 | Meow_MWAI_Logging::error( |
| 358 | Meow_MWAI_FunctionCallException::missing_function_handler( |
| 359 | $needFeedback['name'] |
| 360 | )->getMessage() |
| 361 | ); |
| 362 | } |
| 363 | else { |
| 364 | Meow_MWAI_Logging::warn( "The returned value for '{$needFeedback['name']}' was null." ); |
| 365 | } |
| 366 | $value = '[NO VALUE RETURNED - DO NOT SHOW THIS]'; |
| 367 | } |
| 368 | |
| 369 | // Emit "Got result" event and log for debugging |
| 370 | if ( $this->currentDebugMode ) { |
| 371 | // Format the result preview |
| 372 | $resultPreview = is_array( $value ) ? json_encode( $value ) : (string) $value; |
| 373 | if ( strlen( $resultPreview ) > 100 ) { |
| 374 | $resultPreview = substr( $resultPreview, 0, 100 ) . '...'; |
| 375 | } |
| 376 | |
| 377 | // Log the function result for debugging |
| 378 | Meow_MWAI_Logging::log( "Function '{$needFeedback['name']}' returned: " . $resultPreview ); |
| 379 | |
| 380 | // Emit function result event if we have a callback |
| 381 | if ( !empty( $streamCallback ) ) { |
| 382 | // Load event helper if not already loaded |
| 383 | if ( !class_exists( 'Meow_MWAI_Event' ) ) { |
| 384 | require_once MWAI_PATH . '/classes/event.php'; |
| 385 | } |
| 386 | |
| 387 | $functionName = $needFeedback['name']; |
| 388 | |
| 389 | $event = Meow_MWAI_Event::function_result( $functionName ) |
| 390 | ->set_metadata( 'result', $resultPreview ) |
| 391 | ->set_metadata( 'tool_id', $needFeedback['toolId'] ?? null ); |
| 392 | call_user_func( $streamCallback, $event ); |
| 393 | } |
| 394 | } |
| 395 | |
| 396 | // Add the feedback information to the appropriate feedback block |
| 397 | $feedback_blocks[$rawMessageKey]['feedbacks'][] = [ |
| 398 | 'request' => $needFeedback, |
| 399 | 'reply' => [ 'value' => $value ] |
| 400 | ]; |
| 401 | } |
| 402 | |
| 403 | $query->clear_feedback_blocks(); |
| 404 | foreach ( $feedback_blocks as $feedback_block ) { |
| 405 | $query->add_feedback_block( $feedback_block ); |
| 406 | } |
| 407 | |
| 408 | // Log feedback query if debug is enabled |
| 409 | if ( $queries_debug ) { |
| 410 | error_log( '[AI Engine Queries] Created ' . count( $feedback_blocks ) . ' feedback blocks from ' . count( $reply->needFeedbacks ) . ' function calls' ); |
| 411 | foreach ( $feedback_blocks as $key => $block ) { |
| 412 | error_log( '[AI Engine Queries] Block ' . substr( $key, 0, 8 ) . ' has ' . count( $block['feedbacks'] ) . ' feedbacks' ); |
| 413 | } |
| 414 | } |
| 415 | |
| 416 | // Static-only turn: functions executed above for their side effects; |
| 417 | // skip the recursive AI feedback so the original reply stands. |
| 418 | if ( $all_static ) { |
| 419 | if ( $queries_debug ) { |
| 420 | error_log( '[AI Engine Queries] All function calls are static. Skipping AI feedback round-trip.' ); |
| 421 | } |
| 422 | return $reply; |
| 423 | } |
| 424 | |
| 425 | // Run the feedback query |
| 426 | $reply = $this->run( $query, $streamCallback, $maxDepth - 1 ); |
| 427 | } |
| 428 | |
| 429 | return $reply; |
| 430 | } |
| 431 | |
| 432 | public function retrieve_model_info( $model ) { |
| 433 | $models = $this->get_models(); |
| 434 | foreach ( $models as $currentModel ) { |
| 435 | if ( $currentModel['model'] === $model ) { |
| 436 | return $currentModel; |
| 437 | } |
| 438 | } |
| 439 | return false; |
| 440 | } |
| 441 | |
| 442 | public function final_checks( Meow_MWAI_Query_Base $query ) { |
| 443 | $query->final_checks(); |
| 444 | //$found = false; |
| 445 | |
| 446 | // Check if the model is available, except if it's an assistant |
| 447 | if ( !( $query instanceof Meow_MWAI_Query_Assistant ) ) { |
| 448 | // TODO: Remove the ft: bypass after 2027-02 (OpenAI ends fine-tune job creation on 2027-01-06). Until then we skip model validation for fine-tunes so user-created models still resolve. |
| 449 | if ( substr( $query->model, 0, 3 ) === 'ft:' ) { |
| 450 | return; |
| 451 | } |
| 452 | $model_info = $this->retrieve_model_info( $query->model ); |
| 453 | if ( $model_info === false ) { |
| 454 | // Provide a more helpful error message for embeddings queries without a configured environment |
| 455 | if ( $query instanceof Meow_MWAI_Query_Embed && empty( $query->envId ) ) { |
| 456 | throw new Exception( __( 'No embeddings environment is configured. Please go to Settings > Default Environments for AI > Embeddings and select an environment.', 'ai-engine' ) ); |
| 457 | } |
| 458 | throw new Exception( sprintf( __( "The model '%s' is not available.", 'ai-engine' ), $query->model ) ); |
| 459 | } |
| 460 | if ( isset( $model_info['mode'] ) ) { |
| 461 | $query->mode = $model_info['mode']; |
| 462 | } |
| 463 | } |
| 464 | } |
| 465 | |
| 466 | // Streamline the messages: |
| 467 | // - Concatenate consecutive model messages into a single message for the model role |
| 468 | // - Make sure the first message is a user message |
| 469 | // - Make sure the last message is a user message |
| 470 | protected function streamline_messages( $messages, $systemRole = 'assistant', $messageType = 'content' ) { |
| 471 | $processedMessages = []; |
| 472 | $lastRole = ''; |
| 473 | $concatenatedText = ''; |
| 474 | |
| 475 | // Determine the way to access message content based on messageType |
| 476 | $getContent = function ( $message ) use ( $messageType ) { |
| 477 | if ( $messageType == 'parts' ) { |
| 478 | return $message['parts'][0]['text']; |
| 479 | } |
| 480 | else { // Default to 'content' |
| 481 | return $message['content']; |
| 482 | } |
| 483 | }; |
| 484 | |
| 485 | // Set content to a message depending on the messageType |
| 486 | $setContent = function ( &$message, $content ) use ( $messageType ) { |
| 487 | if ( $messageType == 'parts' ) { |
| 488 | $message['parts'] = [['text' => $content]]; |
| 489 | } |
| 490 | else { // Default to 'content' |
| 491 | $message['content'] = $content; |
| 492 | } |
| 493 | }; |
| 494 | |
| 495 | // Concatenate consecutive model messages into a single message for the model role |
| 496 | foreach ( $messages as $message ) { |
| 497 | if ( $message['role'] == $systemRole ) { |
| 498 | if ( $lastRole == $systemRole ) { |
| 499 | $concatenatedText .= "\n" . $getContent( $message ); |
| 500 | } |
| 501 | else { |
| 502 | if ( $concatenatedText !== '' ) { |
| 503 | $newMessage = [ 'role' => $systemRole ]; |
| 504 | $setContent( $newMessage, $concatenatedText ); |
| 505 | $processedMessages[] = $newMessage; |
| 506 | } |
| 507 | $concatenatedText = $getContent( $message ); |
| 508 | } |
| 509 | } |
| 510 | else { |
| 511 | if ( $lastRole == $systemRole ) { |
| 512 | $newMessage = [ 'role' => $systemRole ]; |
| 513 | $setContent( $newMessage, $concatenatedText ); |
| 514 | $processedMessages[] = $newMessage; |
| 515 | $concatenatedText = ''; |
| 516 | } |
| 517 | $processedMessages[] = $message; |
| 518 | } |
| 519 | $lastRole = $message['role']; |
| 520 | } |
| 521 | if ( $lastRole == $systemRole && $concatenatedText !== '' ) { |
| 522 | $newMessage = [ 'role' => $systemRole ]; |
| 523 | $setContent( $newMessage, $concatenatedText ); |
| 524 | $processedMessages[] = $newMessage; |
| 525 | } |
| 526 | |
| 527 | // Make sure the last message is a user message, if not, throw an exception |
| 528 | if ( end( $processedMessages )['role'] !== 'user' ) { |
| 529 | throw new Exception( __( 'The last message must be a user message.', 'ai-engine' ) ); |
| 530 | } |
| 531 | |
| 532 | // Make sure the first message is a user message, if not, add an empty user message |
| 533 | if ( $processedMessages[0]['role'] !== 'user' ) { |
| 534 | $newMessage = [ 'role' => 'user' ]; |
| 535 | $setContent( $newMessage, '' ); |
| 536 | array_unshift( $processedMessages, $newMessage ); |
| 537 | } |
| 538 | |
| 539 | return $processedMessages; |
| 540 | } |
| 541 | |
| 542 | // Check for a JSON-formatted error in the data, and throw an exception if it's the case. |
| 543 | public function stream_error_check( $data ) { |
| 544 | if ( strpos( $data, 'error' ) === false ) { |
| 545 | return; |
| 546 | } |
| 547 | |
| 548 | $data = trim( $data ); |
| 549 | $jsonPart = $data; |
| 550 | if ( strpos( $jsonPart, 'data:' ) === 0 ) { |
| 551 | $jsonPart = trim( substr( $jsonPart, strlen( 'data:' ) ) ); |
| 552 | } |
| 553 | |
| 554 | $json = json_decode( $jsonPart, true ); |
| 555 | if ( json_last_error() !== JSON_ERROR_NONE ) { |
| 556 | return; // not valid JSON, nothing to do |
| 557 | } |
| 558 | // 1. OpenAI style: { error: {...} } |
| 559 | $error = null; |
| 560 | if ( isset( $json['error'] ) ) { |
| 561 | $error = $json['error']; |
| 562 | } |
| 563 | // 2. Google style: [ { error: {...} } ] |
| 564 | else if ( is_array( $json ) ) { |
| 565 | foreach ( $json as $item ) { |
| 566 | if ( isset( $item['error'] ) ) { |
| 567 | $error = $item['error']; |
| 568 | break; |
| 569 | } |
| 570 | } |
| 571 | } |
| 572 | // 3. Some APIs return { type: "error", message: ... } |
| 573 | else if ( isset( $json['type'] ) && $json['type'] === 'error' ) { |
| 574 | $error = $json; |
| 575 | } |
| 576 | |
| 577 | if ( is_null( $error ) ) { |
| 578 | return; |
| 579 | } |
| 580 | |
| 581 | $message = $error['message'] ?? ( is_string( $error ) ? $error : null ); |
| 582 | $code = $error['code'] ?? null; |
| 583 | // Google uses "status" instead of "type" – accept both |
| 584 | $type = $error['type'] ?? ( $error['status'] ?? null ); |
| 585 | if ( is_null( $message ) ) { |
| 586 | throw new Exception( 'Unknown error (stream_error_check).' ); |
| 587 | } |
| 588 | |
| 589 | $errorMessage = "Error: $message"; |
| 590 | if ( !is_null( $code ) ) { |
| 591 | $errorMessage .= " ($code)"; |
| 592 | } |
| 593 | if ( !is_null( $type ) ) { |
| 594 | $errorMessage .= " ($type)"; |
| 595 | } |
| 596 | |
| 597 | throw new Exception( $errorMessage ); |
| 598 | } |
| 599 | |
| 600 | protected function init_debug_mode( $query ) { |
| 601 | // Check if server debug mode or event logs are enabled in settings |
| 602 | $this->currentDebugMode = ( $this->core->get_option( 'module_devtools' ) && $this->core->get_option( 'server_debug_mode' ) ) || $this->core->get_option( 'event_logs' ); |
| 603 | $this->currentQuery = $query; |
| 604 | } |
| 605 | |
| 606 | public function stream_handler( $handle, $args, $url ) { |
| 607 | curl_setopt( $handle, CURLOPT_SSL_VERIFYPEER, false ); |
| 608 | curl_setopt( $handle, CURLOPT_SSL_VERIFYHOST, false ); |
| 609 | |
| 610 | curl_setopt( $handle, CURLOPT_WRITEFUNCTION, function ( $curl, $data ) use ( $url ) { |
| 611 | $length = strlen( $data ); |
| 612 | |
| 613 | // Log streaming data if queries debug is enabled |
| 614 | $queries_debug = $this->core->get_option( 'queries_debug_mode' ); |
| 615 | static $logged_url = false; |
| 616 | if ( $queries_debug && !$logged_url ) { |
| 617 | error_log( '[AI Engine Queries] Streaming from: ' . $url ); |
| 618 | $logged_url = true; |
| 619 | } |
| 620 | |
| 621 | // Bufferize the unfinished stream (if it's the case) |
| 622 | $this->streamTemporaryBuffer .= $data; |
| 623 | $this->streamBuffer .= $data; |
| 624 | |
| 625 | // Error Management |
| 626 | $this->stream_error_check( $this->streamBuffer ); |
| 627 | |
| 628 | $lines = explode( "\n", $this->streamTemporaryBuffer ); |
| 629 | if ( substr( $this->streamTemporaryBuffer, -1 ) !== "\n" ) { |
| 630 | $this->streamTemporaryBuffer = array_pop( $lines ); |
| 631 | } |
| 632 | else { |
| 633 | $this->streamTemporaryBuffer = ''; |
| 634 | } |
| 635 | |
| 636 | foreach ( $lines as $line ) { |
| 637 | if ( $line === '' ) { |
| 638 | continue; |
| 639 | } |
| 640 | if ( strpos( $line, 'data:' ) === 0 ) { |
| 641 | $line = trim( substr( $line, 5 ) ); |
| 642 | $json = json_decode( trim( $line ), true ); |
| 643 | |
| 644 | if ( json_last_error() === JSON_ERROR_NONE ) { |
| 645 | // Log individual streaming event if queries debug is enabled |
| 646 | static $event_count = 0; |
| 647 | if ( $queries_debug && $event_count < 10 ) { |
| 648 | // Log only the event type and key data, not the entire response |
| 649 | $event_log = [ |
| 650 | 'type' => $json['type'] ?? 'unknown' |
| 651 | ]; |
| 652 | |
| 653 | // Add specific details based on event type |
| 654 | if ( isset( $json['type'] ) ) { |
| 655 | if ( $json['type'] === 'response.output_item.added' && isset( $json['item'] ) ) { |
| 656 | $event_log['item_type'] = $json['item']['type'] ?? 'unknown'; |
| 657 | $event_log['name'] = $json['item']['name'] ?? null; |
| 658 | $event_log['call_id'] = $json['item']['call_id'] ?? null; |
| 659 | } |
| 660 | elseif ( strpos( $json['type'], 'response.function_call' ) === 0 ) { |
| 661 | $event_log['call_id'] = $json['call_id'] ?? $json['item_id'] ?? null; |
| 662 | } |
| 663 | elseif ( $json['type'] === 'response.output_item.done' && isset( $json['item'] ) ) { |
| 664 | $event_log['item_type'] = $json['item']['type'] ?? 'unknown'; |
| 665 | if ( isset( $json['item']['call_id'] ) ) { |
| 666 | $event_log['call_id'] = $json['item']['call_id']; |
| 667 | } |
| 668 | } |
| 669 | } |
| 670 | |
| 671 | error_log( '[AI Engine Queries] Event: ' . json_encode( $event_log ) ); |
| 672 | $event_count++; |
| 673 | } |
| 674 | |
| 675 | $content = $this->stream_data_handler( $json ); |
| 676 | if ( !is_null( $content ) ) { |
| 677 | |
| 678 | // Check if content is an Event object |
| 679 | if ( is_object( $content ) && $content instanceof Meow_MWAI_Event ) { |
| 680 | // For Event objects, pass the object directly to callback |
| 681 | // Don't accumulate in streamContent as it's not regular text |
| 682 | call_user_func( $this->streamCallback, $content ); |
| 683 | } |
| 684 | else if ( !empty( $content ) || $content === '0' ) { |
| 685 | // For regular string content - only process non-empty strings (but allow '0') |
| 686 | |
| 687 | // TO CHECK: Not sure why we need to do this to make sure there is a line return in the chatbot |
| 688 | // If we don't do this, HuggingFace streams "\n" as a token without anything else, and the |
| 689 | // chatbot doesn't display it. |
| 690 | if ( $content === "\n" ) { |
| 691 | $content = " \n"; |
| 692 | } |
| 693 | |
| 694 | $this->streamContent .= $content; |
| 695 | call_user_func( $this->streamCallback, $content ); |
| 696 | } |
| 697 | } |
| 698 | } |
| 699 | else if ( $line !== '[DONE]' && !empty( $line ) ) { |
| 700 | $this->streamTemporaryBuffer .= $line . "\n"; |
| 701 | } |
| 702 | } |
| 703 | } |
| 704 | return $length; |
| 705 | } ); |
| 706 | } |
| 707 | |
| 708 | protected function stream_header_handler( $header ) { |
| 709 | |
| 710 | } |
| 711 | |
| 712 | protected function stream_data_handler( $json ) { |
| 713 | throw new Exception( 'Not implemented.' ); |
| 714 | } |
| 715 | |
| 716 | public function get_models() { |
| 717 | throw new Exception( 'Not implemented.' ); |
| 718 | } |
| 719 | |
| 720 | public function retrieve_models() { |
| 721 | throw new Exception( 'Not implemented.' ); |
| 722 | } |
| 723 | |
| 724 | public function run_completion_query( Meow_MWAI_Query_Base $query, $streamCallback = null ): Meow_MWAI_Reply { |
| 725 | throw new Exception( 'Not implemented.' ); |
| 726 | } |
| 727 | |
| 728 | public function run_assistant_query( Meow_MWAI_Query_Assistant $query, $streamCallback = null ): Meow_MWAI_Reply { |
| 729 | throw new Exception( 'Not implemented, or not supported in this version of AI Engine.' ); |
| 730 | } |
| 731 | |
| 732 | public function run_embedding_query( Meow_MWAI_Query_Base $query ) { |
| 733 | throw new Exception( 'Not implemented.' ); |
| 734 | } |
| 735 | |
| 736 | public function run_image_query( Meow_MWAI_Query_Base $query, $streamCallback = null ) { |
| 737 | throw new Exception( 'Not implemented.' ); |
| 738 | } |
| 739 | |
| 740 | public function run_editimage_query( Meow_MWAI_Query_Base $query ) { |
| 741 | throw new Exception( 'Not implemented.' ); |
| 742 | } |
| 743 | |
| 744 | public function run_transcribe_query( Meow_MWAI_Query_Base $query ) { |
| 745 | throw new Exception( 'Not implemented.' ); |
| 746 | } |
| 747 | |
| 748 | public function get_price( Meow_MWAI_Query_Base $query, Meow_MWAI_Reply $reply ) { |
| 749 | throw new Exception( 'Not implemented.' ); |
| 750 | } |
| 751 | |
| 752 | /** |
| 753 | * Check the connection to the AI service. |
| 754 | * This should be a minimal, cost-free API call to verify credentials and connectivity. |
| 755 | * |
| 756 | * @return array { |
| 757 | * @type bool $success Whether the connection test was successful |
| 758 | * @type string $service The service name (e.g., 'OpenAI', 'Anthropic') |
| 759 | * @type string $message A human-readable message about the test result |
| 760 | * @type array $details Additional service-specific details |
| 761 | * @type string $error Error message if the test failed |
| 762 | * } |
| 763 | */ |
| 764 | public function connection_check() { |
| 765 | throw new Exception( 'Connection check not implemented for this service.' ); |
| 766 | } |
| 767 | } |
| 768 |