anthropic.php
1 year ago
core.php
1 year ago
factory.php
1 year ago
google.php
1 year ago
huggingface.php
1 year ago
openai.php
1 year ago
openrouter.php
1 year ago
replicate.php
1 year ago
core.php
381 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 | public function __construct( $core, $env ) { |
| 17 | $this->core = $core; |
| 18 | $this->env = $env; |
| 19 | $this->envId = isset( $env['id'] ) ? $env['id'] : null; |
| 20 | $this->envType = isset( $env['type'] ) ? $env['type'] : null; |
| 21 | } |
| 22 | |
| 23 | public function run( $query, $streamCallback = null, $maxDepth = 5 ) { |
| 24 | |
| 25 | // Check if the query is allowed. |
| 26 | $limits = $this->core->get_option( 'limits' ); |
| 27 | $allowed = apply_filters( 'mwai_ai_allowed', true, $query, $limits ); |
| 28 | if ( $allowed !== true ) { |
| 29 | $message = is_string( $allowed ) ? $allowed : 'Unauthorized query.'; |
| 30 | throw new Exception( $message ); |
| 31 | } |
| 32 | |
| 33 | // Allow to modify the query before it is sent. |
| 34 | // Embedding and Feedback queries are not allowed to be modified. |
| 35 | if ( !( $query instanceof Meow_MWAI_Query_Embed ) && !( $query instanceof Meow_MWAI_Query_Feedback ) ) { |
| 36 | $query = apply_filters( 'mwai_ai_query', $query ); |
| 37 | } |
| 38 | |
| 39 | // Important as it makes sure everything is consolidated in the query and the engine. |
| 40 | $this->final_checks( $query ); |
| 41 | |
| 42 | // Run the query |
| 43 | $reply = null; |
| 44 | if ( $query instanceof Meow_MWAI_Query_Text || $query instanceof Meow_MWAI_Query_Feedback ) { |
| 45 | $reply = $this->run_completion_query( $query, $streamCallback ); |
| 46 | } |
| 47 | else if ( $query instanceof Meow_MWAI_Query_Assistant || $query instanceof Meow_MWAI_Query_AssistFeedback ) { |
| 48 | $reply = $this->run_assistant_query( $query, $streamCallback ); |
| 49 | if ( $reply === null ) { |
| 50 | throw new Exception( 'Assistants are not supported in this version of AI Engine.' ); |
| 51 | } |
| 52 | } |
| 53 | else if ( $query instanceof Meow_MWAI_Query_Embed ) { |
| 54 | $reply = $this->run_embedding_query( $query ); |
| 55 | } |
| 56 | else if ( $query instanceof Meow_MWAI_Query_Image ) { |
| 57 | $reply = $this->run_image_query( $query ); |
| 58 | } |
| 59 | else if ( $query instanceof Meow_MWAI_Query_Transcribe ) { |
| 60 | $reply = $this->run_transcribe_query( $query ); |
| 61 | } |
| 62 | else { |
| 63 | throw new Exception( 'Unknown query type.' ); |
| 64 | } |
| 65 | |
| 66 | // Allow to modify the reply before it is sent. |
| 67 | $reply = apply_filters( 'mwai_ai_reply', $reply, $query ); |
| 68 | |
| 69 | // Function Call |
| 70 | if ( !empty( $reply->needFeedbacks ) ) { |
| 71 | |
| 72 | // Check if we reached the maximum depth |
| 73 | if ( $maxDepth <= 0 ) { |
| 74 | throw new Exception( 'AI Engine: There seems to be a loop in the function/tools calls.' ); |
| 75 | } |
| 76 | |
| 77 | // We should use a feedback query around the original query |
| 78 | if ( !( $query instanceof Meow_MWAI_Query_AssistFeedback) && !( $query instanceof Meow_MWAI_Query_Feedback ) ) { |
| 79 | $queryClass = $query instanceof Meow_MWAI_Query_Assistant ? |
| 80 | Meow_MWAI_Query_AssistFeedback::class : Meow_MWAI_Query_Feedback::class; |
| 81 | $query = new $queryClass( $reply, $reply->query ); |
| 82 | } |
| 83 | |
| 84 | // The engine for the model will handle the feedback query nicely |
| 85 | // In the case of Anthropic, it's like a "discussion" between the user (= WordPress's AI Engine |
| 86 | // and the model (= Anthropic). The feedback is a way to communicate between the two. |
| 87 | $feedback_blocks = []; |
| 88 | foreach ( $reply->needFeedbacks as $needFeedback ) { |
| 89 | $rawMessageKey = md5( serialize( $needFeedback['rawMessage'] ) ); |
| 90 | |
| 91 | // Initialize the feedback block for this rawMessage if it hasn't been initialized yet |
| 92 | if ( !isset( $feedback_blocks[$rawMessageKey] ) ) { |
| 93 | $feedback_blocks[$rawMessageKey] = [ |
| 94 | 'rawMessage' => $needFeedback['rawMessage'], |
| 95 | 'feedbacks' => [] |
| 96 | ]; |
| 97 | } |
| 98 | |
| 99 | // Get the value related to this feedback (usually, a function call) |
| 100 | $value = apply_filters( 'mwai_ai_feedback', null, $needFeedback, $reply ); |
| 101 | |
| 102 | if ( $value === null ) { |
| 103 | Meow_MWAI_Logging::warn( "The returned value for '{$needFeedback['name']}' was null." ); |
| 104 | $value = "[NO VALUE RETURNED - DO NOT SHOW THIS]"; |
| 105 | } |
| 106 | |
| 107 | // Add the feedback information to the appropriate feedback block |
| 108 | $feedback_blocks[$rawMessageKey]['feedbacks'][] = [ |
| 109 | 'request' => $needFeedback, // TODO: Meow_MWAI_Feedback_Request |
| 110 | 'reply' => [ 'value' => $value ] // TODO: Meow_MWAI_Feedback_Reply |
| 111 | ]; |
| 112 | } |
| 113 | |
| 114 | foreach ( $feedback_blocks as $feedback_block ) { |
| 115 | $query->add_feedback_block( $feedback_block ); |
| 116 | } |
| 117 | |
| 118 | // Run the feedback query |
| 119 | $reply = $this->run( $query, $streamCallback, $maxDepth - 1 ); |
| 120 | } |
| 121 | |
| 122 | return $reply; |
| 123 | } |
| 124 | |
| 125 | public function retrieve_model_info( $model ) { |
| 126 | $models = $this->get_models(); |
| 127 | foreach ( $models as $currentModel ) { |
| 128 | if ( $currentModel['model'] === $model ) { |
| 129 | return $currentModel; |
| 130 | } |
| 131 | } |
| 132 | return false; |
| 133 | } |
| 134 | |
| 135 | public function final_checks( Meow_MWAI_Query_Base $query ) { |
| 136 | $query->final_checks(); |
| 137 | //$found = false; |
| 138 | |
| 139 | // Check if the model is available, except if it's an assistant |
| 140 | if ( !( $query instanceof Meow_MWAI_Query_Assistant ) ) { |
| 141 | // TODO: Avoid checking on the finetuned models for now. |
| 142 | if ( substr( $query->model, 0, 3 ) === 'ft:' ) { |
| 143 | return; |
| 144 | } |
| 145 | $model_info = $this->retrieve_model_info( $query->model ); |
| 146 | if ( $model_info === false ) { |
| 147 | throw new Exception( "AI Engine: The model '{$query->model}' is not available." ); |
| 148 | } |
| 149 | if ( isset( $model_info['mode'] ) ) { |
| 150 | $query->mode = $model_info['mode']; |
| 151 | } |
| 152 | } |
| 153 | } |
| 154 | |
| 155 | // Streamline the messages: |
| 156 | // - Concatenate consecutive model messages into a single message for the model role |
| 157 | // - Make sure the first message is a user message |
| 158 | // - Make sure the last message is a user message |
| 159 | protected function streamline_messages( $messages, $systemRole = 'assistant', $messageType = 'content' ) |
| 160 | { |
| 161 | $processedMessages = []; |
| 162 | $lastRole = ''; |
| 163 | $concatenatedText = ''; |
| 164 | |
| 165 | // Determine the way to access message content based on messageType |
| 166 | $getContent = function( $message ) use ( $messageType ) { |
| 167 | if ( $messageType == 'parts' ) { |
| 168 | return $message['parts'][0]['text']; |
| 169 | } |
| 170 | else { // Default to 'content' |
| 171 | return $message['content']; |
| 172 | } |
| 173 | }; |
| 174 | |
| 175 | // Set content to a message depending on the messageType |
| 176 | $setContent = function( &$message, $content ) use ( $messageType ) { |
| 177 | if ( $messageType == 'parts' ) { |
| 178 | $message['parts'] = [['text' => $content]]; |
| 179 | } |
| 180 | else { // Default to 'content' |
| 181 | $message['content'] = $content; |
| 182 | } |
| 183 | }; |
| 184 | |
| 185 | // Concatenate consecutive model messages into a single message for the model role |
| 186 | foreach ( $messages as $message ) { |
| 187 | if ( $message['role'] == $systemRole ) { |
| 188 | if ( $lastRole == $systemRole ) { |
| 189 | $concatenatedText .= "\n" . $getContent( $message ); |
| 190 | } |
| 191 | else { |
| 192 | if ( $concatenatedText !== '' ) { |
| 193 | $newMessage = [ 'role' => $systemRole ]; |
| 194 | $setContent( $newMessage, $concatenatedText ); |
| 195 | $processedMessages[] = $newMessage; |
| 196 | } |
| 197 | $concatenatedText = $getContent( $message ); |
| 198 | } |
| 199 | } |
| 200 | else { |
| 201 | if ( $lastRole == $systemRole ) { |
| 202 | $newMessage = [ 'role' => $systemRole ]; |
| 203 | $setContent( $newMessage, $concatenatedText ); |
| 204 | $processedMessages[] = $newMessage; |
| 205 | $concatenatedText = ''; |
| 206 | } |
| 207 | $processedMessages[] = $message; |
| 208 | } |
| 209 | $lastRole = $message['role']; |
| 210 | } |
| 211 | if ( $lastRole == $systemRole && $concatenatedText !== '' ) { |
| 212 | $newMessage = [ 'role' => $systemRole ]; |
| 213 | $setContent( $newMessage, $concatenatedText ); |
| 214 | $processedMessages[] = $newMessage; |
| 215 | } |
| 216 | |
| 217 | // Make sure the last message is a user message, if not, throw an exception |
| 218 | if ( end( $processedMessages )['role'] !== 'user' ) { |
| 219 | throw new Exception( 'The last message must be a user message.' ); |
| 220 | } |
| 221 | |
| 222 | // Make sure the first message is a user message, if not, add an empty user message |
| 223 | if ( $processedMessages[0]['role'] !== 'user' ) { |
| 224 | $newMessage = [ 'role' => 'user' ]; |
| 225 | $setContent( $newMessage, '' ); |
| 226 | array_unshift( $processedMessages, $newMessage ); |
| 227 | } |
| 228 | |
| 229 | return $processedMessages; |
| 230 | } |
| 231 | |
| 232 | // Check for a JSON-formatted error in the data, and throw an exception if it's the case. |
| 233 | function stream_error_check( $data ) { |
| 234 | if ( strpos( $data, 'error' ) === false ) { |
| 235 | return; |
| 236 | } |
| 237 | $data = trim( $data ); |
| 238 | $jsonPart = $data; |
| 239 | if ( strpos( $jsonPart, 'data:' ) === 0 ) { |
| 240 | $jsonPart = trim( substr( $jsonPart, strlen( 'data:' ) ) ); |
| 241 | } |
| 242 | $json = json_decode( $jsonPart, true ); |
| 243 | if ( json_last_error() === JSON_ERROR_NONE ) { |
| 244 | if ( isset( $json['error'] ) ) { |
| 245 | $error = $json['error']; |
| 246 | $code = null; |
| 247 | $type = null; |
| 248 | $message = null; |
| 249 | if ( isset( $error['message'] ) ) { |
| 250 | $message = $error['message']; |
| 251 | } |
| 252 | else if ( is_string( $error ) ) { |
| 253 | throw new Exception( "Error: $error" ); |
| 254 | } |
| 255 | else { |
| 256 | throw new Exception( "Unknown error (stream_error_check)." ); |
| 257 | } |
| 258 | if ( isset( $error['code'] ) ) { |
| 259 | $code = $error['code']; |
| 260 | } |
| 261 | if ( isset( $error['type'] ) ) { |
| 262 | $type = $error['type']; |
| 263 | } |
| 264 | $errorMessage = "Error: $message"; |
| 265 | if ( !is_null( $code ) ) { |
| 266 | $errorMessage .= " ($code)"; |
| 267 | } |
| 268 | if ( !is_null( $type ) ) { |
| 269 | $errorMessage .= " ($type)"; |
| 270 | } |
| 271 | throw new Exception( $errorMessage ); |
| 272 | } |
| 273 | else if ( isset( $json['type'] ) && $json['type'] === 'error' ) { |
| 274 | $type = $json['error']['type']; |
| 275 | $message = $json['error']['message']; |
| 276 | throw new Exception( "Error: $message ($type)" ); |
| 277 | } |
| 278 | } |
| 279 | } |
| 280 | |
| 281 | public function stream_handler( $handle, $args, $url ) { |
| 282 | curl_setopt( $handle, CURLOPT_SSL_VERIFYPEER, false ); |
| 283 | curl_setopt( $handle, CURLOPT_SSL_VERIFYHOST, false ); |
| 284 | |
| 285 | // TODO: This is breaking the response. We need to find a way to handle the headers. |
| 286 | // curl_setopt( $handle, CURLOPT_HEADERFUNCTION, function ( $curl, $header ) { |
| 287 | // $length = strlen( $header ); |
| 288 | // $this->streamHeaders[] = $header; |
| 289 | // $this->stream_header_handler( $header ); |
| 290 | // return $length; |
| 291 | // }); |
| 292 | |
| 293 | curl_setopt( $handle, CURLOPT_WRITEFUNCTION, function ( $curl, $data ) { |
| 294 | $length = strlen( $data ); |
| 295 | |
| 296 | // Bufferize the unfinished stream (if it's the case) |
| 297 | $this->streamTemporaryBuffer .= $data; |
| 298 | $this->streamBuffer .= $data; |
| 299 | |
| 300 | // Error Management |
| 301 | $this->stream_error_check( $this->streamBuffer ); |
| 302 | |
| 303 | $lines = explode( "\n", $this->streamTemporaryBuffer ); |
| 304 | if ( substr( $this->streamTemporaryBuffer, -1 ) !== "\n" ) { |
| 305 | $this->streamTemporaryBuffer = array_pop( $lines ); |
| 306 | } |
| 307 | else { |
| 308 | $this->streamTemporaryBuffer = ""; |
| 309 | } |
| 310 | |
| 311 | foreach ( $lines as $line ) { |
| 312 | if ( $line === "" ) { continue; } |
| 313 | if ( strpos( $line, 'data:' ) === 0 ) { |
| 314 | $line = trim( substr( $line, 5 ) ); |
| 315 | $json = json_decode( trim( $line ), true ); |
| 316 | |
| 317 | if ( json_last_error() === JSON_ERROR_NONE ) { |
| 318 | $content = $this->stream_data_handler( $json ); |
| 319 | if ( !is_null( $content ) ) { |
| 320 | |
| 321 | // TO CHECK: Not sure why we need to do this to make sure there is a line return in the chatbot |
| 322 | // If we don't do this, HuggingFace streams "\n" as a token without anything else, and the |
| 323 | // chatbot doesn't display it. |
| 324 | if ( $content === "\n" ) { |
| 325 | $content = " \n"; |
| 326 | } |
| 327 | |
| 328 | $this->streamContent .= $content; |
| 329 | call_user_func( $this->streamCallback, $content ); |
| 330 | } |
| 331 | } |
| 332 | else if ( $line !== '[DONE]' && !empty( $line ) ) { |
| 333 | $this->streamTemporaryBuffer .= $line . "\n"; |
| 334 | } |
| 335 | } |
| 336 | } |
| 337 | return $length; |
| 338 | }); |
| 339 | } |
| 340 | |
| 341 | protected function stream_header_handler( $header ) { |
| 342 | |
| 343 | } |
| 344 | |
| 345 | protected function stream_data_handler( $json ) { |
| 346 | throw new Exception( 'Not implemented.' ); |
| 347 | } |
| 348 | |
| 349 | public function get_models() { |
| 350 | throw new Exception( 'Not implemented.' ); |
| 351 | } |
| 352 | |
| 353 | public function retrieve_models() { |
| 354 | throw new Exception( 'Not implemented.' ); |
| 355 | } |
| 356 | |
| 357 | public function run_completion_query( Meow_MWAI_Query_Base $query, $streamCallback = null ) : Meow_MWAI_Reply { |
| 358 | throw new Exception( 'Not implemented.' ); |
| 359 | } |
| 360 | |
| 361 | public function run_assistant_query( Meow_MWAI_Query_Assistant $query, $streamCallback = null ) : Meow_MWAI_Reply { |
| 362 | throw new Exception( 'Not implemented, or not supported in this version of AI Engine.' ); |
| 363 | } |
| 364 | |
| 365 | public function run_embedding_query( Meow_MWAI_Query_Base $query ) { |
| 366 | throw new Exception( 'Not implemented.' ); |
| 367 | } |
| 368 | |
| 369 | public function run_image_query( Meow_MWAI_Query_Base $query ) { |
| 370 | throw new Exception( 'Not implemented.' ); |
| 371 | } |
| 372 | |
| 373 | public function run_transcribe_query( Meow_MWAI_Query_Base $query ) { |
| 374 | throw new Exception( 'Not implemented.' ); |
| 375 | } |
| 376 | |
| 377 | public function get_price( Meow_MWAI_Query_Base $query, Meow_MWAI_Reply $reply ) { |
| 378 | throw new Exception( 'Not implemented.' ); |
| 379 | } |
| 380 | } |
| 381 |