engines
2 years ago
modules
2 years ago
queries
2 years ago
admin.php
2 years ago
api.php
2 years ago
core.php
2 years ago
init.php
3 years ago
reply.php
2 years ago
rest.php
2 years ago
core.php
994 lines
| 1 | <?php |
| 2 | |
| 3 | require_once( MWAI_PATH . '/vendor/autoload.php' ); |
| 4 | require_once( MWAI_PATH . '/constants/init.php' ); |
| 5 | |
| 6 | define( 'MWAI_IMG_WAND', MWAI_URL . '/images/wand.png' ); |
| 7 | define( 'MWAI_IMG_WAND_HTML', "<img style='height: 22px; margin-bottom: -5px; margin-right: 8px;' |
| 8 | src='" . MWAI_IMG_WAND . "' alt='AI Wand' />" ); |
| 9 | define( 'MWAI_IMG_WAND_HTML_XS', "<img style='height: 16px; margin-bottom: -2px;' |
| 10 | src='" . MWAI_IMG_WAND . "' alt='AI Wand' />" ); |
| 11 | |
| 12 | class Meow_MWAI_Core |
| 13 | { |
| 14 | public $admin = null; |
| 15 | public $is_rest = false; |
| 16 | public $is_cli = false; |
| 17 | public $site_url = null; |
| 18 | public $files = null; |
| 19 | public $tasks = null; |
| 20 | private $option_name = 'mwai_options'; |
| 21 | private $themes_option_name = 'mwai_themes'; |
| 22 | private $chatbots_option_name = 'mwai_chatbots'; |
| 23 | private $nonce = null; |
| 24 | |
| 25 | public $chatbot = null; |
| 26 | public $discussions = null; |
| 27 | |
| 28 | public function __construct() { |
| 29 | $this->site_url = get_site_url(); |
| 30 | $this->is_rest = MeowCommon_Helpers::is_rest(); |
| 31 | $this->is_cli = defined( 'WP_CLI' ); |
| 32 | $this->files = new Meow_MWAI_Modules_Files( $this ); |
| 33 | $this->tasks = new Meow_MWAI_Modules_Tasks( $this ); |
| 34 | |
| 35 | add_action( 'plugins_loaded', array( $this, 'init' ) ); |
| 36 | add_action( 'wp_register_script', array( $this, 'register_scripts' ) ); |
| 37 | add_action( 'wp_enqueue_scripts', array( $this, 'register_scripts' ) ); |
| 38 | add_action( 'admin_enqueue_scripts', array( $this, 'enqueue_scripts' ) ); |
| 39 | } |
| 40 | |
| 41 | #region Init & Scripts |
| 42 | function init() { |
| 43 | global $mwai; |
| 44 | $this->chatbot = null; |
| 45 | $this->discussions = null; |
| 46 | new Meow_MWAI_Modules_Security( $this ); |
| 47 | if ( $this->is_rest ) { |
| 48 | new Meow_MWAI_Rest( $this ); |
| 49 | } |
| 50 | if ( is_admin() ) { |
| 51 | new Meow_MWAI_Admin( $this ); |
| 52 | new Meow_MWAI_Modules_Utilities( $this ); |
| 53 | } |
| 54 | if ( $this->get_option( 'shortcode_chat' ) ) { |
| 55 | $this->chatbot = new Meow_MWAI_Modules_Chatbot(); |
| 56 | $this->discussions = new Meow_MWAI_Modules_Discussions(); |
| 57 | } |
| 58 | |
| 59 | // Advanced core |
| 60 | if ( class_exists( 'MeowPro_MWAI_Core' ) ) { |
| 61 | new MeowPro_MWAI_Core( $this ); |
| 62 | } |
| 63 | |
| 64 | $mwai = new Meow_MWAI_API( $this->chatbot, $this->discussions ); |
| 65 | } |
| 66 | |
| 67 | public function register_scripts() { |
| 68 | wp_register_script( 'mwai_highlight', MWAI_URL . 'vendor/highlightjs/highlight.min.js', [], '11.7', false ); |
| 69 | } |
| 70 | |
| 71 | public function enqueue_scripts() { |
| 72 | $this->register_scripts(); |
| 73 | wp_enqueue_script( "mwai_highlight" ); |
| 74 | } |
| 75 | |
| 76 | #endregion |
| 77 | |
| 78 | #region Roles & Capabilities |
| 79 | |
| 80 | function can_access_settings() { |
| 81 | return apply_filters( 'mwai_allow_setup', current_user_can( 'manage_options' ) ); |
| 82 | } |
| 83 | |
| 84 | function can_access_features() { |
| 85 | $editor_or_admin = current_user_can( 'editor' ) || current_user_can( 'administrator' ); |
| 86 | return apply_filters( 'mwai_allow_usage', $editor_or_admin ); |
| 87 | } |
| 88 | |
| 89 | function can_access_public_api( $feature, $extra ) { |
| 90 | $logged_in = is_user_logged_in(); |
| 91 | return apply_filters( 'mwai_allow_public_api', $logged_in, $feature, $extra ); |
| 92 | } |
| 93 | |
| 94 | #endregion |
| 95 | |
| 96 | #region AI-Related Helpers |
| 97 | function run_query( $query, $streamCallback = null, $markdown = false ) { |
| 98 | $envId = !empty( $query->envId ) ? $query->envId : null; |
| 99 | $engine = Meow_MWAI_Engines_Factory::get( $this, $envId ); |
| 100 | |
| 101 | // If the engine is not set, we need to set it to the default one. |
| 102 | if ( !$envId || !$engine->retrieve_model_info( $query->model ) ) { |
| 103 | if ( $query instanceof Meow_MWAI_Query_Text ) { |
| 104 | $this->set_if_empty_defaults( $query, 'ai_default_env', 'ai_default_model' ); |
| 105 | } |
| 106 | if ( $query instanceof Meow_MWAI_Query_Embed ) { |
| 107 | $this->set_if_empty_defaults( $query, 'ai_embeddings_default_env', 'ai_embeddings_default_model' ); |
| 108 | } |
| 109 | else if ( $query instanceof Meow_MWAI_Query_Image ) { |
| 110 | $this->set_if_empty_defaults( $query, 'ai_images_default_env', 'ai_images_default_model' ); |
| 111 | } |
| 112 | else if ( $query instanceof Meow_MWAI_Query_Transcribe ) { |
| 113 | $this->set_if_empty_defaults( $query, 'ai_audio_default_env', 'ai_audio_default_model' ); |
| 114 | } |
| 115 | $engine = Meow_MWAI_Engines_Factory::get( $this, $query->envId ); |
| 116 | } |
| 117 | |
| 118 | // Let's run the query. |
| 119 | $reply = $engine->run( $query, $streamCallback ); |
| 120 | |
| 121 | // Let's allow to modify the reply before it is sent. |
| 122 | if ( $markdown ) { |
| 123 | if ( $query instanceof Meow_MWAI_Query_Image ) { |
| 124 | $reply->result = ""; |
| 125 | foreach ( $reply->results as $result ) { |
| 126 | $reply->result .= "\n"; |
| 127 | } |
| 128 | } |
| 129 | } |
| 130 | |
| 131 | return $reply; |
| 132 | } |
| 133 | |
| 134 | private function set_if_empty_defaults( $query, $envOption, $modelOption ) { |
| 135 | $defaultEnv = $this->get_option( $envOption ); |
| 136 | $defaultModel = $this->get_option( $modelOption ); |
| 137 | if ( empty( $defaultEnv ) || empty( $defaultModel ) ) { |
| 138 | throw new Exception( 'AI Engine: The default environment and model are not set.' ); |
| 139 | } |
| 140 | $query->set_env_id( $defaultEnv ); |
| 141 | $query->set_model( $defaultModel ); |
| 142 | } |
| 143 | #endregion |
| 144 | |
| 145 | #region Text-Related Helpers |
| 146 | |
| 147 | // Clean the text perfectly, resolve shortcodes, etc, etc. |
| 148 | function clean_text( $rawText = "" ) { |
| 149 | $text = html_entity_decode( $rawText ); |
| 150 | $text = wp_strip_all_tags( $text ); |
| 151 | $text = preg_replace( '/[\r\n]+/', "\n", $text ); |
| 152 | $text = preg_replace( '/\n+/', "\n", $text ); |
| 153 | $text = preg_replace( '/\t+/', "\t", $text ); |
| 154 | return $text . " "; |
| 155 | } |
| 156 | |
| 157 | // Make sure there are no duplicate sentences, and keep the length under a maximum length. |
| 158 | function clean_sentences( $text, $maxLength = null ) { |
| 159 | $maxLength = (int)($maxLength ? $maxLength : $this->get_option( 'context_max_length', 4096 )); |
| 160 | $sentences = preg_split('/(?<=[.?!。．！？])+/u', $text); |
| 161 | $hashes = array(); |
| 162 | $uniqueSentences = array(); |
| 163 | $total = 0; |
| 164 | foreach ( $sentences as $sentence ) { |
| 165 | $sentence = preg_replace( '/^[\pZ\pC]+|[\pZ\pC]+$/u', '', $sentence ); |
| 166 | $hash = md5( $sentence ); |
| 167 | if ( !in_array( $hash, $hashes ) ) { |
| 168 | $length = strlen( $sentence ); |
| 169 | if ( $total + $length > $maxLength ) { |
| 170 | continue; |
| 171 | } |
| 172 | $hashes[] = $hash; |
| 173 | $uniqueSentences[] = $sentence; |
| 174 | $total += $length; |
| 175 | } |
| 176 | } |
| 177 | $freshText = implode( " ", $uniqueSentences ); |
| 178 | $freshText = preg_replace( '/^[\pZ\pC]+|[\pZ\pC]+$/u', '', $freshText ); |
| 179 | return $freshText; |
| 180 | } |
| 181 | |
| 182 | function get_post_content( $postId ) { |
| 183 | $post = get_post( $postId ); |
| 184 | if ( !$post ) { |
| 185 | return false; |
| 186 | } |
| 187 | $text = apply_filters( 'mwai_pre_post_content', $post->post_content, $postId ); |
| 188 | $pattern = '/\[mwai_.*?\]/'; |
| 189 | $text = preg_replace( $pattern, '', $text ); |
| 190 | if ( $this->get_option( 'resolve_shortcodes' ) ) { |
| 191 | $text = apply_filters( 'the_content', $text ); |
| 192 | } |
| 193 | else { |
| 194 | $pattern = "/\[[^\]]+\]/"; |
| 195 | $text = preg_replace( $pattern, '', $text ); |
| 196 | $pattern = "/<!--\s*\/?wp:[^\>]+-->/"; |
| 197 | $text = preg_replace( $pattern, '', $text ); |
| 198 | } |
| 199 | $text = $this->clean_text( $text ); |
| 200 | $text = $this->clean_sentences( $text ); |
| 201 | $text = apply_filters( 'mwai_post_content', $text, $postId ); |
| 202 | return $text; |
| 203 | } |
| 204 | |
| 205 | function markdown_to_html( $content ) { |
| 206 | $Parsedown = new Parsedown(); |
| 207 | $content = $Parsedown->text( $content ); |
| 208 | return $content; |
| 209 | } |
| 210 | |
| 211 | function get_post_language( $postId ) { |
| 212 | $locale = get_locale(); |
| 213 | $code = strtolower( substr( $locale, 0, 2 ) ); |
| 214 | $humanLanguage = strtr( $code, MWAI_ALL_LANGUAGES ); |
| 215 | $lang = apply_filters( 'wpml_post_language_details', null, $postId ); |
| 216 | if ( !empty( $lang ) ) { |
| 217 | $locale = $lang['locale']; |
| 218 | $humanLanguage = $lang['display_name']; |
| 219 | } |
| 220 | return strtolower( "$locale ($humanLanguage)" ); |
| 221 | } |
| 222 | #endregion |
| 223 | |
| 224 | #region Image-Related Helpers |
| 225 | function download_image( $url ) { |
| 226 | $args = array( 'timeout' => 60, ); |
| 227 | $response = wp_safe_remote_get( $url, $args ); |
| 228 | if ( is_wp_error( $response ) ) { |
| 229 | throw new Exception( $response->get_error_message() ); |
| 230 | } |
| 231 | $output = wp_remote_retrieve_body( $response ); |
| 232 | if ( is_wp_error( $output ) ) { |
| 233 | throw new Exception( $output->get_error_message() ); |
| 234 | } |
| 235 | return $output; |
| 236 | } |
| 237 | |
| 238 | /** |
| 239 | * Add an image from a URL to the Media Library. |
| 240 | * @param string $url The URL of the image to be downloaded. |
| 241 | * @param string $filename The filename of the image, if not set, it will be the basename of the URL. |
| 242 | * @param string $title The title of the image. |
| 243 | * @param string $description The description of the image. |
| 244 | * @param string $caption The caption of the image. |
| 245 | * @param string $alt The alt text of the image. |
| 246 | * @return int The attachment ID of the image. |
| 247 | */ |
| 248 | public function add_image_from_url( $url, $filename = null, $title = null, |
| 249 | $description = null, $caption = null, $alt = null ) { |
| 250 | $path_parts = pathinfo( parse_url( $url, PHP_URL_PATH ) ); |
| 251 | $url_filename = $path_parts['basename']; |
| 252 | $file_type = wp_check_filetype( $url_filename, null ); |
| 253 | $allowed_types = get_allowed_mime_types(); |
| 254 | if ( !$file_type || !in_array( $file_type['type'], $allowed_types ) ) { |
| 255 | throw new Exception( 'Invalid file type.' ); |
| 256 | } |
| 257 | $extension = $file_type['ext']; |
| 258 | $image_data = $this->download_image( $url ); |
| 259 | if ( !$image_data ) { |
| 260 | throw new Exception( 'Could not download the image.' ); |
| 261 | } |
| 262 | $upload_dir = wp_upload_dir(); |
| 263 | |
| 264 | // If filename is not set or starts with 'generated_', we will generate a new filename. |
| 265 | if ( empty( $filename ) ) { |
| 266 | $filename = sanitize_file_name( $url_filename ); |
| 267 | $extension = pathinfo( $filename, PATHINFO_EXTENSION ); |
| 268 | if ( empty( $extension ) ) { |
| 269 | $extension = $file_type['ext']; |
| 270 | } |
| 271 | if ( strlen( $filename ) > 32 || strlen( $filename ) < 4 || strpos( $filename, 'generated_' ) === 0 ) { |
| 272 | $filename = $this->get_random_id( 16 ) . '.' . $extension; |
| 273 | } |
| 274 | if ( strpos( $filename, '.' ) === false ) { |
| 275 | $filename .= '.' . $extension; |
| 276 | } |
| 277 | } |
| 278 | if ( wp_mkdir_p( $upload_dir['path'] ) ) { |
| 279 | $file = $upload_dir['path'] . '/' . $filename; |
| 280 | } |
| 281 | else { |
| 282 | $file = $upload_dir['basedir'] . '/' . $filename; |
| 283 | } |
| 284 | |
| 285 | // Make sure the file is unique, if not, add a number to the end of the file before the extension |
| 286 | $i = 1; |
| 287 | $parts = pathinfo( $file ); |
| 288 | while ( file_exists( $file ) ) { |
| 289 | $file = $parts['dirname'] . '/' . $parts['filename'] . '-' . $i . '.' . $parts['extension']; |
| 290 | $i++; |
| 291 | } |
| 292 | |
| 293 | // Write the file |
| 294 | file_put_contents( $file, $image_data ); |
| 295 | $attachment = [ |
| 296 | 'post_mime_type' => $file_type['type'], |
| 297 | 'post_title' => $title ?? '', |
| 298 | 'post_content' => $description ?? '', |
| 299 | 'post_excerpt' => $caption ?? '', |
| 300 | 'post_status' => 'inherit' |
| 301 | ]; |
| 302 | // Register the file as a Media Library attachment |
| 303 | $attachmentId = wp_insert_attachment( $attachment, $file ); |
| 304 | require_once( ABSPATH . 'wp-admin/includes/image.php' ); |
| 305 | $attachment_data = wp_generate_attachment_metadata( $attachmentId, $file ); |
| 306 | wp_update_attachment_metadata( $attachmentId, $attachment_data ); |
| 307 | update_post_meta( $attachmentId, '_wp_attachment_image_alt', $alt ); |
| 308 | return $attachmentId; |
| 309 | } |
| 310 | #endregion |
| 311 | |
| 312 | #region Context-Related Helpers |
| 313 | function retrieve_context( $params, $query ) { |
| 314 | $contextMaxLength = $params['contextMaxLength'] ?? $this->get_option( 'context_max_length', 4096 ); |
| 315 | $embeddingsEnvId = $params['embeddingsEnvId'] ?? null; |
| 316 | $embeddingsIndex = $params['embeddingsIndex'] ?? null; |
| 317 | $embeddingsNamespace = $params['embeddingsNamespace'] ?? null; |
| 318 | $context = apply_filters( 'mwai_context_search', [], $query, [ |
| 319 | 'embeddingsEnvId' => $embeddingsEnvId, |
| 320 | 'embeddingsIndex' => $embeddingsIndex, |
| 321 | 'embeddingsNamespace' => $embeddingsNamespace |
| 322 | ]); |
| 323 | if ( empty( $context ) ) { |
| 324 | return null; |
| 325 | } |
| 326 | else if ( !isset( $context['content'] ) ) { |
| 327 | error_log( "AI Engine: A context without content was returned." ); |
| 328 | return null; |
| 329 | } |
| 330 | $context['content'] = $this->clean_sentences( $context['content'], $contextMaxLength ); |
| 331 | $context['length'] = strlen( $context['content'] ); |
| 332 | return $context; |
| 333 | } |
| 334 | #endregion |
| 335 | |
| 336 | #region Users/Sessions Helpers |
| 337 | |
| 338 | function get_nonce() { |
| 339 | // if ( !is_user_logged_in() ) { |
| 340 | // return null; |
| 341 | // } |
| 342 | if ( isset( $this->nonce ) ) { |
| 343 | return $this->nonce; |
| 344 | } |
| 345 | $this->nonce = wp_create_nonce( 'wp_rest' ); |
| 346 | return $this->nonce; |
| 347 | } |
| 348 | |
| 349 | function get_session_id() { |
| 350 | if ( isset( $_COOKIE['mwai_session_id'] ) ) { |
| 351 | return $_COOKIE['mwai_session_id']; |
| 352 | } |
| 353 | return "N/A"; |
| 354 | } |
| 355 | |
| 356 | // Get the UserID from the data, or from the current user |
| 357 | function get_user_id( $data = null ) { |
| 358 | // TODO: Not sure if that's the best way, but we should probably use an admin user as a fallback for CRON. |
| 359 | if ( defined( 'DOING_CRON' ) && DOING_CRON ) { |
| 360 | $admin = get_users( [ 'role' => 'administrator' ] ); |
| 361 | if ( !empty( $admin ) ) { |
| 362 | return $admin[0]->ID; |
| 363 | } |
| 364 | } |
| 365 | if ( isset( $data ) && isset( $data['userId'] ) ) { |
| 366 | return (int)$data['userId']; |
| 367 | } |
| 368 | if ( is_user_logged_in() ) { |
| 369 | $current_user = wp_get_current_user(); |
| 370 | if ( $current_user->ID > 0 ) { |
| 371 | return $current_user->ID; |
| 372 | } |
| 373 | } |
| 374 | return null; |
| 375 | } |
| 376 | |
| 377 | function get_admin_user() { |
| 378 | $admin = get_users( [ 'role' => 'administrator' ] ); |
| 379 | if ( !empty( $admin ) ) { |
| 380 | return $admin[0]; |
| 381 | } |
| 382 | return null; |
| 383 | } |
| 384 | |
| 385 | function get_user_data() { |
| 386 | $user = wp_get_current_user(); |
| 387 | if ( empty( $user ) || empty( $user->ID ) ) { |
| 388 | return null; |
| 389 | } |
| 390 | $placeholders = array( |
| 391 | 'FIRST_NAME' => get_user_meta( $user->ID, 'first_name', true ), |
| 392 | 'LAST_NAME' => get_user_meta( $user->ID, 'last_name', true ), |
| 393 | 'USER_LOGIN' => isset( $user ) && isset($user->data) && isset( $user->data->user_login ) ? |
| 394 | $user->data->user_login : null, |
| 395 | 'DISPLAY_NAME' => isset( $user ) && isset( $user->data ) && isset( $user->data->display_name ) ? |
| 396 | $user->data->display_name : null, |
| 397 | 'AVATAR_URL' => get_avatar_url( get_current_user_id() ), |
| 398 | ); |
| 399 | return $placeholders; |
| 400 | } |
| 401 | |
| 402 | function get_ip_address( $params = null ) { |
| 403 | $ip = '127.0.0.1'; |
| 404 | $headers = [ |
| 405 | 'HTTP_TRUE_CLIENT_IP', |
| 406 | 'HTTP_CF_CONNECTING_IP', |
| 407 | 'HTTP_X_REAL_IP', |
| 408 | 'HTTP_CLIENT_IP', |
| 409 | 'HTTP_X_FORWARDED_FOR', |
| 410 | 'HTTP_X_FORWARDED', |
| 411 | 'HTTP_X_CLUSTER_CLIENT_IP', |
| 412 | 'HTTP_FORWARDED_FOR', |
| 413 | 'HTTP_FORWARDED', |
| 414 | 'REMOTE_ADDR', |
| 415 | ]; |
| 416 | |
| 417 | if ( isset( $params ) && isset( $params[ 'ip' ] ) ) { |
| 418 | $ip = ( string )$params[ 'ip' ]; |
| 419 | } else { |
| 420 | foreach ( $headers as $header ) { |
| 421 | if ( array_key_exists( $header, $_SERVER ) && !empty( $_SERVER[ $header ] && $_SERVER[ $header ] != '::1' ) ) { |
| 422 | $address_chain = explode( ',', wp_unslash( $_SERVER [ $header ] ) ); |
| 423 | $ip = filter_var( trim( $address_chain[ 0 ] ), FILTER_VALIDATE_IP ); |
| 424 | break; |
| 425 | } |
| 426 | } |
| 427 | } |
| 428 | |
| 429 | return filter_var( apply_filters( 'mwai_get_ip_address', $ip ), FILTER_VALIDATE_IP ); |
| 430 | } |
| 431 | |
| 432 | #endregion |
| 433 | |
| 434 | #region Other Helpers |
| 435 | |
| 436 | public function check_rest_nonce( $request ) { |
| 437 | $nonce = $request->get_header( 'X-WP-Nonce' ); |
| 438 | $rest_nonce = wp_verify_nonce( $nonce, 'wp_rest' ); |
| 439 | return apply_filters( 'mwai_rest_authorized', $rest_nonce, $request ); |
| 440 | } |
| 441 | |
| 442 | function get_random_id( $length = 8, $excludeIds = [] ) { |
| 443 | $characters = '0123456789abcdefghijklmnopqrstuvwxyz'; |
| 444 | $charactersLength = strlen( $characters ); |
| 445 | $randomId = ''; |
| 446 | for ( $i = 0; $i < $length; $i++ ) { |
| 447 | $randomId .= $characters[rand( 0, $charactersLength - 1 )]; |
| 448 | } |
| 449 | if ( in_array( $randomId, $excludeIds ) ) { |
| 450 | return $this->get_random_id( $length, $excludeIds ); |
| 451 | } |
| 452 | return $randomId; |
| 453 | } |
| 454 | |
| 455 | function is_url( $url ) { |
| 456 | return strpos( $url, 'http' ) === 0 ? true : false; |
| 457 | } |
| 458 | |
| 459 | function get_post_types() { |
| 460 | $excluded = array( 'attachment', 'revision', 'nav_menu_item' ); |
| 461 | $post_types = array(); |
| 462 | $types = get_post_types( [], 'objects' ); |
| 463 | |
| 464 | // Let's get the Post Types that are enabled for Embeddings Sync |
| 465 | $embeddingsSettings = $this->get_option( 'embeddings' ); |
| 466 | $syncPostTypes = isset( $embeddingsSettings['syncPostTypes'] ) ? $embeddingsSettings['syncPostTypes'] : []; |
| 467 | |
| 468 | foreach ( $types as $type ) { |
| 469 | $forced = in_array( $type->name, $syncPostTypes ); |
| 470 | // Should not be excluded. |
| 471 | if ( !$forced && in_array( $type->name, $excluded ) ) { |
| 472 | continue; |
| 473 | } |
| 474 | // Should be public. |
| 475 | if ( !$forced && !$type->public ) { |
| 476 | continue; |
| 477 | } |
| 478 | $post_types[] = array( |
| 479 | 'name' => $type->labels->name, |
| 480 | 'type' => $type->name, |
| 481 | ); |
| 482 | } |
| 483 | |
| 484 | // Let's get the Post Types that are enabled for Embeddings Sync |
| 485 | $embeddingsSettings = $this->get_option( 'embeddings' ); |
| 486 | $syncPostTypes = isset( $embeddingsSettings['syncPostTypes'] ) ? $embeddingsSettings['syncPostTypes'] : []; |
| 487 | |
| 488 | return $post_types; |
| 489 | } |
| 490 | |
| 491 | function get_post( $post ) { |
| 492 | if ( is_numeric( $post ) ) { |
| 493 | $post = get_post( $post ); |
| 494 | } |
| 495 | if ( is_object( $post ) ) { |
| 496 | $post = (array)$post; |
| 497 | } |
| 498 | if ( !is_array( $post ) ) { |
| 499 | return null; |
| 500 | } |
| 501 | $language = $this->get_post_language( $post['ID'] ); |
| 502 | $content = $this->get_post_content( $post['ID'] ); |
| 503 | $title = $post['post_title']; |
| 504 | $excerpt = $post['post_excerpt']; |
| 505 | $url = get_permalink( $post['ID'] ); |
| 506 | $checksum = wp_hash( $content . $title . $url ); |
| 507 | return [ |
| 508 | 'postId' => (int)$post['ID'], |
| 509 | 'title' => $title, |
| 510 | 'content' => $content, |
| 511 | 'excerpt' => $excerpt, |
| 512 | 'url' => $url, |
| 513 | 'language' => $language ?? 'english', |
| 514 | 'checksum' => $checksum, |
| 515 | ]; |
| 516 | } |
| 517 | #endregion |
| 518 | |
| 519 | #region Usage & Costs |
| 520 | |
| 521 | // Quick and dirty token estimation |
| 522 | // Let's keep this synchronized with Helpers in JS |
| 523 | static function estimate_tokens( ...$args ): int { |
| 524 | $text = ""; |
| 525 | foreach ( $args as $arg ) { |
| 526 | if ( is_array( $arg ) ) { |
| 527 | foreach ( $arg as $message ) { |
| 528 | $text .= isset( $message['content']['text'] ) ? $message['content']['text'] : ""; |
| 529 | $text .= isset( $message['content'] ) && is_string( $message['content'] ) ? $message['content'] : ""; |
| 530 | } |
| 531 | } |
| 532 | else if ( is_string( $arg ) ) { |
| 533 | $text .= $arg; |
| 534 | } |
| 535 | } |
| 536 | $averageTokenLength = 4; |
| 537 | $words = preg_split( '/\s+/', trim( $text ) ); |
| 538 | $tokenCount = 0; |
| 539 | foreach ( $words as $word ) { |
| 540 | $tokenCount += ceil( strlen( $word ) / $averageTokenLength ); |
| 541 | } |
| 542 | return apply_filters( 'mwai_estimate_tokens', $tokenCount, $text ); |
| 543 | } |
| 544 | |
| 545 | public function record_tokens_usage( $model, $in_tokens, $out_tokens = 0 ) { |
| 546 | if ( !is_numeric( $in_tokens ) ) { |
| 547 | throw new Exception( 'AI Engine: in_tokens must be a number.' ); |
| 548 | } |
| 549 | if ( !is_numeric( $out_tokens ) ) { |
| 550 | $out_tokens = 0; |
| 551 | } |
| 552 | if ( !$model ) { |
| 553 | throw new Exception( 'AI Engine: model is required.' ); |
| 554 | } |
| 555 | $usage = $this->get_option( 'openai_usage' ); |
| 556 | $month = date( 'Y-m' ); |
| 557 | if ( !isset( $usage[$month] ) ) { |
| 558 | $usage[$month] = array(); |
| 559 | } |
| 560 | if ( !isset( $usage[$month][$model] ) ) { |
| 561 | $usage[$month][$model] = array( 'prompt_tokens' => 0, 'completion_tokens' => 0, 'total_tokens' => 0 ); |
| 562 | } |
| 563 | $usage[$month][$model]['prompt_tokens'] += $in_tokens; |
| 564 | $usage[$month][$model]['completion_tokens'] += $out_tokens; |
| 565 | $usage[$month][$model]['total_tokens'] += $in_tokens + $out_tokens; |
| 566 | $this->update_option( 'openai_usage', $usage ); |
| 567 | return [ |
| 568 | 'prompt_tokens' => $in_tokens, |
| 569 | 'completion_tokens' => $out_tokens, |
| 570 | 'total_tokens' => $in_tokens + $out_tokens |
| 571 | ]; |
| 572 | } |
| 573 | |
| 574 | public function record_audio_usage( $model, $seconds ) { |
| 575 | if ( !is_numeric( $seconds ) ) { |
| 576 | throw new Exception( 'AI Engine: seconds must be a number.' ); |
| 577 | } |
| 578 | if ( !$model ) { |
| 579 | throw new Exception( 'AI Engine: model is required.' ); |
| 580 | } |
| 581 | $usage = $this->get_option( 'openai_usage' ); |
| 582 | $month = date( 'Y-m' ); |
| 583 | if ( !isset( $usage[$month] ) ) { |
| 584 | $usage[$month] = array(); |
| 585 | } |
| 586 | if ( !isset( $usage[$month][$model] ) ) { |
| 587 | $usage[$month][$model] = array( 'seconds' => 0 ); |
| 588 | } |
| 589 | $usage[$month][$model]['seconds'] += $seconds; |
| 590 | $this->update_option( 'openai_usage', $usage ); |
| 591 | return [ 'seconds' => $seconds ]; |
| 592 | } |
| 593 | |
| 594 | public function record_images_usage( $model, $resolution, $images ) { |
| 595 | if ( !$model || !$resolution || !$images ) { |
| 596 | throw new Exception( 'Missing parameters for record_image_usage.' ); |
| 597 | } |
| 598 | $usage = $this->get_option( 'openai_usage' ); |
| 599 | $month = date( 'Y-m' ); |
| 600 | if ( !isset( $usage[$month] ) ) { |
| 601 | $usage[$month] = array(); |
| 602 | } |
| 603 | if ( !isset( $usage[$month][$model] ) ) { |
| 604 | $usage[$month][$model] = array( 'resolution' => array(), 'images' => 0 ); |
| 605 | } |
| 606 | if ( !isset( $usage[$month][$model]['resolution'][$resolution] ) ) { |
| 607 | $usage[$month][$model]['resolution'][$resolution] = 0; |
| 608 | } |
| 609 | $usage[$month][$model]['resolution'][$resolution] += $images; |
| 610 | $usage[$month][$model]['images'] += $images; |
| 611 | $this->update_option( 'openai_usage', $usage ); |
| 612 | return [ 'resolution' => $resolution, 'images' => $images ]; |
| 613 | } |
| 614 | |
| 615 | #endregion |
| 616 | |
| 617 | #region Streaming |
| 618 | public function stream_push( $data ) { |
| 619 | $out = "data: " . json_encode( $data ); |
| 620 | echo $out; |
| 621 | echo "\n\n"; |
| 622 | if ( ob_get_level() > 0 ) { |
| 623 | ob_end_flush(); |
| 624 | } |
| 625 | flush(); |
| 626 | } |
| 627 | #endregion |
| 628 | |
| 629 | #region Options |
| 630 | function get_themes() { |
| 631 | $themes = get_option( $this->themes_option_name, [] ); |
| 632 | $themes = empty( $themes ) ? [] : $themes; |
| 633 | |
| 634 | $internalThemes = [ |
| 635 | 'chatgpt' => [ |
| 636 | 'type' => 'internal', 'name' => 'ChatGPT', 'themeId' => 'chatgpt', |
| 637 | 'settings' => [], 'style' => "" |
| 638 | ], |
| 639 | 'messages' => [ |
| 640 | 'type' => 'internal', 'name' => 'Messages', 'themeId' => 'messages', |
| 641 | 'settings' => [], 'style' => "" |
| 642 | ], |
| 643 | ]; |
| 644 | $customThemes = []; |
| 645 | foreach ( $themes as $theme ) { |
| 646 | if ( isset( $internalThemes[$theme['themeId']] ) ) { |
| 647 | $internalThemes[$theme['themeId']] = $theme; |
| 648 | continue; |
| 649 | } |
| 650 | $customThemes[] = $theme; |
| 651 | } |
| 652 | return array_merge(array_values($internalThemes), $customThemes); |
| 653 | } |
| 654 | |
| 655 | function update_themes( $themes ) { |
| 656 | update_option( $this->themes_option_name, $themes ); |
| 657 | return $themes; |
| 658 | } |
| 659 | |
| 660 | function get_chatbots() { |
| 661 | $chatbots = get_option( $this->chatbots_option_name, [] ); |
| 662 | $hasChanges = false; |
| 663 | if ( empty( $chatbots ) ) { |
| 664 | $chatbots = [ array_merge( MWAI_CHATBOT_DEFAULT_PARAMS, ['name' => 'Default', 'botId' => 'default' ] ) ]; |
| 665 | } |
| 666 | foreach ( $chatbots as &$chatbot ) { |
| 667 | foreach ( MWAI_CHATBOT_DEFAULT_PARAMS as $key => $value ) { |
| 668 | // Use default value if not set. |
| 669 | if ( !isset( $chatbot[$key] ) ) { |
| 670 | $chatbot[$key] = $value; |
| 671 | } |
| 672 | } |
| 673 | // After September 2023, let's remove this if statement. |
| 674 | // if ( isset( $chatbot['chatId'] ) ) { |
| 675 | // $chatbot['botId'] = $chatbot['chatId']; |
| 676 | // unset( $chatbot['chatId'] ); |
| 677 | // $hasChanges = true; |
| 678 | // } |
| 679 | // After September 2023, let's remove this if statement. |
| 680 | // if ( empty( $chatbot['botId'] ) && $chatbot['name'] === 'default' ) { |
| 681 | // $chatbot['botId'] = sanitize_title( $chatbot['name'] ); |
| 682 | // $hasChanges = true; |
| 683 | // } |
| 684 | } |
| 685 | if ( $hasChanges ) { |
| 686 | update_option( $this->chatbots_option_name, $chatbots ); |
| 687 | } |
| 688 | return $chatbots; |
| 689 | } |
| 690 | |
| 691 | function get_chatbot( $botId ) { |
| 692 | $chatbots = $this->get_chatbots(); |
| 693 | foreach ( $chatbots as $chatbot ) { |
| 694 | if ( $chatbot['botId'] === (string)$botId ) { |
| 695 | return $chatbot; |
| 696 | } |
| 697 | } |
| 698 | return null; |
| 699 | } |
| 700 | |
| 701 | function get_embeddings_env( $envId ) { |
| 702 | $envs = $this->get_option( 'embeddings_envs' ); |
| 703 | foreach ( $envs as $env ) { |
| 704 | if ( $env['id'] === $envId ) { |
| 705 | return $env; |
| 706 | } |
| 707 | } |
| 708 | return null; |
| 709 | } |
| 710 | |
| 711 | function get_ai_env( $envId ) { |
| 712 | $envs = $this->get_option( 'ai_envs' ); |
| 713 | foreach ( $envs as $env ) { |
| 714 | if ( $env['id'] === $envId ) { |
| 715 | return $env; |
| 716 | } |
| 717 | } |
| 718 | return null; |
| 719 | } |
| 720 | |
| 721 | function get_assistant( $envId, $assistantId ) { |
| 722 | $env = $this->get_ai_env( $envId ); |
| 723 | if ( !$env ) { |
| 724 | return null; |
| 725 | } |
| 726 | $assistants = $env['assistants']; |
| 727 | foreach ( $assistants as $assistant ) { |
| 728 | if ( $assistant['id'] === $assistantId ) { |
| 729 | return $assistant; |
| 730 | } |
| 731 | } |
| 732 | return null; |
| 733 | } |
| 734 | |
| 735 | function get_theme( $themeId ) { |
| 736 | $themes = $this->get_themes(); |
| 737 | foreach ( $themes as $theme ) { |
| 738 | if ( $theme['themeId'] === $themeId ) { |
| 739 | return $theme; |
| 740 | } |
| 741 | } |
| 742 | return null; |
| 743 | } |
| 744 | |
| 745 | function update_chatbots( $chatbots ) { |
| 746 | $htmlFields = [ 'textCompliance', 'aiName', 'userName', 'startSentence' ]; |
| 747 | $whiteSpacedFields = [ 'context' ]; |
| 748 | foreach ( $chatbots as &$chatbot ) { |
| 749 | foreach ( $chatbot as $key => &$value ) { |
| 750 | if ( in_array( $key, $htmlFields ) ) { |
| 751 | $value = wp_kses_post( $value ); |
| 752 | } |
| 753 | else if ( in_array( $key, $whiteSpacedFields ) ) { |
| 754 | $value = sanitize_textarea_field( $value ); |
| 755 | } |
| 756 | else { |
| 757 | $value = sanitize_text_field( $value ); |
| 758 | } |
| 759 | } |
| 760 | } |
| 761 | |
| 762 | update_option( $this->chatbots_option_name, $chatbots ); |
| 763 | return $chatbots; |
| 764 | } |
| 765 | |
| 766 | function get_all_options( $force = false ) { |
| 767 | // We could cache options this way, but if we do, the apply_filters seems to be called too early. |
| 768 | // That causes issues with the mwai_languages filter. |
| 769 | // if ( !$force && !is_null( $this->options ) ) { |
| 770 | // return $this->options; |
| 771 | // } |
| 772 | $options = get_option( $this->option_name, [] ); |
| 773 | $options = $this->sanitize_options( $options ); |
| 774 | foreach ( MWAI_OPTIONS as $key => $value ) { |
| 775 | if ( !isset( $options[$key] ) ) { |
| 776 | $options[$key] = $value; |
| 777 | } |
| 778 | if ( $key === 'languages' ) { |
| 779 | // NOTE: If we decide to make a set of options for languages, we can keep it in the settings |
| 780 | $options[$key] = apply_filters( 'mwai_languages', MWAI_LANGUAGES ); |
| 781 | } |
| 782 | } |
| 783 | $options['chatbot_defaults'] = MWAI_CHATBOT_DEFAULT_PARAMS; |
| 784 | $options['default_limits'] = MWAI_LIMITS; |
| 785 | $options['openai_models'] = apply_filters( 'mwai_openai_models', Meow_MWAI_Engines_OpenAI::get_models_static() ); |
| 786 | $options['fallback_model'] = MWAI_FALLBACK_MODEL; |
| 787 | |
| 788 | //$this->options = $options; |
| 789 | return $options; |
| 790 | } |
| 791 | |
| 792 | // Sanitize options when we update the plugi or perform some updates |
| 793 | // if we change the structure of the options. |
| 794 | function sanitize_options( $options ) { |
| 795 | $needs_update = false; |
| 796 | |
| 797 | // This list was updated on December 11, 2023. After May 2024, let's remove this. |
| 798 | $old_options = [ |
| 799 | 'shortcode_chat_default_params', |
| 800 | 'shortcode_chat_params_override', |
| 801 | 'module_legacy_finetunes', |
| 802 | 'shortcode_chat_legacy', |
| 803 | 'shortcode_chat_inject', |
| 804 | 'shortcode_chat_styles', |
| 805 | 'dynamic_max_tokens', |
| 806 | 'shortcode_chat_formatting', |
| 807 | 'shortcode_forms_legacy', |
| 808 | ]; |
| 809 | foreach ( $old_options as $old_option ) { |
| 810 | if ( isset( $options[$old_option] ) ) { |
| 811 | unset( $options[$old_option] ); |
| 812 | $needs_update = true; |
| 813 | } |
| 814 | } |
| 815 | |
| 816 | // This upgrades namespace to multi-namespaces (June 2023) |
| 817 | // After January 2024, let's remove this. |
| 818 | if ( isset( $options['pinecone'] ) && isset( $options['pinecone']['namespace'] ) ) { |
| 819 | $options['pinecone']['namespaces'] = [ $options['pinecone']['namespace'] ]; |
| 820 | unset( $options['pinecone']['namespace'] ); |
| 821 | $needs_update = true; |
| 822 | } |
| 823 | // Support for Multi Vector DB Environments |
| 824 | // After June 2024, let's remove this. |
| 825 | if ( !isset( $options['embeddings_envs'] ) ) { |
| 826 | $options['embeddings_envs'] = []; |
| 827 | $default_id = $this->get_random_id(); |
| 828 | $pinecone = isset( $options['pinecone'] ) ? $options['pinecone'] : []; |
| 829 | $options['embeddings_envs'][] = [ |
| 830 | 'id' => $default_id, |
| 831 | 'name' => 'Pinecone', |
| 832 | 'type' => 'pinecone', |
| 833 | 'apikey' => isset( $pinecone['apikey'] ) ? $pinecone['apikey'] : '', |
| 834 | 'server' => isset( $pinecone['server'] ) ? $pinecone['server'] : 'gcp-starter', |
| 835 | 'indexes' => isset( $pinecone['indexes'] ) ? $pinecone['indexes'] : [], |
| 836 | 'namespaces' => isset( $pinecone['namespaces'] ) ? $pinecone['namespaces'] : [], |
| 837 | 'index' => isset( $pinecone['index'] ) ? $pinecone['index'] : null, |
| 838 | ]; |
| 839 | $options['embeddings_default_env'] = $default_id; |
| 840 | $needs_update = true; |
| 841 | } |
| 842 | if ( isset( $options['pinecone'] ) ) { |
| 843 | unset( $options['pinecone'] ); |
| 844 | $needs_update = true; |
| 845 | } |
| 846 | // Support for Multi AI Environments |
| 847 | // After June 2024, let's remove this. |
| 848 | if ( !isset( $options['ai_envs'] ) ) { |
| 849 | $options['ai_envs'] = []; |
| 850 | $default_openai_id = $this->get_random_id(); |
| 851 | $default_azure_id = $this->get_random_id(); |
| 852 | $openai_service = isset( $options['openai_service'] ) ? $options['openai_service'] : 'openai'; |
| 853 | $openai_apikey = isset( $options['openai_apikey'] ) ? $options['openai_apikey'] : ''; |
| 854 | $azure_endpoint = isset( $options['openai_azure_endpoint'] ) ? $options['openai_azure_endpoint'] : ''; |
| 855 | |
| 856 | // OpenAI |
| 857 | // We create a default OpenAI environment if the API Key is set, or if the Azure Endpoint is not set. |
| 858 | if ( !empty( $openai_apikey ) || empty( $azure_endpoint ) ) { |
| 859 | $openai_finetunes = isset( $options['openai_finetunes'] ) ? $options['openai_finetunes'] : []; |
| 860 | $openai_finetunes_deleted = isset( $options['openai_finetunes_deleted'] ) ? |
| 861 | $options['openai_finetunes_deleted'] : []; |
| 862 | $openai_legacy_finetunes = isset( $options['openai_legacy_finetunes'] ) ? |
| 863 | $options['openai_legacy_finetunes'] : []; |
| 864 | $openai_legacy_finetunes_deleted = isset( $options['openai_legacy_finetunes_deleted'] ) ? |
| 865 | $options['openai_legacy_finetunes_deleted'] : []; |
| 866 | $options['ai_envs'][] = [ |
| 867 | 'id' => $default_openai_id, |
| 868 | 'name' => 'OpenAI', |
| 869 | 'type' => 'openai', |
| 870 | 'apikey' => $openai_apikey, |
| 871 | 'finetunes' => $openai_finetunes, |
| 872 | 'finetunes_deleted' => $openai_finetunes_deleted, |
| 873 | 'legacy_finetunes' => $openai_legacy_finetunes, |
| 874 | 'legacy_finetunes_deleted' => $openai_legacy_finetunes_deleted |
| 875 | ]; |
| 876 | } |
| 877 | |
| 878 | // Azure |
| 879 | if ( !empty( $azure_endpoint ) ) { |
| 880 | $azure_apikey = isset( $options['openai_azure_apikey'] ) ? $options['openai_azure_apikey'] : ''; |
| 881 | $azure_deployments = isset( $options['openai_azure_deployments'] ) ? $options['openai_azure_deployments'] : []; |
| 882 | $options['ai_envs'][] = [ |
| 883 | 'id' => $default_azure_id, |
| 884 | 'name' => 'Azure', |
| 885 | 'type' => 'azure', |
| 886 | 'apikey' => $azure_apikey, |
| 887 | 'endpoint' => $azure_endpoint, |
| 888 | 'deployments' => $azure_deployments, |
| 889 | ]; |
| 890 | } |
| 891 | |
| 892 | $options['ai_default_env'] = $default_openai_id; |
| 893 | if ( $openai_service === 'azure' ) { |
| 894 | $options['ai_default_env'] = $default_azure_id; |
| 895 | } |
| 896 | $needs_update = true; |
| 897 | } |
| 898 | if ( !empty( $options['openai_apikey'] ) || !empty( $options['openai_azure_apikey'] ) ) { |
| 899 | unset( $options['openai_apikey'] ); |
| 900 | unset( $options['openai_finetunes'] ); |
| 901 | unset( $options['openai_finetunes_deleted'] ); |
| 902 | unset( $options['openai_legacy_finetunes'] ); |
| 903 | unset( $options['openai_legacy_finetunes_deleted'] ); |
| 904 | unset( $options['openai_azure_apikey'] ); |
| 905 | unset( $options['openai_azure_endpoint'] ); |
| 906 | unset( $options['openai_azure_deployments'] ); |
| 907 | unset( $options['openai_service'] ); |
| 908 | $needs_update = true; |
| 909 | } |
| 910 | |
| 911 | // The IDs for the embeddings environments are generated here. |
| 912 | // TODO: We should handle this more gracefully via an option in the Embeddings Settings. |
| 913 | $embeddings_default_exists = false; |
| 914 | if ( isset( $options['embeddings_envs'] ) ) { |
| 915 | foreach ( $options['embeddings_envs'] as &$env ) { |
| 916 | if ( !isset( $env['id'] ) ) { |
| 917 | $env['id'] = $this->get_random_id(); |
| 918 | $needs_update = true; |
| 919 | } |
| 920 | if ( $env['id'] === $options['embeddings_default_env'] ) { |
| 921 | $embeddings_default_exists = true; |
| 922 | } |
| 923 | } |
| 924 | } |
| 925 | if ( !$embeddings_default_exists ) { |
| 926 | $options['embeddings_default_env'] = $options['embeddings_envs'][0]['id'] ?? null; |
| 927 | $needs_update = true; |
| 928 | } |
| 929 | |
| 930 | // The IDs for the AI environments are generated here. |
| 931 | $ai_default_exists = false; |
| 932 | if ( isset( $options['ai_envs'] ) ) { |
| 933 | foreach ( $options['ai_envs'] as &$env ) { |
| 934 | if ( !isset( $env['id'] ) ) { |
| 935 | $env['id'] = $this->get_random_id(); |
| 936 | $needs_update = true; |
| 937 | } |
| 938 | if ( $env['id'] === $options['ai_default_env'] ) { |
| 939 | $ai_default_exists = true; |
| 940 | } |
| 941 | } |
| 942 | } |
| 943 | if ( !$ai_default_exists ) { |
| 944 | $options['ai_default_env'] = $options['ai_envs'][0]['id'] ?? null; |
| 945 | $needs_update = true; |
| 946 | } |
| 947 | |
| 948 | if ( $needs_update ) { |
| 949 | update_option( $this->option_name, $options, false ); |
| 950 | } |
| 951 | |
| 952 | return $options; |
| 953 | } |
| 954 | |
| 955 | function update_options( $options ) { |
| 956 | if ( !update_option( $this->option_name, $options, false ) ) { |
| 957 | return false; |
| 958 | } |
| 959 | $options = $this->get_all_options( true ); |
| 960 | return $options; |
| 961 | } |
| 962 | |
| 963 | function update_option( $option, $value ) { |
| 964 | $options = $this->get_all_options( true ); |
| 965 | $options[$option] = $value; |
| 966 | return $this->update_options( $options ); |
| 967 | } |
| 968 | |
| 969 | function get_option( $option, $default = null ) { |
| 970 | $options = $this->get_all_options(); |
| 971 | return $options[$option] ?? $default; |
| 972 | } |
| 973 | |
| 974 | function update_ai_env( $env_id, $option, $value ) { |
| 975 | $options = $this->get_all_options( true ); |
| 976 | foreach ( $options['ai_envs'] as &$env ) { |
| 977 | if ( $env['id'] === $env_id ) { |
| 978 | $env[$option] = $value; |
| 979 | break; |
| 980 | } |
| 981 | } |
| 982 | return $this->update_options( $options ); |
| 983 | } |
| 984 | |
| 985 | function reset_options() { |
| 986 | delete_option( $this->themes_option_name ); |
| 987 | delete_option( $this->chatbots_option_name ); |
| 988 | delete_option( $this->option_name ); |
| 989 | return $this->get_all_options( true ); |
| 990 | } |
| 991 | #endregion |
| 992 | } |
| 993 | |
| 994 | ?> |