engines
3 years ago
modules
3 years ago
admin.php
3 years ago
api.php
3 years ago
core.php
3 years ago
init.php
3 years ago
query.php
3 years ago
queryembed.php
3 years ago
queryimage.php
3 years ago
querytext.php
3 years ago
querytranscribe.php
3 years ago
reply.php
3 years ago
rest.php
3 years ago
core.php
494 lines
| 1 | <?php |
| 2 | |
| 3 | require_once( MWAI_PATH . '/vendor/autoload.php' ); |
| 4 | require_once( MWAI_PATH . '/constants/init.php' ); |
| 5 | |
| 6 | use Rahul900day\Gpt3Encoder\Encoder; |
| 7 | |
| 8 | define( 'MWAI_IMG_WAND', MWAI_URL . '/images/wand.png' ); |
| 9 | define( 'MWAI_IMG_WAND_HTML', "<img style='height: 22px; margin-bottom: -5px; margin-right: 8px;' |
| 10 | src='" . MWAI_IMG_WAND . "' alt='AI Wand' />" ); |
| 11 | define( 'MWAI_IMG_WAND_HTML_XS', "<img style='height: 16px; margin-bottom: -2px;' |
| 12 | src='" . MWAI_IMG_WAND . "' alt='AI Wand' />" ); |
| 13 | class Meow_MWAI_Core |
| 14 | { |
| 15 | public $admin = null; |
| 16 | public $is_rest = false; |
| 17 | public $is_cli = false; |
| 18 | public $site_url = null; |
| 19 | public $ai = null; |
| 20 | private $option_name = 'mwai_options'; |
| 21 | private $themes_option_name = 'mwai_themes'; |
| 22 | private $chatbots_option_name = 'mwai_chatbots'; |
| 23 | public $defaultChatbotParams = MWAI_CHATBOT_PARAMS; |
| 24 | |
| 25 | public function __construct() { |
| 26 | $this->site_url = get_site_url(); |
| 27 | $this->is_rest = MeowCommon_Helpers::is_rest(); |
| 28 | $this->is_cli = defined( 'WP_CLI' ); |
| 29 | $this->ai = new Meow_MWAI_Engines_Core( $this ); |
| 30 | add_action( 'plugins_loaded', array( $this, 'init' ) ); |
| 31 | } |
| 32 | |
| 33 | function init() { |
| 34 | global $mwai; |
| 35 | $mwai = new Meow_MWAI_API(); |
| 36 | new Meow_MWAI_Modules_Security( $this ); |
| 37 | if ( $this->is_rest ) { |
| 38 | new Meow_MWAI_Rest( $this ); |
| 39 | } |
| 40 | if ( is_admin() ) { |
| 41 | new Meow_MWAI_Admin( $this ); |
| 42 | new Meow_MWAI_Modules_Assistants( $this ); |
| 43 | } |
| 44 | if ( $this->get_option( 'shortcode_chat' ) ) { |
| 45 | new Meow_MWAI_Modules_Chatbot(); |
| 46 | new Meow_MWAI_Modules_Chatbot_Legacy(); |
| 47 | new Meow_MWAI_Modules_Discussions(); |
| 48 | } |
| 49 | |
| 50 | // Advanced core |
| 51 | if ( class_exists( 'MeowPro_MWAI_Core' ) ) { |
| 52 | new MeowPro_MWAI_Core( $this ); |
| 53 | } |
| 54 | |
| 55 | // Dynamic max tokens |
| 56 | if ( $this->get_option( 'dynamic_max_tokens' ) ) { |
| 57 | add_filter( 'mwai_estimate_tokens', array( $this, 'dynamic_max_tokens' ), 10, 2 ); |
| 58 | } |
| 59 | } |
| 60 | |
| 61 | #region Roles & Capabilities |
| 62 | |
| 63 | function can_access_settings() { |
| 64 | return apply_filters( 'mwai_allow_setup', current_user_can( 'manage_options' ) ); |
| 65 | } |
| 66 | |
| 67 | function can_access_features() { |
| 68 | $editor_or_admin = current_user_can( 'editor' ) || current_user_can( 'administrator' ); |
| 69 | return apply_filters( 'mwai_allow_usage', $editor_or_admin ); |
| 70 | } |
| 71 | |
| 72 | #endregion |
| 73 | |
| 74 | #region Text-Related Helpers |
| 75 | |
| 76 | // Clean the text perfectly, resolve shortcodes, etc, etc. |
| 77 | function cleanText( $rawText = "" ) { |
| 78 | $text = html_entity_decode( $rawText ); |
| 79 | $text = wp_strip_all_tags( $text ); |
| 80 | $text = preg_replace( '/[\r\n]+/', "\n", $text ); |
| 81 | return $text . " "; |
| 82 | |
| 83 | // Before simplification: |
| 84 | // $text = strip_tags( $rawText ); |
| 85 | // $text = strip_shortcodes( $text ); |
| 86 | // $text = html_entity_decode( $text ); |
| 87 | // $text = preg_replace( '/[\r\n]+/', "\n", $text ); |
| 88 | // $sentences = preg_split( '/(?<=[.?!])(?=[a-zA-Z ])/', $text ); |
| 89 | // foreach ( $sentences as $key => $sentence ) { |
| 90 | // $sentences[$key] = trim( $sentence ); |
| 91 | // } |
| 92 | // $text = implode( " ", $sentences ); |
| 93 | // $text = preg_replace( '/^[\pZ\pC]+|[\pZ\pC]+$/u', '', $text ); |
| 94 | // return $text . " "; |
| 95 | } |
| 96 | |
| 97 | // Make sure there are no duplicate sentences, and keep the length under a maximum length. |
| 98 | function cleanSentences( $text, $maxTokens = null ) { |
| 99 | //$sentences = preg_split( '/(?<=[.?!])(?=[a-zA-Z ])/', $text ); |
| 100 | $maxTokens = $maxTokens ? $maxTokens : $this->get_option( 'context_max_tokens', 1024 ); |
| 101 | $sentences = preg_split('/(?<=[.?!。．！？])+/u', $text); |
| 102 | $hashes = array(); |
| 103 | $uniqueSentences = array(); |
| 104 | $length = 0; |
| 105 | foreach ( $sentences as $sentence ) { |
| 106 | $sentence = preg_replace( '/^[\pZ\pC]+|[\pZ\pC]+$/u', '', $sentence ); |
| 107 | $hash = md5( $sentence ); |
| 108 | if ( !in_array( $hash, $hashes ) ) { |
| 109 | $tokensCount = apply_filters( 'mwai_estimate_tokens', 0, $sentence ); |
| 110 | if ( $length + $tokensCount > $maxTokens ) { |
| 111 | continue; |
| 112 | } |
| 113 | $hashes[] = $hash; |
| 114 | $uniqueSentences[] = $sentence; |
| 115 | $length += $tokensCount; |
| 116 | } |
| 117 | } |
| 118 | $freshText = implode( " ", $uniqueSentences ); |
| 119 | $freshText = preg_replace( '/^[\pZ\pC]+|[\pZ\pC]+$/u', '', $freshText ); |
| 120 | return $freshText; |
| 121 | } |
| 122 | |
| 123 | function getCleanPostContent( $postId ) { |
| 124 | $post = get_post( $postId ); |
| 125 | if ( !$post ) { |
| 126 | return false; |
| 127 | } |
| 128 | $text = $post->post_content; |
| 129 | $pattern = '/\[mwai_.*?\]/'; |
| 130 | $text = preg_replace( $pattern, '', $text ); |
| 131 | if ( $this->get_option( 'resolve_shortcodes' ) ) { |
| 132 | $text = apply_filters( 'the_content', $text ); |
| 133 | } |
| 134 | $text = $this->cleanText( $text ); |
| 135 | $text = $this->cleanSentences( $text ); |
| 136 | return $text; |
| 137 | } |
| 138 | |
| 139 | function markdown_to_html( $content ) { |
| 140 | $Parsedown = new Parsedown(); |
| 141 | $content = $Parsedown->text( $content ); |
| 142 | return $content; |
| 143 | } |
| 144 | |
| 145 | function get_post_language( $postId ) { |
| 146 | $locale = get_locale(); |
| 147 | $code = strtolower( substr( $locale, 0, 2 ) ); |
| 148 | $humanLanguage = strtr( $code, MWAI_ALL_LANGUAGES ); |
| 149 | $lang = apply_filters( 'wpml_post_language_details', null, $postId ); |
| 150 | if ( !empty( $lang ) ) { |
| 151 | $locale = $lang['locale']; |
| 152 | $humanLanguage = $lang['display_name']; |
| 153 | } |
| 154 | return strtolower( "$locale ($humanLanguage)" ); |
| 155 | } |
| 156 | #endregion |
| 157 | |
| 158 | #region Users/Sessions Helpers |
| 159 | |
| 160 | function get_session_id() { |
| 161 | if ( isset( $_COOKIE['mwai_session_id'] ) ) { |
| 162 | return $_COOKIE['mwai_session_id']; |
| 163 | } |
| 164 | return "N/A"; |
| 165 | } |
| 166 | |
| 167 | // Get the UserID from the data, or from the current user |
| 168 | function get_user_id( $data = null ) { |
| 169 | if ( isset( $data ) && isset( $data['userId'] ) ) { |
| 170 | return (int)$data['userId']; |
| 171 | } |
| 172 | if ( is_user_logged_in() ) { |
| 173 | $current_user = wp_get_current_user(); |
| 174 | if ( $current_user->ID > 0 ) { |
| 175 | return $current_user->ID; |
| 176 | } |
| 177 | } |
| 178 | return null; |
| 179 | } |
| 180 | |
| 181 | function getUserData() { |
| 182 | $user = wp_get_current_user(); |
| 183 | if ( empty( $user ) || empty( $user->ID ) ) { |
| 184 | return null; |
| 185 | } |
| 186 | $placeholders = array( |
| 187 | 'FIRST_NAME' => get_user_meta( $user->ID, 'first_name', true ), |
| 188 | 'LAST_NAME' => get_user_meta( $user->ID, 'last_name', true ), |
| 189 | 'USER_LOGIN' => isset( $user ) && isset($user->data) && isset( $user->data->user_login ) ? |
| 190 | $user->data->user_login : null, |
| 191 | 'DISPLAY_NAME' => isset( $user ) && isset( $user->data ) && isset( $user->data->display_name ) ? |
| 192 | $user->data->display_name : null, |
| 193 | 'AVATAR_URL' => get_avatar_url( get_current_user_id() ), |
| 194 | ); |
| 195 | return $placeholders; |
| 196 | } |
| 197 | |
| 198 | function get_ip_address( $data = null ) { |
| 199 | if ( isset( $data ) && isset( $data['ip'] ) ) { |
| 200 | $data['ip'] = (string)$data['ip']; |
| 201 | } |
| 202 | else { |
| 203 | if ( isset( $_SERVER['REMOTE_ADDR'] ) ) { |
| 204 | $data['ip'] = sanitize_text_field( $_SERVER['REMOTE_ADDR'] ); |
| 205 | } |
| 206 | else if ( isset( $_SERVER['HTTP_CLIENT_IP'] ) ) { |
| 207 | $data['ip'] = sanitize_text_field( $_SERVER['HTTP_CLIENT_IP'] ); |
| 208 | } |
| 209 | else if ( isset( $_SERVER['HTTP_X_FORWARDED_ FOR'] ) ) { |
| 210 | $data['ip'] = sanitize_text_field( $_SERVER['HTTP_X_FORWARDED_FOR'] ); |
| 211 | } |
| 212 | } |
| 213 | $ip = apply_filters( 'mwai_get_ip_address', $data['ip'] ); |
| 214 | return $ip; |
| 215 | } |
| 216 | |
| 217 | #endregion |
| 218 | |
| 219 | #region Other Helpers |
| 220 | |
| 221 | function isUrl( $url ) { |
| 222 | return strpos( $url, 'http' ) === 0 ? true : false; |
| 223 | } |
| 224 | |
| 225 | function getPostTypes() { |
| 226 | $excluded = array( 'attachment', 'revision', 'nav_menu_item' ); |
| 227 | $post_types = array(); |
| 228 | $types = get_post_types( array( 'public' => true ), 'objects' ); |
| 229 | foreach ( $types as $type ) { |
| 230 | if ( in_array( $type->name, $excluded ) ) { |
| 231 | continue; |
| 232 | } |
| 233 | $post_types[] = array( |
| 234 | 'name' => $type->labels->name, |
| 235 | 'type' => $type->name, |
| 236 | ); |
| 237 | } |
| 238 | return $post_types; |
| 239 | } |
| 240 | |
| 241 | function getCleanPost( $post ) { |
| 242 | if ( is_object( $post ) ) { |
| 243 | $post = (array)$post; |
| 244 | } |
| 245 | $language = $this->get_post_language( $post['ID'] ); |
| 246 | $content = apply_filters( 'mwai_pre_post_content', $post['post_content'], $post['ID'] ); |
| 247 | $content = $this->cleanText( $content ); |
| 248 | $content = apply_filters( 'mwai_post_content', $content, $post['ID'] ); |
| 249 | $title = $post['post_title']; |
| 250 | $excerpt = $post['post_excerpt']; |
| 251 | $url = get_permalink( $post['ID'] ); |
| 252 | $checksum = wp_hash( $content . $title . $url ); |
| 253 | return [ |
| 254 | 'postId' => $post['ID'], |
| 255 | 'title' => $title, |
| 256 | 'content' => $content, |
| 257 | 'excerpt' => $excerpt, |
| 258 | 'url' => $url, |
| 259 | 'language' => $language, |
| 260 | 'checksum' => $checksum, |
| 261 | ]; |
| 262 | } |
| 263 | |
| 264 | #endregion |
| 265 | |
| 266 | #region Usage & Costs |
| 267 | |
| 268 | public function dynamic_max_tokens( $tokens, $text ) { |
| 269 | // Approximation (fast, no lib) |
| 270 | $asciiCount = 0; |
| 271 | $nonAsciiCount = 0; |
| 272 | for ( $i = 0; $i < mb_strlen( $text ); $i++ ) { |
| 273 | $char = mb_substr( $text, $i, 1 ); |
| 274 | if ( ord( $char ) < 128 ) { |
| 275 | $asciiCount++; |
| 276 | } |
| 277 | else { |
| 278 | $nonAsciiCount++; |
| 279 | } |
| 280 | } |
| 281 | $asciiTokens = $asciiCount / 3.5; |
| 282 | $nonAsciiTokens = $nonAsciiCount * 2.5; |
| 283 | $tokens = $asciiTokens + $nonAsciiTokens; |
| 284 | |
| 285 | // More exact (slower, and lib) |
| 286 | if ( PHP_VERSION_ID >= 70400 && function_exists( 'mb_convert_encoding' ) ) { |
| 287 | try { |
| 288 | $token_array = Encoder::encode( $text ); |
| 289 | if ( !empty( $token_array ) ) { |
| 290 | $tokens = count( $token_array ); |
| 291 | } |
| 292 | } |
| 293 | catch ( Exception $e ) { |
| 294 | error_log( $e->getMessage() ); |
| 295 | } |
| 296 | } |
| 297 | |
| 298 | $tokens = $tokens; |
| 299 | return (int)$tokens; |
| 300 | } |
| 301 | |
| 302 | public function record_tokens_usage( $model, $prompt_tokens, $completion_tokens = 0 ) { |
| 303 | if ( !is_numeric( $prompt_tokens ) ) { |
| 304 | throw new Exception( 'Record usage: prompt_tokens is not a number.' ); |
| 305 | } |
| 306 | if ( !is_numeric( $completion_tokens ) ) { |
| 307 | $completion_tokens = 0; |
| 308 | } |
| 309 | if ( !$model ) { |
| 310 | throw new Exception( 'Record usage: model is missing.' ); |
| 311 | } |
| 312 | $usage = $this->get_option( 'openai_usage' ); |
| 313 | $month = date( 'Y-m' ); |
| 314 | if ( !isset( $usage[$month] ) ) { |
| 315 | $usage[$month] = array(); |
| 316 | } |
| 317 | if ( !isset( $usage[$month][$model] ) ) { |
| 318 | $usage[$month][$model] = array( |
| 319 | 'prompt_tokens' => 0, |
| 320 | 'completion_tokens' => 0, |
| 321 | 'total_tokens' => 0 |
| 322 | ); |
| 323 | } |
| 324 | $usage[$month][$model]['prompt_tokens'] += $prompt_tokens; |
| 325 | $usage[$month][$model]['completion_tokens'] += $completion_tokens; |
| 326 | $usage[$month][$model]['total_tokens'] += $prompt_tokens + $completion_tokens; |
| 327 | $this->update_option( 'openai_usage', $usage ); |
| 328 | return [ |
| 329 | 'prompt_tokens' => $prompt_tokens, |
| 330 | 'completion_tokens' => $completion_tokens, |
| 331 | 'total_tokens' => $prompt_tokens + $completion_tokens |
| 332 | ]; |
| 333 | } |
| 334 | |
| 335 | public function record_images_usage( $model, $resolution, $images ) { |
| 336 | if ( !$model || !$resolution || !$images ) { |
| 337 | throw new Exception( 'Missing parameters for record_image_usage.' ); |
| 338 | } |
| 339 | $usage = $this->get_option( 'openai_usage' ); |
| 340 | $month = date( 'Y-m' ); |
| 341 | if ( !isset( $usage[$month] ) ) { |
| 342 | $usage[$month] = array(); |
| 343 | } |
| 344 | if ( !isset( $usage[$month][$model] ) ) { |
| 345 | $usage[$month][$model] = array( |
| 346 | 'resolution' => array(), |
| 347 | 'images' => 0 |
| 348 | ); |
| 349 | } |
| 350 | if ( !isset( $usage[$month][$model]['resolution'][$resolution] ) ) { |
| 351 | $usage[$month][$model]['resolution'][$resolution] = 0; |
| 352 | } |
| 353 | $usage[$month][$model]['resolution'][$resolution] += $images; |
| 354 | $usage[$month][$model]['images'] += $images; |
| 355 | $this->update_option( 'openai_usage', $usage ); |
| 356 | return [ |
| 357 | 'resolution' => $resolution, |
| 358 | 'images' => $images |
| 359 | ]; |
| 360 | } |
| 361 | |
| 362 | #endregion |
| 363 | |
| 364 | #region Options |
| 365 | function getThemes() |
| 366 | { |
| 367 | $themes = get_option( $this->themes_option_name, [] ); |
| 368 | $themes = empty( $themes ) ? [] : $themes; |
| 369 | |
| 370 | $internalThemes = [ |
| 371 | 'chatgpt' => [ |
| 372 | 'type' => 'internal', 'name' => 'ChatGPT', 'themeId' => 'chatgpt', |
| 373 | 'settings' => [], 'style' => "" |
| 374 | ], |
| 375 | 'messages' => [ |
| 376 | 'type' => 'internal', 'name' => 'Messages', 'themeId' => 'messages', |
| 377 | 'settings' => [], 'style' => "" |
| 378 | ], |
| 379 | ]; |
| 380 | $customThemes = []; |
| 381 | foreach ( $themes as $theme ) { |
| 382 | if ( isset( $internalThemes[$theme['themeId']] ) ) { |
| 383 | $internalThemes[$theme['themeId']] = $theme; |
| 384 | continue; |
| 385 | } |
| 386 | $customThemes[] = $theme; |
| 387 | } |
| 388 | return array_merge(array_values($internalThemes), $customThemes); |
| 389 | } |
| 390 | |
| 391 | function updateThemes( $themes ) { |
| 392 | update_option( $this->themes_option_name, $themes ); |
| 393 | return $themes; |
| 394 | } |
| 395 | |
| 396 | function getChatbots() { |
| 397 | $chatbots = get_option( $this->chatbots_option_name, [] ); |
| 398 | if ( empty( $chatbots ) ) { |
| 399 | $chatbots = [ array_merge( MWAI_CHATBOT_DEFAULT_PARAMS, ['name' => 'Default', 'chatId' => 'default' ] ) ]; |
| 400 | } |
| 401 | foreach ( $chatbots as &$chatbot ) { |
| 402 | foreach ( MWAI_CHATBOT_DEFAULT_PARAMS as $key => $value ) { |
| 403 | if ( !isset( $chatbot[$key] ) ) { |
| 404 | $chatbot[$key] = $value; |
| 405 | } |
| 406 | } |
| 407 | } |
| 408 | return $chatbots; |
| 409 | } |
| 410 | |
| 411 | function getChatbot( $chatId ) { |
| 412 | $chatbots = $this->getChatbots(); |
| 413 | foreach ( $chatbots as $chatbot ) { |
| 414 | if ( $chatbot['chatId'] === $chatId ) { |
| 415 | // Somehow, the default was set to "openai" when creating a new chatbot, but that overrided |
| 416 | // the default value in the Settings. It should be always empty here (except if we add this |
| 417 | // into the Settings of the chatbot). |
| 418 | $chatbot['service'] = null; |
| 419 | return $chatbot; |
| 420 | } |
| 421 | } |
| 422 | return null; |
| 423 | } |
| 424 | |
| 425 | function getTheme( $themeId ) { |
| 426 | $themes = $this->getThemes(); |
| 427 | foreach ( $themes as $theme ) { |
| 428 | if ( $theme['themeId'] === $themeId ) { |
| 429 | return $theme; |
| 430 | } |
| 431 | } |
| 432 | return null; |
| 433 | } |
| 434 | |
| 435 | function updateChatbots( $chatbots ) { |
| 436 | update_option( $this->chatbots_option_name, $chatbots ); |
| 437 | return $chatbots; |
| 438 | } |
| 439 | |
| 440 | function get_all_options() { |
| 441 | $options = get_option( $this->option_name, null ); |
| 442 | foreach ( MWAI_OPTIONS as $key => $value ) { |
| 443 | if ( !isset( $options[$key] ) ) { |
| 444 | $options[$key] = $value; |
| 445 | } |
| 446 | if ( $key === 'languages' ) { |
| 447 | // NOTE: If we decide to make a set of options for languages, we can keep it in the settings |
| 448 | $options[$key] = MWAI_LANGUAGES; |
| 449 | $options[$key] = apply_filters( 'mwai_languages', $options[$key] ); |
| 450 | } |
| 451 | } |
| 452 | $options['shortcode_chat_default_params'] = MWAI_CHATBOT_PARAMS; |
| 453 | $options['chatbot_defaults'] = MWAI_CHATBOT_DEFAULT_PARAMS; |
| 454 | $options['default_limits'] = MWAI_LIMITS; |
| 455 | $options['openai_models'] = MWAI_OPENAI_MODELS; |
| 456 | return $options; |
| 457 | } |
| 458 | |
| 459 | // Validate and keep the options clean and logical. |
| 460 | function sanitize_options() { |
| 461 | $options = $this->get_all_options(); |
| 462 | $needs_update = false; |
| 463 | |
| 464 | // We can sanitize our future options here, let's always remember it. |
| 465 | // Now, it is empty... |
| 466 | |
| 467 | if ( $needs_update ) { |
| 468 | update_option( $this->option_name, $options, false ); |
| 469 | } |
| 470 | return $options; |
| 471 | } |
| 472 | |
| 473 | function update_options( $options ) { |
| 474 | if ( !update_option( $this->option_name, $options, false ) ) { |
| 475 | return false; |
| 476 | } |
| 477 | $options = $this->sanitize_options(); |
| 478 | return $options; |
| 479 | } |
| 480 | |
| 481 | function update_option( $option, $value ) { |
| 482 | $options = $this->get_all_options(); |
| 483 | $options[$option] = $value; |
| 484 | return $this->update_options( $options ); |
| 485 | } |
| 486 | |
| 487 | function get_option( $option, $default = null ) { |
| 488 | $options = $this->get_all_options(); |
| 489 | return $options[$option] ?? $default; |
| 490 | } |
| 491 | #endregion |
| 492 | } |
| 493 | |
| 494 | ?> |