| @@ -5,409 +5,8 @@ | ||
| 5 | 5 | |
| 6 | 6 | class MxChat_Utils { |
| 7 | 7 | |
| 8 | 8 | /** |
| 9 | - * Validate a client-supplied session id (plan-mxchat-20260731-d42bec). | |
| 10 | - * | |
| 11 | - * sanitize_text_field() — which every session_id read site used before this — | |
| 12 | - * preserves '/' and '..'. Harmless where the value is only an option or | |
| 13 | - * transient key suffix, but mxchat_send_delayed_transcript() interpolates it | |
| 14 | - * into a filesystem path, so '../../../../path/x' wrote, emailed and deleted a | |
| 15 | - * file outside the uploads dir. | |
| 16 | - * | |
| 17 | - * REJECTS rather than rewrites: a silently-stripped id would orphan the | |
| 18 | - * conversation it belongs to, which is harder to diagnose than a clean refusal. | |
| 19 | - * Returns '' for anything malformed, so call sites fall into the empty-session | |
| 20 | - * error paths they already have. | |
| 21 | - * | |
| 22 | - * The generator only ever emits 'mxchat_chat_' + 32 hex chars | |
| 23 | - * (class-mxchat-integrator.php, js/chat-script.js), so this is not restrictive | |
| 24 | - * in practice. Length ceiling is deliberate — session ids are also used as | |
| 25 | - * option-name suffixes, and WP option names cap at 191 chars. | |
| 26 | - * | |
| 27 | - * @param mixed $raw Raw request value. | |
| 28 | - * @return string The id if well-formed, '' otherwise. | |
| 29 | - */ | |
| 30 | -public static function sanitize_session_id($raw) { | |
| 31 | - if (!is_scalar($raw)) { | |
| 32 | - return ''; | |
| 33 | - } | |
| 34 | - $val = trim((string) $raw); | |
| 35 | - if ($val === '') { | |
| 36 | - return ''; | |
| 37 | - } | |
| 38 | - return preg_match('/\A[A-Za-z0-9_-]{1,128}\z/', $val) ? $val : ''; | |
| 39 | -} | |
| 40 | - | |
| 41 | -/** | |
| 42 | - * Per-request cache for get_session_history(). Mirrors get_option()'s | |
| 43 | - * request-scoped caching, which the mxchat_history_ option reads got for | |
| 44 | - * free before plan 839c4c moved history reads onto the transcripts table. | |
| 45 | - */ | |
| 46 | -private static $history_cache = array(); | |
| 47 | - | |
| 48 | -/** | |
| 49 | - * Session chat history read from the transcripts table, in the exact array | |
| 50 | - * shape the legacy mxchat_history_<sid> option stored (plan 839c4c). The | |
| 51 | - * option was a second copy of state the table already held — measured | |
| 52 | - * byte-identical in role/content/order on 174 of 177 real sessions, with | |
| 53 | - * the table a superset on the rest — at up to 64 KB per option row. The | |
| 54 | - * table is now the single store; nothing writes the option any more. | |
| 55 | - * | |
| 56 | - * Shape notes, load-bearing for the consumers: | |
| 57 | - * - id: the transcripts row id (int). Integer ids make the pollers' | |
| 58 | - * ">" comparisons correct where the old uniqid() strings only worked by | |
| 59 | - * accident of hex ordering. | |
| 60 | - * - timestamp: milliseconds, derived from the table's second-resolution GMT | |
| 61 | - * column (x1000). Consumers comparing against a real-millisecond client | |
| 62 | - * cutoff MUST floor the cutoff to the second and err inclusive — see the | |
| 63 | - * persistence-off filters in class-mxchat-integrator.php. | |
| 64 | - * - agent_name: the row's user_identifier, which the writer sets to the | |
| 65 | - * same displayed_name value the option carried (agent name when present, | |
| 66 | - * else email, else identifier). | |
| 67 | - * | |
| 68 | - * Public and static so mxchat-woo / mxchat-forms can call the same accessor | |
| 69 | - * as core, guarded with method_exists against an older mxchat-basic. | |
| 70 | - * | |
| 71 | - * @param string $session_id | |
| 72 | - * @return array[] Chronological entries: id, role, content, timestamp, agent_name. | |
| 73 | - */ | |
| 74 | -public static function get_session_history($session_id) { | |
| 75 | - global $wpdb; | |
| 76 | - | |
| 77 | - $session_id = self::sanitize_session_id($session_id); | |
| 78 | - if ($session_id === '') { | |
| 79 | - return array(); | |
| 80 | - } | |
| 81 | - | |
| 82 | - if (array_key_exists($session_id, self::$history_cache)) { | |
| 83 | - return self::$history_cache[$session_id]; | |
| 84 | - } | |
| 85 | - | |
| 86 | - $table = $wpdb->prefix . 'mxchat_chat_transcripts'; | |
| 87 | - | |
| 88 | - // No SHOW TABLES guard: this is the chat hot path and the table is | |
| 89 | - // created on activation (with an admin-load safety net). A genuinely | |
| 90 | - // missing table fails the query and yields the same empty history the | |
| 91 | - // old option read produced on a fresh session. | |
| 92 | - $rows = $wpdb->get_results( | |
| 93 | - $wpdb->prepare( | |
| 94 | - "SELECT id, role, message, user_identifier, timestamp | |
| 95 | - FROM `$table` WHERE session_id = %s ORDER BY id ASC", | |
| 96 | - $session_id | |
| 97 | - ), | |
| 98 | - ARRAY_A | |
| 99 | - ); | |
| 100 | - | |
| 101 | - $history = array(); | |
| 102 | - if (is_array($rows)) { | |
| 103 | - foreach ($rows as $row) { | |
| 104 | - // The column stores GMT (current_time('mysql', 1) at the writer), | |
| 105 | - // so pin the parse to UTC rather than the site timezone. | |
| 106 | - $ts = strtotime($row['timestamp'] . ' +0000'); | |
| 107 | - $history[] = array( | |
| 108 | - 'id' => (int) $row['id'], | |
| 109 | - 'role' => (string) $row['role'], | |
| 110 | - 'content' => (string) $row['message'], | |
| 111 | - 'timestamp' => ($ts ? $ts : 0) * 1000, | |
| 112 | - 'agent_name' => (string) $row['user_identifier'], | |
| 113 | - ); | |
| 114 | - } | |
| 115 | - } | |
| 116 | - | |
| 117 | - self::$history_cache[$session_id] = $history; | |
| 118 | - | |
| 119 | - return $history; | |
| 120 | -} | |
| 121 | - | |
| 122 | -/** | |
| 123 | - * Drop the cached history for one session (or all). The writer calls this | |
| 124 | - * after every insert so a later read in the same request — e.g. the AI | |
| 125 | - * context build that follows saving the user's message — sees the new row, | |
| 126 | - * matching the read-your-own-write behavior update_option() gave the old | |
| 127 | - * option copy. | |
| 128 | - * | |
| 129 | - * @param string|null $session_id Null flushes everything (test seam). | |
| 130 | - */ | |
| 131 | -public static function flush_session_history_cache($session_id = null) { | |
| 132 | - if ($session_id === null) { | |
| 133 | - self::$history_cache = array(); | |
| 134 | - return; | |
| 135 | - } | |
| 136 | - | |
| 137 | - unset(self::$history_cache[(string) $session_id]); | |
| 138 | -} | |
| 139 | - | |
| 140 | -/** | |
| 141 | - * Most recipients the Notification Email field will accept (plan 2f131a). | |
| 142 | - * A settings field is not a mailing list. | |
| 143 | - */ | |
| 144 | -const NOTIFICATION_EMAIL_MAX = 5; | |
| 145 | - | |
| 146 | -/** | |
| 147 | - * Parse the Notification Email field into a list of recipients (plan 2f131a). | |
| 148 | - * | |
| 149 | - * THE TRAP THIS EXISTS TO CLOSE: sanitize_email() cannot be the validator for | |
| 150 | - * this field, because its output for the failing input is VALID. WordPress | |
| 151 | - * strips the separator and the surplus '@' and concatenates the remains: | |
| 152 | - * | |
| 153 | - * support@acme.com, sales@acme.com -> support@acme.comsalesacme.com | |
| 154 | - * | |
| 155 | - * and is_email() then returns true on that. So every guard in the plugin passed, | |
| 156 | - * the address was stored, the autosave ticked green, and both the new-session | |
| 157 | - * notification and the auto-emailed transcript went to a domain that does not | |
| 158 | - * exist — with no error anywhere. Validating the RAW part BEFORE sanitizing is | |
| 159 | - * the whole point; reversing those two lines silently restores the bug. | |
| 160 | - * | |
| 161 | - * All-or-nothing by design: if any entry is bad the caller must store NOTHING. | |
| 162 | - * A partial accept — keeping the good addresses and dropping the bad one — is | |
| 163 | - * the same defect in a new costume, because the owner still believes everyone | |
| 164 | - * on their list is being notified. | |
| 165 | - * | |
| 166 | - * @param mixed $raw Raw field value, exactly as submitted. | |
| 167 | - * @return array{emails: string[], error: string} Empty emails + empty error | |
| 168 | - * means the field was empty. | |
| 169 | - */ | |
| 170 | -public static function parse_notification_emails($raw) { | |
| 171 | - $out = array('emails' => array(), 'error' => ''); | |
| 172 | - | |
| 173 | - if (!is_scalar($raw)) { | |
| 174 | - $out['error'] = __('The notification email could not be read.', 'mxchat'); | |
| 175 | - return $out; | |
| 176 | - } | |
| 177 | - | |
| 178 | - $raw = trim((string) $raw); | |
| 179 | - if ($raw === '') { | |
| 180 | - return $out; // genuinely empty — the caller falls back to admin_email | |
| 181 | - } | |
| 182 | - | |
| 183 | - $seen = array(); | |
| 184 | - foreach (preg_split('/[,;]/', $raw) as $part) { | |
| 185 | - $part = trim($part); | |
| 186 | - if ($part === '') { | |
| 187 | - // A trailing or doubled separator carries no address, so skipping it | |
| 188 | - // cannot silently drop a recipient. This is the ONLY thing tolerated. | |
| 189 | - continue; | |
| 190 | - } | |
| 191 | - | |
| 192 | - // RAW first. See the note above — order is load-bearing. | |
| 193 | - $clean = is_email($part) ? sanitize_email($part) : ''; | |
| 194 | - if ($clean === '' || !is_email($clean)) { | |
| 195 | - return array( | |
| 196 | - 'emails' => array(), | |
| 197 | - 'error' => sprintf( | |
| 198 | - /* translators: %s: the email address the owner typed. */ | |
| 199 | - __('"%s" is not a valid email address, so nothing was saved. Separate multiple addresses with a comma.', 'mxchat'), | |
| 200 | - esc_html($part) | |
| 201 | - ), | |
| 202 | - ); | |
| 203 | - } | |
| 204 | - | |
| 205 | - $key = strtolower($clean); | |
| 206 | - if (isset($seen[$key])) { | |
| 207 | - continue; // same address twice would simply mail them twice | |
| 208 | - } | |
| 209 | - $seen[$key] = true; | |
| 210 | - $out['emails'][] = $clean; | |
| 211 | - } | |
| 212 | - | |
| 213 | - if (count($out['emails']) > self::NOTIFICATION_EMAIL_MAX) { | |
| 214 | - return array( | |
| 215 | - 'emails' => array(), | |
| 216 | - 'error' => sprintf( | |
| 217 | - /* translators: %d: maximum number of notification recipients. */ | |
| 218 | - __('Enter at most %d email addresses, separated by commas.', 'mxchat'), | |
| 219 | - self::NOTIFICATION_EMAIL_MAX | |
| 220 | - ), | |
| 221 | - ); | |
| 222 | - } | |
| 223 | - | |
| 224 | - return $out; | |
| 225 | -} | |
| 226 | - | |
| 227 | -/** | |
| 228 | - * The stored recipient list, ready to hand to wp_mail() (plan 2f131a). | |
| 229 | - * | |
| 230 | - * Fallback rule, and it is narrow on purpose: an EMPTY field falls back to the | |
| 231 | - * site admin address, because that is the documented behaviour and an owner who | |
| 232 | - * never filled the field in still wants their notifications. A field holding | |
| 233 | - * something unusable does NOT fall back — it sends nowhere, exactly as before | |
| 234 | - * this plan. Falling back on bad input would mean a typo silently redirects a | |
| 235 | - * store's transcripts to a different mailbox than the one on screen. | |
| 236 | - * | |
| 237 | - * @param array|null $options mxchat_transcripts_options, or null to read it. | |
| 238 | - * @return string[] Recipients; empty means do not send. | |
| 239 | - */ | |
| 240 | -public static function notification_recipients($options = null) { | |
| 241 | - if (!is_array($options)) { | |
| 242 | - $options = get_option('mxchat_transcripts_options', array()); | |
| 243 | - if (!is_array($options)) { | |
| 244 | - $options = array(); | |
| 245 | - } | |
| 246 | - } | |
| 247 | - | |
| 248 | - $raw = isset($options['mxchat_notification_email']) ? $options['mxchat_notification_email'] : ''; | |
| 249 | - $raw = is_scalar($raw) ? trim((string) $raw) : ''; | |
| 250 | - | |
| 251 | - if ($raw === '') { | |
| 252 | - $admin = get_option('admin_email'); | |
| 253 | - return is_email($admin) ? array($admin) : array(); | |
| 254 | - } | |
| 255 | - | |
| 256 | - $parsed = self::parse_notification_emails($raw); | |
| 257 | - return $parsed['error'] === '' ? $parsed['emails'] : array(); | |
| 258 | -} | |
| 259 | - | |
| 260 | -/** | |
| 261 | - * Centralized embedding model registry. Single source of truth for dimensions | |
| 262 | - * and provider, so model-switch protection logic doesn't drift across files. | |
| 263 | - */ | |
| 264 | -public static function embedding_model_registry() { | |
| 265 | - return array( | |
| 266 | - 'text-embedding-ada-002' => array('dims' => 1536, 'provider' => 'openai', 'label' => 'Ada 2'), | |
| 267 | - 'text-embedding-3-small' => array('dims' => 1536, 'provider' => 'openai', 'label' => 'TE3 Small'), | |
| 268 | - 'text-embedding-3-large' => array('dims' => 3072, 'provider' => 'openai', 'label' => 'TE3 Large'), | |
| 269 | - 'voyage-3-large' => array('dims' => 2048, 'provider' => 'voyage', 'label' => 'Voyage-3 Large'), | |
| 270 | - 'gemini-embedding-001' => array('dims' => 1536, 'provider' => 'gemini', 'label' => 'Gemini Embedding'), | |
| 271 | - ); | |
| 272 | -} | |
| 273 | - | |
| 274 | -public static function embedding_model_dimensions($model) { | |
| 275 | - $registry = self::embedding_model_registry(); | |
| 276 | - return isset($registry[$model]) ? (int) $registry[$model]['dims'] : 0; | |
| 277 | -} | |
| 278 | - | |
| 279 | -public static function embedding_model_label($model) { | |
| 280 | - if (is_string($model) && strpos($model, 'custom:') === 0) { | |
| 281 | - /* translators: %s: the embedding model name configured on the custom provider */ | |
| 282 | - return sprintf(__('%s (custom provider)', 'mxchat'), substr($model, 7)); | |
| 283 | - } | |
| 284 | - $registry = self::embedding_model_registry(); | |
| 285 | - return isset($registry[$model]) ? $registry[$model]['label'] : $model; | |
| 286 | -} | |
| 287 | - | |
| 288 | -/** | |
| 289 | - * Returns the model that was last used to actually write embeddings into the | |
| 290 | - * KB. Differs from the user-selected setting once a switch has happened but | |
| 291 | - * no re-embed has occurred yet — that's the mismatch state we warn about. | |
| 292 | - */ | |
| 293 | -public static function get_active_embedding_model() { | |
| 294 | - return get_option('mxchat_active_embedding_model', ''); | |
| 295 | -} | |
| 296 | - | |
| 297 | -/** | |
| 298 | - * Stamp the model that produced the most recent successful embedding. Called | |
| 299 | - * from generate_embedding() right after the API responds with a valid vector. | |
| 300 | - */ | |
| 301 | -public static function stamp_active_embedding_model($model) { | |
| 302 | - if (!empty($model) && $model !== self::get_active_embedding_model()) { | |
| 303 | - update_option('mxchat_active_embedding_model', $model, false); | |
| 304 | - } | |
| 305 | -} | |
| 306 | - | |
| 307 | -/** | |
| 308 | - * The model name the custom-provider embedding path will send, mirroring the | |
| 309 | - * fallback chain the request itself uses: dedicated custom embedding model, | |
| 310 | - * else the custom chat model, else 'default'. Single source shared by | |
| 311 | - * generate_embedding_custom() and the mismatch-warning "selected" side so the | |
| 312 | - * two can never drift (plan ae02cb). | |
| 313 | - */ | |
| 314 | -public static function resolve_custom_embedding_model($options) { | |
| 315 | - if (isset($options['custom_provider_embedding_model']) && trim((string) $options['custom_provider_embedding_model']) !== '') { | |
| 316 | - return trim((string) $options['custom_provider_embedding_model']); | |
| 317 | - } | |
| 318 | - if (isset($options['custom_provider_model']) && trim((string) $options['custom_provider_model']) !== '') { | |
| 319 | - return trim((string) $options['custom_provider_model']); | |
| 320 | - } | |
| 321 | - return 'default'; | |
| 322 | -} | |
| 323 | - | |
| 324 | -/** | |
| 325 | - * The EFFECTIVE selected embedding model — what the next embed will actually | |
| 326 | - * use. With custom-provider embeddings on this is the custom identity in the | |
| 327 | - * same 'custom:<model>' form stamp_active_embedding_model() records, not the | |
| 328 | - * inert standard dropdown value. Mismatch-warning comparisons must read this, | |
| 329 | - * never $options['embedding_model'] directly — the dropdown cannot be | |
| 330 | - * deselected, so reading it raw flags every correctly-configured custom setup. | |
| 331 | - */ | |
| 332 | -public static function get_selected_embedding_model($options = null) { | |
| 333 | - if (!is_array($options)) { | |
| 334 | - $options = get_option('mxchat_options', array()); | |
| 335 | - } | |
| 336 | - if (isset($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') { | |
| 337 | - return 'custom:' . self::resolve_custom_embedding_model($options); | |
| 338 | - } | |
| 339 | - return $options['embedding_model'] ?? ''; | |
| 340 | -} | |
| 341 | - | |
| 342 | -/** | |
| 343 | - * Extract the 11-character YouTube video ID from a URL, or '' if the URL is | |
| 344 | - * not a single-video YouTube link. Single source of truth for both the KB | |
| 345 | - * ingestion side and the chat render side — do not duplicate this parsing. | |
| 346 | - * Channel, playlist, and search URLs deliberately return '' (only a URL that | |
| 347 | - * identifies one video can be embedded). | |
| 348 | - */ | |
| 349 | -public static function parse_youtube_id($url) { | |
| 350 | - if (!is_string($url) || $url === '' || !preg_match('#^https?://#i', $url)) { | |
| 351 | - return ''; | |
| 352 | - } | |
| 353 | - $host = strtolower((string) wp_parse_url($url, PHP_URL_HOST)); | |
| 354 | - $host = preg_replace('/^(www|m)\./', '', $host); | |
| 355 | - $path = (string) wp_parse_url($url, PHP_URL_PATH); | |
| 356 | - $id = ''; | |
| 357 | - if ($host === 'youtu.be') { | |
| 358 | - $segments = explode('/', ltrim($path, '/')); | |
| 359 | - $id = $segments[0] ?? ''; | |
| 360 | - } elseif (in_array($host, array('youtube.com', 'youtube-nocookie.com'), true)) { | |
| 361 | - if (preg_match('#^/(?:shorts|embed|live|v)/([A-Za-z0-9_-]+)#', $path, $m)) { | |
| 362 | - $id = $m[1]; | |
| 363 | - } elseif ($path === '/watch') { | |
| 364 | - parse_str((string) wp_parse_url($url, PHP_URL_QUERY), $query_vars); | |
| 365 | - $id = isset($query_vars['v']) ? (string) $query_vars['v'] : ''; | |
| 366 | - } | |
| 367 | - } | |
| 368 | - $id = preg_replace('/[^A-Za-z0-9_-]/', '', (string) $id); | |
| 369 | - return (strlen($id) === 11) ? $id : ''; | |
| 370 | -} | |
| 371 | - | |
| 372 | -/** | |
| 373 | - * plan-mxchat-20260813-f52492 — video-card gating. | |
| 374 | - * | |
| 375 | - * Two standalone options (NOT mxchat_options — they skip the sanitize/autosave | |
| 376 | - * traps entirely), read here so the gate in the integrator and the fields on | |
| 377 | - * Knowledge -> Chunking & Retrieval can never disagree about a default. | |
| 378 | - * | |
| 379 | - * Master switch. Default ON: the card is existing behavior, and this is an | |
| 380 | - * opt-out for owners who never want one, not a new feature to opt into. | |
| 381 | - */ | |
| 382 | -public static function video_embed_enabled() { | |
| 383 | - return get_option('mxchat_video_embed_enabled', 'on') === 'on'; | |
| 384 | -} | |
| 385 | - | |
| 386 | -/** | |
| 387 | - * The video card's OWN confidence floor, as a 0-1 cosine — deliberately not | |
| 388 | - * the site-wide Similarity Threshold (default 35). "Good enough to quote in | |
| 389 | - * the answer" and "good enough to put a video on screen" are different | |
| 390 | - * questions: retrieval is allowed to be generous because the model still | |
| 391 | - * decides what to say, whereas the card is asserted to the visitor with no | |
| 392 | - * such filter. Stored as an int percentage to match the site-wide slider's | |
| 393 | - * convention; the default (55) sits above it on purpose. | |
| 394 | - * | |
| 395 | - * MXCHAT_VIDEO_EMBED_THRESHOLD_DEFAULT is the single source of that number. | |
| 396 | - */ | |
| 397 | -public static function video_embed_threshold() { | |
| 398 | - $stored = get_option('mxchat_video_embed_threshold', null); | |
| 399 | - $percent = ($stored === null || $stored === '') | |
| 400 | - ? MXCHAT_VIDEO_EMBED_THRESHOLD_DEFAULT | |
| 401 | - : (int) $stored; | |
| 402 | - if ($percent < 0) { $percent = 0; } | |
| 403 | - if ($percent > 100) { $percent = 100; } | |
| 404 | - // Cast: PHP evaluates 100/100 to int(1), so an unclamped return type would | |
| 405 | - // vary with the stored value. Callers compare against a cosine — keep it float. | |
| 406 | - return (float) $percent / 100; | |
| 407 | -} | |
| 408 | - | |
| 409 | -/** | |
| 410 | 9 | * UPDATED: Submit or update content (and its embedding) in the database. |
| 411 | 10 | * Stores in Pinecone if enabled, otherwise stores in WordPress DB. |
| 412 | 11 | * |
| 413 | 12 | * @param string $content The content to be embedded. |
| @@ -449,13 +48,10 @@ | ||
| 449 | 48 | // UPDATED: Generate the embedding using bot-specific configuration |
| 450 | 49 | $embedding_vector = self::generate_embedding($content, $api_key, $bot_id); |
| 451 | 50 | |
| 452 | 51 | if (!is_array($embedding_vector)) { |
| 453 | - // Surface the provider's real reason instead of a fixed string (4a7c0a). | |
| 454 | - $reason = is_wp_error($embedding_vector) | |
| 455 | - ? $embedding_vector->get_error_message() | |
| 456 | - : 'Failed to generate embedding for content'; | |
| 457 | - return new WP_Error('embedding_failed', $reason); | |
| 52 | + //error_log('[MXCHAT-DB] Error: Embedding generation failed'); | |
| 53 | + return new WP_Error('embedding_failed', 'Failed to generate embedding for content'); | |
| 458 | 54 | } |
| 459 | 55 | |
| 460 | 56 | //error_log('[MXCHAT-DB] Embedding generated successfully'); |
| 461 | 57 | |
| @@ -775,9 +371,9 @@ | ||
| 775 | 371 | 'source_url' => $url, // Can be empty for manual content |
| 776 | 372 | 'type' => $content_type, // Now supports: post, page, pdf, url, manual, product, etc. |
| 777 | 373 | 'last_updated' => time(), |
| 778 | 374 | 'created_at' => time(), // Add creation timestamp |
| 779 | - 'bot_id' => $bot_id, // Add bot identification | |
| 375 | + 'bot_id' => $bot_id // Add bot identification | |
| 780 | 376 | ); |
| 781 | 377 | |
| 782 | 378 | $vector_data = array( |
| 783 | 379 | 'id' => $vector_id, |
| @@ -834,104 +430,8 @@ | ||
| 834 | 430 | return true; |
| 835 | 431 | } |
| 836 | 432 | |
| 837 | 433 | /** |
| 838 | - * Caller-side pre-flight for KB ingestion: can an embedding request be made | |
| 839 | - * with these options, and which API key should travel downstream? | |
| 840 | - * | |
| 841 | - * Custom-provider-aware — generate_embedding() below routes to the custom | |
| 842 | - * endpoint FIRST and ignores the passed cloud key entirely when | |
| 843 | - * custom_provider_for_embeddings is on, so on that branch the only real | |
| 844 | - * requirement is a Base URL. Ingestion callers that gated on a cloud API key | |
| 845 | - * were killing keyless custom-embeddings sites (local Ollama / LM Studio | |
| 846 | - * class) before the embed layer could route (plan cbd5fd). | |
| 847 | - * | |
| 848 | - * NOTE: reads $options['embedding_model'] raw on purpose — this mirrors | |
| 849 | - * generate_embedding()'s own routing read, NOT the mismatch-banner's | |
| 850 | - * "selected" chain (get_selected_embedding_model). The helper must predict | |
| 851 | - * what the very next embed call will do, byte-for-byte. | |
| 852 | - * | |
| 853 | - * Decision only — callers keep their own error-surfacing shape (admin-notice | |
| 854 | - * transient + redirect, wp_send_json_error, WP_Error, silent return). | |
| 855 | - * | |
| 856 | - * @param array|null $options Resolved options (bot-specific where the caller | |
| 857 | - * has them); null loads the default bot's options. | |
| 858 | - * @return array { | |
| 859 | - * @type bool $ok Whether ingestion can proceed. | |
| 860 | - * @type string $api_key Key to pass downstream ('' on the custom branch — | |
| 861 | - * generate_embedding() ignores it there). | |
| 862 | - * @type string $reason Human-readable blocker; '' when $ok. | |
| 863 | - * @type string $provider Short provider label ('OpenAI', 'Voyage AI', | |
| 864 | - * 'Google Gemini', 'Custom Provider'). | |
| 865 | - * } | |
| 866 | - */ | |
| 867 | -public static function embedding_preflight($options = null) { | |
| 868 | - if (!is_array($options)) { | |
| 869 | - $options = get_option('mxchat_options'); | |
| 870 | - $options = is_array($options) ? $options : array(); | |
| 871 | - } | |
| 872 | - | |
| 873 | - // Custom branch mirrors generate_embedding()'s routing order (custom first). | |
| 874 | - if (isset($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') { | |
| 875 | - $base_url = isset($options['custom_provider_base_url']) ? rtrim(trim((string) $options['custom_provider_base_url']), '/') : ''; | |
| 876 | - if ($base_url === '') { | |
| 877 | - return array( | |
| 878 | - 'ok' => false, | |
| 879 | - 'api_key' => '', | |
| 880 | - // Same string generate_embedding_custom() returns for this state. | |
| 881 | - 'reason' => __('Custom provider Base URL is not configured.', 'mxchat'), | |
| 882 | - 'provider' => 'Custom Provider', | |
| 883 | - ); | |
| 884 | - } | |
| 885 | - return array('ok' => true, 'api_key' => '', 'reason' => '', 'provider' => 'Custom Provider'); | |
| 886 | - } | |
| 887 | - | |
| 888 | - $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; | |
| 889 | - if (strpos($selected_model, 'voyage') === 0) { | |
| 890 | - $api_key = $options['voyage_api_key'] ?? ''; | |
| 891 | - $provider = 'Voyage AI'; | |
| 892 | - } elseif (strpos($selected_model, 'gemini-embedding') === 0) { | |
| 893 | - $api_key = $options['gemini_api_key'] ?? ''; | |
| 894 | - $provider = 'Google Gemini'; | |
| 895 | - } else { | |
| 896 | - $api_key = $options['api_key'] ?? ''; | |
| 897 | - $provider = 'OpenAI'; | |
| 898 | - } | |
| 899 | - | |
| 900 | - if (empty($api_key)) { | |
| 901 | - return array( | |
| 902 | - 'ok' => false, | |
| 903 | - 'api_key' => '', | |
| 904 | - 'reason' => sprintf( | |
| 905 | - /* translators: %s: embedding provider name */ | |
| 906 | - __('%s API key is not configured. Please add your API key in the settings before submitting content.', 'mxchat'), | |
| 907 | - $provider | |
| 908 | - ), | |
| 909 | - 'provider' => $provider, | |
| 910 | - ); | |
| 911 | - } | |
| 912 | - | |
| 913 | - return array('ok' => true, 'api_key' => $api_key, 'reason' => '', 'provider' => $provider); | |
| 914 | -} | |
| 915 | - | |
| 916 | -/** | |
| 917 | - * Public QUERY-side entry point (plan 876edb). The chat pipeline's | |
| 918 | - * MxChat_Integrator::mxchat_generate_embedding() adapter routes through here | |
| 919 | - * so the query and index sides share ONE provider-routing implementation — | |
| 920 | - * the same endpoints, request bodies, and stamping semantics. The Integrator | |
| 921 | - * keeps its own error vocabulary by translating the WP_Error this returns | |
| 922 | - * (see the structured error data on every failure path below). | |
| 923 | - * | |
| 924 | - * @param string $text The text to be embedded. | |
| 925 | - * @param string $api_key Caller-resolved API key (per-bot on the query side). | |
| 926 | - * @param string $bot_id The bot ID for multi-bot support. | |
| 927 | - * @return array|WP_Error The embedding vector, or WP_Error carrying the reason. | |
| 928 | - */ | |
| 929 | -public static function generate_query_embedding($text, $api_key, $bot_id = 'default') { | |
| 930 | - return self::generate_embedding($text, $api_key, $bot_id); | |
| 931 | -} | |
| 932 | - | |
| 933 | -/** | |
| 934 | 434 | * UPDATED: Generate an embedding for the given text using bot-specific configuration. |
| 935 | 435 | * |
| 936 | 436 | * @param string $text The text to be embedded. |
| 937 | 437 | * @param string $api_key The API key used for generating embeddings. |
| @@ -945,24 +445,11 @@ | ||
| 945 | 445 | } else { |
| 946 | 446 | $bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 947 | 447 | $options = !empty($bot_options) ? $bot_options : get_option('mxchat_options'); |
| 948 | 448 | } |
| 949 | - | |
| 950 | - // Opt-in: when the custom provider is selected for embeddings, route the KB | |
| 951 | - // INDEX side through the same custom endpoint the query side uses, so stored | |
| 952 | - // vectors and query vectors come from the same model. Default-off behavior | |
| 953 | - // below is untouched. | |
| 954 | - if (isset($options['custom_provider_for_embeddings']) && $options['custom_provider_for_embeddings'] === 'on') { | |
| 955 | - $custom = self::generate_embedding_custom($text, $options); | |
| 956 | - // The custom path already returns a human-readable error string — | |
| 957 | - // carry it instead of collapsing to null (plan 4a7c0a). The 'custom' | |
| 958 | - // branch marker lets the Integrator adapter map the string back onto | |
| 959 | - // its own error codes (876edb). | |
| 960 | - return is_array($custom) ? $custom : new WP_Error('embedding_failed', (string) $custom, array('branch' => 'custom')); | |
| 961 | - } | |
| 962 | - | |
| 449 | + | |
| 963 | 450 | $selected_model = $options['embedding_model'] ?? 'text-embedding-ada-002'; |
| 964 | - | |
| 451 | + | |
| 965 | 452 | // Determine endpoint and API key based on model |
| 966 | 453 | if (strpos($selected_model, 'voyage') === 0) { |
| 967 | 454 | $endpoint = 'https://api.voyageai.com/v1/embeddings'; |
| 968 | 455 | $api_key = $options['voyage_api_key'] ?? ''; |
| @@ -970,13 +457,10 @@ | ||
| 970 | 457 | $endpoint = 'https://generativelanguage.googleapis.com/v1beta/models/' . $selected_model . ':embedContent'; |
| 971 | 458 | $api_key = $options['gemini_api_key'] ?? ''; |
| 972 | 459 | } else { |
| 973 | 460 | $endpoint = 'https://api.openai.com/v1/embeddings'; |
| 974 | - // Prefer the caller-resolved key when one was passed — the query side | |
| 975 | - // resolves per-bot keys at its call sites (integrator adapter, 876edb). | |
| 976 | - // Index callers pass the preflight key, which equals this options read, | |
| 977 | - // so nothing changes for them. | |
| 978 | - $api_key = !empty($api_key) ? $api_key : ($options['api_key'] ?? ''); | |
| 461 | + // Use the bot-specific API key or fallback to passed API key | |
| 462 | + $api_key = $options['api_key'] ?? $api_key; | |
| 979 | 463 | } |
| 980 | 464 | |
| 981 | 465 | // Prepare request body based on provider |
| 982 | 466 | if (strpos($selected_model, 'gemini-embedding') === 0) { |
| @@ -1027,208 +511,35 @@ | ||
| 1027 | 511 | |
| 1028 | 512 | $response = wp_remote_post($endpoint, $args); |
| 1029 | 513 | |
| 1030 | 514 | if (is_wp_error($response)) { |
| 1031 | - $message = 'Embedding request failed (connection): ' . $response->get_error_message(); | |
| 1032 | - if (class_exists('MxChat_Admin')) { | |
| 1033 | - MxChat_Admin::mxchat_log_debug('embedding_error', $message, array('model' => $selected_model, 'bot_id' => $bot_id)); | |
| 1034 | - } | |
| 1035 | - return new WP_Error('embedding_failed', $message, array( | |
| 1036 | - 'branch' => 'cloud', | |
| 1037 | - 'kind' => 'connection', | |
| 1038 | - 'reason' => $response->get_error_message(), | |
| 1039 | - 'model' => $selected_model, | |
| 1040 | - )); | |
| 515 | + //error_log('Error generating embedding for bot ' . $bot_id . ': ' . $response->get_error_message()); | |
| 516 | + return null; | |
| 1041 | 517 | } |
| 1042 | - | |
| 518 | + | |
| 1043 | 519 | $response_body = json_decode(wp_remote_retrieve_body($response), true); |
| 1044 | - | |
| 520 | + | |
| 1045 | 521 | // Handle different response formats based on provider |
| 1046 | 522 | if (strpos($selected_model, 'gemini-embedding') === 0) { |
| 1047 | 523 | // Gemini API response format |
| 1048 | 524 | if (isset($response_body['embedding']['values']) && is_array($response_body['embedding']['values'])) { |
| 1049 | - self::stamp_active_embedding_model($selected_model); | |
| 1050 | 525 | return $response_body['embedding']['values']; |
| 1051 | 526 | } else { |
| 1052 | - return self::embedding_failure_error($response, $selected_model, $api_key, $bot_id); | |
| 527 | + //error_log('Invalid response received from Gemini embedding API for bot ' . $bot_id . ': ' . wp_json_encode($response_body)); | |
| 528 | + return null; | |
| 1053 | 529 | } |
| 1054 | 530 | } else { |
| 1055 | 531 | // OpenAI/Voyage API response format |
| 1056 | 532 | if (isset($response_body['data'][0]['embedding']) && is_array($response_body['data'][0]['embedding'])) { |
| 1057 | - self::stamp_active_embedding_model($selected_model); | |
| 1058 | 533 | return $response_body['data'][0]['embedding']; |
| 1059 | 534 | } else { |
| 1060 | - return self::embedding_failure_error($response, $selected_model, $api_key, $bot_id); | |
| 535 | + //error_log('Invalid response received from embedding API for bot ' . $bot_id . ': ' . wp_json_encode($response_body)); | |
| 536 | + return null; | |
| 1061 | 537 | } |
| 1062 | 538 | } |
| 1063 | 539 | } |
| 1064 | 540 | |
| 1065 | 541 | /** |
| 1066 | - * Build a WP_Error carrying the embedding provider's REAL failure reason, | |
| 1067 | - * and record it in the Debug Mode log. Previously every failure path | |
| 1068 | - * returned bare null, so customers saw only "Failed to generate embedding | |
| 1069 | - * for content" / "Failed to store any chunks" with no cause (plan 4a7c0a). | |
| 1070 | - * | |
| 1071 | - * The API key never appears in provider response bodies (it travels in the | |
| 1072 | - * request headers), but the reason is scrubbed for it anyway before it can | |
| 1073 | - * reach a notice or the debug log. | |
| 1074 | - */ | |
| 1075 | -private static function embedding_failure_error($response, $selected_model, $api_key, $bot_id) { | |
| 1076 | - $status = (int) wp_remote_retrieve_response_code($response); | |
| 1077 | - $raw = (string) wp_remote_retrieve_body($response); | |
| 1078 | - $decoded = json_decode($raw, true); | |
| 1079 | - | |
| 1080 | - // Provider error shapes: OpenAI + Gemini use {"error":{"message":…}}; | |
| 1081 | - // Voyage uses {"detail":…}. | |
| 1082 | - $reason = ''; | |
| 1083 | - if (is_array($decoded)) { | |
| 1084 | - if (isset($decoded['error']['message']) && is_string($decoded['error']['message'])) { | |
| 1085 | - $reason = $decoded['error']['message']; | |
| 1086 | - } elseif (isset($decoded['detail']) && is_string($decoded['detail'])) { | |
| 1087 | - $reason = $decoded['detail']; | |
| 1088 | - } | |
| 1089 | - } | |
| 1090 | - if ($reason === '') { | |
| 1091 | - $reason = ($raw !== '') ? substr($raw, 0, 200) : 'empty or malformed response'; | |
| 1092 | - } | |
| 1093 | - if (is_string($api_key) && $api_key !== '') { | |
| 1094 | - $reason = str_replace($api_key, '[redacted]', $reason); | |
| 1095 | - } | |
| 1096 | - $reason = substr($reason, 0, 300); | |
| 1097 | - $message = sprintf('Embedding failed (%s, HTTP %d): %s', $selected_model, $status, $reason); | |
| 1098 | - | |
| 1099 | - if (class_exists('MxChat_Admin')) { | |
| 1100 | - MxChat_Admin::mxchat_log_debug('embedding_error', $message, array( | |
| 1101 | - 'model' => $selected_model, | |
| 1102 | - 'status' => $status, | |
| 1103 | - 'bot_id' => $bot_id, | |
| 1104 | - )); | |
| 1105 | - } | |
| 1106 | - | |
| 1107 | - // Structured data so the Integrator's query-side adapter can rebuild its | |
| 1108 | - // typed error contract (auth/rate-limit/quota/invalid-response) without a | |
| 1109 | - // second transport implementation (876edb). Additive — message unchanged. | |
| 1110 | - return new WP_Error('embedding_failed', $message, array( | |
| 1111 | - 'branch' => 'cloud', | |
| 1112 | - 'status' => $status, | |
| 1113 | - 'error_type' => (is_array($decoded) && isset($decoded['error']['type']) && is_string($decoded['error']['type'])) ? $decoded['error']['type'] : '', | |
| 1114 | - 'reason' => $reason, | |
| 1115 | - 'model' => $selected_model, | |
| 1116 | - )); | |
| 1117 | -} | |
| 1118 | - | |
| 1119 | -/** | |
| 1120 | - * Generate an embedding via a Custom (OpenAI-compatible) provider's /embeddings route. | |
| 1121 | - * Shared by every embedding entry point so the KNOWLEDGE-BASE INDEX side and the | |
| 1122 | - * QUERY side route through the same model when the opt-in | |
| 1123 | - * 'custom_provider_for_embeddings' setting is on. Mirrors the query-path logic in | |
| 1124 | - * MxChat_Integrator::mxchat_generate_embedding_custom() but takes an explicit | |
| 1125 | - * $options array so it is callable statically from utils + knowledge-manager. | |
| 1126 | - * | |
| 1127 | - * Returns a numeric array (the embedding vector) on success, or a human-readable | |
| 1128 | - * error string on failure (so callers expecting a string error, like the | |
| 1129 | - * knowledge-manager, can surface it directly; callers expecting array|null wrap it). | |
| 1130 | - * | |
| 1131 | - * @param string $text Text to embed. | |
| 1132 | - * @param array $options The resolved mxchat options (must contain the custom_provider_* keys). | |
| 1133 | - * @return array|string Embedding vector on success; error string on failure. | |
| 1134 | - */ | |
| 1135 | -public static function generate_embedding_custom($text, $options) { | |
| 1136 | - if (empty($text)) { | |
| 1137 | - return 'No text provided for embedding generation'; | |
| 1138 | - } | |
| 1139 | - | |
| 1140 | - $base_url = isset($options['custom_provider_base_url']) ? rtrim(trim((string) $options['custom_provider_base_url']), '/') : ''; | |
| 1141 | - if (empty($base_url)) { | |
| 1142 | - return 'Custom provider Base URL is not configured.'; | |
| 1143 | - } | |
| 1144 | - | |
| 1145 | - $api_key = isset($options['custom_provider_api_key']) ? trim((string) $options['custom_provider_api_key']) : ''; | |
| 1146 | - $auth_scheme = isset($options['custom_provider_auth_scheme']) ? $options['custom_provider_auth_scheme'] : 'bearer'; | |
| 1147 | - $api_version = isset($options['custom_provider_api_version']) ? trim((string) $options['custom_provider_api_version']) : ''; | |
| 1148 | - | |
| 1149 | - // Embedding model: shared resolver (dedicated embedding model -> chat model | |
| 1150 | - // -> 'default') — the mismatch warning's "selected" side reads the same chain. | |
| 1151 | - $model = self::resolve_custom_embedding_model($options); | |
| 1152 | - | |
| 1153 | - $embed_url = $base_url . '/embeddings'; | |
| 1154 | - if (!empty($api_version)) { | |
| 1155 | - $embed_url .= (strpos($embed_url, '?') === false ? '?' : '&') . 'api-version=' . rawurlencode($api_version); | |
| 1156 | - } | |
| 1157 | - | |
| 1158 | - $headers = ['Content-Type' => 'application/json']; | |
| 1159 | - if (!empty($api_key)) { | |
| 1160 | - if ($auth_scheme === 'api-key') { | |
| 1161 | - $headers['api-key'] = $api_key; | |
| 1162 | - } else { | |
| 1163 | - $headers['Authorization'] = 'Bearer ' . $api_key; | |
| 1164 | - } | |
| 1165 | - } | |
| 1166 | - | |
| 1167 | - $response = wp_remote_post($embed_url, [ | |
| 1168 | - 'headers' => $headers, | |
| 1169 | - 'body' => wp_json_encode(['input' => $text, 'model' => $model]), | |
| 1170 | - 'timeout' => 60, | |
| 1171 | - ]); | |
| 1172 | - if (is_wp_error($response)) { | |
| 1173 | - return self::log_custom_embedding_failure( | |
| 1174 | - 'Connection error when generating embeddings (custom provider): ' . $response->get_error_message(), | |
| 1175 | - $model, | |
| 1176 | - $api_key | |
| 1177 | - ); | |
| 1178 | - } | |
| 1179 | - | |
| 1180 | - $status = wp_remote_retrieve_response_code($response); | |
| 1181 | - $body = json_decode(wp_remote_retrieve_body($response), true); | |
| 1182 | - if ($status !== 200) { | |
| 1183 | - $msg = isset($body['error']['message']) ? $body['error']['message'] : 'HTTP ' . $status; | |
| 1184 | - return self::log_custom_embedding_failure( | |
| 1185 | - 'Custom embedding endpoint error: ' . $msg, | |
| 1186 | - $model, | |
| 1187 | - $api_key, | |
| 1188 | - (int) $status | |
| 1189 | - ); | |
| 1190 | - } | |
| 1191 | - if (isset($body['data'][0]['embedding']) && is_array($body['data'][0]['embedding'])) { | |
| 1192 | - // Stamp the custom model identity so the active-embedding-model mismatch | |
| 1193 | - // warning reflects the real (custom) model rather than the built-in setting. | |
| 1194 | - self::stamp_active_embedding_model('custom:' . $model); | |
| 1195 | - return $body['data'][0]['embedding']; | |
| 1196 | - } | |
| 1197 | - return self::log_custom_embedding_failure('Invalid embedding response from custom provider.', $model, $api_key); | |
| 1198 | -} | |
| 1199 | - | |
| 1200 | -/** | |
| 1201 | - * Record a custom-provider embedding failure in the Debug Mode log, then | |
| 1202 | - * return the message unchanged so callers keep their string-error contract. | |
| 1203 | - * The cloud branch has logged its failures since 4a7c0a; the custom branch | |
| 1204 | - * never did, so chat-side failures on Custom-provider installs were | |
| 1205 | - * invisible to Debug Mode despite the 3.2.18 readme saying otherwise | |
| 1206 | - * (plan 71e4b6). Same scrub-then-log shape as embedding_failure_error(). | |
| 1207 | - * | |
| 1208 | - * @param string $message Human-readable failure (the caller's return value). | |
| 1209 | - * @param string $model Resolved custom embedding model. | |
| 1210 | - * @param string $api_key Scrubbed out of the logged message if it ever appears. | |
| 1211 | - * @param int $status HTTP status when one was received, 0 otherwise. | |
| 1212 | - * @return string The (scrubbed) message. | |
| 1213 | - */ | |
| 1214 | -private static function log_custom_embedding_failure($message, $model, $api_key, $status = 0) { | |
| 1215 | - if (is_string($api_key) && $api_key !== '') { | |
| 1216 | - $message = str_replace($api_key, '[redacted]', $message); | |
| 1217 | - } | |
| 1218 | - | |
| 1219 | - if (class_exists('MxChat_Admin')) { | |
| 1220 | - $context = array('model' => 'custom:' . $model); | |
| 1221 | - if ($status > 0) { | |
| 1222 | - $context['status'] = $status; | |
| 1223 | - } | |
| 1224 | - MxChat_Admin::mxchat_log_debug('embedding_error', $message, $context); | |
| 1225 | - } | |
| 1226 | - | |
| 1227 | - return $message; | |
| 1228 | -} | |
| 1229 | - | |
| 1230 | -/** | |
| 1231 | 542 | * Submit content as multiple chunks |
| 1232 | 543 | * |
| 1233 | 544 | * Splits large content into chunks, generates embeddings for each, |
| 1234 | 545 | * and stores them with chunk metadata for later reassembly. |
| @@ -1270,44 +581,13 @@ | ||
| 1270 | 581 | return new WP_Error('chunking_failed', 'Content could not be split into chunks'); |
| 1271 | 582 | } |
| 1272 | 583 | |
| 1273 | 584 | $errors = array(); |
| 1274 | - $embed_failures = 0; | |
| 1275 | - $first_embed_reason = ''; | |
| 1276 | - $first_store_reason = ''; | |
| 1277 | 585 | $is_pinecone = self::is_pinecone_enabled_for_bot($bot_id); |
| 1278 | 586 | |
| 1279 | 587 | foreach ($chunks as $index => $chunk_text) { |
| 1280 | 588 | // Generate chunk metadata |
| 1281 | 589 | $chunk_metadata = MxChat_Chunker::create_chunk_metadata($index, $total_chunks, $source_url); |
| 1282 | - | |
| 1283 | - // AI-Engine-style aliases so external consumers (Pinecone/Qdrant/Chroma) can rely on | |
| 1284 | - // a stable shorthand ('source'/'part_index'/'part_total') without parsing our internal names. | |
| 1285 | - $chunk_metadata['source'] = $source_url; | |
| 1286 | - $chunk_metadata['part_index'] = (int) $index; | |
| 1287 | - $chunk_metadata['part_total'] = (int) $total_chunks; | |
| 1288 | - | |
| 1289 | - /** | |
| 1290 | - * Filter the per-chunk metadata blob before it's written to the KB store. | |
| 1291 | - * | |
| 1292 | - * @param array $chunk_metadata Metadata array (source, part_index, part_total, chunk_index, total_chunks, source_url, parent_url_hash, document_type, ...). | |
| 1293 | - * @param string $chunk_text The chunk text being stored. | |
| 1294 | - * @param array $context ['bot_id' => string, 'content_type' => string, 'source_url' => string, 'part_index' => int, 'part_total' => int] | |
| 1295 | - * @return array Updated metadata array. | |
| 1296 | - */ | |
| 1297 | - $chunk_metadata = apply_filters( | |
| 1298 | - 'mxchat_embedding_chunk_metadata', | |
| 1299 | - $chunk_metadata, | |
| 1300 | - $chunk_text, | |
| 1301 | - array( | |
| 1302 | - 'bot_id' => $bot_id, | |
| 1303 | - 'content_type' => $content_type, | |
| 1304 | - 'source_url' => $source_url, | |
| 1305 | - 'part_index' => (int) $index, | |
| 1306 | - 'part_total' => (int) $total_chunks, | |
| 1307 | - ) | |
| 1308 | - ); | |
| 1309 | - | |
| 1310 | 590 | $chunk_vector_id = MxChat_Chunker::generate_chunk_vector_id($source_url, $index); |
| 1311 | 591 | |
| 1312 | 592 | //error_log('[MXCHAT-CHUNK] Processing chunk ' . ($index + 1) . '/' . $total_chunks . ' (ID: ' . $chunk_vector_id . ')'); |
| 1313 | 593 | |
| @@ -1314,17 +594,10 @@ | ||
| 1314 | 594 | // Generate embedding for this chunk |
| 1315 | 595 | $embedding_vector = self::generate_embedding($chunk_text, $api_key, $bot_id); |
| 1316 | 596 | |
| 1317 | 597 | if (!is_array($embedding_vector)) { |
| 1318 | - // Track embedding failures separately from storage failures, and | |
| 1319 | - // keep the first provider reason seen — the two failure classes | |
| 1320 | - // have opposite remedies (API key vs Pinecone/DB) (plan 4a7c0a). | |
| 1321 | - $embed_failures++; | |
| 1322 | - $reason = is_wp_error($embedding_vector) ? $embedding_vector->get_error_message() : ''; | |
| 1323 | - if ($reason !== '' && $first_embed_reason === '') { | |
| 1324 | - $first_embed_reason = $reason; | |
| 1325 | - } | |
| 1326 | - $errors[] = new WP_Error('embedding_failed', 'Failed to generate embedding for chunk ' . $index . ($reason !== '' ? ' — ' . $reason : '')); | |
| 598 | + $errors[] = new WP_Error('embedding_failed', 'Failed to generate embedding for chunk ' . $index); | |
| 599 | + //error_log('[MXCHAT-CHUNK] Failed to generate embedding for chunk ' . $index); | |
| 1327 | 600 | continue; |
| 1328 | 601 | } |
| 1329 | 602 | |
| 1330 | 603 | if ($is_pinecone) { |
| @@ -1354,44 +627,20 @@ | ||
| 1354 | 627 | } |
| 1355 | 628 | |
| 1356 | 629 | if (is_wp_error($result)) { |
| 1357 | 630 | $errors[] = $result; |
| 1358 | - if ($first_store_reason === '') { | |
| 1359 | - $first_store_reason = $result->get_error_message(); | |
| 1360 | - } | |
| 631 | + //error_log('[MXCHAT-CHUNK] Failed to store chunk ' . $index . ': ' . $result->get_error_message()); | |
| 1361 | 632 | } |
| 1362 | 633 | } |
| 1363 | 634 | |
| 1364 | 635 | if (count($errors) === $total_chunks) { |
| 1365 | - // Say WHICH stage failed — "failed to store" used to cover pure | |
| 1366 | - // embedding failures too, sending customers to debug Pinecone when | |
| 1367 | - // the problem was their embedding API key (plan 4a7c0a). | |
| 1368 | - if ($embed_failures === $total_chunks) { | |
| 1369 | - return new WP_Error('chunking_failed', | |
| 1370 | - 'Failed to store any chunks — every chunk failed to embed' | |
| 1371 | - . ($first_embed_reason !== '' ? ': ' . $first_embed_reason : '') | |
| 1372 | - . ' Check the embedding provider API key and model under MxChat Settings.'); | |
| 1373 | - } | |
| 1374 | - if ($embed_failures === 0) { | |
| 1375 | - return new WP_Error('chunking_failed', | |
| 1376 | - 'Failed to store any chunks — embeddings generated but storage failed' | |
| 1377 | - . ($first_store_reason !== '' ? ': ' . $first_store_reason : '') | |
| 1378 | - . ' Check the knowledge base storage (Pinecone index or database).'); | |
| 1379 | - } | |
| 1380 | - return new WP_Error('chunking_failed', sprintf( | |
| 1381 | - 'Failed to store any chunks — %d failed to embed%s and %d failed to store%s', | |
| 1382 | - $embed_failures, | |
| 1383 | - $first_embed_reason !== '' ? ' (' . $first_embed_reason . ')' : '', | |
| 1384 | - $total_chunks - $embed_failures, | |
| 1385 | - $first_store_reason !== '' ? ' (' . $first_store_reason . ')' : '' | |
| 1386 | - )); | |
| 636 | + return new WP_Error('chunking_failed', 'Failed to store any chunks'); | |
| 1387 | 637 | } |
| 1388 | 638 | |
| 1389 | 639 | if (!empty($errors)) { |
| 1390 | - $detail = $first_embed_reason !== '' ? $first_embed_reason : $first_store_reason; | |
| 640 | + //error_log('[MXCHAT-CHUNK] Completed with ' . count($errors) . ' errors out of ' . $total_chunks . ' chunks'); | |
| 1391 | 641 | return new WP_Error('chunking_partial_failure', |
| 1392 | - sprintf('Failed to store %d of %d chunks', count($errors), $total_chunks) | |
| 1393 | - . ($detail !== '' ? ' — first error: ' . $detail : '')); | |
| 642 | + sprintf('Failed to store %d of %d chunks', count($errors), $total_chunks)); | |
| 1394 | 643 | } |
| 1395 | 644 | |
| 1396 | 645 | //error_log('[MXCHAT-CHUNK] Successfully stored all ' . $total_chunks . ' chunks'); |
| 1397 | 646 | return true; |
| @@ -1437,9 +686,9 @@ | ||
| 1437 | 686 | 'total_chunks' => $chunk_metadata['total_chunks'], |
| 1438 | 687 | 'parent_url_hash' => $chunk_metadata['parent_url_hash'], |
| 1439 | 688 | 'last_updated' => time(), |
| 1440 | 689 | 'created_at' => time(), |
| 1441 | - 'bot_id' => $bot_id, | |
| 690 | + 'bot_id' => $bot_id | |
| 1442 | 691 | ); |
| 1443 | 692 | |
| 1444 | 693 | $vector_data = array( |
| 1445 | 694 | 'id' => $vector_id, |
| @@ -1555,34 +804,31 @@ | ||
| 1555 | 804 | |
| 1556 | 805 | // Add the original single-vector ID (for non-chunked content) |
| 1557 | 806 | $vectors_to_delete[] = $base_vector_id; |
| 1558 | 807 | |
| 1559 | - // Pinecone /vectors/list is a GET endpoint with query-string parameters; a POST here returns a | |
| 1560 | - // non-200 silently and we end up only deleting the base vector, leaving chunks orphaned. | |
| 1561 | - $query_params = array( | |
| 808 | + // Use Pinecone list API to find all chunk vectors with this prefix | |
| 809 | + $list_url = "https://{$host}/vectors/list"; | |
| 810 | + | |
| 811 | + $list_body = array( | |
| 1562 | 812 | 'prefix' => $base_vector_id . '_chunk_', |
| 1563 | - 'limit' => 100, | |
| 813 | + 'limit' => 100 | |
| 1564 | 814 | ); |
| 815 | + | |
| 1565 | 816 | if (!empty($namespace)) { |
| 1566 | - $query_params['namespace'] = $namespace; | |
| 817 | + $list_body['namespace'] = $namespace; | |
| 1567 | 818 | } |
| 1568 | 819 | |
| 1569 | - $list_url = "https://{$host}/vectors/list?" . http_build_query($query_params); | |
| 820 | + $list_response = wp_remote_post($list_url, array( | |
| 821 | + 'headers' => array( | |
| 822 | + 'Api-Key' => $api_key, | |
| 823 | + 'accept' => 'application/json', | |
| 824 | + 'content-type' => 'application/json' | |
| 825 | + ), | |
| 826 | + 'body' => wp_json_encode($list_body), | |
| 827 | + 'timeout' => 30 | |
| 828 | + )); | |
| 1570 | 829 | |
| 1571 | - // Paginate in case a URL has more than 100 chunks. | |
| 1572 | - do { | |
| 1573 | - $list_response = wp_remote_get($list_url, array( | |
| 1574 | - 'headers' => array( | |
| 1575 | - 'Api-Key' => $api_key, | |
| 1576 | - 'accept' => 'application/json', | |
| 1577 | - ), | |
| 1578 | - 'timeout' => 30, | |
| 1579 | - )); | |
| 1580 | - | |
| 1581 | - if (is_wp_error($list_response) || wp_remote_retrieve_response_code($list_response) !== 200) { | |
| 1582 | - break; | |
| 1583 | - } | |
| 1584 | - | |
| 830 | + if (!is_wp_error($list_response)) { | |
| 1585 | 831 | $list_data = json_decode(wp_remote_retrieve_body($list_response), true); |
| 1586 | 832 | if (!empty($list_data['vectors'])) { |
| 1587 | 833 | foreach ($list_data['vectors'] as $vector) { |
| 1588 | 834 | if (isset($vector['id'])) { |
| @@ -1589,18 +835,10 @@ | ||
| 1589 | 835 | $vectors_to_delete[] = $vector['id']; |
| 1590 | 836 | } |
| 1591 | 837 | } |
| 1592 | 838 | } |
| 839 | + } | |
| 1593 | 840 | |
| 1594 | - $next_token = $list_data['pagination']['next'] ?? ''; | |
| 1595 | - if (empty($next_token)) { | |
| 1596 | - break; | |
| 1597 | - } | |
| 1598 | - | |
| 1599 | - $query_params['paginationToken'] = $next_token; | |
| 1600 | - $list_url = "https://{$host}/vectors/list?" . http_build_query($query_params); | |
| 1601 | - } while (true); | |
| 1602 | - | |
| 1603 | 841 | if (empty($vectors_to_delete)) { |
| 1604 | 842 | //error_log('[MXCHAT-CHUNK-DELETE] No vectors found to delete'); |
| 1605 | 843 | return true; |
| 1606 | 844 | } |
| @@ -1659,297 +897,6 @@ | ||
| 1659 | 897 | } |
| 1660 | 898 | |
| 1661 | 899 | //error_log('[MXCHAT-CHUNK-DELETE] Deleted ' . $result . ' rows from WordPress DB'); |
| 1662 | 900 | return true; |
| 1663 | -} | |
| 1664 | - | |
| 1665 | -/** | |
| 1666 | - * Hybrid keyword boost (plan-38ffa1): detect whether the WP-DB knowledge | |
| 1667 | - * table can serve the keyword leg via a MySQL FULLTEXT index, creating the | |
| 1668 | - * index if needed. Detection runs once and caches the answer in the | |
| 1669 | - * mxchat_hybrid_keyword_capability option ('fulltext' | 'like'); pass | |
| 1670 | - * $force to re-detect. LIKE is the graceful fallback for shared hosts | |
| 1671 | - * whose ALTER fails — the feature works either way, FULLTEXT just ranks | |
| 1672 | - * better and scales. | |
| 1673 | - * | |
| 1674 | - * @param bool $force Re-run detection even if a cached answer exists. | |
| 1675 | - * @return string 'fulltext' or 'like' | |
| 1676 | - */ | |
| 1677 | -public static function mxchat_hybrid_detect_capability($force = false) { | |
| 1678 | - $cached = get_option('mxchat_hybrid_keyword_capability', ''); | |
| 1679 | - if (!$force && in_array($cached, array('fulltext', 'like'), true)) { | |
| 1680 | - return $cached; | |
| 1681 | - } | |
| 1682 | - | |
| 1683 | - global $wpdb; | |
| 1684 | - $table = $wpdb->prefix . 'mxchat_system_prompt_content'; | |
| 1685 | - | |
| 1686 | - $index_exists = $wpdb->get_var("SHOW INDEX FROM {$table} WHERE Key_name = 'mxchat_content_ft'"); | |
| 1687 | - if (!$index_exists) { | |
| 1688 | - // Suppress the visible error on hosts where this is not permitted — | |
| 1689 | - // failure is an expected, handled outcome (LIKE fallback). | |
| 1690 | - $suppress = $wpdb->suppress_errors(true); | |
| 1691 | - $wpdb->query("ALTER TABLE {$table} ADD FULLTEXT INDEX mxchat_content_ft (article_content)"); | |
| 1692 | - $wpdb->suppress_errors($suppress); | |
| 1693 | - $index_exists = $wpdb->get_var("SHOW INDEX FROM {$table} WHERE Key_name = 'mxchat_content_ft'"); | |
| 1694 | - } | |
| 1695 | - | |
| 1696 | - $capability = $index_exists ? 'fulltext' : 'like'; | |
| 1697 | - update_option('mxchat_hybrid_keyword_capability', $capability); | |
| 1698 | - return $capability; | |
| 1699 | -} | |
| 1700 | - | |
| 1701 | -/** | |
| 1702 | - * Public entry for re-embedding already-stored content in place (wp mxchat | |
| 1703 | - * rtl-repair, plan d1e6f7). Thin wrapper so the repair CLI gets the exact | |
| 1704 | - * provider routing the import path uses — the repaired vector must come from | |
| 1705 | - * the same model family the bot indexes with, or retrieval stays broken. | |
| 1706 | - */ | |
| 1707 | -public static function regenerate_embedding($text, $api_key, $bot_id = 'default') { | |
| 1708 | - return self::generate_embedding($text, $api_key, $bot_id); | |
| 1709 | -} | |
| 1710 | - | |
| 1711 | -/** | |
| 1712 | - * Restore logical character order in PDF-extracted RTL text (plan 32bf9e). | |
| 1713 | - * | |
| 1714 | - * The bundled Smalot parser only un-reverses text runs tagged with the | |
| 1715 | - * ReversedChars marked-content operator (Word emits it; LibreOffice and most | |
| 1716 | - * other producers do not), so their Hebrew/Arabic PDFs extract in visual | |
| 1717 | - * (reversed) order and embed/search as garbage. This is OUR post-processing | |
| 1718 | - * seam over getText() — the parser itself is never patched (it gets replaced | |
| 1719 | - * wholesale on library updates). | |
| 1720 | - * | |
| 1721 | - * Heuristic and deliberately conservative, per line: | |
| 1722 | - * - lines without strong RTL codepoints are untouched (a fully-Latin line in | |
| 1723 | - * an RTL document therefore stays as extracted — accepted limitation); | |
| 1724 | - * - Arabic presentation forms are a definitive visual-order signal (they only | |
| 1725 | - * appear in shaped output): de-shape to base letters and reverse; | |
| 1726 | - * - otherwise flip only on positive evidence — Hebrew final-letter position | |
| 1727 | - * (a sofit at word START only happens in reversed text) or sentence | |
| 1728 | - * punctuation position (leading in visual order, trailing in logical); | |
| 1729 | - * - ambiguous lines are left alone: a conservative miss beats corrupting a | |
| 1730 | - * Word-produced extraction the parser already handled (the double-flip | |
| 1731 | - * guard this plan's approval named mandatory). | |
| 1732 | - * | |
| 1733 | - * @param string $text One extracted page string, straight from getText(). | |
| 1734 | - * @param string $context Caller tag for the Debug Mode entry (site + page). | |
| 1735 | - * @return string Text with RTL lines restored to logical order. | |
| 1736 | - */ | |
| 1737 | -public static function normalize_pdf_rtl($text, $context = '') { | |
| 1738 | - if (!is_string($text) || '' === $text) { | |
| 1739 | - return $text; | |
| 1740 | - } | |
| 1741 | - // Escape hatch for sites whose PDFs already extract logically. | |
| 1742 | - if (!apply_filters('mxchat_pdf_rtl_normalize', true, $text)) { | |
| 1743 | - return $text; | |
| 1744 | - } | |
| 1745 | - // Fast bail: nothing RTL anywhere in the page. | |
| 1746 | - if (!preg_match('/[\x{0590}-\x{05FF}\x{0600}-\x{06FF}\x{0750}-\x{077F}\x{FB50}-\x{FDFF}\x{FE70}-\x{FEFF}]/u', $text)) { | |
| 1747 | - return $text; | |
| 1748 | - } | |
| 1749 | - | |
| 1750 | - $parts = preg_split('/(\R)/u', $text, -1, PREG_SPLIT_DELIM_CAPTURE); | |
| 1751 | - if (false === $parts) { | |
| 1752 | - return $text; | |
| 1753 | - } | |
| 1754 | - | |
| 1755 | - $flipped_lines = 0; | |
| 1756 | - $deshaped_lines = 0; | |
| 1757 | - foreach ($parts as $i => $part) { | |
| 1758 | - if ('' === $part || preg_match('/^\R$/u', $part)) { | |
| 1759 | - continue; | |
| 1760 | - } | |
| 1761 | - $was_flipped = false; | |
| 1762 | - $was_deshaped = false; | |
| 1763 | - $new = self::pdf_rtl_normalize_line($part, $was_flipped, $was_deshaped); | |
| 1764 | - if ($new !== $part) { | |
| 1765 | - $parts[$i] = $new; | |
| 1766 | - } | |
| 1767 | - if ($was_flipped) { | |
| 1768 | - $flipped_lines++; | |
| 1769 | - } | |
| 1770 | - if ($was_deshaped) { | |
| 1771 | - $deshaped_lines++; | |
| 1772 | - } | |
| 1773 | - } | |
| 1774 | - | |
| 1775 | - if (($flipped_lines || $deshaped_lines) && class_exists('MxChat_Admin')) { | |
| 1776 | - MxChat_Admin::mxchat_log_debug('pdf_rtl_normalized', 'RTL PDF text restored to logical order', array( | |
| 1777 | - 'context' => (string) $context, | |
| 1778 | - 'lines_flipped' => $flipped_lines, | |
| 1779 | - 'lines_deshaped' => $deshaped_lines, | |
| 1780 | - 'decision' => 'visual-order extraction detected', | |
| 1781 | - )); | |
| 1782 | - } | |
| 1783 | - | |
| 1784 | - return implode('', $parts); | |
| 1785 | -} | |
| 1786 | - | |
| 1787 | -/** | |
| 1788 | - * Normalize one line. Sets $flipped/$deshaped for the caller's debug entry. | |
| 1789 | - */ | |
| 1790 | -private static function pdf_rtl_normalize_line($line, &$flipped, &$deshaped) { | |
| 1791 | - $flipped = false; | |
| 1792 | - $deshaped = false; | |
| 1793 | - | |
| 1794 | - if (!preg_match('/[\x{0590}-\x{05FF}\x{0600}-\x{06FF}\x{0750}-\x{077F}\x{FB50}-\x{FDFF}\x{FE70}-\x{FEFF}]/u', $line)) { | |
| 1795 | - return $line; | |
| 1796 | - } | |
| 1797 | - | |
| 1798 | - $has_forms = (bool) preg_match('/[\x{FB50}-\x{FDFF}\x{FE70}-\x{FEFF}]/u', $line); | |
| 1799 | - $work = $line; | |
| 1800 | - if ($has_forms) { | |
| 1801 | - $work = strtr($work, self::pdf_rtl_deshape_map()); | |
| 1802 | - $deshaped = ($work !== $line); | |
| 1803 | - } | |
| 1804 | - | |
| 1805 | - $verdict = 'ambiguous'; | |
| 1806 | - if ($has_forms) { | |
| 1807 | - // Shaped glyph codepoints only exist in visual-order output. | |
| 1808 | - $verdict = 'visual'; | |
| 1809 | - } else { | |
| 1810 | - // Strong-direction dominance gate first: an LTR-dominant line with an | |
| 1811 | - // embedded RTL word is not flip material. | |
| 1812 | - $rtl_count = preg_match_all('/[\x{0590}-\x{05FF}\x{0600}-\x{06FF}\x{0750}-\x{077F}]/u', $work, $m_rtl); | |
| 1813 | - $ltr_count = preg_match_all('/[A-Za-z]/u', $work, $m_ltr); | |
| 1814 | - if ($rtl_count < 1 || $rtl_count <= $ltr_count) { | |
| 1815 | - return $line; | |
| 1816 | - } | |
| 1817 | - | |
| 1818 | - // Hebrew final letters (ך ם ן ף ץ) end words in logical text; one at | |
| 1819 | - // a word START (Hebrew letter follows, none precedes) is reversal | |
| 1820 | - // evidence. Positional, so it survives the line being reversed. | |
| 1821 | - $sofit_initial = preg_match_all('/(?<![\x{05D0}-\x{05EA}])[\x{05DA}\x{05DD}\x{05DF}\x{05E3}\x{05E5}](?=[\x{05D0}-\x{05EA}])/u', $work, $m_i); | |
| 1822 | - $sofit_terminal = preg_match_all('/(?<=[\x{05D0}-\x{05EA}])[\x{05DA}\x{05DD}\x{05DF}\x{05E3}\x{05E5}](?![\x{05D0}-\x{05EA}])/u', $work, $m_t); | |
| 1823 | - if ($sofit_initial > $sofit_terminal) { | |
| 1824 | - $verdict = 'visual'; | |
| 1825 | - } elseif ($sofit_terminal > $sofit_initial) { | |
| 1826 | - $verdict = 'logical'; | |
| 1827 | - } else { | |
| 1828 | - // Sentence punctuation lands at the visual LEFT edge of an RTL | |
| 1829 | - // line, i.e. the START of a visual-order extraction. | |
| 1830 | - $trimmed = trim($work); | |
| 1831 | - $starts_punct = (bool) preg_match('/^[.?!:;,]/u', $trimmed); | |
| 1832 | - $ends_punct = (bool) preg_match('/[.?!:;,]$/u', $trimmed); | |
| 1833 | - if ($starts_punct && !$ends_punct) { | |
| 1834 | - $verdict = 'visual'; | |
| 1835 | - } elseif ($ends_punct && !$starts_punct) { | |
| 1836 | - $verdict = 'logical'; | |
| 1837 | - } | |
| 1838 | - } | |
| 1839 | - } | |
| 1840 | - | |
| 1841 | - if ('visual' !== $verdict) { | |
| 1842 | - // Ambiguous or logical: hand back the original line UNLESS we | |
| 1843 | - // de-shaped (de-shaping alone is always safe — same letters, same | |
| 1844 | - // order, un-ligated). | |
| 1845 | - return $deshaped ? $work : $line; | |
| 1846 | - } | |
| 1847 | - | |
| 1848 | - $flipped = true; | |
| 1849 | - return self::pdf_rtl_flip_line($work); | |
| 1850 | -} | |
| 1851 | - | |
| 1852 | -/** | |
| 1853 | - * Reverse a visual-order line into logical order: full character reversal, | |
| 1854 | - * mirror paired punctuation, then re-reverse embedded LTR runs (Latin words | |
| 1855 | - * and digit sequences, incl. Arabic-Indic digits) so they stay readable. | |
| 1856 | - */ | |
| 1857 | -private static function pdf_rtl_flip_line($line) { | |
| 1858 | - $chars = preg_split('//u', $line, -1, PREG_SPLIT_NO_EMPTY); | |
| 1859 | - if (false === $chars) { | |
| 1860 | - return $line; | |
| 1861 | - } | |
| 1862 | - $reversed = implode('', array_reverse($chars)); | |
| 1863 | - $reversed = strtr($reversed, array( | |
| 1864 | - '(' => ')', ')' => '(', | |
| 1865 | - '[' => ']', ']' => '[', | |
| 1866 | - '{' => '}', '}' => '{', | |
| 1867 | - '<' => '>', '>' => '<', | |
| 1868 | - )); | |
| 1869 | - $restored = preg_replace_callback( | |
| 1870 | - '/[0-9A-Za-z\x{0660}-\x{0669}\x{06F0}-\x{06F9}](?:[0-9A-Za-z\x{0660}-\x{0669}\x{06F0}-\x{06F9} .,\'"%\-:\/]*[0-9A-Za-z\x{0660}-\x{0669}\x{06F0}-\x{06F9}])?/u', | |
| 1871 | - function ($m) { | |
| 1872 | - $run = preg_split('//u', $m[0], -1, PREG_SPLIT_NO_EMPTY); | |
| 1873 | - return false === $run ? $m[0] : implode('', array_reverse($run)); | |
| 1874 | - }, | |
| 1875 | - $reversed | |
| 1876 | - ); | |
| 1877 | - return null === $restored ? $reversed : $restored; | |
| 1878 | -} | |
| 1879 | - | |
| 1880 | -/** | |
| 1881 | - * Arabic presentation forms (A + B) -> base letters. Built once from range | |
| 1882 | - * specs rather than ~120 hand-written literal entries; every codepoint in a | |
| 1883 | - * range maps to the same base sequence (isolated/final/initial/medial forms | |
| 1884 | - * of one letter are contiguous in the FE70 block). | |
| 1885 | - */ | |
| 1886 | -private static function pdf_rtl_deshape_map() { | |
| 1887 | - static $map = null; | |
| 1888 | - if (null !== $map) { | |
| 1889 | - return $map; | |
| 1890 | - } | |
| 1891 | - $ranges = array( | |
| 1892 | - // Form B harakat (each pair = standalone + tatweel-joined form). | |
| 1893 | - array(0xFE70, 0xFE71, array(0x064B)), array(0xFE72, 0xFE72, array(0x064C)), | |
| 1894 | - array(0xFE74, 0xFE74, array(0x064D)), array(0xFE76, 0xFE77, array(0x064E)), | |
| 1895 | - array(0xFE78, 0xFE79, array(0x064F)), array(0xFE7A, 0xFE7B, array(0x0650)), | |
| 1896 | - array(0xFE7C, 0xFE7D, array(0x0651)), array(0xFE7E, 0xFE7F, array(0x0652)), | |
| 1897 | - // Form B letters. | |
| 1898 | - array(0xFE80, 0xFE80, array(0x0621)), array(0xFE81, 0xFE82, array(0x0622)), | |
| 1899 | - array(0xFE83, 0xFE84, array(0x0623)), array(0xFE85, 0xFE86, array(0x0624)), | |
| 1900 | - array(0xFE87, 0xFE88, array(0x0625)), array(0xFE89, 0xFE8C, array(0x0626)), | |
| 1901 | - array(0xFE8D, 0xFE8E, array(0x0627)), array(0xFE8F, 0xFE92, array(0x0628)), | |
| 1902 | - array(0xFE93, 0xFE94, array(0x0629)), array(0xFE95, 0xFE98, array(0x062A)), | |
| 1903 | - array(0xFE99, 0xFE9C, array(0x062B)), array(0xFE9D, 0xFEA0, array(0x062C)), | |
| 1904 | - array(0xFEA1, 0xFEA4, array(0x062D)), array(0xFEA5, 0xFEA8, array(0x062E)), | |
| 1905 | - array(0xFEA9, 0xFEAA, array(0x062F)), array(0xFEAB, 0xFEAC, array(0x0630)), | |
| 1906 | - array(0xFEAD, 0xFEAE, array(0x0631)), array(0xFEAF, 0xFEB0, array(0x0632)), | |
| 1907 | - array(0xFEB1, 0xFEB4, array(0x0633)), array(0xFEB5, 0xFEB8, array(0x0634)), | |
| 1908 | - array(0xFEB9, 0xFEBC, array(0x0635)), array(0xFEBD, 0xFEC0, array(0x0636)), | |
| 1909 | - array(0xFEC1, 0xFEC4, array(0x0637)), array(0xFEC5, 0xFEC8, array(0x0638)), | |
| 1910 | - array(0xFEC9, 0xFECC, array(0x0639)), array(0xFECD, 0xFED0, array(0x063A)), | |
| 1911 | - array(0xFED1, 0xFED4, array(0x0641)), array(0xFED5, 0xFED8, array(0x0642)), | |
| 1912 | - array(0xFED9, 0xFEDC, array(0x0643)), array(0xFEDD, 0xFEE0, array(0x0644)), | |
| 1913 | - array(0xFEE1, 0xFEE4, array(0x0645)), array(0xFEE5, 0xFEE8, array(0x0646)), | |
| 1914 | - array(0xFEE9, 0xFEEC, array(0x0647)), array(0xFEED, 0xFEEE, array(0x0648)), | |
| 1915 | - array(0xFEEF, 0xFEF0, array(0x0649)), array(0xFEF1, 0xFEF4, array(0x064A)), | |
| 1916 | - // Form B lam-alef ligatures decompose to two letters. | |
| 1917 | - array(0xFEF5, 0xFEF6, array(0x0644, 0x0622)), array(0xFEF7, 0xFEF8, array(0x0644, 0x0623)), | |
| 1918 | - array(0xFEF9, 0xFEFA, array(0x0644, 0x0625)), array(0xFEFB, 0xFEFC, array(0x0644, 0x0627)), | |
| 1919 | - // Form A: Persian / Urdu letters in common use. | |
| 1920 | - array(0xFB56, 0xFB59, array(0x067E)), array(0xFB66, 0xFB69, array(0x0679)), | |
| 1921 | - array(0xFB7A, 0xFB7D, array(0x0686)), array(0xFB88, 0xFB89, array(0x0688)), | |
| 1922 | - array(0xFB8A, 0xFB8B, array(0x0698)), array(0xFB8E, 0xFB91, array(0x06A9)), | |
| 1923 | - array(0xFB92, 0xFB95, array(0x06AF)), array(0xFBA6, 0xFBA9, array(0x06C1)), | |
| 1924 | - array(0xFBAA, 0xFBAD, array(0x06BE)), array(0xFBAE, 0xFBAF, array(0x06D2)), | |
| 1925 | - array(0xFBFC, 0xFBFF, array(0x06CC)), | |
| 1926 | - ); | |
| 1927 | - $map = array(); | |
| 1928 | - foreach ($ranges as $range) { | |
| 1929 | - $base = ''; | |
| 1930 | - foreach ($range[2] as $cp) { | |
| 1931 | - $base .= self::pdf_rtl_cp_to_utf8($cp); | |
| 1932 | - } | |
| 1933 | - for ($cp = $range[0]; $cp <= $range[1]; $cp++) { | |
| 1934 | - $map[self::pdf_rtl_cp_to_utf8($cp)] = $base; | |
| 1935 | - } | |
| 1936 | - } | |
| 1937 | - return $map; | |
| 1938 | -} | |
| 1939 | - | |
| 1940 | -/** | |
| 1941 | - * Codepoint to UTF-8 without ext-intl / mbstring entity tricks (PHP 7.2 floor). | |
| 1942 | - */ | |
| 1943 | -private static function pdf_rtl_cp_to_utf8($cp) { | |
| 1944 | - if ($cp < 0x80) { | |
| 1945 | - return chr($cp); | |
| 1946 | - } | |
| 1947 | - if ($cp < 0x800) { | |
| 1948 | - return chr(0xC0 | ($cp >> 6)) . chr(0x80 | ($cp & 0x3F)); | |
| 1949 | - } | |
| 1950 | - if ($cp < 0x10000) { | |
| 1951 | - return chr(0xE0 | ($cp >> 12)) . chr(0x80 | (($cp >> 6) & 0x3F)) . chr(0x80 | ($cp & 0x3F)); | |
| 1952 | - } | |
| 1953 | - return chr(0xF0 | ($cp >> 18)) . chr(0x80 | (($cp >> 12) & 0x3F)) . chr(0x80 | (($cp >> 6) & 0x3F)) . chr(0x80 | ($cp & 0x3F)); | |
| 1954 | 901 | } |
| 1955 | 902 | } |
