| 1 |
<?php |
| 2 |
/** |
| 3 |
* MxChat REST API |
| 4 |
* |
| 5 |
* Bearer-token-authenticated REST endpoints exposing core primitives: |
| 6 |
* |
| 7 |
* GET /wp-json/mxchat/v1/transcripts — read chat transcripts (filterable) |
| 8 |
* DELETE /wp-json/mxchat/v1/transcripts — delete by session_ids (cascades) |
| 9 |
* POST /wp-json/mxchat/v1/knowledge — push content into the knowledge base |
| 10 |
* DELETE /wp-json/mxchat/v1/knowledge — remove entries by source_url |
| 11 |
* GET /wp-json/mxchat/v1/leads — read lead-capture submissions |
| 12 |
* GET /wp-json/mxchat/v1/health — connectivity + capability check |
| 13 |
* |
| 14 |
* These are general-purpose primitives. They power the official MxChat FAQ |
| 15 |
* agent, but anyone running MxChat can use them for analytics exports, |
| 16 |
* external automations (n8n, Zapier, Make), data migrations, custom RAG |
| 17 |
* pipelines, etc. |
| 18 |
* |
| 19 |
* Auth: |
| 20 |
* Bearer token stored in wp_options under `mxchat_api_token`. |
| 21 |
* - When the token is empty (default), all authenticated routes are |
| 22 |
* locked. The endpoints simply refuse with 401, so the surface area |
| 23 |
* is zero until the site owner explicitly enables it from the |
| 24 |
* "MxChat → API Access" admin page (or via WP-CLI). |
| 25 |
* - Comparison uses hash_equals() for constant-time safety. |
| 26 |
* |
| 27 |
* Privacy: |
| 28 |
* The /transcripts endpoint returns user-submitted chat data. It is |
| 29 |
* intentionally gated behind a token the site owner generates. No data |
| 30 |
* leaves the site unsolicited. |
| 31 |
* |
| 32 |
* @package MxChat |
| 33 |
* @since 3.2.5 |
| 34 |
*/ |
| 35 |
|
| 36 |
if (!defined('ABSPATH')) { |
| 37 |
exit; |
| 38 |
} |
| 39 |
|
| 40 |
class MxChat_Rest_Api { |
| 41 |
|
| 42 |
const REST_NAMESPACE = 'mxchat/v1'; |
| 43 |
const TOKEN_OPTION = 'mxchat_api_token'; |
| 44 |
|
| 45 |
public function __construct() { |
| 46 |
add_action('rest_api_init', array($this, 'register_routes')); |
| 47 |
} |
| 48 |
|
| 49 |
/** |
| 50 |
* Register the REST routes. |
| 51 |
*/ |
| 52 |
public function register_routes() { |
| 53 |
register_rest_route(self::REST_NAMESPACE, '/health', array( |
| 54 |
array( |
| 55 |
'methods' => WP_REST_Server::READABLE, |
| 56 |
'callback' => array($this, 'handle_health'), |
| 57 |
'permission_callback' => '__return_true', |
| 58 |
), |
| 59 |
)); |
| 60 |
|
| 61 |
register_rest_route(self::REST_NAMESPACE, '/transcripts', array( |
| 62 |
array( |
| 63 |
'methods' => WP_REST_Server::READABLE, |
| 64 |
'callback' => array($this, 'handle_get_transcripts'), |
| 65 |
'permission_callback' => array($this, 'check_bearer_token'), |
| 66 |
'args' => array( |
| 67 |
'since' => array( |
| 68 |
'description' => __('Only return rows with timestamp >= this value. Accepts ISO 8601 (2026-05-07T00:00:00Z) or any strtotime-compatible string.', 'mxchat'), |
| 69 |
'type' => 'string', |
| 70 |
'required' => false, |
| 71 |
'sanitize_callback' => 'sanitize_text_field', |
| 72 |
), |
| 73 |
'until' => array( |
| 74 |
'description' => __('Only return rows with timestamp <= this value. Same format as `since`.', 'mxchat'), |
| 75 |
'type' => 'string', |
| 76 |
'required' => false, |
| 77 |
'sanitize_callback' => 'sanitize_text_field', |
| 78 |
), |
| 79 |
'session_id' => array( |
| 80 |
'description' => __('Filter to a single conversation.', 'mxchat'), |
| 81 |
'type' => 'string', |
| 82 |
'required' => false, |
| 83 |
'sanitize_callback' => 'sanitize_text_field', |
| 84 |
), |
| 85 |
'role' => array( |
| 86 |
'description' => __('Filter by role. One of: user, assistant, all (default: all).', 'mxchat'), |
| 87 |
'type' => 'string', |
| 88 |
'required' => false, |
| 89 |
'enum' => array('user', 'assistant', 'all'), |
| 90 |
'default' => 'all', |
| 91 |
), |
| 92 |
'has_rag_context' => array( |
| 93 |
'description' => __('Filter by whether the row has retrieved RAG context. Useful for finding "no-knowledge-hit" answers. One of: yes, no, any (default: any).', 'mxchat'), |
| 94 |
'type' => 'string', |
| 95 |
'required' => false, |
| 96 |
'enum' => array('yes', 'no', 'any'), |
| 97 |
'default' => 'any', |
| 98 |
), |
| 99 |
'limit' => array( |
| 100 |
'description' => __('Max number of rows to return. 1-1000, default 100.', 'mxchat'), |
| 101 |
'type' => 'integer', |
| 102 |
'required' => false, |
| 103 |
'default' => 100, |
| 104 |
'minimum' => 1, |
| 105 |
'maximum' => 1000, |
| 106 |
), |
| 107 |
'offset' => array( |
| 108 |
'description' => __('Skip this many rows (for pagination).', 'mxchat'), |
| 109 |
'type' => 'integer', |
| 110 |
'required' => false, |
| 111 |
'default' => 0, |
| 112 |
'minimum' => 0, |
| 113 |
), |
| 114 |
'order' => array( |
| 115 |
'description' => __('Sort order on timestamp. asc or desc (default: desc).', 'mxchat'), |
| 116 |
'type' => 'string', |
| 117 |
'required' => false, |
| 118 |
'enum' => array('asc', 'desc'), |
| 119 |
'default' => 'desc', |
| 120 |
), |
| 121 |
), |
| 122 |
), |
| 123 |
)); |
| 124 |
|
| 125 |
register_rest_route(self::REST_NAMESPACE, '/transcripts', array( |
| 126 |
array( |
| 127 |
'methods' => WP_REST_Server::DELETABLE, |
| 128 |
'callback' => array($this, 'handle_delete_transcripts'), |
| 129 |
'permission_callback' => array($this, 'check_bearer_token'), |
| 130 |
'args' => array( |
| 131 |
'session_ids' => array( |
| 132 |
'description' => __('Array of session_ids to delete. Required and non-empty — there is intentionally no "delete all" shorthand.', 'mxchat'), |
| 133 |
'type' => 'array', |
| 134 |
'required' => true, |
| 135 |
'items' => array('type' => 'string'), |
| 136 |
), |
| 137 |
'cascade' => array( |
| 138 |
'description' => __('If true (default), also deletes related rows in mxchat_transcript_translations and mxchat_url_clicks for each session_id.', 'mxchat'), |
| 139 |
'type' => 'boolean', |
| 140 |
'required' => false, |
| 141 |
'default' => true, |
| 142 |
), |
| 143 |
), |
| 144 |
), |
| 145 |
)); |
| 146 |
|
| 147 |
register_rest_route(self::REST_NAMESPACE, '/knowledge', array( |
| 148 |
array( |
| 149 |
'methods' => WP_REST_Server::CREATABLE, |
| 150 |
'callback' => array($this, 'handle_post_knowledge'), |
| 151 |
'permission_callback' => array($this, 'check_bearer_token'), |
| 152 |
'args' => array( |
| 153 |
'content' => array( |
| 154 |
'description' => __('The content body to embed and store.', 'mxchat'), |
| 155 |
'type' => 'string', |
| 156 |
'required' => true, |
| 157 |
), |
| 158 |
'source_url' => array( |
| 159 |
'description' => __('Canonical URL the content came from. Used as the dedupe key — submitting the same URL replaces the existing entry.', 'mxchat'), |
| 160 |
'type' => 'string', |
| 161 |
'required' => true, |
| 162 |
'sanitize_callback' => 'esc_url_raw', |
| 163 |
), |
| 164 |
'bot_id' => array( |
| 165 |
'description' => __('Bot ID (for multi-bot installs). Defaults to "default".', 'mxchat'), |
| 166 |
'type' => 'string', |
| 167 |
'required' => false, |
| 168 |
'default' => 'default', |
| 169 |
'sanitize_callback' => 'sanitize_key', |
| 170 |
), |
| 171 |
'content_type' => array( |
| 172 |
'description' => __('Free-form content type label, e.g. content, manual, faq, page, post. Stored alongside the entry for filtering. Defaults to "manual".', 'mxchat'), |
| 173 |
'type' => 'string', |
| 174 |
'required' => false, |
| 175 |
'default' => 'manual', |
| 176 |
'sanitize_callback' => 'sanitize_key', |
| 177 |
), |
| 178 |
), |
| 179 |
), |
| 180 |
)); |
| 181 |
|
| 182 |
register_rest_route(self::REST_NAMESPACE, '/knowledge', array( |
| 183 |
array( |
| 184 |
'methods' => WP_REST_Server::DELETABLE, |
| 185 |
'callback' => array($this, 'handle_delete_knowledge'), |
| 186 |
'permission_callback' => array($this, 'check_bearer_token'), |
| 187 |
'args' => array( |
| 188 |
'source_url' => array( |
| 189 |
'description' => __('The source_url of the entry to remove — the same value used when it was pushed. Accepts a single string or an array of up to 50 URLs so a sync job can batch. Removing a URL that is not in the knowledge base is not an error: the result reports deleted:false for it.', 'mxchat'), |
| 190 |
'required' => true, |
| 191 |
), |
| 192 |
'bot_id' => array( |
| 193 |
'description' => __('Bot ID (for multi-bot installs). Defaults to "default".', 'mxchat'), |
| 194 |
'type' => 'string', |
| 195 |
'required' => false, |
| 196 |
'default' => 'default', |
| 197 |
'sanitize_callback' => 'sanitize_key', |
| 198 |
), |
| 199 |
), |
| 200 |
), |
| 201 |
)); |
| 202 |
|
| 203 |
register_rest_route(self::REST_NAMESPACE, '/leads', array( |
| 204 |
array( |
| 205 |
'methods' => WP_REST_Server::READABLE, |
| 206 |
'callback' => array($this, 'handle_get_leads'), |
| 207 |
'permission_callback' => array($this, 'check_bearer_token'), |
| 208 |
'args' => array( |
| 209 |
'since' => array( |
| 210 |
'description' => __('Only return leads last seen at or after this value. Accepts ISO 8601 (2026-05-07T00:00:00Z) or any strtotime-compatible string. Orphan leads (captured but never chatted) have no timestamp and are omitted when this is set.', 'mxchat'), |
| 211 |
'type' => 'string', |
| 212 |
'required' => false, |
| 213 |
'sanitize_callback' => 'sanitize_text_field', |
| 214 |
), |
| 215 |
'search' => array( |
| 216 |
'description' => __('Substring match on email or name.', 'mxchat'), |
| 217 |
'type' => 'string', |
| 218 |
'required' => false, |
| 219 |
'sanitize_callback' => 'sanitize_text_field', |
| 220 |
), |
| 221 |
'status' => array( |
| 222 |
'description' => __('Filter by lead source. One of: all, orphan (captured but never chatted), chat_deleted (conversation removed by an admin). Default: all.', 'mxchat'), |
| 223 |
'type' => 'string', |
| 224 |
'required' => false, |
| 225 |
'enum' => array('all', 'orphan', 'chat_deleted'), |
| 226 |
'default' => 'all', |
| 227 |
), |
| 228 |
'page' => array( |
| 229 |
'description' => __('1-based page number.', 'mxchat'), |
| 230 |
'type' => 'integer', |
| 231 |
'required' => false, |
| 232 |
'default' => 1, |
| 233 |
'minimum' => 1, |
| 234 |
), |
| 235 |
'per_page' => array( |
| 236 |
'description' => __('Leads per page. 1-200, default 100.', 'mxchat'), |
| 237 |
'type' => 'integer', |
| 238 |
'required' => false, |
| 239 |
'default' => 100, |
| 240 |
'minimum' => 1, |
| 241 |
'maximum' => 200, |
| 242 |
), |
| 243 |
'order' => array( |
| 244 |
'description' => __('Sort order on last_seen. asc or desc (default: desc).', 'mxchat'), |
| 245 |
'type' => 'string', |
| 246 |
'required' => false, |
| 247 |
'enum' => array('asc', 'desc'), |
| 248 |
'default' => 'desc', |
| 249 |
), |
| 250 |
), |
| 251 |
), |
| 252 |
)); |
| 253 |
} |
| 254 |
|
| 255 |
/** |
| 256 |
* Permission callback: validate bearer token in constant time. |
| 257 |
* Returns true on success, WP_Error on failure. |
| 258 |
* |
| 259 |
* @param WP_REST_Request $request |
| 260 |
* @return bool|WP_Error |
| 261 |
*/ |
| 262 |
public function check_bearer_token($request) { |
| 263 |
$stored = (string) get_option(self::TOKEN_OPTION, ''); |
| 264 |
if ($stored === '') { |
| 265 |
return new WP_Error( |
| 266 |
'mxchat_rest_disabled', |
| 267 |
__('MxChat REST API is disabled. Generate a token from MxChat → API Access in the WordPress admin.', 'mxchat'), |
| 268 |
array('status' => 401) |
| 269 |
); |
| 270 |
} |
| 271 |
|
| 272 |
$auth = $request->get_header('authorization'); |
| 273 |
if (!is_string($auth) || $auth === '') { |
| 274 |
return new WP_Error( |
| 275 |
'mxchat_rest_no_auth', |
| 276 |
__('Missing Authorization header. Expected: Authorization: Bearer <token>', 'mxchat'), |
| 277 |
array('status' => 401) |
| 278 |
); |
| 279 |
} |
| 280 |
|
| 281 |
if (!preg_match('/^Bearer\s+(.+)$/i', $auth, $matches)) { |
| 282 |
return new WP_Error( |
| 283 |
'mxchat_rest_bad_auth', |
| 284 |
__('Malformed Authorization header. Expected: Authorization: Bearer <token>', 'mxchat'), |
| 285 |
array('status' => 401) |
| 286 |
); |
| 287 |
} |
| 288 |
|
| 289 |
$provided = trim($matches[1]); |
| 290 |
if ($provided === '' || !hash_equals($stored, $provided)) { |
| 291 |
return new WP_Error( |
| 292 |
'mxchat_rest_bad_token', |
| 293 |
__('Invalid API token.', 'mxchat'), |
| 294 |
array('status' => 401) |
| 295 |
); |
| 296 |
} |
| 297 |
|
| 298 |
return true; |
| 299 |
} |
| 300 |
|
| 301 |
/** |
| 302 |
* GET /health — public, no auth. |
| 303 |
* Reports whether the API is configured and which capabilities are present. |
| 304 |
*/ |
| 305 |
public function handle_health($request) { |
| 306 |
$token_set = (string) get_option(self::TOKEN_OPTION, '') !== ''; |
| 307 |
|
| 308 |
$options = get_option('mxchat_options', array()); |
| 309 |
// The effective model: custom:<model> when Custom Provider embeddings are on (plan 418afd). |
| 310 |
$embedding_model = class_exists('MxChat_Utils') |
| 311 |
? (string) MxChat_Utils::get_selected_embedding_model(is_array($options) ? $options : array()) |
| 312 |
: (isset($options['embedding_model']) ? (string) $options['embedding_model'] : ''); |
| 313 |
|
| 314 |
return rest_ensure_response(array( |
| 315 |
'ok' => true, |
| 316 |
'plugin_version' => defined('MXCHAT_VERSION') ? MXCHAT_VERSION : null, |
| 317 |
'token_set' => $token_set, |
| 318 |
'embedding_model' => $embedding_model, |
| 319 |
'utils_loaded' => class_exists('MxChat_Utils'), |
| 320 |
'namespace' => self::REST_NAMESPACE, |
| 321 |
)); |
| 322 |
} |
| 323 |
|
| 324 |
/** |
| 325 |
* GET /transcripts — return transcript rows with optional filters. |
| 326 |
* Always uses prepared statements; user-supplied strings never concatenated into SQL. |
| 327 |
*/ |
| 328 |
public function handle_get_transcripts($request) { |
| 329 |
global $wpdb; |
| 330 |
$table = $wpdb->prefix . 'mxchat_chat_transcripts'; |
| 331 |
|
| 332 |
if ($wpdb->get_var($wpdb->prepare("SHOW TABLES LIKE %s", $table)) !== $table) { |
| 333 |
return new WP_Error( |
| 334 |
'mxchat_rest_no_table', |
| 335 |
__('Transcripts table does not exist on this site.', 'mxchat'), |
| 336 |
array('status' => 500) |
| 337 |
); |
| 338 |
} |
| 339 |
|
| 340 |
$since = (string) $request->get_param('since'); |
| 341 |
$until = (string) $request->get_param('until'); |
| 342 |
$session_id = (string) $request->get_param('session_id'); |
| 343 |
$role = (string) $request->get_param('role'); |
| 344 |
$has_rag_context = (string) $request->get_param('has_rag_context'); |
| 345 |
$limit = (int) $request->get_param('limit'); |
| 346 |
$offset = (int) $request->get_param('offset'); |
| 347 |
$order = strtolower((string) $request->get_param('order')) === 'asc' ? 'ASC' : 'DESC'; |
| 348 |
|
| 349 |
$where = array('1=1'); |
| 350 |
$params = array(); |
| 351 |
|
| 352 |
if ($since !== '') { |
| 353 |
$ts = $this->parse_datetime($since); |
| 354 |
if ($ts === null) { |
| 355 |
return new WP_Error('mxchat_rest_bad_since', __('Could not parse `since` parameter as a date.', 'mxchat'), array('status' => 400)); |
| 356 |
} |
| 357 |
$where[] = 'timestamp >= %s'; |
| 358 |
$params[] = gmdate('Y-m-d H:i:s', $ts); |
| 359 |
} |
| 360 |
if ($until !== '') { |
| 361 |
$ts = $this->parse_datetime($until); |
| 362 |
if ($ts === null) { |
| 363 |
return new WP_Error('mxchat_rest_bad_until', __('Could not parse `until` parameter as a date.', 'mxchat'), array('status' => 400)); |
| 364 |
} |
| 365 |
$where[] = 'timestamp <= %s'; |
| 366 |
$params[] = gmdate('Y-m-d H:i:s', $ts); |
| 367 |
} |
| 368 |
if ($session_id !== '') { |
| 369 |
$where[] = 'session_id = %s'; |
| 370 |
$params[] = $session_id; |
| 371 |
} |
| 372 |
if ($role === 'user' || $role === 'assistant') { |
| 373 |
$where[] = 'role = %s'; |
| 374 |
$params[] = $role; |
| 375 |
} |
| 376 |
if ($has_rag_context === 'yes') { |
| 377 |
$where[] = "(rag_context IS NOT NULL AND rag_context <> '')"; |
| 378 |
} elseif ($has_rag_context === 'no') { |
| 379 |
$where[] = "(rag_context IS NULL OR rag_context = '')"; |
| 380 |
} |
| 381 |
|
| 382 |
$where_sql = implode(' AND ', $where); |
| 383 |
|
| 384 |
// Fixed column list — never user-controlled. |
| 385 |
$columns = 'id, user_id, session_id, role, message, user_email, user_name, user_identifier, originating_page_url, originating_page_title, rag_context, timestamp'; |
| 386 |
|
| 387 |
// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared |
| 388 |
// $where_sql, $columns, $order, $table are all built from server-side |
| 389 |
// whitelisted values; no user input is concatenated. All variable |
| 390 |
// values flow through $wpdb->prepare() via $params + LIMIT/OFFSET. |
| 391 |
$sql = "SELECT $columns FROM $table WHERE $where_sql ORDER BY timestamp $order LIMIT %d OFFSET %d"; |
| 392 |
$params[] = $limit; |
| 393 |
$params[] = $offset; |
| 394 |
|
| 395 |
$prepared = $wpdb->prepare($sql, $params); |
| 396 |
$rows = $wpdb->get_results($prepared, ARRAY_A); |
| 397 |
// phpcs:enable |
| 398 |
|
| 399 |
// Total count (respecting filters, ignoring limit/offset) for pagination. |
| 400 |
$count_sql = "SELECT COUNT(*) FROM $table WHERE $where_sql"; |
| 401 |
if (!empty(array_slice($params, 0, count($params) - 2))) { |
| 402 |
$count_prepared = $wpdb->prepare($count_sql, array_slice($params, 0, count($params) - 2)); |
| 403 |
} else { |
| 404 |
$count_prepared = $count_sql; |
| 405 |
} |
| 406 |
$total = (int) $wpdb->get_var($count_prepared); |
| 407 |
|
| 408 |
$items = array(); |
| 409 |
if (is_array($rows)) { |
| 410 |
foreach ($rows as $row) { |
| 411 |
$items[] = array( |
| 412 |
'id' => (int) $row['id'], |
| 413 |
'user_id' => (int) $row['user_id'], |
| 414 |
'session_id' => (string) $row['session_id'], |
| 415 |
'role' => (string) $row['role'], |
| 416 |
'message' => (string) $row['message'], |
| 417 |
'user_email' => $row['user_email'] !== null ? (string) $row['user_email'] : null, |
| 418 |
'user_name' => $row['user_name'] !== null ? (string) $row['user_name'] : null, |
| 419 |
'user_identifier' => $row['user_identifier'] !== null ? (string) $row['user_identifier'] : null, |
| 420 |
'originating_page_url' => $row['originating_page_url'] !== null ? (string) $row['originating_page_url'] : null, |
| 421 |
'originating_page_title' => $row['originating_page_title'] !== null ? (string) $row['originating_page_title'] : null, |
| 422 |
'has_rag_context' => !empty($row['rag_context']), |
| 423 |
'rag_context' => $row['rag_context'] !== null ? (string) $row['rag_context'] : null, |
| 424 |
'timestamp' => (string) $row['timestamp'], |
| 425 |
); |
| 426 |
} |
| 427 |
} |
| 428 |
|
| 429 |
return rest_ensure_response(array( |
| 430 |
'total' => $total, |
| 431 |
'count' => count($items), |
| 432 |
'limit' => $limit, |
| 433 |
'offset' => $offset, |
| 434 |
'items' => $items, |
| 435 |
)); |
| 436 |
} |
| 437 |
|
| 438 |
/** |
| 439 |
* DELETE /transcripts — delete chat data for one or more session_ids. |
| 440 |
* |
| 441 |
* Body (JSON): |
| 442 |
* { |
| 443 |
* "session_ids": ["sid1", "sid2", ...], // required, non-empty |
| 444 |
* "cascade": true // optional, default true |
| 445 |
* } |
| 446 |
* |
| 447 |
* Cascading deletes (when cascade=true) also remove rows in: |
| 448 |
* - wp_mxchat_transcript_translations (any saved translations) |
| 449 |
* - wp_mxchat_url_clicks (link-click tracking rows) |
| 450 |
* |
| 451 |
* Hard caps for safety: |
| 452 |
* - max 1000 session_ids per call |
| 453 |
* - each session_id must be a non-empty string |
| 454 |
* |
| 455 |
* Response: |
| 456 |
* { |
| 457 |
* "session_ids_requested": N, |
| 458 |
* "transcripts_deleted": N, |
| 459 |
* "translations_deleted": N, |
| 460 |
* "url_clicks_deleted": N, |
| 461 |
* "cascade": bool |
| 462 |
* } |
| 463 |
*/ |
| 464 |
public function handle_delete_transcripts($request) { |
| 465 |
global $wpdb; |
| 466 |
|
| 467 |
$body = $request->get_json_params(); |
| 468 |
if (!is_array($body)) { |
| 469 |
$body = array(); |
| 470 |
} |
| 471 |
|
| 472 |
$session_ids_raw = isset($body['session_ids']) ? $body['session_ids'] : null; |
| 473 |
$cascade = isset($body['cascade']) ? (bool) $body['cascade'] : true; |
| 474 |
|
| 475 |
if (!is_array($session_ids_raw) || empty($session_ids_raw)) { |
| 476 |
return new WP_Error( |
| 477 |
'mxchat_rest_no_session_ids', |
| 478 |
__('session_ids is required and must be a non-empty array.', 'mxchat'), |
| 479 |
array('status' => 400) |
| 480 |
); |
| 481 |
} |
| 482 |
|
| 483 |
if (count($session_ids_raw) > 1000) { |
| 484 |
return new WP_Error( |
| 485 |
'mxchat_rest_too_many', |
| 486 |
__('Too many session_ids in a single request. Cap is 1000; split into multiple calls.', 'mxchat'), |
| 487 |
array('status' => 400) |
| 488 |
); |
| 489 |
} |
| 490 |
|
| 491 |
// Sanitize: must be non-empty strings, dedupe, drop bad values. |
| 492 |
$session_ids = array(); |
| 493 |
foreach ($session_ids_raw as $sid) { |
| 494 |
if (is_string($sid) && trim($sid) !== '') { |
| 495 |
$clean = sanitize_text_field($sid); |
| 496 |
if ($clean !== '') { |
| 497 |
$session_ids[] = $clean; |
| 498 |
} |
| 499 |
} |
| 500 |
} |
| 501 |
$session_ids = array_values(array_unique($session_ids)); |
| 502 |
|
| 503 |
if (empty($session_ids)) { |
| 504 |
return new WP_Error( |
| 505 |
'mxchat_rest_no_valid_session_ids', |
| 506 |
__('session_ids must contain at least one valid non-empty string.', 'mxchat'), |
| 507 |
array('status' => 400) |
| 508 |
); |
| 509 |
} |
| 510 |
|
| 511 |
$transcripts_table = $wpdb->prefix . 'mxchat_chat_transcripts'; |
| 512 |
$translations_table = $wpdb->prefix . 'mxchat_transcript_translations'; |
| 513 |
$url_clicks_table = $wpdb->prefix . 'mxchat_url_clicks'; |
| 514 |
|
| 515 |
// Verify the main table exists; the others are best-effort. |
| 516 |
if ($wpdb->get_var($wpdb->prepare("SHOW TABLES LIKE %s", $transcripts_table)) !== $transcripts_table) { |
| 517 |
return new WP_Error( |
| 518 |
'mxchat_rest_no_table', |
| 519 |
__('Transcripts table does not exist on this site.', 'mxchat'), |
| 520 |
array('status' => 500) |
| 521 |
); |
| 522 |
} |
| 523 |
|
| 524 |
$placeholders = implode(',', array_fill(0, count($session_ids), '%s')); |
| 525 |
|
| 526 |
// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared |
| 527 |
// $placeholders is a server-built list of literal "%s" tokens — never user-controlled. |
| 528 |
// All values flow through $wpdb->prepare() via $session_ids. |
| 529 |
$deleted_transcripts = (int) $wpdb->query( |
| 530 |
$wpdb->prepare( |
| 531 |
"DELETE FROM $transcripts_table WHERE session_id IN ($placeholders)", |
| 532 |
$session_ids |
| 533 |
) |
| 534 |
); |
| 535 |
|
| 536 |
$deleted_translations = 0; |
| 537 |
$deleted_url_clicks = 0; |
| 538 |
|
| 539 |
if ($cascade) { |
| 540 |
if ($wpdb->get_var($wpdb->prepare("SHOW TABLES LIKE %s", $translations_table)) === $translations_table) { |
| 541 |
$deleted_translations = (int) $wpdb->query( |
| 542 |
$wpdb->prepare( |
| 543 |
"DELETE FROM $translations_table WHERE session_id IN ($placeholders)", |
| 544 |
$session_ids |
| 545 |
) |
| 546 |
); |
| 547 |
} |
| 548 |
if ($wpdb->get_var($wpdb->prepare("SHOW TABLES LIKE %s", $url_clicks_table)) === $url_clicks_table) { |
| 549 |
$deleted_url_clicks = (int) $wpdb->query( |
| 550 |
$wpdb->prepare( |
| 551 |
"DELETE FROM $url_clicks_table WHERE session_id IN ($placeholders)", |
| 552 |
$session_ids |
| 553 |
) |
| 554 |
); |
| 555 |
} |
| 556 |
} |
| 557 |
// phpcs:enable |
| 558 |
|
| 559 |
return rest_ensure_response(array( |
| 560 |
'session_ids_requested' => count($session_ids), |
| 561 |
'transcripts_deleted' => $deleted_transcripts, |
| 562 |
'translations_deleted' => $deleted_translations, |
| 563 |
'url_clicks_deleted' => $deleted_url_clicks, |
| 564 |
'cascade' => $cascade, |
| 565 |
)); |
| 566 |
} |
| 567 |
|
| 568 |
/** |
| 569 |
* POST /knowledge — embed + store content in the KB. |
| 570 |
* Wraps MxChat_Utils::submit_content_to_db() so external tools can push |
| 571 |
* content the same way an admin would via the Knowledge UI. |
| 572 |
*/ |
| 573 |
public function handle_post_knowledge($request) { |
| 574 |
if (!class_exists('MxChat_Utils')) { |
| 575 |
return new WP_Error( |
| 576 |
'mxchat_rest_no_utils', |
| 577 |
__('MxChat_Utils is not loaded.', 'mxchat'), |
| 578 |
array('status' => 500) |
| 579 |
); |
| 580 |
} |
| 581 |
|
| 582 |
$body = $request->get_json_params(); |
| 583 |
if (!is_array($body)) { |
| 584 |
$body = array(); |
| 585 |
} |
| 586 |
|
| 587 |
$raw_content = isset($body['content']) ? (string) $body['content'] : ''; |
| 588 |
$content = wp_kses_post(wp_unslash($raw_content)); |
| 589 |
|
| 590 |
$source_url = isset($body['source_url']) ? esc_url_raw((string) $body['source_url']) : ''; |
| 591 |
$bot_id = isset($body['bot_id']) ? sanitize_key((string) $body['bot_id']) : 'default'; |
| 592 |
$content_type = isset($body['content_type']) ? sanitize_key((string) $body['content_type']) : 'manual'; |
| 593 |
|
| 594 |
if ($content === '') { |
| 595 |
return new WP_Error('mxchat_rest_no_content', __('content is required.', 'mxchat'), array('status' => 400)); |
| 596 |
} |
| 597 |
if ($source_url === '') { |
| 598 |
return new WP_Error('mxchat_rest_no_url', __('source_url is required and must be a valid URL.', 'mxchat'), array('status' => 400)); |
| 599 |
} |
| 600 |
if ($bot_id === '') { |
| 601 |
$bot_id = 'default'; |
| 602 |
} |
| 603 |
|
| 604 |
// Pull the embedding API key from MxChat options (same logic as the |
| 605 |
// admin form handler in MxChat_Knowledge_Manager::mxchat_handle_content_submission). |
| 606 |
$bot_options = $this->get_bot_options($bot_id); |
| 607 |
$options = !empty($bot_options) ? $bot_options : get_option('mxchat_options', array()); |
| 608 |
// The same decision every admin importer makes (plan 418afd): a site |
| 609 |
// that embeds through a Custom Provider needs no OpenAI, Voyage or |
| 610 |
// Gemini key, and the key check used to refuse it. |
| 611 |
$preflight = MxChat_Utils::embedding_preflight(is_array($options) ? $options : array()); |
| 612 |
if (!$preflight['ok']) { |
| 613 |
return new WP_Error('mxchat_rest_no_api_key', $preflight['reason'], array('status' => 500)); |
| 614 |
} |
| 615 |
$api_key = $preflight['api_key']; |
| 616 |
|
| 617 |
// Embedding + chunking can take 10-60s for large documents. |
| 618 |
if (function_exists('set_time_limit')) { |
| 619 |
@set_time_limit(300); |
| 620 |
} |
| 621 |
@ignore_user_abort(true); |
| 622 |
|
| 623 |
$result = MxChat_Utils::submit_content_to_db($content, $source_url, $api_key, null, $bot_id, $content_type); |
| 624 |
|
| 625 |
if (is_wp_error($result)) { |
| 626 |
return new WP_Error( |
| 627 |
'mxchat_rest_kb_failed', |
| 628 |
$result->get_error_message(), |
| 629 |
array('status' => 500) |
| 630 |
); |
| 631 |
} |
| 632 |
|
| 633 |
return rest_ensure_response(array( |
| 634 |
'success' => true, |
| 635 |
'source_url' => $source_url, |
| 636 |
'bot_id' => $bot_id, |
| 637 |
'content_type' => $content_type, |
| 638 |
'bytes' => strlen($content), |
| 639 |
)); |
| 640 |
} |
| 641 |
|
| 642 |
/** |
| 643 |
* DELETE /knowledge — remove entries from the KB by source_url. |
| 644 |
* |
| 645 |
* Wraps MxChat_Utils::delete_chunks_for_url(), which is the same entry |
| 646 |
* point the admin's delete-by-URL action uses: it mirrors the removal to |
| 647 |
* the OpenAI Vector Store and then clears whichever backend holds the |
| 648 |
* content (Pinecone chunks, or the WordPress rows). There is deliberately |
| 649 |
* no second delete path here. |
| 650 |
* |
| 651 |
* Idempotent by design: a URL that is not in the knowledge base comes back |
| 652 |
* as deleted:false rather than an error, so a sync job can re-run cleanly. |
| 653 |
*/ |
| 654 |
public function handle_delete_knowledge($request) { |
| 655 |
if (!class_exists('MxChat_Utils')) { |
| 656 |
return new WP_Error( |
| 657 |
'mxchat_rest_no_utils', |
| 658 |
__('MxChat_Utils is not loaded.', 'mxchat'), |
| 659 |
array('status' => 500) |
| 660 |
); |
| 661 |
} |
| 662 |
|
| 663 |
// get_param() covers both a JSON body and a query string, so an |
| 664 |
// automation can send either. |
| 665 |
$raw = $request->get_param('source_url'); |
| 666 |
if ($raw === null) { |
| 667 |
$body = $request->get_json_params(); |
| 668 |
if (is_array($body) && isset($body['source_url'])) { |
| 669 |
$raw = $body['source_url']; |
| 670 |
} |
| 671 |
} |
| 672 |
|
| 673 |
$bot_id = $request->get_param('bot_id'); |
| 674 |
$bot_id = is_string($bot_id) ? sanitize_key($bot_id) : ''; |
| 675 |
if ($bot_id === '') { |
| 676 |
$bot_id = 'default'; |
| 677 |
} |
| 678 |
|
| 679 |
if (is_string($raw)) { |
| 680 |
$raw = array($raw); |
| 681 |
} |
| 682 |
|
| 683 |
if (!is_array($raw) || empty($raw)) { |
| 684 |
return new WP_Error( |
| 685 |
'mxchat_rest_no_source_url', |
| 686 |
__('source_url is required — a string, or a non-empty array of strings.', 'mxchat'), |
| 687 |
array('status' => 400) |
| 688 |
); |
| 689 |
} |
| 690 |
|
| 691 |
if (count($raw) > 50) { |
| 692 |
return new WP_Error( |
| 693 |
'mxchat_rest_too_many', |
| 694 |
__('Too many source_urls in a single request. Cap is 50; split into multiple calls.', 'mxchat'), |
| 695 |
array('status' => 400) |
| 696 |
); |
| 697 |
} |
| 698 |
|
| 699 |
// Normalise the same way the content was normalised on the way in, so |
| 700 |
// a URL that POST accepted is the URL DELETE matches. Dedupe, keep order. |
| 701 |
$urls = array(); |
| 702 |
foreach ($raw as $candidate) { |
| 703 |
if (!is_string($candidate)) { |
| 704 |
continue; |
| 705 |
} |
| 706 |
$clean = $this->normalize_source_url($candidate); |
| 707 |
if ($clean !== '' && !in_array($clean, $urls, true)) { |
| 708 |
$urls[] = $clean; |
| 709 |
} |
| 710 |
} |
| 711 |
|
| 712 |
if (empty($urls)) { |
| 713 |
return new WP_Error( |
| 714 |
'mxchat_rest_no_valid_source_url', |
| 715 |
__('source_url must contain at least one valid non-empty URL.', 'mxchat'), |
| 716 |
array('status' => 400) |
| 717 |
); |
| 718 |
} |
| 719 |
|
| 720 |
// Pinecone listing + deletion is several round trips per URL. |
| 721 |
if (function_exists('set_time_limit')) { |
| 722 |
@set_time_limit(300); |
| 723 |
} |
| 724 |
@ignore_user_abort(true); |
| 725 |
|
| 726 |
$results = array(); |
| 727 |
$urls_deleted = 0; |
| 728 |
$entries_removed = 0; |
| 729 |
|
| 730 |
foreach ($urls as $url) { |
| 731 |
// Counted before the delete, against the same backend the delete |
| 732 |
// will act on, so the number reported is the number removed. |
| 733 |
$before = (int) MxChat_Utils::count_chunks_for_url($url, $bot_id); |
| 734 |
|
| 735 |
// Called even at zero: the entry may still have a mirrored Vector |
| 736 |
// Store file, and that must not outlive it. |
| 737 |
$outcome = MxChat_Utils::delete_chunks_for_url($url, $bot_id); |
| 738 |
|
| 739 |
$row = array( |
| 740 |
'source_url' => $url, |
| 741 |
'deleted' => false, |
| 742 |
'entries_removed' => 0, |
| 743 |
); |
| 744 |
|
| 745 |
if (is_wp_error($outcome)) { |
| 746 |
$row['error'] = $outcome->get_error_message(); |
| 747 |
if (class_exists('MxChat_Admin')) { |
| 748 |
MxChat_Admin::mxchat_log_debug( |
| 749 |
'rest_knowledge_delete', |
| 750 |
'DELETE /knowledge failed: ' . $outcome->get_error_message(), |
| 751 |
array('source_url' => $url, 'bot_id' => $bot_id) |
| 752 |
); |
| 753 |
} |
| 754 |
} elseif ($before > 0) { |
| 755 |
$row['deleted'] = true; |
| 756 |
$row['entries_removed'] = $before; |
| 757 |
$urls_deleted++; |
| 758 |
$entries_removed += $before; |
| 759 |
} |
| 760 |
|
| 761 |
$results[] = $row; |
| 762 |
} |
| 763 |
|
| 764 |
if (class_exists('MxChat_Admin')) { |
| 765 |
MxChat_Admin::mxchat_log_debug( |
| 766 |
'rest_knowledge_delete', |
| 767 |
'DELETE /knowledge removed ' . $entries_removed . ' entries across ' . $urls_deleted . ' of ' . count($urls) . ' URLs', |
| 768 |
array('bot_id' => $bot_id) |
| 769 |
); |
| 770 |
} |
| 771 |
|
| 772 |
return rest_ensure_response(array( |
| 773 |
'source_urls_requested' => count($urls), |
| 774 |
'urls_deleted' => $urls_deleted, |
| 775 |
'entries_removed' => $entries_removed, |
| 776 |
'bot_id' => $bot_id, |
| 777 |
'results' => $results, |
| 778 |
)); |
| 779 |
} |
| 780 |
|
| 781 |
/** |
| 782 |
* GET /leads — lead-capture submissions, the same rows the Leads tab shows. |
| 783 |
* |
| 784 |
* Reads through MxChat_Admin::mxchat_collect_leads() rather than issuing |
| 785 |
* its own query, so it inherits the address-less lead key (a name-only |
| 786 |
* capture is keyed "session:<id>") instead of re-opening that blind spot. |
| 787 |
*/ |
| 788 |
public function handle_get_leads($request) { |
| 789 |
if (!class_exists('MxChat_Admin') || !method_exists('MxChat_Admin', 'mxchat_collect_leads')) { |
| 790 |
return new WP_Error( |
| 791 |
'mxchat_rest_no_leads_source', |
| 792 |
__('The leads reader is not available on this install.', 'mxchat'), |
| 793 |
array('status' => 500) |
| 794 |
); |
| 795 |
} |
| 796 |
|
| 797 |
$page = (int) $request->get_param('page'); |
| 798 |
$per_page = (int) $request->get_param('per_page'); |
| 799 |
$page = $page > 0 ? $page : 1; |
| 800 |
$per_page = $per_page > 0 ? min(200, $per_page) : 100; |
| 801 |
|
| 802 |
$since_raw = (string) $request->get_param('since'); |
| 803 |
$since_ts = $this->parse_datetime($since_raw); |
| 804 |
if ($since_raw !== '' && $since_ts === null) { |
| 805 |
return new WP_Error( |
| 806 |
'mxchat_rest_bad_since', |
| 807 |
__('since could not be parsed as a date.', 'mxchat'), |
| 808 |
array('status' => 400) |
| 809 |
); |
| 810 |
} |
| 811 |
|
| 812 |
$status = (string) $request->get_param('status'); |
| 813 |
if (!in_array($status, array('all', 'orphan', 'chat_deleted'), true)) { |
| 814 |
$status = 'all'; |
| 815 |
} |
| 816 |
|
| 817 |
$payload = MxChat_Admin::mxchat_collect_leads(array( |
| 818 |
'page' => $page, |
| 819 |
'per_page' => $per_page, |
| 820 |
'search' => (string) $request->get_param('search'), |
| 821 |
// Timestamps in the transcripts table are UTC, as gmdate() writes |
| 822 |
// them everywhere else the cutoff is built. |
| 823 |
'date_cutoff' => $since_ts === null ? '' : gmdate('Y-m-d H:i:s', $since_ts), |
| 824 |
'status' => $status, |
| 825 |
'sort' => 'last_seen', |
| 826 |
'sort_dir' => $request->get_param('order') === 'asc' ? 'asc' : 'desc', |
| 827 |
)); |
| 828 |
|
| 829 |
$leads = array(); |
| 830 |
foreach ((array) ($payload['leads'] ?? array()) as $lead) { |
| 831 |
$leads[] = array( |
| 832 |
'lead_key' => (string) ($lead['lead_key'] ?? ''), |
| 833 |
'email' => (string) ($lead['email'] ?? ''), |
| 834 |
'name' => (string) ($lead['name'] ?? ''), |
| 835 |
'session_id' => (string) ($lead['latest_session_id'] ?? ''), |
| 836 |
'conversation_count' => (int) ($lead['conversation_count'] ?? 0), |
| 837 |
'first_seen' => (string) ($lead['first_seen'] ?? ''), |
| 838 |
'last_seen' => (string) ($lead['last_seen'] ?? ''), |
| 839 |
'consent' => (string) ($lead['consent'] ?? ''), |
| 840 |
'consent_at' => (string) ($lead['consent_at'] ?? ''), |
| 841 |
'consent_label' => (string) ($lead['consent_label'] ?? ''), |
| 842 |
'originating_page_url' => (string) ($lead['top_page_url'] ?? ''), |
| 843 |
'originating_page_title' => (string) ($lead['top_page_title'] ?? ''), |
| 844 |
'status' => (string) ($lead['status'] ?? 'active'), |
| 845 |
'is_orphan' => !empty($lead['is_orphan']), |
| 846 |
); |
| 847 |
} |
| 848 |
|
| 849 |
return rest_ensure_response(array( |
| 850 |
'leads' => $leads, |
| 851 |
'page' => (int) ($payload['page'] ?? $page), |
| 852 |
'per_page' => (int) ($payload['per_page'] ?? $per_page), |
| 853 |
'total_count' => (int) ($payload['total_count'] ?? 0), |
| 854 |
'total_pages' => (int) ($payload['total_pages'] ?? 0), |
| 855 |
)); |
| 856 |
} |
| 857 |
|
| 858 |
/** |
| 859 |
* Normalise a source_url the way MxChat_Utils::submit_content_to_db() does |
| 860 |
* before it stores, so DELETE matches what POST created. The pseudo-scheme |
| 861 |
* branch matters: esc_url_raw() EMPTIES mxchat:// and upload:// identities. |
| 862 |
*/ |
| 863 |
private function normalize_source_url($raw) { |
| 864 |
$raw = trim((string) $raw); |
| 865 |
if ($raw === '') { |
| 866 |
return ''; |
| 867 |
} |
| 868 |
if (preg_match('#^(mxchat|upload)://#i', $raw)) { |
| 869 |
return sanitize_text_field($raw); |
| 870 |
} |
| 871 |
return esc_url_raw($raw); |
| 872 |
} |
| 873 |
|
| 874 |
/** |
| 875 |
* Get bot-specific options if the multi-bot add-on is present, else return empty. |
| 876 |
*/ |
| 877 |
private function get_bot_options($bot_id) { |
| 878 |
if ($bot_id === '' || $bot_id === 'default') { |
| 879 |
return array(); |
| 880 |
} |
| 881 |
if (!class_exists('MxChat_Multi_Bot_Manager')) { |
| 882 |
return array(); |
| 883 |
} |
| 884 |
$bot_options = apply_filters('mxchat_get_bot_options', array(), $bot_id); |
| 885 |
return is_array($bot_options) ? $bot_options : array(); |
| 886 |
} |
| 887 |
|
| 888 |
/** |
| 889 |
* Try a few formats before falling back to strtotime. |
| 890 |
* Returns a Unix timestamp or null. |
| 891 |
*/ |
| 892 |
private function parse_datetime($value) { |
| 893 |
$value = trim((string) $value); |
| 894 |
if ($value === '') { |
| 895 |
return null; |
| 896 |
} |
| 897 |
$ts = strtotime($value); |
| 898 |
return ($ts === false || $ts <= 0) ? null : $ts; |
| 899 |
} |
| 900 |
} |
| 901 |
|
