| 1 |
<?php |
| 2 |
|
| 3 |
namespace Extendify\Agent\Controllers; |
| 4 |
|
| 5 |
defined('ABSPATH') || die('No direct access.'); |
| 6 |
|
| 7 |
class ContentController |
| 8 |
{ |
| 9 |
// phpcs:disable PSR12.Properties.ConstantVisibility.NotFound |
| 10 |
// Header and footer text lives in a template part, a synced pattern's in a wp_block. |
| 11 |
const POST_TYPES = ['post', 'page', 'wp_template_part', 'wp_block']; |
| 12 |
|
| 13 |
const FORMATS = ['text', 'blocks', 'excerpt']; |
| 14 |
|
| 15 |
const MAX_SCANNED = 100; |
| 16 |
|
| 17 |
const MAX_POSTS = 25; |
| 18 |
|
| 19 |
const MAX_MATCHES = 3; |
| 20 |
|
| 21 |
const MAX_CONTEXT = 300; |
| 22 |
|
| 23 |
const MAX_CONTENT = 15000; |
| 24 |
|
| 25 |
const MAX_BYTES = 20000; |
| 26 |
|
| 27 |
const MAX_META_KEYS = 50; |
| 28 |
|
| 29 |
const MAX_META_VALUE = 500; |
| 30 |
// phpcs:enable PSR12.Properties.ConstantVisibility.NotFound |
| 31 |
|
| 32 |
/** |
| 33 |
* @param \WP_REST_Request $request The REST API request object. |
| 34 |
* @return \WP_REST_Response |
| 35 |
*/ |
| 36 |
public static function search(\WP_REST_Request $request) |
| 37 |
{ |
| 38 |
$params = self::params($request); |
| 39 |
$query = trim(self::text($params, 'query')); |
| 40 |
if ($query === '') { |
| 41 |
return self::reject('query must say what to look for'); |
| 42 |
} |
| 43 |
|
| 44 |
$types = self::types($params); |
| 45 |
if (!$types) { |
| 46 |
return self::reject('postTypes names no post type this site has'); |
| 47 |
} |
| 48 |
|
| 49 |
$limit = self::number($params, 'limit', 10, self::MAX_POSTS); |
| 50 |
$context = self::number($params, 'context', 60, self::MAX_CONTEXT); |
| 51 |
$candidates = new \WP_Query([ |
| 52 |
'post_type' => $types, |
| 53 |
'post_status' => 'publish', |
| 54 |
'posts_per_page' => self::MAX_SCANNED, |
| 55 |
'orderby' => 'modified', |
| 56 |
'order' => 'DESC', |
| 57 |
'ignore_sticky_posts' => true, |
| 58 |
]); |
| 59 |
|
| 60 |
$results = []; |
| 61 |
$matched = 0; |
| 62 |
$spent = 0; |
| 63 |
foreach ($candidates->posts as $post) { |
| 64 |
$found = self::snippets(self::rendered($post), $query, $context); |
| 65 |
if (!$found['matches']) { |
| 66 |
continue; |
| 67 |
} |
| 68 |
|
| 69 |
$matched++; |
| 70 |
if (count($results) >= $limit || $spent > self::MAX_BYTES) { |
| 71 |
continue; |
| 72 |
} |
| 73 |
|
| 74 |
$result = [ |
| 75 |
'id' => (int) $post->ID, |
| 76 |
'type' => $post->post_type, |
| 77 |
'title' => self::decoded($post->post_title), |
| 78 |
'slug' => $post->post_name, |
| 79 |
'matches' => $found['matches'], |
| 80 |
'moreMatches' => $found['count'] - count($found['matches']), |
| 81 |
]; |
| 82 |
// Escaped JSON counts one Japanese character as six, spending the budget six times over. |
| 83 |
$spent += strlen((string) wp_json_encode($result, JSON_UNESCAPED_UNICODE)); |
| 84 |
$results[] = $result; |
| 85 |
} |
| 86 |
|
| 87 |
$response = [ |
| 88 |
'query' => $query, |
| 89 |
'scanned' => count($candidates->posts), |
| 90 |
'matched' => $matched, |
| 91 |
'results' => $results, |
| 92 |
]; |
| 93 |
$left = self::leftOut($matched - count($results), (int) $candidates->found_posts - count($candidates->posts)); |
| 94 |
if ($left !== '') { |
| 95 |
$response['truncated'] = $left; |
| 96 |
} |
| 97 |
|
| 98 |
return new \WP_REST_Response($response); |
| 99 |
} |
| 100 |
|
| 101 |
/** |
| 102 |
* @param \WP_REST_Request $request The REST API request object. |
| 103 |
* @return \WP_REST_Response |
| 104 |
*/ |
| 105 |
public static function read(\WP_REST_Request $request) |
| 106 |
{ |
| 107 |
$params = self::params($request); |
| 108 |
$id = self::number($params, 'id', 0, PHP_INT_MAX); |
| 109 |
$slug = trim(self::text($params, 'slug')); |
| 110 |
if (!$id && $slug === '') { |
| 111 |
return self::reject('id or slug must say what to read'); |
| 112 |
} |
| 113 |
|
| 114 |
$format = self::text($params, 'format', 'text'); |
| 115 |
if (!in_array($format, self::FORMATS, true)) { |
| 116 |
return self::reject('format must be one of ' . implode(', ', self::FORMATS)); |
| 117 |
} |
| 118 |
|
| 119 |
$post = $id ? self::byId($id) : self::bySlug($slug); |
| 120 |
if (!$post) { |
| 121 |
return self::reject('Nothing readable was found for ' . ($id ? $id : $slug)); |
| 122 |
} |
| 123 |
|
| 124 |
$content = self::content($post, $format); |
| 125 |
$response = [ |
| 126 |
'id' => (int) $post->ID, |
| 127 |
'type' => $post->post_type, |
| 128 |
'title' => self::decoded($post->post_title), |
| 129 |
'slug' => $post->post_name, |
| 130 |
'status' => $post->post_status, |
| 131 |
'date' => $post->post_date, |
| 132 |
'format' => $format, |
| 133 |
'content' => mb_substr($content, 0, self::MAX_CONTENT), |
| 134 |
]; |
| 135 |
if (mb_strlen($content) > self::MAX_CONTENT) { |
| 136 |
$response['truncated'] = 'Cut after ' . self::MAX_CONTENT . ' characters, of ' |
| 137 |
. mb_strlen($content) . ' the post holds.'; |
| 138 |
} |
| 139 |
if (!empty($params['includeMeta'])) { |
| 140 |
$response['meta'] = self::meta($post->ID); |
| 141 |
} |
| 142 |
|
| 143 |
return new \WP_REST_Response($response); |
| 144 |
} |
| 145 |
|
| 146 |
/** |
| 147 |
* @param \WP_REST_Request $request The REST API request object. |
| 148 |
* @return array The request body. |
| 149 |
*/ |
| 150 |
private static function params(\WP_REST_Request $request) |
| 151 |
{ |
| 152 |
$params = $request->get_json_params(); |
| 153 |
|
| 154 |
return is_array($params) ? $params : []; |
| 155 |
} |
| 156 |
|
| 157 |
/** |
| 158 |
* @param string $message What the caller has to change. |
| 159 |
* @return \WP_REST_Response |
| 160 |
*/ |
| 161 |
private static function reject($message) |
| 162 |
{ |
| 163 |
return new \WP_REST_Response(['error' => $message], 400); |
| 164 |
} |
| 165 |
|
| 166 |
/** |
| 167 |
* @param array $params The array to read from. |
| 168 |
* @param string $key The key to read. |
| 169 |
* @param string $default What to use when the key is absent. |
| 170 |
* @return string |
| 171 |
*/ |
| 172 |
private static function text($params, $key, $default = '') |
| 173 |
{ |
| 174 |
return isset($params[$key]) && is_string($params[$key]) ? $params[$key] : $default; |
| 175 |
} |
| 176 |
|
| 177 |
/** |
| 178 |
* @param array $params The array to read from. |
| 179 |
* @param string $key The key to read. |
| 180 |
* @param integer $default What to use when the key is absent. |
| 181 |
* @param integer $most The largest value to allow. |
| 182 |
* @return integer |
| 183 |
*/ |
| 184 |
private static function number($params, $key, $default, $most) |
| 185 |
{ |
| 186 |
$value = isset($params[$key]) && is_numeric($params[$key]) |
| 187 |
? (int) $params[$key] |
| 188 |
: $default; |
| 189 |
|
| 190 |
return max(0, min($value, $most)); |
| 191 |
} |
| 192 |
|
| 193 |
/** |
| 194 |
* @param array $params The request body. |
| 195 |
* @return array The post types to search through. |
| 196 |
*/ |
| 197 |
private static function types($params) |
| 198 |
{ |
| 199 |
$requested = isset($params['postTypes']) && is_array($params['postTypes']) |
| 200 |
? $params['postTypes'] |
| 201 |
: []; |
| 202 |
if (!$requested) { |
| 203 |
return self::POST_TYPES; |
| 204 |
} |
| 205 |
|
| 206 |
return array_values(array_filter($requested, function ($type) { |
| 207 |
return is_string($type) && post_type_exists($type); |
| 208 |
})); |
| 209 |
} |
| 210 |
|
| 211 |
/** |
| 212 |
* An empty result and one nobody looked for read the same to a model. |
| 213 |
* |
| 214 |
* @param integer $matched Matching posts the response left out. |
| 215 |
* @param integer $unscanned Candidates the scan never reached. |
| 216 |
* @return string |
| 217 |
*/ |
| 218 |
private static function leftOut($matched, $unscanned) |
| 219 |
{ |
| 220 |
$left = []; |
| 221 |
if ($matched > 0) { |
| 222 |
$left[] = $matched . ' more posts matched'; |
| 223 |
} |
| 224 |
if ($unscanned > 0) { |
| 225 |
$left[] = $unscanned . ' posts were not searched'; |
| 226 |
} |
| 227 |
|
| 228 |
return $left ? implode(', and ', $left) . '.' : ''; |
| 229 |
} |
| 230 |
|
| 231 |
/** |
| 232 |
* Blocks that read the post they sit in render empty without the global. |
| 233 |
* |
| 234 |
* @param \WP_Post $post The post to render. |
| 235 |
* @return string Its text as a visitor reads it. |
| 236 |
*/ |
| 237 |
private static function rendered($post) |
| 238 |
{ |
| 239 |
$GLOBALS['post'] = $post; |
| 240 |
setup_postdata($post); |
| 241 |
$html = do_shortcode(do_blocks($post->post_content)); |
| 242 |
wp_reset_postdata(); |
| 243 |
|
| 244 |
return trim(preg_replace('/\s+/u', ' ', self::decoded(wp_strip_all_tags($html)))); |
| 245 |
} |
| 246 |
|
| 247 |
/** |
| 248 |
* @param string $text Text as WordPress stored or rendered it. |
| 249 |
* @return string |
| 250 |
*/ |
| 251 |
private static function decoded($text) |
| 252 |
{ |
| 253 |
return html_entity_decode($text, ENT_QUOTES, get_bloginfo('charset')); |
| 254 |
} |
| 255 |
|
| 256 |
/** |
| 257 |
* @param string $text The text to look through. |
| 258 |
* @param string $query What to look for. |
| 259 |
* @param integer $context How much to quote either side of a match. |
| 260 |
* @return array The snippets, and how many times the text says it. |
| 261 |
*/ |
| 262 |
private static function snippets($text, $query, $context) |
| 263 |
{ |
| 264 |
// Turkish İ lower cases to two characters, so a folded copy's offsets miss the match. |
| 265 |
$pattern = '/' . preg_quote($query, '/') . '/iu'; |
| 266 |
if (!preg_match_all($pattern, $text, $found, PREG_OFFSET_CAPTURE)) { |
| 267 |
return ['matches' => [], 'count' => 0]; |
| 268 |
} |
| 269 |
|
| 270 |
$end = strlen($text); |
| 271 |
$matches = []; |
| 272 |
$quoted = 0; |
| 273 |
foreach ($found[0] as $occurrence) { |
| 274 |
list($match, $at) = $occurrence; |
| 275 |
if (count($matches) >= self::MAX_MATCHES || $at < $quoted) { |
| 276 |
continue; |
| 277 |
} |
| 278 |
|
| 279 |
$before = self::quoteBefore($text, $at, $context); |
| 280 |
$after = self::quoteAfter($text, $at + strlen($match), $context); |
| 281 |
$quoted = $at + strlen($match) + strlen($after); |
| 282 |
$matches[] = ($at > strlen($before) ? '…' : '') |
| 283 |
. $before . $match . $after |
| 284 |
. ($quoted < $end ? '…' : ''); |
| 285 |
} |
| 286 |
|
| 287 |
return ['matches' => $matches, 'count' => count($found[0])]; |
| 288 |
} |
| 289 |
|
| 290 |
/** |
| 291 |
* Up to $context characters of $text ending at byte offset $at. |
| 292 |
* |
| 293 |
* @param string $text The text a match was found in. |
| 294 |
* @param integer $at Where the match begins, in bytes. |
| 295 |
* @param integer $context How many characters to quote. |
| 296 |
* @return string |
| 297 |
*/ |
| 298 |
private static function quoteBefore($text, $at, $context) |
| 299 |
{ |
| 300 |
// A character is at most four bytes, so the window always holds enough. |
| 301 |
$slice = substr($text, max(0, $at - (4 * $context)), min($at, 4 * $context)); |
| 302 |
// A slice starting mid-character would fail the /u match and empty the quote. |
| 303 |
$slice = (string) preg_replace('/^[\x80-\xBF]+/', '', $slice); |
| 304 |
preg_match('/.{0,' . $context . '}$/su', $slice, $match); |
| 305 |
|
| 306 |
return $match[0] ?? ''; |
| 307 |
} |
| 308 |
|
| 309 |
/** |
| 310 |
* Up to $context characters of $text starting at byte offset $at. |
| 311 |
* |
| 312 |
* @param string $text The text a match was found in. |
| 313 |
* @param integer $at Where the match ends, in bytes. |
| 314 |
* @param integer $context How many characters to quote. |
| 315 |
* @return string |
| 316 |
*/ |
| 317 |
private static function quoteAfter($text, $at, $context) |
| 318 |
{ |
| 319 |
// A slice ending mid-character would fail the /u match and empty the quote. |
| 320 |
$slice = (string) preg_replace( |
| 321 |
'/(?:[\xC2-\xDF]|[\xE0-\xEF][\x80-\xBF]?|[\xF0-\xF4][\x80-\xBF]{0,2})$/D', |
| 322 |
'', |
| 323 |
substr($text, $at, 4 * $context) |
| 324 |
); |
| 325 |
preg_match('/^.{0,' . $context . '}/su', $slice, $match); |
| 326 |
|
| 327 |
return $match[0] ?? ''; |
| 328 |
} |
| 329 |
|
| 330 |
/** |
| 331 |
* @param integer $id A post id. |
| 332 |
* @return \WP_Post|null |
| 333 |
*/ |
| 334 |
private static function byId($id) |
| 335 |
{ |
| 336 |
$post = get_post($id); |
| 337 |
if (!$post || in_array($post->post_status, ['auto-draft', 'trash'], true)) { |
| 338 |
return null; |
| 339 |
} |
| 340 |
|
| 341 |
return $post; |
| 342 |
} |
| 343 |
|
| 344 |
/** |
| 345 |
* @param string $slug A post slug. |
| 346 |
* @return \WP_Post|null |
| 347 |
*/ |
| 348 |
private static function bySlug($slug) |
| 349 |
{ |
| 350 |
$posts = get_posts([ |
| 351 |
'name' => $slug, |
| 352 |
'post_type' => self::POST_TYPES, |
| 353 |
'post_status' => 'any', |
| 354 |
'posts_per_page' => 1, |
| 355 |
]); |
| 356 |
|
| 357 |
return $posts ? $posts[0] : null; |
| 358 |
} |
| 359 |
|
| 360 |
/** |
| 361 |
* @param \WP_Post $post The post being read. |
| 362 |
* @param string $format Which of self::FORMATS to return. |
| 363 |
* @return string |
| 364 |
*/ |
| 365 |
private static function content($post, $format) |
| 366 |
{ |
| 367 |
if ($format === 'blocks') { |
| 368 |
return $post->post_content; |
| 369 |
} |
| 370 |
|
| 371 |
if ($format === 'excerpt') { |
| 372 |
return get_the_excerpt($post); |
| 373 |
} |
| 374 |
|
| 375 |
return self::rendered($post); |
| 376 |
} |
| 377 |
|
| 378 |
/** |
| 379 |
* Keys under an underscore are WordPress and plugin bookkeeping. |
| 380 |
* |
| 381 |
* @param integer $id The post to read meta from. |
| 382 |
* @return array Name => value pairs. |
| 383 |
*/ |
| 384 |
private static function meta($id) |
| 385 |
{ |
| 386 |
$meta = []; |
| 387 |
foreach (get_post_meta($id) as $key => $stored) { |
| 388 |
if (strpos($key, '_') === 0 || count($meta) >= self::MAX_META_KEYS) { |
| 389 |
continue; |
| 390 |
} |
| 391 |
|
| 392 |
$value = maybe_unserialize(isset($stored[0]) ? $stored[0] : ''); |
| 393 |
$meta[$key] = is_string($value) && mb_strlen($value) > self::MAX_META_VALUE |
| 394 |
? mb_substr($value, 0, self::MAX_META_VALUE) |
| 395 |
: $value; |
| 396 |
} |
| 397 |
|
| 398 |
return $meta; |
| 399 |
} |
| 400 |
} |
| 401 |
|
