| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Published-content lookup queries used by the suggestion engine to find |
| 9 |
* matching destinations for 404 URLs (posts, pages, images, tags, categories). |
| 10 |
* |
| 11 |
* Split from DataAccessTrait_Redirects in 4.1.x to keep both files under the |
| 12 |
* 1500-line modularity cap. These methods are semantically distinct from |
| 13 |
* redirect lifecycle code: they read WordPress core tables (wp_posts, |
| 14 |
* wp_terms) to enumerate candidate destinations, not the plugin's own |
| 15 |
* redirects table. |
| 16 |
*/ |
| 17 |
trait ABJ_404_Solution_DataAccess_PublishedContentTrait { |
| 18 |
|
| 19 |
/** Returns rows with the IDs of the published items. |
| 20 |
* @global type $wpdb |
| 21 |
* @global type $abj404logic |
| 22 |
* @global type $abj404dao |
| 23 |
* @global type $abj404logging |
| 24 |
* @param string $slug only get results for this slug. (empty means all posts) |
| 25 |
* @param string $searchTerm use this string in a LIKE on the sql. |
| 26 |
* @param string $limitResults |
| 27 |
* @param string $orderResults |
| 28 |
* @param string $extraWhereClause use this string in a where on the sql. |
| 29 |
* @return array<int, object> |
| 30 |
*/ |
| 31 |
function getPublishedPagesAndPostsIDs($slug = '', $searchTerm = '', |
| 32 |
$limitResults = '', $orderResults = '', $extraWhereClause = '') { |
| 33 |
global $wpdb; |
| 34 |
$abj404logic = abj_service('plugin_logic'); |
| 35 |
|
| 36 |
// Fix for missing table error (reported by 2 users - 4% of errors) |
| 37 |
// Check if wp_posts table exists before querying |
| 38 |
if (!$this->tableExists($wpdb->posts)) { |
| 39 |
$this->logger->errorMessage("WordPress posts table not found: " . $wpdb->posts . |
| 40 |
". This may indicate an incorrect table prefix or database configuration issue."); |
| 41 |
return array(); // Return empty array instead of crashing |
| 42 |
} |
| 43 |
|
| 44 |
// get the valid post types |
| 45 |
$options = $abj404logic->getOptions(); |
| 46 |
$recognizedPostTypes = $this->buildPostTypeSqlList($options); |
| 47 |
if ($recognizedPostTypes === '') { |
| 48 |
return array(); |
| 49 |
} |
| 50 |
// ---------------- |
| 51 |
|
| 52 |
if ($slug != "") { |
| 53 |
// Sanitize invalid UTF-8 before SQL to prevent database errors |
| 54 |
// (fixes bug: URLs like %9F%9F%9F%9F-%9F%9F%9F-1.png cause "invalid data" errors) |
| 55 |
$slug = $this->f->sanitizeInvalidUTF8($slug); |
| 56 |
|
| 57 |
// Check if the post_name column supports utf8mb4 collation |
| 58 |
// (fixes bug: Arabic sites on latin1 databases get "invalid data" errors) |
| 59 |
// Note: Check actual column collation, not database default - on mixed setups |
| 60 |
// the database may be latin1 but wp_posts.post_name is utf8mb4 |
| 61 |
$collationResult = $this->queryAndGetResults( |
| 62 |
"SELECT COLLATION_NAME FROM INFORMATION_SCHEMA.COLUMNS |
| 63 |
WHERE TABLE_SCHEMA = DATABASE() |
| 64 |
AND TABLE_NAME = %s |
| 65 |
AND COLUMN_NAME = 'post_name'", |
| 66 |
array('query_params' => array($wpdb->posts), 'log_errors' => false) |
| 67 |
); |
| 68 |
$collationRows = isset($collationResult['rows']) && is_array($collationResult['rows']) ? $collationResult['rows'] : array(); |
| 69 |
$columnCollation = null; |
| 70 |
if (!empty($collationRows) && is_array($collationRows[0])) { |
| 71 |
$first = reset($collationRows[0]); |
| 72 |
$columnCollation = is_scalar($first) ? (string)$first : null; |
| 73 |
} |
| 74 |
if ($columnCollation !== null && strpos(strtolower($columnCollation), 'utf8mb4') !== false) { |
| 75 |
// Column supports utf8mb4 - use CAST for proper Unicode comparison |
| 76 |
$resolvedCollation = $this->sanitizeCollationIdentifier($columnCollation); |
| 77 |
if ($resolvedCollation === '') { |
| 78 |
$resolvedCollation = $this->getPreferredUtf8mb4Collation(); |
| 79 |
} |
| 80 |
$specifiedSlug = " */\n and CAST(wp_posts.post_name AS CHAR CHARACTER SET utf8mb4) COLLATE utf8mb4_unicode_ci = " |
| 81 |
. "'" . esc_sql($slug) . "' \n "; |
| 82 |
$specifiedSlug = str_replace('utf8mb4_unicode_ci', $resolvedCollation, $specifiedSlug); |
| 83 |
} else { |
| 84 |
// Legacy column (latin1, utf8, etc.) - use simple comparison. |
| 85 |
// 4-byte UTF-8 characters (emoji, rare CJK, etc.) cannot exist in a |
| 86 |
// utf8mb3/latin1 column, so skip the slug comparison entirely to avoid |
| 87 |
// "Illegal mix of collations" errors. |
| 88 |
if ($this->f->containsUtf8mb4Characters($slug)) { |
| 89 |
$specifiedSlug = ''; |
| 90 |
} else { |
| 91 |
$specifiedSlug = " */\n and wp_posts.post_name = " |
| 92 |
. "'" . esc_sql($slug) . "' \n "; |
| 93 |
} |
| 94 |
} |
| 95 |
} else { |
| 96 |
$specifiedSlug = ''; |
| 97 |
} |
| 98 |
|
| 99 |
if ($searchTerm != "") { |
| 100 |
$searchTerm = " */\n and lower(wp_posts.post_title) like " |
| 101 |
. "'%" . esc_sql($this->f->strtolower($searchTerm)) . "%' \n "; |
| 102 |
} else { |
| 103 |
$searchTerm = ''; |
| 104 |
} |
| 105 |
|
| 106 |
if ($extraWhereClause != "") { |
| 107 |
$extraWhereClause = " */\n " . $extraWhereClause; |
| 108 |
} |
| 109 |
|
| 110 |
if (!empty($limitResults)) { |
| 111 |
$limitResults = " */\n limit " . $limitResults; |
| 112 |
} |
| 113 |
if (!empty($orderResults)) { |
| 114 |
$orderResults = " */\n order by " . $orderResults; |
| 115 |
} |
| 116 |
|
| 117 |
// load the query and do the replacements. |
| 118 |
$query = ABJ_404_Solution_Functions::readFileContents(__DIR__ . "/sql/getPublishedPagesAndPostsIDs.sql"); |
| 119 |
$query = $this->doTableNameReplacements($query); |
| 120 |
$query = $this->f->str_replace('{recognizedPostTypes}', $recognizedPostTypes, $query); |
| 121 |
$query = $this->f->str_replace('{specifiedSlug}', $specifiedSlug, $query); |
| 122 |
$query = $this->f->str_replace('{searchTerm}', $searchTerm, $query); |
| 123 |
$query = $this->f->str_replace('{extraWhereClause}', $extraWhereClause, $query); |
| 124 |
$query = $this->f->str_replace('{limit-results}', $limitResults, $query); |
| 125 |
$query = $this->f->str_replace('{order-results}', $orderResults, $query); |
| 126 |
|
| 127 |
$result = $this->queryAndGetResults($query, array('result_type' => OBJECT)); |
| 128 |
$queryError = is_string($result['last_error'] ?? '') ? ($result['last_error'] ?? '') : ''; |
| 129 |
$rows = is_array($result['rows']) ? $result['rows'] : array(); |
| 130 |
|
| 131 |
// Collation-error fallback: if CONVERT(... USING utf8mb4) COLLATE still fails |
| 132 |
// (e.g. MySQL version quirk), retry without any COLLATE forcing. This is the |
| 133 |
// pre-4.1.4 behavior that relies on implicit collation resolution. |
| 134 |
if (!empty($queryError) && $this->isCollationError($queryError)) { |
| 135 |
$fpreg = ABJ_404_Solution_FunctionsPreg::getInstance(); |
| 136 |
$fallbackQuery = $fpreg->regexReplace( |
| 137 |
'CONVERT\(wpt\.name USING utf8mb4\) COLLATE [A-Za-z0-9_]+', |
| 138 |
'wpt.name', $query); |
| 139 |
$fallbackQuery = $fpreg->regexReplace( |
| 140 |
'CONVERT\(usefulterms\.grouped_terms USING utf8mb4\) COLLATE [A-Za-z0-9_]+', |
| 141 |
'usefulterms.grouped_terms', is_string($fallbackQuery) ? $fallbackQuery : $query); |
| 142 |
$fallbackResult = $this->queryAndGetResults( |
| 143 |
is_string($fallbackQuery) ? $fallbackQuery : $query, |
| 144 |
array('result_type' => OBJECT, 'log_errors' => false)); |
| 145 |
$queryError = is_string($fallbackResult['last_error'] ?? '') ? ($fallbackResult['last_error'] ?? '') : ''; |
| 146 |
if (empty($queryError)) { |
| 147 |
$rows = is_array($fallbackResult['rows']) ? $fallbackResult['rows'] : array(); |
| 148 |
} |
| 149 |
} |
| 150 |
|
| 151 |
if (!empty($queryError) && $this->isInvalidDataError($queryError) && |
| 152 |
$slug != "" && strpos($query, 'CAST(wp_posts.post_name AS CHAR CHARACTER SET utf8mb4)') !== false) { |
| 153 |
// Compatibility fallback: retry once without CAST/COLLATE for environments |
| 154 |
// where mixed encodings still reject utf8mb4 coercion. |
| 155 |
$fallbackSpecifiedSlug = " */\n and wp_posts.post_name = '" . esc_sql($slug) . "' \n "; |
| 156 |
$fallbackQuery = ABJ_404_Solution_Functions::readFileContents(__DIR__ . "/sql/getPublishedPagesAndPostsIDs.sql"); |
| 157 |
$fallbackQuery = $this->doTableNameReplacements($fallbackQuery); |
| 158 |
$fallbackQuery = $this->f->str_replace('{recognizedPostTypes}', $recognizedPostTypes, $fallbackQuery); |
| 159 |
$fallbackQuery = $this->f->str_replace('{specifiedSlug}', $fallbackSpecifiedSlug, $fallbackQuery); |
| 160 |
$fallbackQuery = $this->f->str_replace('{searchTerm}', $searchTerm, $fallbackQuery); |
| 161 |
$fallbackQuery = $this->f->str_replace('{extraWhereClause}', $extraWhereClause, $fallbackQuery); |
| 162 |
$fallbackQuery = $this->f->str_replace('{limit-results}', $limitResults, $fallbackQuery); |
| 163 |
$fallbackQuery = $this->f->str_replace('{order-results}', $orderResults, $fallbackQuery); |
| 164 |
$fallbackResult = $this->queryAndGetResults($fallbackQuery, array('result_type' => OBJECT, 'log_errors' => false)); |
| 165 |
$fallbackError = is_string($fallbackResult['last_error'] ?? '') ? ($fallbackResult['last_error'] ?? '') : ''; |
| 166 |
if (empty($fallbackError)) { |
| 167 |
$queryError = ''; |
| 168 |
$rows = is_array($fallbackResult['rows']) ? $fallbackResult['rows'] : array(); |
| 169 |
} |
| 170 |
} |
| 171 |
|
| 172 |
// check for errors (use $queryError which tracks the latest attempt) |
| 173 |
if ($queryError) { |
| 174 |
// "Unknown column 'plc.content_keywords'" occurs during the DB migration window |
| 175 |
// when the column hasn't been added yet (e.g. sync lock was stuck for ~24h). |
| 176 |
// Degrade to warning so it doesn't generate email reports for every 404 hit. |
| 177 |
if (stripos($queryError, 'unknown column') !== false && |
| 178 |
stripos($queryError, 'content_keywords') !== false) { |
| 179 |
$this->logger->warn("content_keywords column not yet available (DB migration pending): " . $queryError); |
| 180 |
} else if (!$this->classifyAndHandleInfrastructureError($queryError)) { |
| 181 |
$this->logger->errorMessage("Error executing query. Err: " . $queryError . ", Query: " . $query); |
| 182 |
} |
| 183 |
} |
| 184 |
|
| 185 |
return $rows; |
| 186 |
} |
| 187 |
|
| 188 |
/** Returns rows with the IDs of the published images. |
| 189 |
* @return array<int, object> |
| 190 |
*/ |
| 191 |
function getPublishedImagesIDs() { |
| 192 |
global $wpdb; |
| 193 |
$abj404logic = abj_service('plugin_logic'); |
| 194 |
|
| 195 |
// get the valid post types |
| 196 |
$options = $abj404logic->getOptions(); |
| 197 |
$recognizedPostTypes = $this->buildPostTypeSqlList($options); |
| 198 |
if ($recognizedPostTypes === '') { |
| 199 |
return array(); |
| 200 |
} |
| 201 |
// ---------------- |
| 202 |
|
| 203 |
// load the query and do the replacements. |
| 204 |
$query = ABJ_404_Solution_Functions::readFileContents(__DIR__ . "/sql/getPublishedImageIDs.sql"); |
| 205 |
$query = $this->doTableNameReplacements($query); |
| 206 |
$query = $this->f->str_replace('{recognizedPostTypes}', $recognizedPostTypes, $query); |
| 207 |
|
| 208 |
$result = $this->queryAndGetResults($query, array('result_type' => OBJECT)); |
| 209 |
$queryError = is_string($result['last_error'] ?? '') ? ($result['last_error'] ?? '') : ''; |
| 210 |
if ($queryError && !$this->classifyAndHandleInfrastructureError($queryError)) { |
| 211 |
$this->logger->errorMessage("Error executing query. Err: " . $queryError . ", Query: " . $query); |
| 212 |
} |
| 213 |
|
| 214 |
return is_array($result['rows']) ? $result['rows'] : array(); |
| 215 |
} |
| 216 |
|
| 217 |
/** Returns rows with the defined terms (tags). |
| 218 |
* @param string|null $slug |
| 219 |
* @param int|null $limit |
| 220 |
* @return array<int, object> |
| 221 |
*/ |
| 222 |
function getPublishedTags($slug = null, $limit = null) { |
| 223 |
global $wpdb; |
| 224 |
$abj404logic = abj_service('plugin_logic'); |
| 225 |
|
| 226 |
// get the valid post types |
| 227 |
$options = $abj404logic->getOptions(); |
| 228 |
|
| 229 |
$recognizedCategories = $this->buildCategorySqlList($options); |
| 230 |
|
| 231 |
if ($slug != null) { |
| 232 |
// Sanitize invalid UTF-8 before SQL to prevent database errors |
| 233 |
$slug = $this->f->sanitizeInvalidUTF8($slug); |
| 234 |
$slug = "*/ and wp_terms.slug = '" . esc_sql($slug) . "'\n"; |
| 235 |
} |
| 236 |
|
| 237 |
$limitClause = ''; |
| 238 |
if ($limit !== null && is_numeric($limit) && $limit > 0) { |
| 239 |
$limitClause = "LIMIT " . intval($limit); |
| 240 |
} |
| 241 |
|
| 242 |
// load the query and do the replacements. |
| 243 |
$query = ABJ_404_Solution_Functions::readFileContents(__DIR__ . "/sql/getPublishedTags.sql"); |
| 244 |
$query = $this->f->str_replace('{slug}', $slug, $query); |
| 245 |
$query = $this->f->str_replace('{limit}', $limitClause, $query); |
| 246 |
$query = $this->doTableNameReplacements($query); |
| 247 |
$query = $this->f->str_replace('{recognizedCategories}', $recognizedCategories, $query); |
| 248 |
|
| 249 |
$result = $this->queryAndGetResults($query, array('result_type' => OBJECT)); |
| 250 |
$queryError = is_string($result['last_error'] ?? '') ? ($result['last_error'] ?? '') : ''; |
| 251 |
if ($queryError && !$this->classifyAndHandleInfrastructureError($queryError)) { |
| 252 |
$this->logger->errorMessage("Error executing query. Err: " . $queryError . ", Query: " . $query); |
| 253 |
} |
| 254 |
$rows = is_array($result['rows']) ? $result['rows'] : array(); |
| 255 |
|
| 256 |
$rows = $this->addURLToTermsRows($rows); |
| 257 |
|
| 258 |
return $rows; |
| 259 |
} |
| 260 |
|
| 261 |
/** |
| 262 |
* @param array<int, object> $rows |
| 263 |
* @return array<int, object> |
| 264 |
*/ |
| 265 |
function addURLToTermsRows($rows) { |
| 266 |
// add url data |
| 267 |
global $wp_rewrite; |
| 268 |
$extraPermaStructureCache = array(); |
| 269 |
foreach ($rows as $row) { |
| 270 |
$taxonomy = isset($row->taxonomy) ? (string)$row->taxonomy : ''; |
| 271 |
if (!array_key_exists($taxonomy, $extraPermaStructureCache)) { |
| 272 |
$extraPermaStructureCache[$taxonomy] = $wp_rewrite->get_extra_permastruct($taxonomy); |
| 273 |
} |
| 274 |
$struct = $extraPermaStructureCache[$taxonomy]; |
| 275 |
|
| 276 |
$slug = isset($row->slug) ? (string)$row->slug : ''; |
| 277 |
$url = str_replace('%' . $taxonomy . '%', $slug, $struct); |
| 278 |
|
| 279 |
// TODO verify one of the urls? |
| 280 |
/* |
| 281 |
if (!$verifiedOne) { |
| 282 |
$id = $row->term_id; |
| 283 |
$link = get_tag_link($id); |
| 284 |
$link = get_category_link($id); |
| 285 |
// $link should equal $url |
| 286 |
$verifiedOne = true; |
| 287 |
} |
| 288 |
*/ |
| 289 |
|
| 290 |
/** @var \stdClass $row */ |
| 291 |
$row->url = $url; |
| 292 |
} |
| 293 |
|
| 294 |
return $rows; |
| 295 |
} |
| 296 |
|
| 297 |
/** Returns rows with the defined categories. |
| 298 |
* @param int|null $term_id |
| 299 |
* @param string|null $slug |
| 300 |
* @param int|null $limit |
| 301 |
* @return array<int, object> |
| 302 |
*/ |
| 303 |
function getPublishedCategories($term_id = null, $slug = null, $limit = null) { |
| 304 |
global $wpdb; |
| 305 |
$abj404logic = abj_service('plugin_logic'); |
| 306 |
|
| 307 |
// get the valid post types |
| 308 |
$options = $abj404logic->getOptions(); |
| 309 |
|
| 310 |
$recognizedCategories = $this->buildCategorySqlList($options); |
| 311 |
if ($recognizedCategories === '') { |
| 312 |
$recognizedCategories = "''"; |
| 313 |
} |
| 314 |
|
| 315 |
if ($term_id != null) { |
| 316 |
// Cast to integer for safety even though term_id is currently always null from callers |
| 317 |
$term_id = "*/ and {wp_terms}.term_id = " . intval($term_id) . "\n"; |
| 318 |
} |
| 319 |
|
| 320 |
if ($slug != null) { |
| 321 |
// Sanitize invalid UTF-8 before SQL to prevent database errors |
| 322 |
$slug = $this->f->sanitizeInvalidUTF8($slug); |
| 323 |
$slug = "*/ and {wp_terms}.slug = '" . esc_sql($slug) . "'\n"; |
| 324 |
} |
| 325 |
|
| 326 |
$limitClause = ''; |
| 327 |
if ($limit !== null && is_numeric($limit) && $limit > 0) { |
| 328 |
$limitClause = "LIMIT " . intval($limit); |
| 329 |
} |
| 330 |
|
| 331 |
// load the query and do the replacements. |
| 332 |
$query = ABJ_404_Solution_Functions::readFileContents(__DIR__ . "/sql/getPublishedCategories.sql"); |
| 333 |
$query = $this->f->str_replace('{recognizedCategories}', $recognizedCategories, $query); |
| 334 |
$query = $this->f->str_replace('{term_id}', $term_id !== null ? (string)$term_id : '', $query); |
| 335 |
$query = $this->f->str_replace('{slug}', $slug, $query); |
| 336 |
$query = $this->f->str_replace('{limit}', $limitClause, $query); |
| 337 |
$query = $this->doTableNameReplacements($query); |
| 338 |
|
| 339 |
$result = $this->queryAndGetResults($query, array('result_type' => OBJECT)); |
| 340 |
$queryError = is_string($result['last_error'] ?? '') ? ($result['last_error'] ?? '') : ''; |
| 341 |
if ($queryError && !$this->classifyAndHandleInfrastructureError($queryError)) { |
| 342 |
$this->logger->errorMessage("Error executing query. Err: " . $queryError . ", Query: " . $query); |
| 343 |
} |
| 344 |
$rows = is_array($result['rows']) ? $result['rows'] : array(); |
| 345 |
|
| 346 |
$rows = $this->addURLToTermsRows($rows); |
| 347 |
|
| 348 |
return $rows; |
| 349 |
} |
| 350 |
} |
| 351 |
|
