| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
trait ABJ_404_Solution_DatabaseUpgradesEtc_NGramTrait { |
| 8 |
|
| 9 |
/** @return bool */ |
| 10 |
function scheduleNGramCacheRebuild() { |
| 11 |
global $wpdb; |
| 12 |
|
| 13 |
// MULTISITE: Acquire network-wide lock to prevent race conditions during scheduling |
| 14 |
$lockKey = 'ngram_schedule'; |
| 15 |
$uniqueID = $this->syncUtils->synchronizerAcquireLockTry($lockKey); |
| 16 |
|
| 17 |
if (empty($uniqueID)) { |
| 18 |
$this->logger->debugMessage("N-gram rebuild scheduling: Another process holds the lock. Skipping."); |
| 19 |
return true; // Another site is already handling scheduling |
| 20 |
} |
| 21 |
|
| 22 |
try { |
| 23 |
// MULTISITE: Use network-aware option getter |
| 24 |
$rawCurrentOffset = $this->getNetworkAwareOption('abj404_ngram_rebuild_offset', 0); |
| 25 |
$currentOffset = is_scalar($rawCurrentOffset) ? (int)$rawCurrentOffset : 0; |
| 26 |
|
| 27 |
// MULTISITE: Count pages across all sites if network-activated |
| 28 |
$totalPages = $this->countTotalPagesForNGramRebuild(); |
| 29 |
|
| 30 |
// If offset is between 0 and total (exclusive), rebuild is in progress |
| 31 |
if ($currentOffset > 0 && $currentOffset < $totalPages) { |
| 32 |
$this->logger->debugMessage("N-gram cache rebuild already in progress at offset {$currentOffset} of {$totalPages}"); |
| 33 |
return true; |
| 34 |
} |
| 35 |
|
| 36 |
// Check if already scheduled |
| 37 |
$nextScheduled = wp_next_scheduled('abj404_rebuild_ngram_cache_hook'); |
| 38 |
if ($nextScheduled) { |
| 39 |
$this->logger->debugMessage("N-gram cache rebuild already scheduled for " . date('Y-m-d H:i:s', $nextScheduled)); |
| 40 |
return true; |
| 41 |
} |
| 42 |
|
| 43 |
// MULTISITE: Reset offset using network-aware setter |
| 44 |
$this->updateNetworkAwareOption('abj404_ngram_rebuild_offset', 0); |
| 45 |
|
| 46 |
// Schedule to run in 30 seconds (gives time for activation to complete) |
| 47 |
$scheduleTime = time() + 30; |
| 48 |
$hookName = 'abj404_rebuild_ngram_cache_hook'; |
| 49 |
$scheduled = wp_schedule_single_event($scheduleTime, $hookName); |
| 50 |
|
| 51 |
if ($scheduled === false) { |
| 52 |
// Quick check for DISABLE_WP_CRON as immediate diagnostic |
| 53 |
if (defined('DISABLE_WP_CRON') && DISABLE_WP_CRON) { |
| 54 |
$this->logger->errorMessage( |
| 55 |
"Cannot schedule N-gram cache rebuild: WP-Cron is disabled (DISABLE_WP_CRON=true). " . |
| 56 |
"Consider enabling WP-Cron or using server-side cron with a fallback mechanism." |
| 57 |
); |
| 58 |
return false; |
| 59 |
} |
| 60 |
|
| 61 |
global $wpdb; |
| 62 |
|
| 63 |
// Gather comprehensive diagnostic information for troubleshooting |
| 64 |
$cronDisabled = defined('DISABLE_WP_CRON') && DISABLE_WP_CRON; |
| 65 |
$alreadyScheduled = wp_next_scheduled($hookName); |
| 66 |
$dbError = !empty($wpdb->last_error) ? $wpdb->last_error : 'none'; |
| 67 |
$rawRebuildOffset = $this->getNetworkAwareOption('abj404_ngram_rebuild_offset', 'not set'); |
| 68 |
$rebuildOffset = is_scalar($rawRebuildOffset) ? (string)$rawRebuildOffset : 'not set'; |
| 69 |
$rawCacheInit = $this->getNetworkAwareOption('abj404_ngram_cache_initialized', 'not set'); |
| 70 |
$cacheInitialized = is_scalar($rawCacheInit) ? (string)$rawCacheInit : 'not set'; |
| 71 |
|
| 72 |
$errorMsg = sprintf( |
| 73 |
"Failed to schedule N-gram cache rebuild. Hook: %s, Schedule time: %d (current: %d), " . |
| 74 |
"Already scheduled: %s, WP-Cron disabled: %s, DB error: %s, " . |
| 75 |
"Rebuild offset: %s, Cache initialized: %s, Multisite: %s, Blog ID: %d", |
| 76 |
$hookName, |
| 77 |
$scheduleTime, |
| 78 |
time(), |
| 79 |
$alreadyScheduled ? date('Y-m-d H:i:s', $alreadyScheduled) : 'no', |
| 80 |
$cronDisabled ? 'yes' : 'no', |
| 81 |
$dbError, |
| 82 |
$rebuildOffset, |
| 83 |
$cacheInitialized, |
| 84 |
is_multisite() ? 'yes' : 'no', |
| 85 |
get_current_blog_id() |
| 86 |
); |
| 87 |
|
| 88 |
// Pattern 7 (defense-in-depth): if a concurrent infra-level |
| 89 |
// DB error (disk full, read-only, crashed table) contributed |
| 90 |
// to wp_schedule_single_event() failing, surface it as a |
| 91 |
// plugin-page admin notice. The cron failure itself remains |
| 92 |
// ERROR level — the user must act on a broken cron — but |
| 93 |
// the underlying hosting issue is the actionable cause. |
| 94 |
if (!empty($wpdb->last_error)) { |
| 95 |
$this->dao->classifyAndHandleInfrastructureError($wpdb->last_error); |
| 96 |
} |
| 97 |
|
| 98 |
$this->logger->errorMessage($errorMsg); |
| 99 |
return false; |
| 100 |
} |
| 101 |
|
| 102 |
$context = is_multisite() ? ' (network-wide)' : ''; |
| 103 |
$this->logger->infoMessage("N-gram cache rebuild scheduled to start in 30 seconds{$context}."); |
| 104 |
return true; |
| 105 |
|
| 106 |
} finally { |
| 107 |
// Always release the lock |
| 108 |
$this->syncUtils->synchronizerReleaseLock($uniqueID, $lockKey); |
| 109 |
} |
| 110 |
} |
| 111 |
|
| 112 |
/** |
| 113 |
* WP-Cron callback: Rebuild N-gram cache in batches (async). |
| 114 |
* |
| 115 |
* Lock Acquisition Flow (per-batch): |
| 116 |
* 1. Create unique ID |
| 117 |
* 2. Write unique ID to lock if empty |
| 118 |
* 3. Sleep 30ms (allows race condition resolution) |
| 119 |
* 4. Read lock back and verify ownership |
| 120 |
* 5. Process batch only if lock belongs to this process |
| 121 |
* 6. Release lock in finally block |
| 122 |
* |
| 123 |
* MULTISITE BEHAVIOR (FIXED): |
| 124 |
* - Processes one site at a time completely before moving to the next site |
| 125 |
* - Uses network options to track: pending sites, current site, and offset within current site |
| 126 |
* - Switches to each site's blog context before processing its pages |
| 127 |
* - Prevents the bug where only the first site got cache entries |
| 128 |
* - Progress tracking shows per-site and network-wide completion status |
| 129 |
* |
| 130 |
* SINGLE SITE BEHAVIOR: |
| 131 |
* - Uses simple offset tracking with network-aware options |
| 132 |
* - Processes all pages in batches until complete |
| 133 |
* |
| 134 |
* @param int $offset Current batch offset (default: 0, overridden by network options) |
| 135 |
* @return void |
| 136 |
*/ |
| 137 |
function rebuildNGramCacheAsync($offset = 0) { |
| 138 |
global $wpdb; |
| 139 |
|
| 140 |
// Acquire lock using SynchronizationUtils (per-batch lock) |
| 141 |
$uniqueID = $this->syncUtils->synchronizerAcquireLockTry('ngram_rebuild'); |
| 142 |
if (empty($uniqueID)) { |
| 143 |
$this->logger->debugMessage("N-gram async rebuild batch already processing (another process holds lock). Skipping."); |
| 144 |
return; |
| 145 |
} |
| 146 |
|
| 147 |
try { |
| 148 |
$batchSize = 50; // Smaller batches for async processing |
| 149 |
$maxBatchesPerRun = 20; // Process up to 1000 pages per cron run |
| 150 |
|
| 151 |
// MULTISITE: Process one site at a time to ensure all sites get cache entries |
| 152 |
if ($this->isNetworkActivated()) { |
| 153 |
// Get or initialize list of pending sites |
| 154 |
$pendingSitesRaw = $this->getNetworkAwareOption('abj404_ngram_pending_sites', null); |
| 155 |
/** @var array<int, int> $pendingSites */ |
| 156 |
$pendingSites = is_array($pendingSitesRaw) ? $pendingSitesRaw : []; |
| 157 |
|
| 158 |
if ($pendingSitesRaw === null) { |
| 159 |
// First run: Initialize site list and tracking |
| 160 |
$sites = get_sites(array('fields' => 'ids', 'number' => 0)); |
| 161 |
$this->updateNetworkAwareOption('abj404_ngram_pending_sites', $sites); |
| 162 |
$this->updateNetworkAwareOption('abj404_ngram_total_sites', count($sites)); |
| 163 |
$this->updateNetworkAwareOption('abj404_ngram_current_site_offset', 0); |
| 164 |
$pendingSites = $sites; |
| 165 |
} |
| 166 |
|
| 167 |
if (empty($pendingSites)) { |
| 168 |
// All sites processed! |
| 169 |
$this->updateNetworkAwareOption('abj404_ngram_cache_initialized', '1'); |
| 170 |
$this->updateNetworkAwareOption('abj404_ngram_pending_sites', null); |
| 171 |
$this->updateNetworkAwareOption('abj404_ngram_total_sites', null); |
| 172 |
$this->updateNetworkAwareOption('abj404_ngram_current_site_offset', null); |
| 173 |
$this->logger->infoMessage("N-gram cache rebuild complete for all sites in network!"); |
| 174 |
return; |
| 175 |
} |
| 176 |
|
| 177 |
// Get current site to process |
| 178 |
$currentSiteId = (int)$pendingSites[0]; |
| 179 |
$rawOffset = $this->getNetworkAwareOption('abj404_ngram_current_site_offset', 0); |
| 180 |
$offset = is_scalar($rawOffset) ? (int)$rawOffset : 0; |
| 181 |
$rawTotalSites = $this->getNetworkAwareOption('abj404_ngram_total_sites', count($pendingSites)); |
| 182 |
$totalSites = is_scalar($rawTotalSites) ? (int)$rawTotalSites : count($pendingSites); |
| 183 |
$completedSites = $totalSites - count($pendingSites); |
| 184 |
|
| 185 |
// Switch to the site being processed |
| 186 |
switch_to_blog($currentSiteId); |
| 187 |
|
| 188 |
// Count pages for THIS site only |
| 189 |
$permalinkCacheTable = $this->dao->getPrefixedTableName('abj404_permalink_cache'); |
| 190 |
$sitePages = $this->dao->queryScalarInt("SELECT COUNT(*) AS c FROM {$permalinkCacheTable}"); |
| 191 |
|
| 192 |
if ($sitePages == 0) { |
| 193 |
// This site has no pages, move to next site |
| 194 |
array_shift($pendingSites); |
| 195 |
$this->updateNetworkAwareOption('abj404_ngram_pending_sites', $pendingSites); |
| 196 |
$this->updateNetworkAwareOption('abj404_ngram_current_site_offset', 0); |
| 197 |
restore_current_blog(); |
| 198 |
|
| 199 |
$this->logger->infoMessage(sprintf( |
| 200 |
"Site %d has no pages. Moving to next site. Progress: %d/%d sites completed.", |
| 201 |
$currentSiteId, |
| 202 |
$completedSites + 1, |
| 203 |
$totalSites |
| 204 |
)); |
| 205 |
|
| 206 |
// Reschedule immediately for next site |
| 207 |
wp_schedule_single_event(time(), 'abj404_rebuild_ngram_cache_hook'); |
| 208 |
return; |
| 209 |
} |
| 210 |
|
| 211 |
$this->logger->infoMessage(sprintf( |
| 212 |
"Processing N-gram cache for site %d (Site %d of %d): Offset %d of %d pages", |
| 213 |
$currentSiteId, |
| 214 |
$completedSites + 1, |
| 215 |
$totalSites, |
| 216 |
$offset, |
| 217 |
$sitePages |
| 218 |
)); |
| 219 |
|
| 220 |
// Process batches for current site |
| 221 |
$batchesProcessed = 0; |
| 222 |
$totalStats = ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 223 |
|
| 224 |
while ($batchesProcessed < $maxBatchesPerRun && $offset < $sitePages) { |
| 225 |
try { |
| 226 |
// Process batch (already switched to correct blog) |
| 227 |
$stats = $this->ngramFilter->rebuildCache($batchSize, $offset); |
| 228 |
|
| 229 |
$totalStats['processed'] += $stats['processed']; |
| 230 |
$totalStats['success'] += $stats['success']; |
| 231 |
$totalStats['failed'] += $stats['failed']; |
| 232 |
|
| 233 |
$offset += $batchSize; |
| 234 |
$batchesProcessed++; |
| 235 |
|
| 236 |
// Update offset for current site |
| 237 |
$this->updateNetworkAwareOption('abj404_ngram_current_site_offset', $offset); |
| 238 |
|
| 239 |
// Stop if we processed fewer pages than expected (end of site data) |
| 240 |
if ($stats['processed'] < $batchSize) { |
| 241 |
break; |
| 242 |
} |
| 243 |
|
| 244 |
} catch (Exception $e) { |
| 245 |
$this->logger->errorMessage("Error during N-gram rebuild for site {$currentSiteId} at offset {$offset}: " . $e->getMessage()); |
| 246 |
$totalStats['failed'] += $batchSize; |
| 247 |
$offset += $batchSize; |
| 248 |
$batchesProcessed++; |
| 249 |
$this->updateNetworkAwareOption('abj404_ngram_current_site_offset', $offset); |
| 250 |
} |
| 251 |
} |
| 252 |
|
| 253 |
$progress = $sitePages > 0 ? min(100, round(($offset / $sitePages) * 100, 1)) : 100; |
| 254 |
|
| 255 |
$this->logger->infoMessage(sprintf( |
| 256 |
"Site %d progress: %d%% complete (%d/%d pages), %d success, %d failed", |
| 257 |
$currentSiteId, |
| 258 |
$progress, |
| 259 |
$offset, |
| 260 |
$sitePages, |
| 261 |
$totalStats['success'], |
| 262 |
$totalStats['failed'] |
| 263 |
)); |
| 264 |
|
| 265 |
// Check if current site is complete |
| 266 |
if ($offset >= $sitePages) { |
| 267 |
// Site complete! Move to next site |
| 268 |
array_shift($pendingSites); |
| 269 |
$this->updateNetworkAwareOption('abj404_ngram_pending_sites', $pendingSites); |
| 270 |
$this->updateNetworkAwareOption('abj404_ngram_current_site_offset', 0); |
| 271 |
|
| 272 |
$this->logger->infoMessage(sprintf( |
| 273 |
"Site %d complete! Progress: %d/%d sites completed.", |
| 274 |
$currentSiteId, |
| 275 |
$completedSites + 1, |
| 276 |
$totalSites |
| 277 |
)); |
| 278 |
} |
| 279 |
|
| 280 |
restore_current_blog(); |
| 281 |
|
| 282 |
// Reschedule for next batch or next site |
| 283 |
wp_schedule_single_event(time() + 10, 'abj404_rebuild_ngram_cache_hook'); |
| 284 |
|
| 285 |
} else { |
| 286 |
// SINGLE SITE: Use original simple logic |
| 287 |
$rawSingleOffset = $this->getNetworkAwareOption('abj404_ngram_rebuild_offset', 0); |
| 288 |
$offset = is_scalar($rawSingleOffset) ? (int)$rawSingleOffset : 0; |
| 289 |
$permalinkCacheTable = $this->dao->getPrefixedTableName('abj404_permalink_cache'); |
| 290 |
$totalPages = $this->dao->queryScalarInt("SELECT COUNT(*) AS c FROM {$permalinkCacheTable}"); |
| 291 |
|
| 292 |
if ($totalPages == 0) { |
| 293 |
$this->logger->debugMessage("No pages to process. Setting initialized flag."); |
| 294 |
$this->updateNetworkAwareOption('abj404_ngram_cache_initialized', '1'); |
| 295 |
$this->updateNetworkAwareOption('abj404_ngram_rebuild_offset', 0); |
| 296 |
return; |
| 297 |
} |
| 298 |
|
| 299 |
$this->logger->infoMessage(sprintf( |
| 300 |
"Async N-gram rebuild: Processing batch at offset %d of %d total pages", |
| 301 |
$offset, |
| 302 |
$totalPages |
| 303 |
)); |
| 304 |
|
| 305 |
// Process batches |
| 306 |
$batchesProcessed = 0; |
| 307 |
$totalStats = ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 308 |
|
| 309 |
while ($batchesProcessed < $maxBatchesPerRun && $offset < $totalPages) { |
| 310 |
try { |
| 311 |
$stats = $this->ngramFilter->rebuildCache($batchSize, $offset); |
| 312 |
|
| 313 |
$totalStats['processed'] += $stats['processed']; |
| 314 |
$totalStats['success'] += $stats['success']; |
| 315 |
$totalStats['failed'] += $stats['failed']; |
| 316 |
|
| 317 |
$offset += $batchSize; |
| 318 |
$batchesProcessed++; |
| 319 |
|
| 320 |
$this->updateNetworkAwareOption('abj404_ngram_rebuild_offset', $offset); |
| 321 |
|
| 322 |
if ($stats['processed'] < $batchSize) { |
| 323 |
break; |
| 324 |
} |
| 325 |
|
| 326 |
} catch (Exception $e) { |
| 327 |
$this->logger->errorMessage("Error during async N-gram cache rebuild at offset {$offset}: " . $e->getMessage()); |
| 328 |
$totalStats['failed'] += $batchSize; |
| 329 |
$offset += $batchSize; |
| 330 |
$batchesProcessed++; |
| 331 |
$this->updateNetworkAwareOption('abj404_ngram_rebuild_offset', $offset); |
| 332 |
} |
| 333 |
} |
| 334 |
|
| 335 |
$progress = $totalPages > 0 ? min(100, round(($offset / $totalPages) * 100, 1)) : 100; |
| 336 |
|
| 337 |
$this->logger->infoMessage(sprintf( |
| 338 |
"Async N-gram rebuild progress: %d%% complete (%d/%d pages), %d success, %d failed", |
| 339 |
$progress, |
| 340 |
$offset, |
| 341 |
$totalPages, |
| 342 |
$totalStats['success'], |
| 343 |
$totalStats['failed'] |
| 344 |
)); |
| 345 |
|
| 346 |
if ($offset < $totalPages) { |
| 347 |
$scheduleTime = time() + 10; |
| 348 |
$hookName = 'abj404_rebuild_ngram_cache_hook'; |
| 349 |
$scheduled = wp_schedule_single_event($scheduleTime, $hookName, [$offset]); |
| 350 |
|
| 351 |
if ($scheduled === false) { |
| 352 |
// Quick check for DISABLE_WP_CRON as immediate diagnostic |
| 353 |
if (defined('DISABLE_WP_CRON') && DISABLE_WP_CRON) { |
| 354 |
$this->logger->errorMessage( |
| 355 |
"Cannot schedule next N-gram rebuild batch at offset {$offset}: WP-Cron is disabled (DISABLE_WP_CRON=true). " . |
| 356 |
"Consider enabling WP-Cron or using server-side cron with a fallback mechanism." |
| 357 |
); |
| 358 |
// Don't return - let the rebuild complete gracefully, just log the issue |
| 359 |
} else { |
| 360 |
global $wpdb; |
| 361 |
|
| 362 |
// Gather comprehensive diagnostic information for troubleshooting |
| 363 |
$cronDisabled = defined('DISABLE_WP_CRON') && DISABLE_WP_CRON; |
| 364 |
$alreadyScheduled = wp_next_scheduled($hookName, [$offset]); |
| 365 |
$dbError = !empty($wpdb->last_error) ? $wpdb->last_error : 'none'; |
| 366 |
$rawCacheInit2 = $this->getNetworkAwareOption('abj404_ngram_cache_initialized', 'not set'); |
| 367 |
$cacheInitialized = is_scalar($rawCacheInit2) ? (string)$rawCacheInit2 : 'not set'; |
| 368 |
|
| 369 |
$errorMsg = sprintf( |
| 370 |
"Failed to schedule next N-gram rebuild batch at offset %d. Hook: %s, Schedule time: %d (current: %d), " . |
| 371 |
"Already scheduled: %s, WP-Cron disabled: %s, DB error: %s, " . |
| 372 |
"Cache initialized: %s, Progress: %.1f%%, Multisite: %s, Blog ID: %d", |
| 373 |
$offset, |
| 374 |
$hookName, |
| 375 |
$scheduleTime, |
| 376 |
time(), |
| 377 |
$alreadyScheduled ? date('Y-m-d H:i:s', $alreadyScheduled) : 'no', |
| 378 |
$cronDisabled ? 'yes' : 'no', |
| 379 |
$dbError, |
| 380 |
$cacheInitialized, |
| 381 |
$progress, |
| 382 |
is_multisite() ? 'yes' : 'no', |
| 383 |
get_current_blog_id() |
| 384 |
); |
| 385 |
|
| 386 |
// Pattern 7 (defense-in-depth): if a concurrent |
| 387 |
// infra-level DB error contributed to the cron |
| 388 |
// failure, surface the hosting cause as a |
| 389 |
// plugin-page admin notice. |
| 390 |
if (!empty($wpdb->last_error)) { |
| 391 |
$this->dao->classifyAndHandleInfrastructureError($wpdb->last_error); |
| 392 |
} |
| 393 |
|
| 394 |
$this->logger->errorMessage($errorMsg); |
| 395 |
} |
| 396 |
} |
| 397 |
} else { |
| 398 |
// All done! |
| 399 |
$this->updateNetworkAwareOption('abj404_ngram_cache_initialized', '1'); |
| 400 |
$this->updateNetworkAwareOption('abj404_ngram_rebuild_offset', 0); |
| 401 |
$this->logger->infoMessage("N-gram cache rebuild complete! Total: {$totalStats['processed']} processed, {$totalStats['success']} success, {$totalStats['failed']} failed."); |
| 402 |
} |
| 403 |
} |
| 404 |
|
| 405 |
} finally { |
| 406 |
// Always release lock, even if exception occurs |
| 407 |
$this->syncUtils->synchronizerReleaseLock($uniqueID, 'ngram_rebuild'); |
| 408 |
} |
| 409 |
} |
| 410 |
|
| 411 |
/** |
| 412 |
* Rebuild the N-gram cache for all pages (synchronous). |
| 413 |
* |
| 414 |
* WARNING: This method is synchronous and can take minutes on large sites. |
| 415 |
* Use scheduleNGramCacheRebuild() instead for non-blocking background processing. |
| 416 |
* |
| 417 |
* This method is kept for manual rebuilds and testing purposes. |
| 418 |
* |
| 419 |
* @param int $batchSize Number of pages to process per batch (default: 100) |
| 420 |
* @param bool $forceRebuild Force rebuild even if cache is already populated (default: false) |
| 421 |
* @return array<string, mixed> Statistics: ['total_pages' => int, 'processed' => int, 'success' => int, 'failed' => int] |
| 422 |
*/ |
| 423 |
function rebuildNGramCache($batchSize = 100, $forceRebuild = false) { |
| 424 |
global $wpdb; |
| 425 |
|
| 426 |
// Use the same SynchronizationUtils lock as rebuildNGramCacheAsync() to |
| 427 |
// prevent TRUNCATE TABLE from racing with async batch inserts. |
| 428 |
$lockKey = 'ngram_rebuild'; |
| 429 |
$uniqueID = $this->syncUtils->synchronizerAcquireLockTry($lockKey); |
| 430 |
if (empty($uniqueID)) { |
| 431 |
$this->logger->infoMessage("N-gram rebuild already in progress (locked). Skipping."); |
| 432 |
return [ |
| 433 |
'total_pages' => 0, |
| 434 |
'processed' => 0, |
| 435 |
'success' => 0, |
| 436 |
'failed' => 0, |
| 437 |
'locked' => true |
| 438 |
]; |
| 439 |
} |
| 440 |
|
| 441 |
try { |
| 442 |
$ngramTable = $this->dao->getPrefixedTableName('abj404_ngram_cache'); |
| 443 |
$permalinkCacheTable = $this->dao->getPrefixedTableName('abj404_permalink_cache'); |
| 444 |
|
| 445 |
// Check if cache is already populated (unless force rebuild) |
| 446 |
if (!$forceRebuild) { |
| 447 |
$existingCount = $this->dao->queryScalarInt("SELECT COUNT(*) AS c FROM {$ngramTable}"); |
| 448 |
if ($existingCount > 0) { |
| 449 |
$this->logger->debugMessage("N-gram cache already contains {$existingCount} entries. Skipping rebuild (use forceRebuild=true to override)."); |
| 450 |
return [ |
| 451 |
'total_pages' => $existingCount, |
| 452 |
'processed' => 0, |
| 453 |
'success' => $existingCount, |
| 454 |
'failed' => 0, |
| 455 |
'skipped' => true |
| 456 |
]; |
| 457 |
} |
| 458 |
} |
| 459 |
|
| 460 |
$this->logger->debugMessage("Starting N-gram cache rebuild..."); |
| 461 |
|
| 462 |
// Clear existing N-gram cache (only if force rebuild or empty). |
| 463 |
// skip_repair: TRUNCATE itself is the recovery path during rebuild; |
| 464 |
// we must not recurse into the missing-table repairer here. |
| 465 |
$truncateResult = $this->dao->queryAndGetResults( |
| 466 |
"TRUNCATE TABLE {$ngramTable}", |
| 467 |
['skip_repair' => true] |
| 468 |
); |
| 469 |
$truncateError = isset($truncateResult['last_error']) && is_string($truncateResult['last_error']) ? $truncateResult['last_error'] : ''; |
| 470 |
if ($truncateError !== '') { |
| 471 |
if (!$this->dao->classifyAndHandleInfrastructureError($truncateError)) { |
| 472 |
$this->logger->errorMessage("Failed to truncate N-gram cache table: " . $truncateError); |
| 473 |
} |
| 474 |
return ['total_pages' => 0, 'processed' => 0, 'success' => 0, 'failed' => 1, 'error' => $truncateError]; |
| 475 |
} |
| 476 |
|
| 477 |
// Invalidate coverage ratio caches immediately after truncate |
| 478 |
// This prevents stale transient data from making SpellChecker believe |
| 479 |
// the cache is populated when it's actually empty |
| 480 |
$this->ngramFilter->invalidateCoverageCaches(); |
| 481 |
|
| 482 |
// Get total page count from permalink cache |
| 483 |
$totalPagesResult = $this->dao->queryAndGetResults("SELECT COUNT(*) AS c FROM {$permalinkCacheTable}"); |
| 484 |
$totalPagesRows = isset($totalPagesResult['rows']) && is_array($totalPagesResult['rows']) ? $totalPagesResult['rows'] : []; |
| 485 |
$totalPagesRow = $totalPagesRows[0] ?? null; |
| 486 |
|
| 487 |
if (!is_array($totalPagesRow) || !isset($totalPagesRow['c'])) { |
| 488 |
$countError = isset($totalPagesResult['last_error']) && is_string($totalPagesResult['last_error']) ? $totalPagesResult['last_error'] : ''; |
| 489 |
if (!$this->dao->classifyAndHandleInfrastructureError($countError)) { |
| 490 |
$this->logger->errorMessage("Failed to query permalink cache table: " . $countError); |
| 491 |
} |
| 492 |
return ['total_pages' => 0, 'processed' => 0, 'success' => 0, 'failed' => 1, 'error' => $countError]; |
| 493 |
} |
| 494 |
$totalPages = is_scalar($totalPagesRow['c']) ? (int)$totalPagesRow['c'] : 0; |
| 495 |
|
| 496 |
if ($totalPages == 0) { |
| 497 |
$this->logger->debugMessage("No pages in permalink cache. N-gram cache rebuild skipped (will rebuild when pages are added)."); |
| 498 |
return ['total_pages' => 0, 'processed' => 0, 'success' => 0, 'failed' => 0]; |
| 499 |
} |
| 500 |
|
| 501 |
$this->logger->infoMessage("Rebuilding N-gram cache for {$totalPages} pages in batches of {$batchSize}..."); |
| 502 |
|
| 503 |
// Process in batches |
| 504 |
$offset = 0; |
| 505 |
$totalStats = ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 506 |
|
| 507 |
while ($offset < $totalPages) { |
| 508 |
try { |
| 509 |
$stats = $this->ngramFilter->rebuildCache($batchSize, $offset); |
| 510 |
|
| 511 |
$totalStats['processed'] += $stats['processed']; |
| 512 |
$totalStats['success'] += $stats['success']; |
| 513 |
$totalStats['failed'] += $stats['failed']; |
| 514 |
|
| 515 |
$offset += $batchSize; |
| 516 |
|
| 517 |
// Stop if we processed fewer pages than expected (end of data) |
| 518 |
if ($stats['processed'] < $batchSize) { |
| 519 |
break; |
| 520 |
} |
| 521 |
|
| 522 |
} catch (Exception $e) { |
| 523 |
$this->logger->errorMessage("Error during N-gram cache rebuild at offset {$offset}: " . $e->getMessage()); |
| 524 |
$totalStats['failed'] += $batchSize; // Mark batch as failed |
| 525 |
$offset += $batchSize; // Continue to next batch |
| 526 |
} |
| 527 |
} |
| 528 |
|
| 529 |
$totalStats['total_pages'] = $totalPages; |
| 530 |
|
| 531 |
$successRate = $totalStats['processed'] > 0 ? |
| 532 |
round(($totalStats['success'] / $totalStats['processed']) * 100, 1) : 0; |
| 533 |
|
| 534 |
$this->logger->infoMessage(sprintf( |
| 535 |
"N-gram cache rebuild complete: %d pages processed, %d success, %d failed (%.1f%% success rate)", |
| 536 |
$totalStats['processed'], |
| 537 |
$totalStats['success'], |
| 538 |
$totalStats['failed'], |
| 539 |
$successRate |
| 540 |
)); |
| 541 |
|
| 542 |
return $totalStats; |
| 543 |
|
| 544 |
} finally { |
| 545 |
// Always release the lock |
| 546 |
$this->syncUtils->synchronizerReleaseLock($uniqueID, $lockKey); |
| 547 |
} |
| 548 |
} |
| 549 |
|
| 550 |
/** |
| 551 |
* Sync missing ngram entries for posts/pages and categories that don't have them yet. |
| 552 |
* This runs as a background task to add entries for newly published content. |
| 553 |
* |
| 554 |
* Uses the same lock as rebuildNGramCache to prevent concurrent execution. |
| 555 |
* |
| 556 |
* @param int $batchSize Number of entries to process per batch (default: 50) |
| 557 |
* @return array<string, mixed> Statistics: ['posts_added' => int, 'posts_failed' => int, 'categories_added' => int, 'categories_failed' => int] |
| 558 |
*/ |
| 559 |
function syncMissingNGrams($batchSize = 50) { |
| 560 |
global $wpdb; |
| 561 |
|
| 562 |
// Use the same SynchronizationUtils lock as rebuildNGramCacheAsync() and |
| 563 |
// rebuildNGramCache() to prevent concurrent modification of the ngram table. |
| 564 |
$lockKey = 'ngram_rebuild'; |
| 565 |
$uniqueID = $this->syncUtils->synchronizerAcquireLockTry($lockKey); |
| 566 |
if (empty($uniqueID)) { |
| 567 |
$this->logger->debugMessage("Ngram sync skipped - rebuild/sync already in progress."); |
| 568 |
return ['posts_added' => 0, 'posts_failed' => 0, 'categories_added' => 0, 'categories_failed' => 0, 'locked' => true]; |
| 569 |
} |
| 570 |
|
| 571 |
try { |
| 572 |
$ngramTable = $this->dao->getPrefixedTableName('abj404_ngram_cache'); |
| 573 |
$permalinkCacheTable = $this->dao->getPrefixedTableName('abj404_permalink_cache'); |
| 574 |
|
| 575 |
$stats = ['posts_added' => 0, 'posts_failed' => 0, 'categories_added' => 0, 'categories_failed' => 0]; |
| 576 |
|
| 577 |
// ===== SYNC POSTS ===== |
| 578 |
// Find posts in permalink cache that don't have ngram entries |
| 579 |
// Using LEFT JOIN to find missing entries |
| 580 |
$missingResult = $this->dao->queryAndGetResults( |
| 581 |
"SELECT pc.id |
| 582 |
FROM {$permalinkCacheTable} pc |
| 583 |
LEFT JOIN {$ngramTable} ng ON pc.id = ng.id AND ng.type = 'post' |
| 584 |
WHERE ng.id IS NULL |
| 585 |
LIMIT %d", |
| 586 |
['query_params' => [$batchSize]] |
| 587 |
); |
| 588 |
|
| 589 |
$missingError = isset($missingResult['last_error']) && is_string($missingResult['last_error']) ? $missingResult['last_error'] : ''; |
| 590 |
if ($missingError !== '') { |
| 591 |
if (!$this->dao->classifyAndHandleInfrastructureError($missingError)) { |
| 592 |
$this->logger->errorMessage("Failed to query for missing post ngram entries: " . $missingError); |
| 593 |
} |
| 594 |
return array_merge($stats, ['error' => $missingError]); |
| 595 |
} |
| 596 |
$missingRows = isset($missingResult['rows']) && is_array($missingResult['rows']) ? $missingResult['rows'] : []; |
| 597 |
$missingIds = []; |
| 598 |
foreach ($missingRows as $row) { |
| 599 |
if (is_array($row) && isset($row['id'])) { |
| 600 |
$missingIds[] = $row['id']; |
| 601 |
} |
| 602 |
} |
| 603 |
|
| 604 |
if (!empty($missingIds)) { |
| 605 |
$this->logger->infoMessage("Found " . count($missingIds) . " posts missing ngram entries. Adding..."); |
| 606 |
|
| 607 |
// Add ngrams for missing posts |
| 608 |
$result = $this->ngramFilter->updateNGramsForPages($missingIds); |
| 609 |
|
| 610 |
$stats['posts_added'] = $result['success']; |
| 611 |
$stats['posts_failed'] = $result['failed']; |
| 612 |
} else { |
| 613 |
$this->logger->debugMessage("No missing post ngram entries found. All posts are synced."); |
| 614 |
} |
| 615 |
|
| 616 |
// ===== SYNC CATEGORIES ===== |
| 617 |
// Get all published categories |
| 618 |
$categories = $this->dao->getPublishedCategories(); |
| 619 |
|
| 620 |
if (!empty($categories)) { |
| 621 |
$missingCategories = []; |
| 622 |
|
| 623 |
// Check which categories are missing from ngram cache |
| 624 |
foreach ($categories as $category) { |
| 625 |
/** @var object{term_id: int, url: string} $category */ |
| 626 |
$termId = (int)$category->term_id; |
| 627 |
|
| 628 |
// Check if this category already has an ngram entry |
| 629 |
$exists = $this->dao->queryScalarInt( |
| 630 |
"SELECT COUNT(*) AS c FROM {$ngramTable} WHERE id = %d AND type = 'category'", |
| 631 |
['query_params' => [$termId]] |
| 632 |
); |
| 633 |
|
| 634 |
if ($exists == 0) { |
| 635 |
$missingCategories[] = $category; |
| 636 |
} |
| 637 |
} |
| 638 |
|
| 639 |
if (!empty($missingCategories)) { |
| 640 |
$this->logger->infoMessage("Found " . count($missingCategories) . " categories missing ngram entries. Adding..."); |
| 641 |
|
| 642 |
// Add ngrams for missing categories |
| 643 |
foreach ($missingCategories as $category) { |
| 644 |
try { |
| 645 |
/** @var object{term_id: int, url: string} $category */ |
| 646 |
$termId = (int)$category->term_id; |
| 647 |
$url = (string)$category->url; |
| 648 |
|
| 649 |
if (empty($url) || $url === 'in code') { |
| 650 |
$this->logger->debugMessage("Skipping category {$termId} - no valid URL"); |
| 651 |
continue; |
| 652 |
} |
| 653 |
|
| 654 |
// Normalize URL |
| 655 |
$urlNormalized = $this->f->strtolower(trim($url)); |
| 656 |
|
| 657 |
// Extract N-grams |
| 658 |
$ngrams = $this->ngramFilter->extractNGrams($urlNormalized); |
| 659 |
|
| 660 |
// Store with type='category' |
| 661 |
$success = $this->ngramFilter->storeNGrams($termId, $url, $urlNormalized, $ngrams, 'category'); |
| 662 |
|
| 663 |
if ($success) { |
| 664 |
$stats['categories_added']++; |
| 665 |
} else { |
| 666 |
$stats['categories_failed']++; |
| 667 |
} |
| 668 |
} catch (Exception $e) { |
| 669 |
$this->logger->errorMessage("Failed to add ngram for category {$termId}: " . $e->getMessage()); |
| 670 |
$stats['categories_failed']++; |
| 671 |
} |
| 672 |
} |
| 673 |
} else { |
| 674 |
$this->logger->debugMessage("No missing category ngram entries found. All categories are synced."); |
| 675 |
} |
| 676 |
} |
| 677 |
|
| 678 |
$this->logger->infoMessage("Ngram sync complete: {$stats['posts_added']} posts added, {$stats['posts_failed']} posts failed, {$stats['categories_added']} categories added, {$stats['categories_failed']} categories failed."); |
| 679 |
|
| 680 |
return $stats; |
| 681 |
|
| 682 |
} finally { |
| 683 |
// Always release the lock |
| 684 |
$this->syncUtils->synchronizerReleaseLock($uniqueID, $lockKey); |
| 685 |
} |
| 686 |
} |
| 687 |
|
| 688 |
/** |
| 689 |
* Cleanup orphaned ngram entries that don't have corresponding posts/pages or categories. |
| 690 |
* This removes stale entries when posts are deleted or categories are removed. |
| 691 |
* |
| 692 |
* @return array<string, mixed> Statistics: ['posts_deleted' => int, 'categories_deleted' => int, 'errors' => int] |
| 693 |
*/ |
| 694 |
function cleanupOrphanedNGrams() { |
| 695 |
$ngramTable = $this->dao->getPrefixedTableName('abj404_ngram_cache'); |
| 696 |
$permalinkCacheTable = $this->dao->getPrefixedTableName('abj404_permalink_cache'); |
| 697 |
|
| 698 |
$this->logger->debugMessage("Checking for orphaned ngram entries..."); |
| 699 |
|
| 700 |
$stats = ['posts_deleted' => 0, 'categories_deleted' => 0, 'errors' => 0]; |
| 701 |
|
| 702 |
// ===== CLEANUP ORPHANED POSTS ===== |
| 703 |
// Find ngram entries for posts that don't exist in permalink cache |
| 704 |
// Using LEFT JOIN to find orphaned entries |
| 705 |
$orphanedResult = $this->dao->queryAndGetResults( |
| 706 |
"SELECT ng.id, ng.type |
| 707 |
FROM {$ngramTable} ng |
| 708 |
LEFT JOIN {$permalinkCacheTable} pc ON ng.id = pc.id AND ng.type = 'post' |
| 709 |
WHERE ng.type = 'post' AND pc.id IS NULL", |
| 710 |
['result_type' => OBJECT] |
| 711 |
); |
| 712 |
|
| 713 |
$orphanedError = isset($orphanedResult['last_error']) && is_string($orphanedResult['last_error']) ? $orphanedResult['last_error'] : ''; |
| 714 |
if ($orphanedError !== '') { |
| 715 |
if (!$this->dao->classifyAndHandleInfrastructureError($orphanedError)) { |
| 716 |
$this->logger->errorMessage("Failed to query for orphaned post ngram entries: " . $orphanedError); |
| 717 |
} |
| 718 |
return array_merge($stats, ['error' => $orphanedError]); |
| 719 |
} |
| 720 |
$orphanedPosts = isset($orphanedResult['rows']) && is_array($orphanedResult['rows']) ? $orphanedResult['rows'] : []; |
| 721 |
|
| 722 |
if (!empty($orphanedPosts)) { |
| 723 |
$this->logger->infoMessage("Found " . count($orphanedPosts) . " orphaned post ngram entries. Deleting..."); |
| 724 |
|
| 725 |
// Delete each orphaned post entry |
| 726 |
foreach ($orphanedPosts as $entry) { |
| 727 |
if (!is_object($entry)) { |
| 728 |
continue; |
| 729 |
} |
| 730 |
/** @var object{id: int, type: string} $entry */ |
| 731 |
$entryId = (int)$entry->id; |
| 732 |
$entryType = (string)$entry->type; |
| 733 |
$deleteResult = $this->dao->queryAndGetResults( |
| 734 |
"DELETE FROM {$ngramTable} WHERE id = %d AND type = %s", |
| 735 |
['query_params' => [$entryId, $entryType]] |
| 736 |
); |
| 737 |
|
| 738 |
$deleteError = isset($deleteResult['last_error']) && is_string($deleteResult['last_error']) ? $deleteResult['last_error'] : ''; |
| 739 |
if ($deleteError !== '') { |
| 740 |
if (!$this->dao->classifyAndHandleInfrastructureError($deleteError)) { |
| 741 |
$this->logger->errorMessage("Failed to delete orphaned post ngram entry ID {$entryId}: " . $deleteError); |
| 742 |
} |
| 743 |
$stats['errors']++; |
| 744 |
} else { |
| 745 |
$stats['posts_deleted']++; |
| 746 |
} |
| 747 |
} |
| 748 |
} else { |
| 749 |
$this->logger->debugMessage("No orphaned post ngram entries found."); |
| 750 |
} |
| 751 |
|
| 752 |
// ===== CLEANUP ORPHANED CATEGORIES ===== |
| 753 |
// Get all published categories |
| 754 |
$publishedCategories = $this->dao->getPublishedCategories(); |
| 755 |
$publishedCategoryIds = []; |
| 756 |
|
| 757 |
if (!empty($publishedCategories)) { |
| 758 |
foreach ($publishedCategories as $category) { |
| 759 |
/** @var object{term_id: int, url: string} $category */ |
| 760 |
$publishedCategoryIds[] = (int)$category->term_id; |
| 761 |
} |
| 762 |
} |
| 763 |
|
| 764 |
// Get all category ngram entries |
| 765 |
$catEntriesResult = $this->dao->queryAndGetResults( |
| 766 |
"SELECT DISTINCT id FROM {$ngramTable} WHERE type = 'category'", |
| 767 |
['result_type' => OBJECT] |
| 768 |
); |
| 769 |
$categoryNGramEntries = isset($catEntriesResult['rows']) && is_array($catEntriesResult['rows']) ? $catEntriesResult['rows'] : []; |
| 770 |
|
| 771 |
if (!empty($categoryNGramEntries)) { |
| 772 |
$orphanedCategories = []; |
| 773 |
|
| 774 |
// Find category ngram entries that don't have corresponding published categories |
| 775 |
foreach ($categoryNGramEntries as $entry) { |
| 776 |
if (!is_object($entry)) { |
| 777 |
continue; |
| 778 |
} |
| 779 |
/** @var object{id: int} $entry */ |
| 780 |
$entId = (int)$entry->id; |
| 781 |
if (!in_array($entId, $publishedCategoryIds)) { |
| 782 |
$orphanedCategories[] = $entId; |
| 783 |
} |
| 784 |
} |
| 785 |
|
| 786 |
if (!empty($orphanedCategories)) { |
| 787 |
$this->logger->infoMessage("Found " . count($orphanedCategories) . " orphaned category ngram entries. Deleting..."); |
| 788 |
|
| 789 |
// Delete orphaned category entries |
| 790 |
foreach ($orphanedCategories as $categoryId) { |
| 791 |
$catDeleteResult = $this->dao->queryAndGetResults( |
| 792 |
"DELETE FROM {$ngramTable} WHERE id = %d AND type = %s", |
| 793 |
['query_params' => [$categoryId, 'category']] |
| 794 |
); |
| 795 |
|
| 796 |
$catDeleteError = isset($catDeleteResult['last_error']) && is_string($catDeleteResult['last_error']) ? $catDeleteResult['last_error'] : ''; |
| 797 |
if ($catDeleteError !== '') { |
| 798 |
if (!$this->dao->classifyAndHandleInfrastructureError($catDeleteError)) { |
| 799 |
$this->logger->errorMessage("Failed to delete orphaned category ngram entry ID {$categoryId}: " . $catDeleteError); |
| 800 |
} |
| 801 |
$stats['errors']++; |
| 802 |
} else { |
| 803 |
$stats['categories_deleted']++; |
| 804 |
} |
| 805 |
} |
| 806 |
} else { |
| 807 |
$this->logger->debugMessage("No orphaned category ngram entries found."); |
| 808 |
} |
| 809 |
} |
| 810 |
|
| 811 |
$this->logger->infoMessage("Orphaned ngram cleanup complete: {$stats['posts_deleted']} posts deleted, {$stats['categories_deleted']} categories deleted, {$stats['errors']} errors."); |
| 812 |
|
| 813 |
return $stats; |
| 814 |
} |
| 815 |
|
| 816 |
/** |
| 817 |
* Build ngrams for all categories. |
| 818 |
* Should be called during initial setup or manual rebuild. |
| 819 |
* |
| 820 |
* @param int $batchSize Number of categories to process per batch (default: 50) |
| 821 |
* @return array<string, int> Statistics: ['processed' => int, 'success' => int, 'failed' => int] |
| 822 |
*/ |
| 823 |
function buildNGramsForCategories($batchSize = 50) { |
| 824 |
$this->logger->debugMessage("Building N-grams for categories..."); |
| 825 |
|
| 826 |
$categories = $this->dao->getPublishedCategories(); |
| 827 |
|
| 828 |
if (empty($categories)) { |
| 829 |
$this->logger->debugMessage("No published categories found."); |
| 830 |
return ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 831 |
} |
| 832 |
|
| 833 |
$stats = ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 834 |
|
| 835 |
foreach ($categories as $category) { |
| 836 |
try { |
| 837 |
/** @var object{term_id: int, url: string} $category */ |
| 838 |
$termId = (int)$category->term_id; |
| 839 |
$url = (string)$category->url; |
| 840 |
|
| 841 |
if (empty($url) || $url === 'in code') { |
| 842 |
$this->logger->debugMessage("Skipping category {$termId} - no valid URL"); |
| 843 |
continue; |
| 844 |
} |
| 845 |
|
| 846 |
// Normalize URL |
| 847 |
$urlNormalized = $this->f->strtolower(trim($url)); |
| 848 |
|
| 849 |
// Extract N-grams |
| 850 |
$ngrams = $this->ngramFilter->extractNGrams($urlNormalized); |
| 851 |
|
| 852 |
// Store with type='category' |
| 853 |
$success = $this->ngramFilter->storeNGrams($termId, $url, $urlNormalized, $ngrams, 'category'); |
| 854 |
|
| 855 |
$stats['processed']++; |
| 856 |
if ($success) { |
| 857 |
$stats['success']++; |
| 858 |
} else { |
| 859 |
$stats['failed']++; |
| 860 |
} |
| 861 |
} catch (Exception $e) { |
| 862 |
$this->logger->errorMessage("Failed to build ngram for category {$termId}: " . $e->getMessage()); |
| 863 |
$stats['processed']++; |
| 864 |
$stats['failed']++; |
| 865 |
} |
| 866 |
} |
| 867 |
|
| 868 |
$this->logger->infoMessage("Category N-grams built: {$stats['processed']} processed, {$stats['success']} success, {$stats['failed']} failed."); |
| 869 |
|
| 870 |
return $stats; |
| 871 |
} |
| 872 |
|
| 873 |
/** |
| 874 |
* Build ngrams for all tags. |
| 875 |
* Should be called during initial setup or manual rebuild. |
| 876 |
* |
| 877 |
* @param int $batchSize Number of tags to process per batch (default: 50) |
| 878 |
* @return array<string, int> Statistics: ['processed' => int, 'success' => int, 'failed' => int] |
| 879 |
*/ |
| 880 |
function buildNGramsForTags($batchSize = 50) { |
| 881 |
$this->logger->debugMessage("Building N-grams for tags..."); |
| 882 |
|
| 883 |
$tags = $this->dao->getPublishedTags(); |
| 884 |
|
| 885 |
if (empty($tags)) { |
| 886 |
$this->logger->debugMessage("No published tags found."); |
| 887 |
return ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 888 |
} |
| 889 |
|
| 890 |
$stats = ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 891 |
|
| 892 |
foreach ($tags as $tag) { |
| 893 |
try { |
| 894 |
/** @var object{term_id: int, url: string} $tag */ |
| 895 |
$termId = (int)$tag->term_id; |
| 896 |
$url = (string)$tag->url; |
| 897 |
|
| 898 |
if (empty($url) || $url === 'in code') { |
| 899 |
$this->logger->debugMessage("Skipping tag {$termId} - no valid URL"); |
| 900 |
continue; |
| 901 |
} |
| 902 |
|
| 903 |
// Normalize URL |
| 904 |
$urlNormalized = $this->f->strtolower(trim($url)); |
| 905 |
|
| 906 |
// Extract N-grams |
| 907 |
$ngrams = $this->ngramFilter->extractNGrams($urlNormalized); |
| 908 |
|
| 909 |
// Store with type='tag' |
| 910 |
$success = $this->ngramFilter->storeNGrams($termId, $url, $urlNormalized, $ngrams, 'tag'); |
| 911 |
|
| 912 |
$stats['processed']++; |
| 913 |
if ($success) { |
| 914 |
$stats['success']++; |
| 915 |
} else { |
| 916 |
$stats['failed']++; |
| 917 |
} |
| 918 |
} catch (Exception $e) { |
| 919 |
$this->logger->errorMessage("Failed to build ngram for tag {$termId}: " . $e->getMessage()); |
| 920 |
$stats['processed']++; |
| 921 |
$stats['failed']++; |
| 922 |
} |
| 923 |
} |
| 924 |
|
| 925 |
$this->logger->infoMessage("Tag N-grams built: {$stats['processed']} processed, {$stats['success']} success, {$stats['failed']} failed."); |
| 926 |
|
| 927 |
return $stats; |
| 928 |
} |
| 929 |
|
| 930 |
/** |
| 931 |
* Build ngrams for all content types (posts, pages, categories, tags). |
| 932 |
* This is the comprehensive rebuild that should be called from the Tools page. |
| 933 |
* |
| 934 |
* @param int $batchSize Number of items to process per batch |
| 935 |
* @return array<string, mixed> Combined statistics |
| 936 |
*/ |
| 937 |
function buildNGramsForAllContent($batchSize = 100) { |
| 938 |
$this->logger->infoMessage("Starting comprehensive N-gram cache build for all content types..."); |
| 939 |
|
| 940 |
// Rebuild posts/pages (existing functionality) |
| 941 |
$postsStats = $this->rebuildNGramCache($batchSize, true); |
| 942 |
|
| 943 |
// Build categories |
| 944 |
$categoriesStats = $this->buildNGramsForCategories($batchSize); |
| 945 |
|
| 946 |
// Build tags |
| 947 |
$tagsStats = $this->buildNGramsForTags($batchSize); |
| 948 |
|
| 949 |
$totalStats = [ |
| 950 |
'posts' => $postsStats, |
| 951 |
'categories' => $categoriesStats, |
| 952 |
'tags' => $tagsStats, |
| 953 |
'total_processed' => ($postsStats['processed'] ?? 0) + ($categoriesStats['processed'] ?? 0) + ($tagsStats['processed'] ?? 0), |
| 954 |
'total_success' => ($postsStats['success'] ?? 0) + ($categoriesStats['success'] ?? 0) + ($tagsStats['success'] ?? 0), |
| 955 |
'total_failed' => ($postsStats['failed'] ?? 0) + ($categoriesStats['failed'] ?? 0) + ($tagsStats['failed'] ?? 0) |
| 956 |
]; |
| 957 |
|
| 958 |
$this->logger->infoMessage("Comprehensive N-gram build complete: {$totalStats['total_processed']} total processed, {$totalStats['total_success']} success, {$totalStats['total_failed']} failed."); |
| 959 |
|
| 960 |
return $totalStats; |
| 961 |
} |
| 962 |
|
| 963 |
/** |
| 964 |
* Check if the plugin is network-activated in a multisite environment. |
| 965 |
* |
| 966 |
* @return bool True if network-activated, false otherwise |
| 967 |
*/ |
| 968 |
private function isNetworkActivated() { |
| 969 |
if (!is_multisite()) { |
| 970 |
return false; |
| 971 |
} |
| 972 |
|
| 973 |
if (!function_exists('is_plugin_active_for_network')) { |
| 974 |
require_once ABSPATH . '/wp-admin/includes/plugin.php'; |
| 975 |
} |
| 976 |
|
| 977 |
return is_plugin_active_for_network(plugin_basename(ABJ404_FILE)); |
| 978 |
} |
| 979 |
|
| 980 |
/** |
| 981 |
* Get an option value, using network-wide storage in multisite when network-activated. |
| 982 |
* |
| 983 |
* MULTISITE BEHAVIOR: |
| 984 |
* - Network-activated: Uses get_site_option() for network-wide state |
| 985 |
* - Single-site or per-site activation: Uses get_option() for site-specific state |
| 986 |
* |
| 987 |
* This ensures that N-gram rebuild state is shared across all sites in network-activated |
| 988 |
* scenarios, preventing race conditions and duplicate work. |
| 989 |
* |
| 990 |
* @param string $option_name The option name |
| 991 |
* @param mixed $default Default value if option doesn't exist |
| 992 |
* @return mixed The option value |
| 993 |
*/ |
| 994 |
private function getNetworkAwareOption($option_name, $default = false) { |
| 995 |
if ($this->isNetworkActivated()) { |
| 996 |
return get_site_option($option_name, $default); |
| 997 |
} |
| 998 |
return get_option($option_name, $default); |
| 999 |
} |
| 1000 |
|
| 1001 |
/** |
| 1002 |
* Update an option value, using network-wide storage in multisite when network-activated. |
| 1003 |
* |
| 1004 |
* MULTISITE BEHAVIOR: |
| 1005 |
* - Network-activated: Uses update_site_option() for network-wide state |
| 1006 |
* - Single-site or per-site activation: Uses update_option() for site-specific state |
| 1007 |
* |
| 1008 |
* This ensures that N-gram rebuild state is shared across all sites in network-activated |
| 1009 |
* scenarios, preventing race conditions and duplicate work. |
| 1010 |
* |
| 1011 |
* @param string $option_name The option name |
| 1012 |
* @param mixed $value The value to store |
| 1013 |
* @return bool True if updated successfully |
| 1014 |
*/ |
| 1015 |
private function updateNetworkAwareOption($option_name, $value) { |
| 1016 |
if ($this->isNetworkActivated()) { |
| 1017 |
return update_site_option($option_name, $value); |
| 1018 |
} |
| 1019 |
return update_option($option_name, $value); |
| 1020 |
} |
| 1021 |
|
| 1022 |
/** |
| 1023 |
* Count total pages for N-gram rebuild across all sites if network-activated. |
| 1024 |
* |
| 1025 |
* MULTISITE BEHAVIOR: |
| 1026 |
* - Network-activated: Counts permalink cache entries across ALL sites in the network |
| 1027 |
* - Single-site: Counts only current site's permalink cache entries |
| 1028 |
* |
| 1029 |
* This allows the rebuild process to accurately track progress when processing |
| 1030 |
* pages from multiple sites. |
| 1031 |
* |
| 1032 |
* @return int Total number of pages to process |
| 1033 |
*/ |
| 1034 |
private function countTotalPagesForNGramRebuild() { |
| 1035 |
if (!$this->isNetworkActivated()) { |
| 1036 |
// Single site: count only current site's pages |
| 1037 |
$permalinkCacheTable = $this->dao->getPrefixedTableName('abj404_permalink_cache'); |
| 1038 |
return $this->dao->queryScalarInt("SELECT COUNT(*) AS c FROM {$permalinkCacheTable}"); |
| 1039 |
} |
| 1040 |
|
| 1041 |
// Multisite network-activated: count pages across all sites |
| 1042 |
$sites = get_sites(array('fields' => 'ids', 'number' => 0)); |
| 1043 |
$totalPages = 0; |
| 1044 |
|
| 1045 |
foreach ($sites as $blog_id) { |
| 1046 |
switch_to_blog($blog_id); |
| 1047 |
$permalinkCacheTable = $this->dao->getPrefixedTableName('abj404_permalink_cache'); |
| 1048 |
$totalPages += $this->dao->queryScalarInt("SELECT COUNT(*) AS c FROM {$permalinkCacheTable}"); |
| 1049 |
restore_current_blog(); |
| 1050 |
} |
| 1051 |
|
| 1052 |
return $totalPages; |
| 1053 |
} |
| 1054 |
} |
| 1055 |
|
