| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Synchronous (blocking) full rebuild of the N-gram cache from |
| 9 |
* permalink_cache. Used by manual rebuilds and by the all-content |
| 10 |
* orchestrator's first phase. |
| 11 |
* |
| 12 |
* WARNING: this path is synchronous and can take minutes on large |
| 13 |
* sites. The cron-driven async rebuild |
| 14 |
* (NGramCacheRebuildScheduler::runAsyncBatch) is the standard path; |
| 15 |
* this collaborator is for manual tools and tests. |
| 16 |
* |
| 17 |
* Lock ownership: the orchestrator (DatabaseUpgradeNGram) acquires |
| 18 |
* the shared 'ngram_rebuild' SyncUtils lock before calling. This |
| 19 |
* collaborator assumes the lock is held and is therefore safe to |
| 20 |
* TRUNCATE the cache table. |
| 21 |
*/ |
| 22 |
class ABJ_404_Solution_NGramCacheSyncRebuilder { |
| 23 |
|
| 24 |
/** @var ABJ_404_Solution_DatabaseCore */ |
| 25 |
private $dbCore; |
| 26 |
|
| 27 |
/** @var mixed */ |
| 28 |
private $rebuilder; |
| 29 |
|
| 30 |
/** @var mixed */ |
| 31 |
private $coveragePolicy; |
| 32 |
|
| 33 |
/** @var ABJ_404_Solution_Logging */ |
| 34 |
private $logger; |
| 35 |
|
| 36 |
/** |
| 37 |
* @param ABJ_404_Solution_DatabaseCore $dbCore |
| 38 |
* @param mixed $rebuilder Object exposing rebuildCache(). |
| 39 |
* @param mixed $coveragePolicy Object exposing invalidateCoverageCaches(). |
| 40 |
* @param ABJ_404_Solution_Logging $logger |
| 41 |
*/ |
| 42 |
public function __construct($dbCore, $rebuilder, $coveragePolicy, $logger) { |
| 43 |
$this->dbCore = $dbCore; |
| 44 |
$this->rebuilder = $rebuilder; |
| 45 |
$this->coveragePolicy = $coveragePolicy; |
| 46 |
$this->logger = $logger; |
| 47 |
} |
| 48 |
|
| 49 |
/** |
| 50 |
* TRUNCATE the cache table and rebuild it from permalink_cache in |
| 51 |
* batches of $batchSize. Skips when the cache is already populated |
| 52 |
* unless $forceRebuild is true. |
| 53 |
* |
| 54 |
* @param int $batchSize |
| 55 |
* @param bool $forceRebuild |
| 56 |
* @return array<string, mixed> ['total_pages' => int, 'processed' => int, 'success' => int, 'failed' => int] |
| 57 |
*/ |
| 58 |
public function rebuild(int $batchSize = 100, bool $forceRebuild = false): array { |
| 59 |
$ngramTable = $this->dbCore->tableNameResolver()->getPrefixedTableName('abj404_ngram_cache'); |
| 60 |
$permalinkCacheTable = $this->dbCore->tableNameResolver()->getPrefixedTableName('abj404_permalink_cache'); |
| 61 |
|
| 62 |
$alreadyPopulated = $this->checkAlreadyPopulated($ngramTable, $forceRebuild); |
| 63 |
if ($alreadyPopulated !== null) { |
| 64 |
return $alreadyPopulated; |
| 65 |
} |
| 66 |
|
| 67 |
$this->logger->debugMessage("Starting N-gram cache rebuild..."); |
| 68 |
|
| 69 |
$truncateError = $this->truncateAndInvalidate($ngramTable); |
| 70 |
if ($truncateError !== null) { |
| 71 |
return $truncateError; |
| 72 |
} |
| 73 |
|
| 74 |
$totalPagesOrError = $this->countTotalPages($permalinkCacheTable); |
| 75 |
if (is_array($totalPagesOrError)) { |
| 76 |
return $totalPagesOrError; |
| 77 |
} |
| 78 |
$totalPages = $totalPagesOrError; |
| 79 |
|
| 80 |
if ($totalPages == 0) { |
| 81 |
$this->logger->debugMessage("No pages in permalink cache. N-gram cache rebuild skipped (will rebuild when pages are added)."); |
| 82 |
return ['total_pages' => 0, 'processed' => 0, 'success' => 0, 'failed' => 0]; |
| 83 |
} |
| 84 |
|
| 85 |
$this->logger->infoMessage("Rebuilding N-gram cache for {$totalPages} pages in batches of {$batchSize}..."); |
| 86 |
|
| 87 |
$totalStats = $this->runBatchedRebuild($batchSize, $totalPages); |
| 88 |
$totalStats['total_pages'] = $totalPages; |
| 89 |
|
| 90 |
$successRate = $totalStats['processed'] > 0 ? |
| 91 |
round(($totalStats['success'] / $totalStats['processed']) * 100, 1) : 0; |
| 92 |
|
| 93 |
$this->logger->infoMessage(sprintf( |
| 94 |
"N-gram cache rebuild complete: %d pages processed, %d success, %d failed (%.1f%% success rate)", |
| 95 |
$totalStats['processed'], |
| 96 |
$totalStats['success'], |
| 97 |
$totalStats['failed'], |
| 98 |
$successRate |
| 99 |
)); |
| 100 |
|
| 101 |
return $totalStats; |
| 102 |
} |
| 103 |
|
| 104 |
/** |
| 105 |
* @return array<string, mixed>|null populated-skip payload, or null when rebuild should proceed |
| 106 |
*/ |
| 107 |
private function checkAlreadyPopulated(string $ngramTable, bool $forceRebuild): ?array { |
| 108 |
if ($forceRebuild) { |
| 109 |
return null; |
| 110 |
} |
| 111 |
$existingCount = $this->dbCore->queryScalarInt("SELECT COUNT(*) AS c FROM {$ngramTable}"); |
| 112 |
if ($existingCount <= 0) { |
| 113 |
return null; |
| 114 |
} |
| 115 |
$this->logger->debugMessage("N-gram cache already contains {$existingCount} entries. Skipping rebuild (use forceRebuild=true to override)."); |
| 116 |
return [ |
| 117 |
'total_pages' => $existingCount, |
| 118 |
'processed' => 0, |
| 119 |
'success' => $existingCount, |
| 120 |
'failed' => 0, |
| 121 |
'skipped' => true, |
| 122 |
]; |
| 123 |
} |
| 124 |
|
| 125 |
/** |
| 126 |
* TRUNCATE the cache table and invalidate coverage caches. |
| 127 |
* |
| 128 |
* @return array<string, mixed>|null error-result payload when truncate fails, or null on success |
| 129 |
*/ |
| 130 |
private function truncateAndInvalidate(string $ngramTable): ?array { |
| 131 |
// skip_repair: TRUNCATE itself is the recovery path during |
| 132 |
// rebuild; we must not recurse into the missing-table |
| 133 |
// repairer here. |
| 134 |
$truncateResult = $this->dbCore->queryAndGetResults( |
| 135 |
"TRUNCATE TABLE {$ngramTable}", |
| 136 |
['skip_repair' => true] |
| 137 |
); |
| 138 |
$truncateError = isset($truncateResult['last_error']) && is_string($truncateResult['last_error']) ? $truncateResult['last_error'] : ''; |
| 139 |
if ($truncateError !== '') { |
| 140 |
if (!$this->dbCore->errorClassifier()->classifyAndHandleInfrastructureError($truncateError)) { |
| 141 |
$this->logger->errorMessage("Failed to truncate N-gram cache table: " . $truncateError); |
| 142 |
} |
| 143 |
return ['total_pages' => 0, 'processed' => 0, 'success' => 0, 'failed' => 1, 'error' => $truncateError]; |
| 144 |
} |
| 145 |
|
| 146 |
// Invalidate coverage ratio caches immediately after truncate |
| 147 |
// so SpellChecker does not see stale transient data while the |
| 148 |
// cache is empty. |
| 149 |
$this->invalidateCoverageCaches(); |
| 150 |
return null; |
| 151 |
} |
| 152 |
|
| 153 |
/** |
| 154 |
* @return int|array<string, mixed> int on success, error-result payload on failure |
| 155 |
*/ |
| 156 |
private function countTotalPages(string $permalinkCacheTable) { |
| 157 |
$totalPagesResult = $this->dbCore->queryAndGetResults("SELECT COUNT(*) AS c FROM {$permalinkCacheTable}"); |
| 158 |
$totalPagesRows = isset($totalPagesResult['rows']) && is_array($totalPagesResult['rows']) ? $totalPagesResult['rows'] : []; |
| 159 |
$totalPagesRow = $totalPagesRows[0] ?? null; |
| 160 |
|
| 161 |
if (!is_array($totalPagesRow) || !isset($totalPagesRow['c'])) { |
| 162 |
$countError = isset($totalPagesResult['last_error']) && is_string($totalPagesResult['last_error']) ? $totalPagesResult['last_error'] : ''; |
| 163 |
if (!$this->dbCore->errorClassifier()->classifyAndHandleInfrastructureError($countError)) { |
| 164 |
$this->logger->errorMessage("Failed to query permalink cache table: " . $countError); |
| 165 |
} |
| 166 |
return ['total_pages' => 0, 'processed' => 0, 'success' => 0, 'failed' => 1, 'error' => $countError]; |
| 167 |
} |
| 168 |
return is_scalar($totalPagesRow['c']) ? (int)$totalPagesRow['c'] : 0; |
| 169 |
} |
| 170 |
|
| 171 |
/** |
| 172 |
* @return array{processed:int, success:int, failed:int} |
| 173 |
*/ |
| 174 |
private function runBatchedRebuild(int $batchSize, int $totalPages): array { |
| 175 |
$offset = 0; |
| 176 |
$totalStats = ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 177 |
|
| 178 |
while ($offset < $totalPages) { |
| 179 |
try { |
| 180 |
$stats = $this->runRebuildBatch($batchSize, $offset); |
| 181 |
|
| 182 |
$totalStats['processed'] += $stats['processed']; |
| 183 |
$totalStats['success'] += $stats['success']; |
| 184 |
$totalStats['failed'] += $stats['failed']; |
| 185 |
|
| 186 |
$offset += $batchSize; |
| 187 |
|
| 188 |
if ($stats['processed'] < $batchSize) { |
| 189 |
break; |
| 190 |
} |
| 191 |
} catch (Exception $e) { |
| 192 |
$this->logger->errorMessage("Error during N-gram cache rebuild at offset {$offset}: " . $e->getMessage()); |
| 193 |
$totalStats['failed'] += $batchSize; |
| 194 |
$offset += $batchSize; |
| 195 |
} |
| 196 |
} |
| 197 |
|
| 198 |
return $totalStats; |
| 199 |
} |
| 200 |
|
| 201 |
/** @return void */ |
| 202 |
private function invalidateCoverageCaches(): void { |
| 203 |
$coveragePolicy = $this->coveragePolicy; |
| 204 |
if (!is_object($coveragePolicy) || !method_exists($coveragePolicy, 'invalidateCoverageCaches')) { |
| 205 |
throw new RuntimeException('NGramCacheSyncRebuilder requires a coverage policy with invalidateCoverageCaches().'); |
| 206 |
} |
| 207 |
$coveragePolicy->invalidateCoverageCaches(); |
| 208 |
} |
| 209 |
|
| 210 |
/** |
| 211 |
* @param int $batchSize |
| 212 |
* @param int $offset |
| 213 |
* @return array{processed: int, success: int, failed: int} |
| 214 |
*/ |
| 215 |
private function runRebuildBatch(int $batchSize, int $offset): array { |
| 216 |
$rebuilder = $this->rebuilder; |
| 217 |
if (!is_object($rebuilder) || !method_exists($rebuilder, 'rebuildCache')) { |
| 218 |
throw new RuntimeException('NGramCacheSyncRebuilder requires a rebuilder with rebuildCache().'); |
| 219 |
} |
| 220 |
$stats = $rebuilder->rebuildCache($batchSize, $offset); |
| 221 |
return [ |
| 222 |
'processed' => is_array($stats) && isset($stats['processed']) && is_numeric($stats['processed']) ? (int)$stats['processed'] : 0, |
| 223 |
'success' => is_array($stats) && isset($stats['success']) && is_numeric($stats['success']) ? (int)$stats['success'] : 0, |
| 224 |
'failed' => is_array($stats) && isset($stats['failed']) && is_numeric($stats['failed']) ? (int)$stats['failed'] : 0, |
| 225 |
]; |
| 226 |
} |
| 227 |
} |
| 228 |
|
