| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Batch + incremental N-gram cache regeneration. |
| 9 |
* |
| 10 |
* Two entry points: rebuildCache() walks the permalink cache in batches |
| 11 |
* for full reseeding, and updateNGramsForPages() handles per-post events |
| 12 |
* (slug change, post insert) from SpellPostListeners. Both compose the |
| 13 |
* extractor and cache repository; coverage invalidation happens once per |
| 14 |
* batch (rebuild) or per-write (update) via the repository's own write |
| 15 |
* hooks. |
| 16 |
*/ |
| 17 |
class ABJ_404_Solution_NGramRebuilder { |
| 18 |
|
| 19 |
/** @var ABJ_404_Solution_DatabaseCore */ |
| 20 |
private $dbCore; |
| 21 |
|
| 22 |
/** @var ABJ_404_Solution_Logging */ |
| 23 |
private $logger; |
| 24 |
|
| 25 |
/** @var ABJ_404_Solution_Functions */ |
| 26 |
private $f; |
| 27 |
|
| 28 |
/** @var ABJ_404_Solution_NGramExtractor */ |
| 29 |
private $extractor; |
| 30 |
|
| 31 |
/** @var ABJ_404_Solution_NGramCacheRepository */ |
| 32 |
private $repo; |
| 33 |
|
| 34 |
/** @var ABJ_404_Solution_NGramCoveragePolicy */ |
| 35 |
private $coveragePolicy; |
| 36 |
|
| 37 |
/** |
| 38 |
* @param ABJ_404_Solution_NGramRebuilderDependencies|null $deps |
| 39 |
*/ |
| 40 |
public function __construct(?ABJ_404_Solution_NGramRebuilderDependencies $deps = null) { |
| 41 |
$deps = $deps ?? new ABJ_404_Solution_NGramRebuilderDependencies(); |
| 42 |
$this->dbCore = $deps->dbCore !== null ? $deps->dbCore : abj_service('db_core'); |
| 43 |
$loggingResolved = $deps->logging !== null ? $deps->logging : abj_service('logging'); |
| 44 |
$functionsResolved = $deps->functions !== null ? $deps->functions : abj_service('functions'); |
| 45 |
$extractorResolved = $deps->extractor !== null ? $deps->extractor : abj_service('ngram_extractor'); |
| 46 |
$repoResolved = $deps->repo !== null ? $deps->repo : abj_service('ngram_cache_repository'); |
| 47 |
$coverageResolved = $deps->coveragePolicy !== null ? $deps->coveragePolicy : abj_service('ngram_coverage_policy'); |
| 48 |
if (!$loggingResolved instanceof ABJ_404_Solution_Logging |
| 49 |
|| !$functionsResolved instanceof ABJ_404_Solution_Functions |
| 50 |
|| !$extractorResolved instanceof ABJ_404_Solution_NGramExtractor |
| 51 |
|| !$repoResolved instanceof ABJ_404_Solution_NGramCacheRepository |
| 52 |
|| !$coverageResolved instanceof ABJ_404_Solution_NGramCoveragePolicy) { |
| 53 |
throw new RuntimeException('NGramRebuilder requires fully-wired collaborators.'); |
| 54 |
} |
| 55 |
$this->logger = $loggingResolved; |
| 56 |
$this->f = $functionsResolved; |
| 57 |
$this->extractor = $extractorResolved; |
| 58 |
$this->repo = $repoResolved; |
| 59 |
$this->coveragePolicy = $coverageResolved; |
| 60 |
} |
| 61 |
|
| 62 |
/** |
| 63 |
* Update N-grams for specific pages (incremental update). |
| 64 |
* |
| 65 |
* @param array<int, int> $pageIds |
| 66 |
* @return array{processed: int, success: int, failed: int} |
| 67 |
*/ |
| 68 |
public function updateNGramsForPages($pageIds) { |
| 69 |
if (empty($pageIds) || !is_array($pageIds)) { |
| 70 |
return ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 71 |
} |
| 72 |
|
| 73 |
$permalinkCacheTable = $this->dbCore->tableNameResolver()->getPrefixedTableName('abj404_permalink_cache'); |
| 74 |
|
| 75 |
$placeholders = implode(',', array_fill(0, count($pageIds), '%d')); |
| 76 |
$pageResult = $this->dbCore->queryAndGetResults( |
| 77 |
"SELECT id, url FROM {$permalinkCacheTable} WHERE id IN ({$placeholders})", |
| 78 |
['query_params' => array_values($pageIds)] |
| 79 |
); |
| 80 |
$pages = isset($pageResult['rows']) && is_array($pageResult['rows']) ? $pageResult['rows'] : []; |
| 81 |
|
| 82 |
if (empty($pages)) { |
| 83 |
return ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 84 |
} |
| 85 |
|
| 86 |
$stats = ['processed' => 0, 'success' => 0, 'failed' => 0]; |
| 87 |
|
| 88 |
foreach ($pages as $page) { |
| 89 |
if (is_object($page)) { |
| 90 |
$page = (array) $page; |
| 91 |
} |
| 92 |
if (!is_array($page) || !isset($page['id'], $page['url'])) { |
| 93 |
continue; |
| 94 |
} |
| 95 |
$pageId = is_scalar($page['id']) ? (int)$page['id'] : 0; |
| 96 |
$url = is_scalar($page['url']) ? (string)$page['url'] : ''; |
| 97 |
|
| 98 |
$urlNormalized = $this->f->strtolower(trim($url)); |
| 99 |
$ngrams = $this->extractor->extractNGrams($urlNormalized); |
| 100 |
$success = $this->repo->storeNGrams($pageId, $url, $urlNormalized, $ngrams); |
| 101 |
|
| 102 |
$stats['processed']++; |
| 103 |
if ($success) { |
| 104 |
$stats['success']++; |
| 105 |
} else { |
| 106 |
$stats['failed']++; |
| 107 |
} |
| 108 |
} |
| 109 |
|
| 110 |
$this->logger->debugMessage(sprintf( |
| 111 |
"Incremental N-gram update: %d pages, %d success, %d failed", |
| 112 |
$stats['processed'], |
| 113 |
$stats['success'], |
| 114 |
$stats['failed'] |
| 115 |
)); |
| 116 |
|
| 117 |
return $stats; |
| 118 |
} |
| 119 |
|
| 120 |
/** |
| 121 |
* Rebuild the N-gram cache for all pages (background batch). |
| 122 |
* |
| 123 |
* @param int $batchSize |
| 124 |
* @param int $offset |
| 125 |
* @return array{processed: int, success: int, failed: int} |
| 126 |
*/ |
| 127 |
public function rebuildCache($batchSize = 100, $offset = 0) { |
| 128 |
$permalinkCacheTable = $this->dbCore->tableNameResolver()->getPrefixedTableName('abj404_permalink_cache'); |
| 129 |
|
| 130 |
$batchResult = $this->dbCore->queryAndGetResults( |
| 131 |
"SELECT id, url FROM {$permalinkCacheTable} LIMIT %d OFFSET %d", |
| 132 |
['query_params' => [$batchSize, $offset]] |
| 133 |
); |
| 134 |
$pages = isset($batchResult['rows']) && is_array($batchResult['rows']) ? $batchResult['rows'] : []; |
| 135 |
|
| 136 |
$stats = [ |
| 137 |
'processed' => 0, |
| 138 |
'success' => 0, |
| 139 |
'failed' => 0 |
| 140 |
]; |
| 141 |
|
| 142 |
foreach ($pages as $page) { |
| 143 |
if (is_object($page)) { |
| 144 |
$page = (array) $page; |
| 145 |
} |
| 146 |
if (!is_array($page) || !isset($page['id'], $page['url'])) { |
| 147 |
continue; |
| 148 |
} |
| 149 |
$pageId = is_scalar($page['id']) ? (int)$page['id'] : 0; |
| 150 |
$url = is_scalar($page['url']) ? (string)$page['url'] : ''; |
| 151 |
|
| 152 |
$urlNormalized = $this->f->strtolower(trim($url)); |
| 153 |
$ngrams = $this->extractor->extractNGrams($urlNormalized); |
| 154 |
|
| 155 |
// Skip per-item invalidation for bulk efficiency; invalidate at batch end. |
| 156 |
$success = $this->repo->storeNGrams($pageId, $url, $urlNormalized, $ngrams, 'post', true); |
| 157 |
|
| 158 |
$stats['processed']++; |
| 159 |
if ($success) { |
| 160 |
$stats['success']++; |
| 161 |
} else { |
| 162 |
$stats['failed']++; |
| 163 |
} |
| 164 |
} |
| 165 |
|
| 166 |
if ($stats['success'] > 0) { |
| 167 |
$this->coveragePolicy->invalidateCoverageCaches(); |
| 168 |
$this->repo->resetMemo(); |
| 169 |
} |
| 170 |
|
| 171 |
// Log every 1000 pages to reduce verbosity |
| 172 |
if ($offset % 1000 == 0) { |
| 173 |
$this->logger->debugMessage(sprintf( |
| 174 |
"N-gram cache rebuild batch (offset %d): %d processed, %d success, %d failed", |
| 175 |
$offset, |
| 176 |
$stats['processed'], |
| 177 |
$stats['success'], |
| 178 |
$stats['failed'] |
| 179 |
)); |
| 180 |
} |
| 181 |
|
| 182 |
return $stats; |
| 183 |
} |
| 184 |
} |
| 185 |
|
