| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
// allow-no-test-found: exercised by RedirectsCanonicalUrlBackfillCompleteTest |
| 8 |
|
| 9 |
require_once __DIR__ . '/LogsHitsCanonicalUrlJoinHelper.php'; |
| 10 |
require_once __DIR__ . '/../database/DatabaseCollationHelper.php'; |
| 11 |
|
| 12 |
/** |
| 13 |
* Two-phase SQL execution engine for the wp_abj404_logs_hits rollup. |
| 14 |
* |
| 15 |
* Extracted from LogsHitsRollupService (i468) so the service stays focused on |
| 16 |
* rebuild coordination (gate / lock / scheduling), staleness signaling, and |
| 17 |
* freshness/id reads, while this class owns the one cohesive responsibility of |
| 18 |
* materializing the rollup and swapping it in. |
| 19 |
* |
| 20 |
* Given a snapshot id range it: |
| 21 |
* 1. (re)creates and truncates the {wp_abj404_logs_hits}_temp staging table, |
| 22 |
* 2. picks the direct INSERT...SELECT path for small ranges or the chunked |
| 23 |
* pre-aggregation path (Phase 1 per-id-range -> Phase 2 JOIN) for large |
| 24 |
* ones, |
| 25 |
* 3. stamps the temp table COMMENT with elapsed|maxLogId and renames it over |
| 26 |
* the live table in a single transaction, |
| 27 |
* 4. records per-phase success/failure into RebuildHealthState and cleans up |
| 28 |
* the pre-aggregation temp table. |
| 29 |
* |
| 30 |
* The min/max log-id snapshot is supplied by the caller (not read here) so the |
| 31 |
* tracking test subclass can control the id range without a live logsv2 table, |
| 32 |
* and so the watermark stored in the table COMMENT is the pre-insert value. |
| 33 |
* |
| 34 |
* Pure data-access / execution layer: no admin-notice policy, no scheduling |
| 35 |
* decision, no runtime-flag bookkeeping. Those stay on the service. |
| 36 |
*/ |
| 37 |
class ABJ_404_Solution_LogsHitsTableRebuilder { |
| 38 |
|
| 39 |
/** @var int Number of logsv2 IDs to process per chunk during pre-aggregation. */ |
| 40 |
const HITS_TABLE_PREAGG_CHUNK_SIZE = 100000; |
| 41 |
/** @var int Direct-path threshold for hits-table rebuild. */ |
| 42 |
const HITS_TABLE_DIRECT_PATH_THRESHOLD = 5000; |
| 43 |
|
| 44 |
/** @var ABJ_404_Solution_DatabaseCore */ |
| 45 |
private $dbCore; |
| 46 |
|
| 47 |
/** @var ABJ_404_Solution_Logging */ |
| 48 |
private $logger; |
| 49 |
|
| 50 |
/** @var ABJ_404_Solution_RebuildHealthState|null */ |
| 51 |
private $rebuildHealth; |
| 52 |
|
| 53 |
/** @var ABJ_404_Solution_LogsHitsCanonicalUrlJoinHelper */ |
| 54 |
private $joinHelper; |
| 55 |
|
| 56 |
/** @var callable():bool|null */ |
| 57 |
private $leaseRenewer; |
| 58 |
|
| 59 |
/** |
| 60 |
* @param ABJ_404_Solution_DatabaseCore $dbCore |
| 61 |
* @param ABJ_404_Solution_Logging $logger |
| 62 |
* @param ABJ_404_Solution_RebuildHealthState|null $rebuildHealth |
| 63 |
* @param ABJ_404_Solution_LogsHitsCanonicalUrlJoinHelper $joinHelper |
| 64 |
* @param callable():bool|null $leaseRenewer |
| 65 |
*/ |
| 66 |
public function __construct( |
| 67 |
ABJ_404_Solution_DatabaseCore $dbCore, |
| 68 |
$logger, |
| 69 |
$rebuildHealth, |
| 70 |
ABJ_404_Solution_LogsHitsCanonicalUrlJoinHelper $joinHelper, |
| 71 |
$leaseRenewer = null |
| 72 |
) { |
| 73 |
$this->dbCore = $dbCore; |
| 74 |
$this->logger = $logger; |
| 75 |
$this->rebuildHealth = $rebuildHealth instanceof ABJ_404_Solution_RebuildHealthState |
| 76 |
? $rebuildHealth |
| 77 |
: null; |
| 78 |
$this->joinHelper = $joinHelper; |
| 79 |
if ($leaseRenewer !== null && !is_callable($leaseRenewer)) { |
| 80 |
throw new InvalidArgumentException( |
| 81 |
'LogsHitsTableRebuilder lease renewer must be callable or null.' |
| 82 |
); |
| 83 |
} |
| 84 |
$this->leaseRenewer = $leaseRenewer; |
| 85 |
} |
| 86 |
|
| 87 |
/** |
| 88 |
* Materialize the rollup over [$minLogId, $maxLogId] into a fresh temp |
| 89 |
* table and rename it over the live wp_abj404_logs_hits table. Records |
| 90 |
* per-phase outcomes into RebuildHealthState and cleans up the |
| 91 |
* pre-aggregation temp table. |
| 92 |
* |
| 93 |
* Expected SQL failures (timeout / last_error) are caught and reported via |
| 94 |
* the return value; unexpected Throwables are recorded and also reported |
| 95 |
* via the return value (never re-thrown) so the caller's lock release runs |
| 96 |
* unconditionally. |
| 97 |
* |
| 98 |
* @param int $minLogId Pre-insert MIN(logsv2.id) snapshot. |
| 99 |
* @param int $maxLogId Pre-insert MAX(logsv2.id) snapshot (stored as the watermark). |
| 100 |
* @return array{refreshed: bool, elapsed_time: float, error: string} |
| 101 |
*/ |
| 102 |
public function rebuildAndSwap(int $minLogId, int $maxLogId): array { |
| 103 |
$preAggTable = $this->dbCore->doTableNameReplacements("{wp_abj404_logs_hits}_preagg"); |
| 104 |
try { |
| 105 |
$finalDestTable = $this->dbCore->doTableNameReplacements("{wp_abj404_logs_hits}"); |
| 106 |
$tempDestTable = $this->dbCore->doTableNameReplacements("{wp_abj404_logs_hits}_temp"); |
| 107 |
$this->renewLeaseOrThrow(); |
| 108 |
$this->dbCore->queryAndGetResults("drop table if exists " . $tempDestTable); |
| 109 |
$this->renewLeaseOrThrow(); |
| 110 |
$resolvedCollation = $this->joinHelper->resolveHitsJoinCollation(); |
| 111 |
$createTempTableQuery = ABJ_404_Solution_FileSystemService::readFileContents(__DIR__ . "/../sql/createLogsHitsTempTable.sql"); |
| 112 |
$createTempTableQuery = $this->dbCore->doTableNameReplacements($createTempTableQuery); |
| 113 |
$createTempTableQuery = $this->applyJoinCharsetCollation(array( |
| 114 |
'ddl' => $createTempTableQuery, |
| 115 |
'rawCollation' => $resolvedCollation, |
| 116 |
)); |
| 117 |
$this->dbCore->queryAndGetResults($createTempTableQuery); |
| 118 |
$this->renewLeaseOrThrow(); |
| 119 |
// @cache-write-audit: opt-out - truncates an unpublished temp table before rebuilding it. |
| 120 |
$this->dbCore->queryAndGetResults("truncate table " . $tempDestTable); |
| 121 |
$idRange = $maxLogId - $minLogId; |
| 122 |
$chunkSize = $this->getHitsRebuildChunkSize($idRange); |
| 123 |
$this->renewLeaseOrThrow(); |
| 124 |
if ($idRange <= self::HITS_TABLE_DIRECT_PATH_THRESHOLD) { $results = $this->hitsTableInsertDirect($tempDestTable); } else { $results = $this->hitsTableInsertChunked($tempDestTable, $preAggTable, $minLogId, $maxLogId, $chunkSize); } |
| 125 |
$this->renewLeaseOrThrow(); |
| 126 |
if ($results === false || !empty($results['timed_out']) || !empty($results['last_error'])) { |
| 127 |
$rawLastError = is_array($results) && isset($results['last_error']) && is_string($results['last_error']) ? $results['last_error'] : ''; |
| 128 |
$errorMessage = $results === false ? 'Hits rebuild phase 1 chunk failed.' : ($rawLastError !== '' ? $rawLastError : 'Hits rebuild timed out.'); |
| 129 |
$this->recordHitsRebuildFailure($errorMessage); |
| 130 |
if ($idRange > self::HITS_TABLE_DIRECT_PATH_THRESHOLD && $results !== false && (!empty($results['timed_out']) || !empty($results['last_error']))) { |
| 131 |
$this->recordHitsChunkFailure(); |
| 132 |
} |
| 133 |
$this->dropScratchTableIfLeaseOwned($tempDestTable); |
| 134 |
$this->logger->debugMessage(__FUNCTION__ . " INSERT timed out or errored; aborting rebuild."); |
| 135 |
return array('refreshed' => false, 'elapsed_time' => 0.0, 'error' => $errorMessage); |
| 136 |
} |
| 137 |
$rawElapsed = $results['elapsed_time'] ?? 0; |
| 138 |
$elapsedTime = is_numeric($rawElapsed) ? (float)$rawElapsed : 0.0; |
| 139 |
$comment = $elapsedTime . '|' . $maxLogId; |
| 140 |
$this->renewLeaseOrThrow(); |
| 141 |
// @utf8-audit: opt-out - rebuild table comment is synthesized from numeric timing and ID values. |
| 142 |
$comment = substr(esc_sql($comment), 0, 2048); |
| 143 |
$this->dbCore->queryAndGetResults(sprintf("ALTER TABLE %s COMMENT '%s'", $tempDestTable, $comment)); |
| 144 |
$this->renewLeaseOrThrow(); |
| 145 |
$statements = array("drop table if exists " . $finalDestTable, "rename table " . $tempDestTable . ' to ' . $finalDestTable); |
| 146 |
$this->dbCore->executeAsTransaction($statements); |
| 147 |
$this->recordHitsRebuildSuccess($chunkSize); |
| 148 |
$this->logger->debugMessage(__FUNCTION__ . " refreshed " . $finalDestTable . " in " . $elapsedTime . " seconds."); |
| 149 |
return array('refreshed' => true, 'elapsed_time' => $elapsedTime, 'error' => ''); |
| 150 |
} catch (Throwable $e) { |
| 151 |
$this->recordHitsRebuildFailure($e->getMessage()); |
| 152 |
$this->logger->errorMessage(__FUNCTION__ . " failed: " . $e->getMessage(), $e instanceof \Exception ? $e : null); |
| 153 |
return array('refreshed' => false, 'elapsed_time' => 0.0, 'error' => $e->getMessage()); |
| 154 |
} finally { |
| 155 |
$this->dropScratchTableIfLeaseOwned($preAggTable); |
| 156 |
} |
| 157 |
} |
| 158 |
|
| 159 |
/** |
| 160 |
* Fill the {CHARSET} / {COLLATION} pair in a staging-table DDL from the one |
| 161 |
* collation the staging table must honour. |
| 162 |
* |
| 163 |
* The rollup's phase-2 JOIN probes redirects.canonical_url, so requested_url |
| 164 |
* has to carry that column's collation or the index cannot serve the probe. |
| 165 |
* The charset therefore cannot be chosen independently: the DDL used to |
| 166 |
* hard-code CHARACTER SET utf8mb4 next to a column collation read from |
| 167 |
* information_schema, and on an install whose redirects table is still |
| 168 |
* latin1 the engine rejects the CREATE outright ("COLLATION |
| 169 |
* 'latin1_swedish_ci' is not valid for CHARACTER SET 'utf8mb4'"), taking the |
| 170 |
* whole rollup rebuild with it. Both halves now come from one pair. |
| 171 |
* |
| 172 |
* @param array{ddl: string, rawCollation: string} $options |
| 173 |
* @return string DDL with a self-consistent charset/collation pair. |
| 174 |
*/ |
| 175 |
private function applyJoinCharsetCollation(array $options): string { |
| 176 |
$pair = ABJ_404_Solution_DatabaseCollationHelper::charsetCollationPair( |
| 177 |
$options['rawCollation'] |
| 178 |
); |
| 179 |
return str_replace( |
| 180 |
array('{CHARSET}', '{COLLATION}'), |
| 181 |
array($pair['charset'], $pair['collation']), |
| 182 |
$options['ddl'] |
| 183 |
); |
| 184 |
} |
| 185 |
|
| 186 |
/** |
| 187 |
* @param string $tempDestTable |
| 188 |
* @return array<string, mixed> |
| 189 |
*/ |
| 190 |
private function hitsTableInsertDirect(string $tempDestTable): array { |
| 191 |
$ttSelectQuery = ABJ_404_Solution_FileSystemService::readFileContents(__DIR__ . "/../sql/getRedirectsForViewTempTable.sql"); |
| 192 |
if ($this->joinHelper->isLogsv2CanonicalUrlBackfillComplete()) { $ttSelectQuery = $this->joinHelper->dropLogsv2CanonicalCoalesceWrap($ttSelectQuery); } |
| 193 |
$ttSelectQuery = str_replace('{redirects_canonical_url_join_rhs}', $this->joinHelper->buildDirectJoinRhs(), $ttSelectQuery); |
| 194 |
$ttSelectQuery = $this->dbCore->doTableNameReplacements($ttSelectQuery); |
| 195 |
$ttInsertQuery = "/* abj404:src=LogsHitsTableRebuilder::hitsTableInsertDirect */ insert into " . $tempDestTable . " (requested_url, logsid, last_used, logshits, failed_hits) \n " . $ttSelectQuery; |
| 196 |
return $this->dbCore->queryAndGetResults($ttInsertQuery, array('log_too_slow' => false, 'timeout' => 60)); |
| 197 |
} |
| 198 |
|
| 199 |
/** @return array<string, mixed>|false */ |
| 200 |
private function hitsTableInsertChunked(string $tempDestTable, string $preAggTable, int $minId, int $maxId, int $chunkSize) { |
| 201 |
$logsv2Table = $this->dbCore->doTableNameReplacements("{wp_abj404_logsv2}"); |
| 202 |
$redirectsTable = $this->dbCore->doTableNameReplacements("{wp_abj404_redirects}"); |
| 203 |
$resolvedCollation = $this->joinHelper->resolveHitsJoinCollation(); |
| 204 |
$startTime = abj_clock()->nowFloat(); |
| 205 |
$this->renewLeaseOrThrow(); |
| 206 |
$this->dbCore->queryAndGetResults("drop table if exists " . $preAggTable); |
| 207 |
$this->renewLeaseOrThrow(); |
| 208 |
$createPreAggQuery = ABJ_404_Solution_FileSystemService::readFileContents(__DIR__ . "/../sql/createLogsHitsPreAggTempTable.sql"); |
| 209 |
$createPreAggQuery = $this->dbCore->doTableNameReplacements($createPreAggQuery); |
| 210 |
$createPreAggQuery = $this->applyJoinCharsetCollation(array( |
| 211 |
'ddl' => $createPreAggQuery, |
| 212 |
'rawCollation' => $resolvedCollation, |
| 213 |
)); |
| 214 |
$this->dbCore->queryAndGetResults($createPreAggQuery); |
| 215 |
$this->renewLeaseOrThrow(); |
| 216 |
$logsv2CanonicalExpr = $this->joinHelper->isLogsv2CanonicalUrlBackfillComplete() ? "canonical_url" : "COALESCE(canonical_url, CONCAT('/', TRIM(BOTH '/' FROM requested_url)))"; |
| 217 |
for ($start = $minId; $start <= $maxId; $start += $chunkSize) { |
| 218 |
$this->renewLeaseOrThrow(); |
| 219 |
$end = $start + $chunkSize; |
| 220 |
$chunkQuery = "/* abj404:src=LogsHitsTableRebuilder::hitsTableInsertChunked#phase1Chunk */ INSERT INTO " . $preAggTable . " (requested_url, logsid, last_used, logshits, failed_hits) SELECT " . $logsv2CanonicalExpr . ", MIN(id), MAX(timestamp), COUNT(*), SUM(CASE WHEN dest_url = '' OR dest_url IS NULL THEN 1 ELSE 0 END) FROM " . $logsv2Table . " WHERE id >= %d AND id < %d GROUP BY " . $logsv2CanonicalExpr; |
| 221 |
$chunkResult = $this->dbCore->queryAndGetResults($chunkQuery, array('log_too_slow' => false, 'timeout' => 10, 'query_params' => array($start, $end))); |
| 222 |
$this->renewLeaseOrThrow(); |
| 223 |
if (!empty($chunkResult['timed_out']) || !empty($chunkResult['last_error'])) { $this->recordHitsChunkFailure(); $this->logger->debugMessage(__FUNCTION__ . " Phase 1 chunk failed at id range [{$start}, {$end}); aborting."); return false; } |
| 224 |
} |
| 225 |
// Defensive form covers legacy and in-progress installs; optimized |
| 226 |
// form lets idx_canonical_url serve the JOIN probe and gets the |
| 227 |
// rebuild under the host's 60s max_statement_time on Bruno-class |
| 228 |
// data (i359). See LogsHitsCanonicalUrlJoinHelper::buildPhase2JoinRhs. |
| 229 |
$joinRhs = $this->joinHelper->buildPhase2JoinRhs($resolvedCollation); |
| 230 |
$this->renewLeaseOrThrow(); |
| 231 |
$phase2Query = "/* abj404:src=LogsHitsTableRebuilder::hitsTableInsertChunked#phase2Aggregate */ INSERT INTO " . $tempDestTable . " (requested_url, logsid, last_used, logshits, failed_hits) SELECT a.requested_url, MIN(a.logsid), MAX(a.last_used), SUM(a.logshits), SUM(a.failed_hits) FROM " . $preAggTable . " a INNER JOIN " . $redirectsTable . " r ON a.requested_url = " . $joinRhs . " GROUP BY a.requested_url"; |
| 232 |
$results = $this->dbCore->queryAndGetResults($phase2Query, array('log_too_slow' => false, 'timeout' => 60)); |
| 233 |
$this->renewLeaseOrThrow(); |
| 234 |
$results['elapsed_time'] = round(abj_clock()->nowFloat() - $startTime, 3); |
| 235 |
return $results; |
| 236 |
} |
| 237 |
|
| 238 |
/** Abort before another request can run beside a worker that lost its lease. */ |
| 239 |
private function renewLeaseOrThrow(): void { |
| 240 |
if (!$this->renewLease()) { |
| 241 |
throw new RuntimeException( |
| 242 |
'The logs-hits rebuild lost its exclusive lease; aborting before staging-table work can overlap.' |
| 243 |
); |
| 244 |
} |
| 245 |
} |
| 246 |
|
| 247 |
/** Return whether this worker still owns (and has renewed) its lease. */ |
| 248 |
private function renewLease(): bool { |
| 249 |
return $this->leaseRenewer === null || (bool)call_user_func($this->leaseRenewer); |
| 250 |
} |
| 251 |
|
| 252 |
/** Never let a superseded worker drop a replacement worker's scratch table. */ |
| 253 |
private function dropScratchTableIfLeaseOwned(string $scratchTable): void { |
| 254 |
if (!$this->renewLease()) { |
| 255 |
$this->logger->debugMessage( |
| 256 |
__FUNCTION__ . " skipped cleanup after lease ownership was lost: " . $scratchTable |
| 257 |
); |
| 258 |
return; |
| 259 |
} |
| 260 |
$this->dbCore->queryAndGetResults("drop table if exists " . $scratchTable); |
| 261 |
} |
| 262 |
|
| 263 |
/** @param int $idRange @return int */ |
| 264 |
private function getHitsRebuildChunkSize(int $idRange): int { |
| 265 |
if ($this->rebuildHealth === null) { |
| 266 |
return self::HITS_TABLE_PREAGG_CHUNK_SIZE; |
| 267 |
} |
| 268 |
return $this->rebuildHealth->getHitsChunkSize($idRange); |
| 269 |
} |
| 270 |
|
| 271 |
/** @return void */ |
| 272 |
private function recordHitsChunkFailure(): void { |
| 273 |
if ($this->rebuildHealth !== null) { |
| 274 |
$this->rebuildHealth->recordHitsChunkFailure(); |
| 275 |
} |
| 276 |
} |
| 277 |
|
| 278 |
/** @param int $chunkSize @return void */ |
| 279 |
private function recordHitsRebuildSuccess(int $chunkSize): void { |
| 280 |
if ($this->rebuildHealth === null) { |
| 281 |
return; |
| 282 |
} |
| 283 |
$this->rebuildHealth->recordFullRebuildSuccess($chunkSize); |
| 284 |
$this->rebuildHealth->recordSuccess(); |
| 285 |
} |
| 286 |
|
| 287 |
/** @param string $message @return void */ |
| 288 |
private function recordHitsRebuildFailure(string $message): void { |
| 289 |
if ($this->rebuildHealth === null) { |
| 290 |
return; |
| 291 |
} |
| 292 |
$this->rebuildHealth->recordFailure($message, $this->rebuildHealth->classifyError($message)); |
| 293 |
} |
| 294 |
} |
| 295 |
|
