| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Reads aggregate statistics used by admin dashboards and periodic summaries. |
| 9 |
*/ |
| 10 |
class ABJ_404_Solution_StatsReadRepository { |
| 11 |
|
| 12 |
/** @var int Max age for cached stats-periodic aggregates. */ |
| 13 |
const PERIODIC_STATS_CACHE_TTL_SECONDS = 300; |
| 14 |
/** @var int Minimum interval before recalculating expensive stats aggregates. */ |
| 15 |
const PERIODIC_STATS_REFRESH_COOLDOWN_SECONDS = 30; |
| 16 |
|
| 17 |
/** @var ABJ_404_Solution_DatabaseQueryInterface */ |
| 18 |
private $dbCore; |
| 19 |
/** @var ABJ_404_Solution_LogsRepositoryInterface */ |
| 20 |
private $logsRepo; |
| 21 |
/** @var ABJ_404_Solution_Logging */ |
| 22 |
private $logger; |
| 23 |
/** @var ABJ_404_Solution_StatsRefreshLock */ |
| 24 |
private $refreshLock; |
| 25 |
|
| 26 |
/** |
| 27 |
* @param ABJ_404_Solution_DatabaseQueryInterface $dbCore |
| 28 |
* @param ABJ_404_Solution_LogsRepositoryInterface $logsRepo |
| 29 |
* @param ABJ_404_Solution_Logging $logging |
| 30 |
* @param ABJ_404_Solution_StatsRefreshLock $refreshLock |
| 31 |
*/ |
| 32 |
public function __construct( |
| 33 |
ABJ_404_Solution_DatabaseQueryInterface $dbCore, |
| 34 |
ABJ_404_Solution_LogsRepositoryInterface $logsRepo, |
| 35 |
$logging, |
| 36 |
ABJ_404_Solution_StatsRefreshLock $refreshLock |
| 37 |
) { |
| 38 |
$this->dbCore = $dbCore; |
| 39 |
$this->logsRepo = $logsRepo; |
| 40 |
$this->logger = $logging; |
| 41 |
$this->refreshLock = $refreshLock; |
| 42 |
} |
| 43 |
|
| 44 |
/** |
| 45 |
* @param string $query |
| 46 |
* @param array<int|string, mixed> $valueParams |
| 47 |
* @return int |
| 48 |
*/ |
| 49 |
public function getStatsCount($query, array $valueParams) { |
| 50 |
if ($query == '') { |
| 51 |
return 0; |
| 52 |
} |
| 53 |
|
| 54 |
$result = $this->dbCore->queryAndGetResults($query, array('query_params' => $valueParams)); |
| 55 |
|
| 56 |
if (!empty($result['timed_out']) || (isset($result['last_error']) && $result['last_error'] != '')) { |
| 57 |
return 0; |
| 58 |
} |
| 59 |
|
| 60 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array(); |
| 61 |
if (empty($rows)) { |
| 62 |
$this->logger->debugMessage("getStatsCount returned no results for query: " . esc_html($query)); |
| 63 |
return 0; |
| 64 |
} |
| 65 |
|
| 66 |
$first = $rows[0]; |
| 67 |
if (is_array($first)) { |
| 68 |
$value = reset($first); |
| 69 |
} else { |
| 70 |
$value = $first; |
| 71 |
} |
| 72 |
return $this->toInt($value, 0); |
| 73 |
} |
| 74 |
|
| 75 |
/** |
| 76 |
* @param int $sinceTimestamp |
| 77 |
* @param string $notFoundDest |
| 78 |
* @return array{disp404:int,distinct404:int,visitors404:int,refer404:int,redirected:int,distinctredirected:int,distinctvisitors:int,distinctrefer:int} |
| 79 |
*/ |
| 80 |
public function getPeriodicStatsSummary($sinceTimestamp, $notFoundDest = '404') { |
| 81 |
$sinceTimestamp = absint($sinceTimestamp); |
| 82 |
$notFoundDest = sanitize_text_field((string)$notFoundDest); |
| 83 |
if ($notFoundDest === '') { |
| 84 |
$notFoundDest = '404'; |
| 85 |
} |
| 86 |
|
| 87 |
$zero = $this->zeroPeriodicStats(); |
| 88 |
|
| 89 |
$logsTable = $this->dbCore->doTableNameReplacements('{wp_abj404_logsv2}'); |
| 90 |
$sql = "SELECT |
| 91 |
COUNT(CASE WHEN dest_url = %s THEN 1 END) AS disp404, |
| 92 |
COUNT(DISTINCT CASE WHEN dest_url = %s THEN requested_url END) AS distinct404, |
| 93 |
COUNT(DISTINCT CASE WHEN dest_url = %s THEN user_ip END) AS visitors404, |
| 94 |
COUNT(DISTINCT CASE WHEN dest_url = %s THEN referrer END) AS refer404, |
| 95 |
COUNT(CASE WHEN dest_url <> %s THEN 1 END) AS redirected, |
| 96 |
COUNT(DISTINCT CASE WHEN dest_url <> %s THEN requested_url END) AS distinctredirected, |
| 97 |
COUNT(DISTINCT CASE WHEN dest_url <> %s THEN user_ip END) AS distinctvisitors, |
| 98 |
COUNT(DISTINCT CASE WHEN dest_url <> %s THEN referrer END) AS distinctrefer |
| 99 |
FROM {$logsTable} |
| 100 |
WHERE timestamp >= %d"; |
| 101 |
|
| 102 |
$result = $this->dbCore->queryAndGetResults($sql, array( |
| 103 |
'query_params' => array( |
| 104 |
$notFoundDest, $notFoundDest, $notFoundDest, $notFoundDest, |
| 105 |
$notFoundDest, $notFoundDest, $notFoundDest, $notFoundDest, |
| 106 |
$sinceTimestamp, |
| 107 |
), |
| 108 |
)); |
| 109 |
|
| 110 |
if (!empty($result['timed_out']) || (isset($result['last_error']) && $result['last_error'] != '')) { |
| 111 |
return $zero; |
| 112 |
} |
| 113 |
|
| 114 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array(); |
| 115 |
if (empty($rows) || !is_array($rows[0] ?? null)) { |
| 116 |
return $zero; |
| 117 |
} |
| 118 |
$row = $rows[0]; |
| 119 |
|
| 120 |
foreach ($zero as $key => $unused) { |
| 121 |
$zero[$key] = array_key_exists($key, $row) ? $this->toInt($row[$key], 0) : 0; |
| 122 |
} |
| 123 |
|
| 124 |
return $zero; |
| 125 |
} |
| 126 |
|
| 127 |
/** |
| 128 |
* @param string $notFoundDest |
| 129 |
* @return array{today:array<string,int>,month:array<string,int>,year:array<string,int>,all:array<string,int>} |
| 130 |
*/ |
| 131 |
public function getPeriodicStatsSummariesCached($notFoundDest = '404') { |
| 132 |
$now = abj_clock()->now(); |
| 133 |
$today = mktime(0, 0, 0, abs(intval(date('m', $now))), abs(intval(date('d', $now))), abs(intval(date('Y', $now)))); |
| 134 |
$firstm = mktime(0, 0, 0, abs(intval(date('m', $now))), 1, abs(intval(date('Y', $now)))); |
| 135 |
$firsty = mktime(0, 0, 0, 1, 1, abs(intval(date('Y', $now)))); |
| 136 |
|
| 137 |
$thresholds = array( |
| 138 |
'today' => intval($today), |
| 139 |
'month' => intval($firstm), |
| 140 |
'year' => intval($firsty), |
| 141 |
'all' => 0, |
| 142 |
); |
| 143 |
|
| 144 |
$emptyPayload = $this->emptyPeriodicPayload(); |
| 145 |
$cacheKey = 'abj404_stats_periodic_v1_' . $this->currentBlogId() . '_' . md5( |
| 146 |
$notFoundDest . '|' . $thresholds['today'] . '|' . $thresholds['month'] . '|' . $thresholds['year'] |
| 147 |
); |
| 148 |
$cached = function_exists('get_transient') ? get_transient($cacheKey) : null; |
| 149 |
|
| 150 |
$isCachedValid = (is_array($cached) && isset($cached['periods']) && is_array($cached['periods'])); |
| 151 |
$cachedPeriods = $isCachedValid ? $this->normalizePeriodicPayload($cached['periods']) : $emptyPayload; |
| 152 |
$currentMaxLogId = -1; |
| 153 |
try { |
| 154 |
$currentMaxLogId = $this->toInt($this->logsRepo->getMaxLogId(), -1); |
| 155 |
} catch (Throwable $unused) { // allow-silent-catch: cache-key derivation; -1 means "no cached entry, recompute" which is the correct degraded behavior |
| 156 |
$currentMaxLogId = -1; |
| 157 |
} |
| 158 |
|
| 159 |
if ($isCachedValid) { |
| 160 |
$refreshedAt = $this->toInt($cached['refreshed_at'] ?? 0, 0); |
| 161 |
$ageSeconds = max(0, abj_clock()->now() - $refreshedAt); |
| 162 |
$cachedMaxLogId = $this->toInt($cached['max_log_id'] ?? -1, -1); |
| 163 |
if ($currentMaxLogId >= 0 && $cachedMaxLogId === $currentMaxLogId) { |
| 164 |
return $cachedPeriods; |
| 165 |
} |
| 166 |
if ($ageSeconds < self::PERIODIC_STATS_REFRESH_COOLDOWN_SECONDS) { |
| 167 |
return $cachedPeriods; |
| 168 |
} |
| 169 |
} |
| 170 |
|
| 171 |
$lockKey = 'stats-periodic:' . $cacheKey; |
| 172 |
$lockAcquired = $this->refreshLock->acquire($lockKey); |
| 173 |
if (!$lockAcquired && $isCachedValid) { |
| 174 |
return $cachedPeriods; |
| 175 |
} |
| 176 |
|
| 177 |
try { |
| 178 |
$result = array( |
| 179 |
'today' => $this->getPeriodicStatsSummary($thresholds['today'], $notFoundDest), |
| 180 |
'month' => $this->getPeriodicStatsSummary($thresholds['month'], $notFoundDest), |
| 181 |
'year' => $this->getPeriodicStatsSummary($thresholds['year'], $notFoundDest), |
| 182 |
'all' => $this->getPeriodicStatsSummary($thresholds['all'], $notFoundDest), |
| 183 |
); |
| 184 |
|
| 185 |
if (function_exists('set_transient')) { |
| 186 |
set_transient( |
| 187 |
$cacheKey, |
| 188 |
array( |
| 189 |
'refreshed_at' => abj_clock()->now(), |
| 190 |
'max_log_id' => $currentMaxLogId, |
| 191 |
'periods' => $result, |
| 192 |
), |
| 193 |
self::PERIODIC_STATS_CACHE_TTL_SECONDS |
| 194 |
); |
| 195 |
} |
| 196 |
|
| 197 |
return $result; |
| 198 |
} finally { |
| 199 |
if ($lockAcquired) { |
| 200 |
$this->refreshLock->release($lockKey); |
| 201 |
} |
| 202 |
} |
| 203 |
} |
| 204 |
|
| 205 |
/** @return int */ |
| 206 |
public function getEarliestLogTimestamp() { |
| 207 |
$query = 'SELECT min(timestamp) as timestamp FROM {wp_abj404_logsv2}'; |
| 208 |
|
| 209 |
$result = $this->dbCore->queryAndGetResults($query); |
| 210 |
|
| 211 |
if (!empty($result['timed_out']) || (isset($result['last_error']) && $result['last_error'] != '')) { |
| 212 |
return -1; |
| 213 |
} |
| 214 |
|
| 215 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array(); |
| 216 |
if (empty($rows)) { |
| 217 |
return -1; |
| 218 |
} |
| 219 |
|
| 220 |
$first = $rows[0]; |
| 221 |
$value = is_array($first) ? reset($first) : $first; |
| 222 |
if ($value === null || $value === false || $value === '') { |
| 223 |
return -1; |
| 224 |
} |
| 225 |
return $this->toInt($value, -1); |
| 226 |
} |
| 227 |
|
| 228 |
/** |
| 229 |
* Count redirect rows grouped into match-confidence bands for the stats |
| 230 |
* page Match Confidence card. |
| 231 |
* |
| 232 |
* A NULL score is the "manual" band (no automated scoring took place); |
| 233 |
* scored rows fall into high/medium/low per |
| 234 |
* {@see ABJ_404_Solution_ScoreThresholds}. Disabled rows and rows with |
| 235 |
* status 0 are excluded. Routed through queryAndGetResults() so the |
| 236 |
* 5x SUM(CASE...) aggregate inherits the centralized 60s SELECT timeout |
| 237 |
* (the redirects table can be very large on busy sites). |
| 238 |
* |
| 239 |
* This is the single owner of the confidence-band SQL and thresholds: |
| 240 |
* the view layer asks for the counts and only formats them. |
| 241 |
* |
| 242 |
* @return array{high:int,medium:int,low:int,manual:int,avg:float|null,total:int}|null |
| 243 |
* Band counts plus the rounded average score (avg is null when no scored |
| 244 |
* rows exist), or null when the query timed out, errored, or returned no |
| 245 |
* aggregate row so the caller can skip rendering the card. |
| 246 |
*/ |
| 247 |
public function getConfidenceBandCounts() { |
| 248 |
$redirectsTable = $this->dbCore->doTableNameReplacements('{wp_abj404_redirects}'); |
| 249 |
|
| 250 |
$high = ABJ_404_Solution_ScoreThresholds::HIGH; |
| 251 |
$medium = ABJ_404_Solution_ScoreThresholds::MEDIUM; |
| 252 |
$sql = "SELECT |
| 253 |
SUM(CASE WHEN score IS NULL THEN 1 ELSE 0 END) AS manual_count, |
| 254 |
SUM(CASE WHEN score >= {$high} THEN 1 ELSE 0 END) AS high_count, |
| 255 |
SUM(CASE WHEN score >= {$medium} AND score < {$high} THEN 1 ELSE 0 END) AS medium_count, |
| 256 |
SUM(CASE WHEN score IS NOT NULL AND score < {$medium} THEN 1 ELSE 0 END) AS low_count, |
| 257 |
AVG(score) AS avg_score |
| 258 |
FROM `{$redirectsTable}` |
| 259 |
WHERE disabled = %d AND status != %d"; |
| 260 |
|
| 261 |
$result = $this->dbCore->queryAndGetResults($sql, array('query_params' => array(0, 0))); |
| 262 |
if (!empty($result['timed_out']) || (isset($result['last_error']) && $result['last_error'] != '')) { |
| 263 |
return null; |
| 264 |
} |
| 265 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array(); |
| 266 |
if (empty($rows) || !is_array($rows[0] ?? null)) { |
| 267 |
return null; |
| 268 |
} |
| 269 |
$row = $rows[0]; |
| 270 |
|
| 271 |
$highCount = $this->toInt($row['high_count'] ?? 0, 0); |
| 272 |
$mediumCount = $this->toInt($row['medium_count'] ?? 0, 0); |
| 273 |
$lowCount = $this->toInt($row['low_count'] ?? 0, 0); |
| 274 |
$manualCount = $this->toInt($row['manual_count'] ?? 0, 0); |
| 275 |
$avgRaw = $row['avg_score'] ?? null; |
| 276 |
$avgScore = is_numeric($avgRaw) ? round((float)$avgRaw, 1) : null; |
| 277 |
|
| 278 |
return array( |
| 279 |
'high' => $highCount, |
| 280 |
'medium' => $mediumCount, |
| 281 |
'low' => $lowCount, |
| 282 |
'manual' => $manualCount, |
| 283 |
'avg' => $avgScore, |
| 284 |
'total' => $highCount + $mediumCount + $lowCount + $manualCount, |
| 285 |
); |
| 286 |
} |
| 287 |
|
| 288 |
/** @return array<string, mixed> */ |
| 289 |
public function buildStatsDashboardSnapshotData() { |
| 290 |
$redirectsTable = $this->dbCore->doTableNameReplacements("{wp_abj404_redirects}"); |
| 291 |
|
| 292 |
$auto301 = $this->getStatsCount( |
| 293 |
"select count(id) from $redirectsTable where disabled = 0 and code = 301 and status = %d", |
| 294 |
array(ABJ404_STATUS_AUTO) |
| 295 |
); |
| 296 |
$auto302 = $this->getStatsCount( |
| 297 |
"select count(id) from $redirectsTable where disabled = 0 and code = 302 and status = %d", |
| 298 |
array(ABJ404_STATUS_AUTO) |
| 299 |
); |
| 300 |
$manual301 = $this->getStatsCount( |
| 301 |
"select count(id) from $redirectsTable where disabled = 0 and code = 301 and status = %d", |
| 302 |
array(ABJ404_STATUS_MANUAL) |
| 303 |
); |
| 304 |
$manual302 = $this->getStatsCount( |
| 305 |
"select count(id) from $redirectsTable where disabled = 0 and code = 302 and status = %d", |
| 306 |
array(ABJ404_STATUS_MANUAL) |
| 307 |
); |
| 308 |
$trashedRedirects = $this->getStatsCount( |
| 309 |
"select count(id) from $redirectsTable where disabled = 1 and (status = %d or status = %d)", |
| 310 |
array(ABJ404_STATUS_AUTO, ABJ404_STATUS_MANUAL) |
| 311 |
); |
| 312 |
|
| 313 |
$captured = $this->getStatsCount( |
| 314 |
"select count(id) from $redirectsTable where disabled = 0 and status = %d", |
| 315 |
array(ABJ404_STATUS_CAPTURED) |
| 316 |
); |
| 317 |
$ignored = $this->getStatsCount( |
| 318 |
"select count(id) from $redirectsTable where disabled = 0 and status in (%d, %d)", |
| 319 |
array(ABJ404_STATUS_IGNORED, ABJ404_STATUS_LATER) |
| 320 |
); |
| 321 |
$trashedCaptured = $this->getStatsCount( |
| 322 |
"select count(id) from $redirectsTable where disabled = 1 and (status in (%d, %d, %d) )", |
| 323 |
array(ABJ404_STATUS_CAPTURED, ABJ404_STATUS_IGNORED, ABJ404_STATUS_LATER) |
| 324 |
); |
| 325 |
|
| 326 |
$now = abj_clock()->now(); |
| 327 |
$thresholds = array( |
| 328 |
'today' => (int)mktime(0, 0, 0, abs(intval(date('m', $now))), abs(intval(date('d', $now))), abs(intval(date('Y', $now)))), |
| 329 |
'month' => (int)mktime(0, 0, 0, abs(intval(date('m', $now))), 1, abs(intval(date('Y', $now)))), |
| 330 |
'year' => (int)mktime(0, 0, 0, 1, 1, abs(intval(date('Y', $now)))), |
| 331 |
'all' => 0, |
| 332 |
); |
| 333 |
$periods = array(); |
| 334 |
foreach ($thresholds as $periodKey => $ts) { |
| 335 |
$periods[$periodKey] = $this->getPeriodicStatsSummary($ts, '404'); |
| 336 |
} |
| 337 |
|
| 338 |
return array( |
| 339 |
'redirects' => array( |
| 340 |
'auto301' => intval($auto301), |
| 341 |
'auto302' => intval($auto302), |
| 342 |
'manual301' => intval($manual301), |
| 343 |
'manual302' => intval($manual302), |
| 344 |
'trashed' => intval($trashedRedirects), |
| 345 |
), |
| 346 |
'captured' => array( |
| 347 |
'captured' => intval($captured), |
| 348 |
'ignored' => intval($ignored), |
| 349 |
'trashed' => intval($trashedCaptured), |
| 350 |
), |
| 351 |
'periods' => $periods, |
| 352 |
); |
| 353 |
} |
| 354 |
|
| 355 |
/** @return array{disp404:int,distinct404:int,visitors404:int,refer404:int,redirected:int,distinctredirected:int,distinctvisitors:int,distinctrefer:int} */ |
| 356 |
private function zeroPeriodicStats(): array { |
| 357 |
return array( |
| 358 |
'disp404' => 0, |
| 359 |
'distinct404' => 0, |
| 360 |
'visitors404' => 0, |
| 361 |
'refer404' => 0, |
| 362 |
'redirected' => 0, |
| 363 |
'distinctredirected' => 0, |
| 364 |
'distinctvisitors' => 0, |
| 365 |
'distinctrefer' => 0, |
| 366 |
); |
| 367 |
} |
| 368 |
|
| 369 |
/** @return array{today:array<string,int>,month:array<string,int>,year:array<string,int>,all:array<string,int>} */ |
| 370 |
private function emptyPeriodicPayload(): array { |
| 371 |
$zero = $this->zeroPeriodicStats(); |
| 372 |
return array( |
| 373 |
'today' => $zero, |
| 374 |
'month' => $zero, |
| 375 |
'year' => $zero, |
| 376 |
'all' => $zero, |
| 377 |
); |
| 378 |
} |
| 379 |
|
| 380 |
/** |
| 381 |
* @param mixed $periods |
| 382 |
* @return array{today:array<string,int>,month:array<string,int>,year:array<string,int>,all:array<string,int>} |
| 383 |
*/ |
| 384 |
private function normalizePeriodicPayload($periods): array { |
| 385 |
$payload = $this->emptyPeriodicPayload(); |
| 386 |
if (!is_array($periods)) { |
| 387 |
return $payload; |
| 388 |
} |
| 389 |
foreach (array('today', 'month', 'year', 'all') as $periodKey) { |
| 390 |
if (isset($periods[$periodKey]) && is_array($periods[$periodKey])) { |
| 391 |
$payload[$periodKey] = $this->normalizePeriodicStats($periods[$periodKey]); |
| 392 |
} |
| 393 |
} |
| 394 |
return $payload; |
| 395 |
} |
| 396 |
|
| 397 |
/** |
| 398 |
* @param array<int|string, mixed> $stats |
| 399 |
* @return array{disp404:int,distinct404:int,visitors404:int,refer404:int,redirected:int,distinctredirected:int,distinctvisitors:int,distinctrefer:int} |
| 400 |
*/ |
| 401 |
private function normalizePeriodicStats(array $stats): array { |
| 402 |
$zero = $this->zeroPeriodicStats(); |
| 403 |
foreach ($zero as $key => $unused) { |
| 404 |
$zero[$key] = array_key_exists($key, $stats) ? $this->toInt($stats[$key], 0) : 0; |
| 405 |
} |
| 406 |
return $zero; |
| 407 |
} |
| 408 |
|
| 409 |
/** @return int */ |
| 410 |
private function currentBlogId(): int { |
| 411 |
$blogId = 1; |
| 412 |
if (function_exists('get_current_blog_id')) { |
| 413 |
$blogId = absint(get_current_blog_id()); |
| 414 |
if ($blogId <= 0) { |
| 415 |
$blogId = 1; |
| 416 |
} |
| 417 |
} |
| 418 |
return $blogId; |
| 419 |
} |
| 420 |
|
| 421 |
/** @param mixed $value @param int $default @return int */ |
| 422 |
private function toInt($value, int $default): int { |
| 423 |
if ($value === null) { |
| 424 |
return $default; |
| 425 |
} |
| 426 |
if (is_scalar($value)) { |
| 427 |
return intval($value); |
| 428 |
} |
| 429 |
return $default; |
| 430 |
} |
| 431 |
} |
| 432 |
|
