| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Admin view query construction for the redirect/captured lists. |
| 9 |
* |
| 10 |
* Builds the SQL for the admin redirect/captured lists: high-impact captured |
| 11 |
* count, regex redirects, optimized count, and the single-table read/count |
| 12 |
* against wp_abj404_redirects (Denorm Step 3b). The staged view_done read path |
| 13 |
* that used to live here was removed when the denorm chain dropped the |
| 14 |
* wp_abj404_view_done table (Step 3e-D / i467); admin reads now serve straight |
| 15 |
* off the redirects row. |
| 16 |
*/ |
| 17 |
class ABJ_404_Solution_ViewQueryBuilder { |
| 18 |
|
| 19 |
/** @var ABJ_404_Solution_DatabaseCore */ |
| 20 |
private $dbCore; |
| 21 |
|
| 22 |
/** @var ABJ_404_Solution_ViewQueryPolicy */ |
| 23 |
private $policy; |
| 24 |
|
| 25 |
/** |
| 26 |
* @param ABJ_404_Solution_DatabaseCore $dbCore |
| 27 |
*/ |
| 28 |
public function __construct(ABJ_404_Solution_DatabaseCore $dbCore) { |
| 29 |
$this->dbCore = $dbCore; |
| 30 |
$this->policy = new ABJ_404_Solution_ViewQueryPolicy(); |
| 31 |
} |
| 32 |
|
| 33 |
/** @return string */ |
| 34 |
public function buildHighImpactCapturedCountQuery(): string { |
| 35 |
// Plain equality gives the optimizer an indexable requested_url probe; |
| 36 |
// the BINARY predicate keeps exact-match URL semantics. |
| 37 |
$query = "SELECT COUNT(*) AS cnt |
| 38 |
FROM {wp_abj404_redirects} r |
| 39 |
INNER JOIN {wp_abj404_logs_hits} h |
| 40 |
ON h.requested_url = |
| 41 |
COALESCE(r.canonical_url, CONCAT('/', TRIM(BOTH '/' FROM r.url))) |
| 42 |
AND BINARY h.requested_url = BINARY |
| 43 |
COALESCE(r.canonical_url, CONCAT('/', TRIM(BOTH '/' FROM r.url))) |
| 44 |
WHERE r.status = " . ABJ404_STATUS_CAPTURED . " AND r.disabled = 0 |
| 45 |
AND h.logshits >= 3"; |
| 46 |
return $this->dbCore->doTableNameReplacements($query); |
| 47 |
} |
| 48 |
|
| 49 |
/** |
| 50 |
* @param int|null $limit |
| 51 |
* @return array<int, array<string, mixed>> |
| 52 |
*/ |
| 53 |
public function queryRegexRedirects(?int $limit = null) { |
| 54 |
$query = "select \n {wp_abj404_redirects}.id,\n {wp_abj404_redirects}.url,\n {wp_abj404_redirects}.status,\n" |
| 55 |
. " {wp_abj404_redirects}.type,\n {wp_abj404_redirects}.final_dest,\n {wp_abj404_redirects}.code,\n" |
| 56 |
. " {wp_abj404_redirects}.timestamp,\n {wp_posts}.id as wp_post_id\n " |
| 57 |
. "from {wp_abj404_redirects}\n " |
| 58 |
. " LEFT OUTER JOIN {wp_posts} \n " |
| 59 |
. " on {wp_abj404_redirects}.final_dest = {wp_posts}.id \n " |
| 60 |
. "where status in (" . ABJ404_STATUS_REGEX . ") \n " |
| 61 |
. " and disabled = 0\n" |
| 62 |
. "order by {wp_abj404_redirects}.id ASC"; |
| 63 |
if ($limit !== null) { |
| 64 |
$query .= "\nlimit " . max(1, intval($limit)); |
| 65 |
} |
| 66 |
|
| 67 |
$results = $this->dbCore->queryAndGetResults($query); |
| 68 |
$rows = is_array($results['rows'] ?? null) ? $results['rows'] : array(); |
| 69 |
/** @var array<int, array<string, mixed>> $rows */ |
| 70 |
return $rows; |
| 71 |
} |
| 72 |
|
| 73 |
/** |
| 74 |
* @param string $sub |
| 75 |
* @param array<string, mixed> $tableOptions |
| 76 |
* @return string |
| 77 |
*/ |
| 78 |
public function getOptimizedRedirectsForViewCountQuery(string $sub, array $tableOptions): string { |
| 79 |
$statusTypes = $this->policy->resolveStatusTypeList($sub, $tableOptions); |
| 80 |
$trashValue = $this->policy->resolveTrashValue($tableOptions); |
| 81 |
$scoreRangeClause = $this->policy->buildScoreRangeClause($tableOptions, 'wp_abj404_redirects.'); // allow-prefix-literal: SQL alias bound by FROM clause |
| 82 |
|
| 83 |
$query = "SELECT COUNT(*) AS count\n" |
| 84 |
. "FROM {wp_abj404_redirects} wp_abj404_redirects\n" // allow-prefix-literal: second token is the SQL alias name, not a table reference |
| 85 |
. "WHERE 1 and status IN (" . $statusTypes . ") AND disabled = " . intval($trashValue) . "\n" |
| 86 |
. $scoreRangeClause; |
| 87 |
|
| 88 |
return $this->dbCore->doTableNameReplacements($query); |
| 89 |
} |
| 90 |
|
| 91 |
/** |
| 92 |
* Execute the single-table redirects read for one page (Denorm Step 3b). |
| 93 |
* |
| 94 |
* Serves rows straight off wp_abj404_redirects, where every sortable column |
| 95 |
* (url, status, type, code, timestamp, score, logshits, last_used, |
| 96 |
* dest_for_view) is a real column -> a single-table filesort over at most one |
| 97 |
* page of rows, no temp-table materialize, no join. The four derived columns |
| 98 |
* are refreshed live per visible row by RedirectsViewLiveResolver after this |
| 99 |
* read; this method only fetches the ordered/filtered page. |
| 100 |
* |
| 101 |
* @param string $sub |
| 102 |
* @param array<string, mixed> $tableOptions |
| 103 |
* @param bool $derivedPresent Whether the four denorm columns exist on the |
| 104 |
* redirects table (schema-drift tolerance: false selects base columns only |
| 105 |
* and the live resolver fills the derived values). |
| 106 |
* @return array<int, array<string, mixed>> |
| 107 |
*/ |
| 108 |
public function readRedirectsSingleTable(string $sub, array $tableOptions, bool $derivedPresent = true): array { |
| 109 |
$query = $this->buildRedirectsSingleTableReadQuery($sub, $tableOptions, $derivedPresent); |
| 110 |
$result = $this->dbCore->queryAndGetResults($query, $this->resolveReadTimeoutOptions($tableOptions)); |
| 111 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array(); |
| 112 |
/** @var array<int, array<string, mixed>> $rows */ |
| 113 |
return $rows; |
| 114 |
} |
| 115 |
|
| 116 |
/** |
| 117 |
* Build the queryAndGetResults options for the single-table read/count from |
| 118 |
* the optional per-request timeout, with a level-9-safe numeric guard. |
| 119 |
* |
| 120 |
* @param array<string, mixed> $tableOptions |
| 121 |
* @return array<string, int> |
| 122 |
*/ |
| 123 |
private function resolveReadTimeoutOptions(array $tableOptions): array { |
| 124 |
$raw = $tableOptions['_abj404_query_timeout'] ?? null; |
| 125 |
if (!is_numeric($raw)) { |
| 126 |
return array(); |
| 127 |
} |
| 128 |
$timeout = (int)$raw; |
| 129 |
return $timeout > 0 ? array('timeout' => $timeout) : array(); |
| 130 |
} |
| 131 |
|
| 132 |
/** |
| 133 |
* Execute the single-table filtered count against wp_abj404_redirects. |
| 134 |
* |
| 135 |
* @param string $sub |
| 136 |
* @param array<string, mixed> $tableOptions |
| 137 |
* @param bool $derivedPresent |
| 138 |
* @return int |
| 139 |
*/ |
| 140 |
public function countRedirectsSingleTable(string $sub, array $tableOptions, bool $derivedPresent = true): int { |
| 141 |
$query = $this->buildRedirectsSingleTableCountQuery($sub, $tableOptions, $derivedPresent); |
| 142 |
$result = $this->dbCore->queryAndGetResults($query, $this->resolveReadTimeoutOptions($tableOptions)); |
| 143 |
$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array(); |
| 144 |
if (empty($rows) || !is_array($rows[0])) { |
| 145 |
return 0; |
| 146 |
} |
| 147 |
$raw = $rows[0]['cnt'] ?? reset($rows[0]); |
| 148 |
return is_scalar($raw) ? intval($raw) : 0; |
| 149 |
} |
| 150 |
|
| 151 |
/** |
| 152 |
* @param string $sub |
| 153 |
* @param array<string, mixed> $tableOptions |
| 154 |
* @param bool $derivedPresent |
| 155 |
* @return string |
| 156 |
*/ |
| 157 |
public function buildRedirectsSingleTableReadQuery(string $sub, array $tableOptions, bool $derivedPresent = true): string { |
| 158 |
$effectiveSort = $this->resolveEffectiveSort($tableOptions, $derivedPresent); |
| 159 |
$orderBy = $effectiveSort['orderby']; |
| 160 |
$order = $effectiveSort['order']; |
| 161 |
|
| 162 |
$rawPaged = $tableOptions['paged'] ?? 1; |
| 163 |
$paged = max(1, is_scalar($rawPaged) ? intval($rawPaged) : 1); |
| 164 |
$rawPerpage = $tableOptions['perpage'] ?? ABJ404_OPTION_DEFAULT_PERPAGE; |
| 165 |
$perpage = max(1, is_scalar($rawPerpage) ? intval($rawPerpage) : (int)ABJ404_OPTION_DEFAULT_PERPAGE); |
| 166 |
$limitStart = ($paged - 1) * $perpage; |
| 167 |
|
| 168 |
// The derived columns are projected only when they exist, purely so the |
| 169 |
// live resolver can dedupe its write-back against the stored values; the |
| 170 |
// rendered values come from the live resolution either way. On a |
| 171 |
// schema-drifted table without them, the base-column projection still |
| 172 |
// renders a complete page. |
| 173 |
$derivedProjection = $derivedPresent |
| 174 |
? ",\n dest_for_view, published_status, logshits, last_used" : ""; |
| 175 |
|
| 176 |
// Tie-break on the PK id in the sort direction (not on url): url is |
| 177 |
// varchar(2048) and only prefix-indexable, so an `ORDER BY <col>, url` |
| 178 |
// can never be index-ordered and always filesorts. With `id <dir>` a |
| 179 |
// single ascending composite index (disabled, <col>, id) serves both |
| 180 |
// ASC and DESC scans, so the Hits / Last Used sorts stop after one page |
| 181 |
// instead of sorting the whole active-redirect set. See |
| 182 |
// RedirectsDerivedSortExplainPlanTest. id is unique, so the ordering is |
| 183 |
// still fully deterministic for pagination. |
| 184 |
return "SELECT id, url, status, type, final_dest, code, timestamp, engine, score" |
| 185 |
. $derivedProjection . "\n" |
| 186 |
. "FROM {wp_abj404_redirects}\n" |
| 187 |
. $this->buildSingleTableWhere($sub, $tableOptions, $derivedPresent) |
| 188 |
. "ORDER BY " . $orderBy . " " . $order . ", id " . $order . "\n" |
| 189 |
. "LIMIT " . $limitStart . ", " . $perpage; |
| 190 |
} |
| 191 |
|
| 192 |
/** |
| 193 |
* @param string $sub |
| 194 |
* @param array<string, mixed> $tableOptions |
| 195 |
* @param bool $derivedPresent |
| 196 |
* @return string |
| 197 |
*/ |
| 198 |
public function buildRedirectsSingleTableCountQuery(string $sub, array $tableOptions, bool $derivedPresent = true): string { |
| 199 |
return "SELECT COUNT(*) AS cnt\n" |
| 200 |
. "FROM {wp_abj404_redirects}\n" |
| 201 |
. $this->buildSingleTableWhere($sub, $tableOptions, $derivedPresent); |
| 202 |
} |
| 203 |
|
| 204 |
/** |
| 205 |
* The shared WHERE body for the single-table read and count: status filter, |
| 206 |
* trash (disabled) filter, score range, and the dest_for_view-aware |
| 207 |
* filterText search. Identical between read and count so a filtered count |
| 208 |
* always equals the unpaginated row set (i457 invariant). The filterText |
| 209 |
* destination-title match drops to url/label matching when the dest_for_view |
| 210 |
* column is absent (schema-drift tolerance). |
| 211 |
* |
| 212 |
* @param string $sub |
| 213 |
* @param array<string, mixed> $tableOptions |
| 214 |
* @param bool $derivedPresent |
| 215 |
* @return string |
| 216 |
*/ |
| 217 |
private function buildSingleTableWhere(string $sub, array $tableOptions, bool $derivedPresent = true): string { |
| 218 |
$statusTypes = $this->policy->resolveStatusTypeList($sub, $tableOptions); |
| 219 |
$trashClause = 'AND disabled = ' . intval($this->policy->resolveTrashValue($tableOptions)); |
| 220 |
$scoreRangeClause = $this->policy->buildScoreRangeClause($tableOptions, ''); |
| 221 |
$filterTextClause = $this->policy->buildFilterTextClause($sub, $tableOptions, true, $derivedPresent); |
| 222 |
|
| 223 |
return "WHERE status IN (" . $statusTypes . ")\n" |
| 224 |
. " " . $trashClause . "\n" |
| 225 |
. " " . $scoreRangeClause . "\n" |
| 226 |
. " " . $filterTextClause . "\n"; |
| 227 |
} |
| 228 |
|
| 229 |
/** |
| 230 |
* Effective (orderby column, direction) for the single-table read ORDER BY. |
| 231 |
* |
| 232 |
* On EITHER tab, a URL or Destination sort that cannot be served index-ordered |
| 233 |
* right now (its narrow sort key is not ready: column missing, composite index |
| 234 |
* missing, or the legacy-row drain not yet converged -- see |
| 235 |
* RedirectsDenormSchemaReadiness::sortKeyReadyForColumn) would force a filesort over |
| 236 |
* the wide source column (varchar(2048), prefix-only). On a large table that |
| 237 |
* scan can exceed a shared host's max_statement_time, the server kills the |
| 238 |
* query, and the tab is stuck on its loading placeholder every load. Until the |
| 239 |
* sort key is ready we therefore serve the always-indexed safe default |
| 240 |
* (timestamp DESC, "newest first") instead. This is display-only: the user's |
| 241 |
* saved sort preference is left untouched and resumes automatically once the |
| 242 |
* sort key is ready -- and the column header is rendered non-sortable with a |
| 243 |
* progress tooltip meanwhile (captured tab: View_CapturedURLsTable; Page |
| 244 |
* Redirects tab: ABJ_404_Solution_AdminTableColumnHeaders). |
| 245 |
* |
| 246 |
* Applied to BOTH tabs (the literal "no wide-column filesort" rule): the |
| 247 |
* Page Redirects status filter does not make the wide-url filesort safe at |
| 248 |
* scale, and the readiness predicate self-heals so the real sort resumes the |
| 249 |
* moment the key is index-ordered. |
| 250 |
* |
| 251 |
* @param array<string, mixed> $tableOptions |
| 252 |
* @param bool $derivedPresent |
| 253 |
* @return array{orderby: string, order: string} |
| 254 |
*/ |
| 255 |
private function resolveEffectiveSort(array $tableOptions, bool $derivedPresent): array { |
| 256 |
$rawOrderBy = strtolower(is_string($tableOptions['orderby'] ?? null) ? $tableOptions['orderby'] : ''); |
| 257 |
if ($this->wideColumnSortPendingBackfill($rawOrderBy, $tableOptions)) { |
| 258 |
return array('orderby' => 'timestamp', 'order' => 'DESC'); |
| 259 |
} |
| 260 |
return array( |
| 261 |
'orderby' => $this->resolveSingleTableOrderByColumn($tableOptions, $derivedPresent), |
| 262 |
'order' => $this->policy->resolveOrderDirection($tableOptions), |
| 263 |
); |
| 264 |
} |
| 265 |
|
| 266 |
/** |
| 267 |
* Whether the requested sort targets a narrow sort-key-backed column |
| 268 |
* (url -> url_sort_key, dest -> dest_sort_key) that cannot be served |
| 269 |
* index-ordered yet. The coordinator sets the matching |
| 270 |
* _abj404_*_sort_key_present flag true only when the column exists AND its |
| 271 |
* composite indexes exist AND its drain latch is set |
| 272 |
* (RedirectsDenormSchemaReadiness::sortKeyReadyForColumn); a falsey flag means the |
| 273 |
* only available ordering is a wide-column filesort. Sorts on real |
| 274 |
* always-populated columns (logshits, last_used, score, code, type, status, |
| 275 |
* timestamp) are never pending and are not substituted. |
| 276 |
* |
| 277 |
* @param string $rawOrderBy Lowercased requested orderby. |
| 278 |
* @param array<string, mixed> $tableOptions |
| 279 |
* @return bool |
| 280 |
*/ |
| 281 |
private function wideColumnSortPendingBackfill(string $rawOrderBy, array $tableOptions): bool { |
| 282 |
if ($rawOrderBy === 'url') { |
| 283 |
return empty($tableOptions['_abj404_url_sort_key_present']); |
| 284 |
} |
| 285 |
if ($rawOrderBy === 'dest' || $rawOrderBy === 'final_dest') { |
| 286 |
return empty($tableOptions['_abj404_dest_sort_key_present']); |
| 287 |
} |
| 288 |
return false; |
| 289 |
} |
| 290 |
|
| 291 |
/** |
| 292 |
* Resolve the ORDER BY column for the single-table read. |
| 293 |
* |
| 294 |
* A sort on a derived column (logshits / last_used / dest) uses the real |
| 295 |
* column whenever the four denorm columns exist. The only fallback is |
| 296 |
* schema-drift tolerance: when the columns are absent entirely (the column-add |
| 297 |
* ALTER never completed) a derived sort would reference a missing column, so |
| 298 |
* it falls back to the always-present native url column. |
| 299 |
* |
| 300 |
* No backfill-completion flag is consulted. Each derived column degrades |
| 301 |
* gracefully for not-yet-backfilled rows by construction, so ordering on it is |
| 302 |
* always meaningful and self-heals as rows are resolved: |
| 303 |
* - logshits is NOT NULL and defaults to 0, so an un-backfilled row simply |
| 304 |
* sorts as 0 hits and rises into place once the rollup is written; |
| 305 |
* - last_used is NULL for no-hit rows, which sort last; |
| 306 |
* - the dest_for_view ordering groups NULL/empty destinations last via its |
| 307 |
* CASE expression (see ViewQueryPolicy::resolveOrderByColumn). |
| 308 |
* |
| 309 |
* The previous implementation gated these sorts on the obsolete denorm |
| 310 |
* backfill-complete option and fell back to url order until it flipped. That |
| 311 |
* option was flipped only when no row had dest_for_view NULL, a condition a |
| 312 |
* live 404 site never reaches: every newly captured 404 is inserted with |
| 313 |
* dest_for_view NULL, so the option stayed false forever and the Hits / Last |
| 314 |
* Used / Destination columns were permanently sorted by url instead of by |
| 315 |
* their own values (the rendered Hits column looked random). |
| 316 |
* |
| 317 |
* @param array<string, mixed> $tableOptions |
| 318 |
* @param bool $derivedPresent Whether the four denorm columns exist on the |
| 319 |
* table. False (schema drift) is the only case that forces the url fallback. |
| 320 |
* @return string |
| 321 |
*/ |
| 322 |
private function resolveSingleTableOrderByColumn(array $tableOptions, bool $derivedPresent = true): string { |
| 323 |
$rawOrderByValue = $tableOptions['orderby'] ?? ''; |
| 324 |
$rawOrderBy = strtolower(is_string($rawOrderByValue) ? $rawOrderByValue : ''); |
| 325 |
$derivedSorts = array('logshits', 'last_used', 'dest', 'final_dest'); |
| 326 |
if (in_array($rawOrderBy, $derivedSorts, true) && !$derivedPresent) { |
| 327 |
return 'url'; |
| 328 |
} |
| 329 |
// Destination sort: ORDER BY the narrow indexable dest_sort_key |
| 330 |
// (uniform direction -> (disabled, dest_sort_key, id) / |
| 331 |
// (status, disabled, dest_sort_key, id) serve it without a filesort) when |
| 332 |
// the column exists. Blank/NULL destinations then sort naturally (first |
| 333 |
// ascending, last descending) rather than the old always-last CASE, which |
| 334 |
// was the very thing that forced a filesort (computed expression over a |
| 335 |
// varchar(2048) prefix-only column). When the column is absent (an install |
| 336 |
// mid-upgrade), fall back to the CASE-on-dest_for_view filesort so the |
| 337 |
// sort still works -- just unindexed -- until the column-add completes. |
| 338 |
if (($rawOrderBy === 'dest' || $rawOrderBy === 'final_dest') |
| 339 |
&& !empty($tableOptions['_abj404_dest_sort_key_present'])) { |
| 340 |
return 'dest_sort_key'; |
| 341 |
} |
| 342 |
// URL sort (incl. the Page Redirects default): ORDER BY the narrow |
| 343 |
// indexable url_sort_key so (disabled, url_sort_key, id) / |
| 344 |
// (status, disabled, url_sort_key, id) serve it without a filesort. url is |
| 345 |
// varchar(2048), prefix-only, so ORDER BY url itself always filesorts even |
| 346 |
// when every value is short (MySQL will not order by a prefix index). When |
| 347 |
// the column is absent (an install mid-upgrade) fall back to raw url so the |
| 348 |
// sort still works -- unindexed -- until the column-add completes. URLs |
| 349 |
// longer than 191 chars sharing a 191-char prefix tie-break by id. |
| 350 |
if ($rawOrderBy === 'url' && !empty($tableOptions['_abj404_url_sort_key_present'])) { |
| 351 |
return 'url_sort_key'; |
| 352 |
} |
| 353 |
return $this->policy->resolveOrderByColumn($tableOptions); |
| 354 |
} |
| 355 |
|
| 356 |
/** |
| 357 |
* @param string $sub |
| 358 |
* @param array<string, mixed> $tableOptions |
| 359 |
* @return string |
| 360 |
*/ |
| 361 |
public function resolveStatusTypeList(string $sub, array $tableOptions): string { |
| 362 |
return $this->policy->resolveStatusTypeList($sub, $tableOptions); |
| 363 |
} |
| 364 |
|
| 365 |
/** |
| 366 |
* @param array<string, mixed> $tableOptions |
| 367 |
* @return string |
| 368 |
*/ |
| 369 |
public function resolveOrderByColumn(array $tableOptions): string { |
| 370 |
return $this->policy->resolveOrderByColumn($tableOptions); |
| 371 |
} |
| 372 |
|
| 373 |
} |
| 374 |
|
