| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Selects checkpoint records that must survive bounded support ranking. |
| 9 |
* |
| 10 |
* Ordinary request-group ranking may omit an early census, a row-progress |
| 11 |
* sample, or a completed browser report even though each is the only record |
| 12 |
* carrying its discriminator fields. This policy reserves one complete record |
| 13 |
* for every such evidence class before the remaining byte budget is ranked. |
| 14 |
*/ |
| 15 |
final class ABJ_404_Solution_RequiredCheckpointEvidence { |
| 16 |
|
| 17 |
/** |
| 18 |
* Latest complete record for every required evidence identity. |
| 19 |
* |
| 20 |
* @param array<int, string> $lines JSONL lines, oldest first. |
| 21 |
* @return array<int, string> |
| 22 |
*/ |
| 23 |
public static function select(array $lines): array { |
| 24 |
if (class_exists('ABJ_404_Solution_ActiveOperationBreadcrumbs')) { |
| 25 |
$lines = ABJ_404_Solution_ActiveOperationBreadcrumbs::compactSupportLines($lines); |
| 26 |
} |
| 27 |
$operationState = array(); |
| 28 |
foreach (array_reverse($lines) as $line) { |
| 29 |
$record = json_decode($line, true); |
| 30 |
if (!is_array($record)) { |
| 31 |
continue; |
| 32 |
} |
| 33 |
$operationKey = self::activeOperationKey($record); |
| 34 |
if ($operationKey !== '' && !isset($operationState[$operationKey])) { |
| 35 |
$operationState[$operationKey] = array('line' => $line, 'record' => $record); |
| 36 |
} |
| 37 |
} |
| 38 |
$required = array_merge( |
| 39 |
ABJ_404_Solution_MalformedCheckpointEvidence::select($lines), |
| 40 |
ABJ_404_Solution_RequiredCheckpointIdentityEvidence::select($lines) |
| 41 |
); |
| 42 |
foreach ($operationState as $latest) { |
| 43 |
$record = $latest['record']; |
| 44 |
$line = $latest['line']; |
| 45 |
if (self::isReservedActiveOperation($record)) { |
| 46 |
$required[] = $line; |
| 47 |
} |
| 48 |
} |
| 49 |
foreach (self::reservedDurableOperationLines($lines) as $line) { |
| 50 |
$required[] = $line; |
| 51 |
} |
| 52 |
foreach (self::unmatchedOperationLines($lines) as $line) { |
| 53 |
$required[] = $line; |
| 54 |
} |
| 55 |
foreach (ABJ_404_Solution_CheckpointIntentCorrelation::unmatchedIntentLines($lines) |
| 56 |
as $line) { |
| 57 |
$required[] = $line; |
| 58 |
} |
| 59 |
foreach (self::lastFullBoundaryLinesForIncompleteRequests($lines) as $line) { |
| 60 |
$required[] = $line; |
| 61 |
} |
| 62 |
return array_values(array_unique($required)); |
| 63 |
} |
| 64 |
|
| 65 |
/** |
| 66 |
* The final full-envelope boundary reached by each unterminated request. |
| 67 |
* |
| 68 |
* High-frequency query and row records can follow the last lifecycle |
| 69 |
* boundary before a worker stalls. Raw head/tail byte trimming therefore |
| 70 |
* treats that boundary as middle content and can discard the exact |
| 71 |
* pre-stall location when host-pressure counters widen the full envelope. |
| 72 |
* Reserve one boundary per incomplete request so variable environment |
| 73 |
* values cannot change whether the failure remains attributable. |
| 74 |
* |
| 75 |
* A request with only frequent records has no boundary to invent. A |
| 76 |
* request with any terminal event is complete and stays in ordinary |
| 77 |
* request-group ranking. |
| 78 |
* |
| 79 |
* @param array<int, string> $lines |
| 80 |
* @return array<int, string> |
| 81 |
*/ |
| 82 |
private static function lastFullBoundaryLinesForIncompleteRequests(array $lines): array { |
| 83 |
$requests = array(); |
| 84 |
foreach ($lines as $line) { |
| 85 |
$record = json_decode($line, true); |
| 86 |
if (!is_array($record) || !is_scalar($record['request_id'] ?? null) |
| 87 |
|| !is_scalar($record['event'] ?? null)) { |
| 88 |
continue; |
| 89 |
} |
| 90 |
$requestId = (string)$record['request_id']; |
| 91 |
$event = (string)$record['event']; |
| 92 |
if ($requestId === '' || $event === '') { |
| 93 |
continue; |
| 94 |
} |
| 95 |
if (!isset($requests[$requestId])) { |
| 96 |
$requests[$requestId] = array( |
| 97 |
'terminal' => false, |
| 98 |
'boundary' => '', |
| 99 |
'boundary_count' => 0, |
| 100 |
); |
| 101 |
} |
| 102 |
if (in_array($event, ABJ_404_Solution_DiagnosticEvidencePriority::TERMINAL_EVENTS, true)) { |
| 103 |
$requests[$requestId]['terminal'] = true; |
| 104 |
} |
| 105 |
if (($record['envelope'] ?? '') |
| 106 |
=== ABJ_404_Solution_CheckpointRecordFactory::ENVELOPE_FULL) { |
| 107 |
$requests[$requestId]['boundary'] = $line; |
| 108 |
$requests[$requestId]['boundary_count']++; |
| 109 |
} |
| 110 |
} |
| 111 |
|
| 112 |
$selected = array(); |
| 113 |
foreach ($requests as $request) { |
| 114 |
// A lone full record is already both the head and tail of its |
| 115 |
// request group, so raw head/tail ranking cannot hide it. Requiring |
| 116 |
// a preceding boundary also keeps a standalone partial evidence |
| 117 |
// sample from being promoted merely because it lacks a terminal. |
| 118 |
if ($request['terminal'] === false && $request['boundary_count'] > 1) { |
| 119 |
$selected[] = $request['boundary']; |
| 120 |
} |
| 121 |
} |
| 122 |
return $selected; |
| 123 |
} |
| 124 |
|
| 125 |
/** |
| 126 |
* Select the latest unresolved fixed-sink state for each operation. |
| 127 |
* |
| 128 |
* @param array<int, string> $lines |
| 129 |
* @return array<int, string> |
| 130 |
*/ |
| 131 |
private static function reservedDurableOperationLines(array $lines): array { |
| 132 |
$latestByOperation = array(); |
| 133 |
foreach ($lines as $line) { |
| 134 |
$record = json_decode($line, true); |
| 135 |
$operationKey = is_array($record) ? self::durableOperationKey($record) : ''; |
| 136 |
if ($operationKey !== '') { |
| 137 |
$latestByOperation[$operationKey] = array('line' => $line, 'record' => $record); |
| 138 |
} |
| 139 |
} |
| 140 |
$selected = array(); |
| 141 |
foreach ($latestByOperation as $latest) { |
| 142 |
$record = $latest['record'] ?? null; |
| 143 |
$line = $latest['line'] ?? null; |
| 144 |
if (is_array($record) && is_string($line) |
| 145 |
&& self::isReservedDurableOperation($record)) { |
| 146 |
$selected[] = $line; |
| 147 |
} |
| 148 |
} |
| 149 |
return $selected; |
| 150 |
} |
| 151 |
|
| 152 |
/** |
| 153 |
* Starts whose matching completion never reached disk. |
| 154 |
* |
| 155 |
* The start/end pairs to reserve are DERIVED from the decisive-record |
| 156 |
* manifest, not hardcoded here: every operation family the manifest marks |
| 157 |
* `reserve` (row-render, rate-limit backend/cache, option persistence, |
| 158 |
* option-hook callback) is preserved by the same request_id + operation_id |
| 159 |
* matching. Enrolling a new decisive record in the manifest extends this |
| 160 |
* reservation automatically, which is the structural fix for the recurring |
| 161 |
* "emitted but un-reserved" gap. The query timeline keeps its own start / |
| 162 |
* clear semantics (a summary with open_query === null cancels the reserve). |
| 163 |
* |
| 164 |
* @param array<int, string> $lines |
| 165 |
* @return array<int, string> |
| 166 |
*/ |
| 167 |
private static function unmatchedOperationLines(array $lines): array { |
| 168 |
return array_merge( |
| 169 |
self::unmatchedReservedStartLines($lines), |
| 170 |
self::unmatchedQueryLines($lines) |
| 171 |
); |
| 172 |
} |
| 173 |
|
| 174 |
/** |
| 175 |
* Reserved operation starts (from the manifest) whose matching end never |
| 176 |
* reached disk. Each family's start/end pair is matched by request_id + |
| 177 |
* operation_id; an end removes its start, so what remains is the hung set. |
| 178 |
* |
| 179 |
* @param array<int, string> $lines |
| 180 |
* @return array<int, string> |
| 181 |
*/ |
| 182 |
private static function unmatchedReservedStartLines(array $lines): array { |
| 183 |
$startEvents = array(); |
| 184 |
$endToStart = array(); |
| 185 |
foreach (ABJ_404_Solution_DecisiveRecordManifest::reservedOperationPairs() as $pair) { |
| 186 |
$startEvents[$pair['start']] = true; |
| 187 |
$endToStart[$pair['end']] = $pair['start']; |
| 188 |
} |
| 189 |
|
| 190 |
$openStarts = array(); |
| 191 |
foreach ($lines as $line) { |
| 192 |
$record = json_decode($line, true); |
| 193 |
if (!is_array($record)) { |
| 194 |
continue; |
| 195 |
} |
| 196 |
$event = is_scalar($record['event'] ?? null) ? (string)$record['event'] : ''; |
| 197 |
$isStart = isset($startEvents[$event]); |
| 198 |
if (!$isStart && !isset($endToStart[$event])) { |
| 199 |
continue; |
| 200 |
} |
| 201 |
$key = self::operationKey($record, $isStart ? $event : $endToStart[$event]); |
| 202 |
if ($key === '') { |
| 203 |
continue; |
| 204 |
} |
| 205 |
if ($isStart) { |
| 206 |
$openStarts[$key] = $line; |
| 207 |
} else { |
| 208 |
unset($openStarts[$key]); |
| 209 |
} |
| 210 |
} |
| 211 |
return array_values($openStarts); |
| 212 |
} |
| 213 |
|
| 214 |
/** |
| 215 |
* The reservation key for a start/end record: its start-event namespace |
| 216 |
* plus request_id and operation_id. Empty when either identifier is absent, |
| 217 |
* so an unattributable record is never reserved. |
| 218 |
* |
| 219 |
* @param array<mixed, mixed> $record |
| 220 |
*/ |
| 221 |
private static function operationKey(array $record, string $startEvent): string { |
| 222 |
$requestId = is_scalar($record['request_id'] ?? null) ? (string)$record['request_id'] : ''; |
| 223 |
$operationId = is_scalar($record['operation_id'] ?? null) ? (string)$record['operation_id'] : ''; |
| 224 |
if ($requestId === '' || $operationId === '') { |
| 225 |
return ''; |
| 226 |
} |
| 227 |
return $startEvent . '|' . $requestId . '|' . $operationId; |
| 228 |
} |
| 229 |
|
| 230 |
/** |
| 231 |
* The last query probe per request whose timeline never closed (no summary |
| 232 |
* with open_query === null). A killed worker mid-query leaves exactly this. |
| 233 |
* |
| 234 |
* @param array<int, string> $lines |
| 235 |
* @return array<int, string> |
| 236 |
*/ |
| 237 |
private static function unmatchedQueryLines(array $lines): array { |
| 238 |
$lastQueries = array(); |
| 239 |
foreach ($lines as $line) { |
| 240 |
$record = json_decode($line, true); |
| 241 |
if (!is_array($record)) { |
| 242 |
continue; |
| 243 |
} |
| 244 |
$requestId = is_scalar($record['request_id'] ?? null) |
| 245 |
? (string)$record['request_id'] : ''; |
| 246 |
$event = is_scalar($record['event'] ?? null) ? (string)$record['event'] : ''; |
| 247 |
if ($event === 'query_probe' && $requestId !== '') { |
| 248 |
$lastQueries[$requestId] = $line; |
| 249 |
} elseif ($event === 'query_timeline_summary' && $requestId !== '' |
| 250 |
&& ($record['open_query'] ?? null) === null) { |
| 251 |
unset($lastQueries[$requestId]); |
| 252 |
} |
| 253 |
} |
| 254 |
return array_values($lastQueries); |
| 255 |
} |
| 256 |
|
| 257 |
/** @param array<mixed, mixed> $record */ |
| 258 |
private static function activeOperationKey(array $record): string { |
| 259 |
if (($record['event'] ?? '') !== 'active_operation_breadcrumb') { |
| 260 |
return ''; |
| 261 |
} |
| 262 |
$requestId = is_scalar($record['request_id'] ?? null) |
| 263 |
? (string)$record['request_id'] : ''; |
| 264 |
$boundary = is_scalar($record['boundary'] ?? null) |
| 265 |
? (string)$record['boundary'] : ''; |
| 266 |
$state = is_scalar($record['state'] ?? null) ? (string)$record['state'] : ''; |
| 267 |
$manifest = self::activeBoundaryManifest(); |
| 268 |
if ($requestId === '' |
| 269 |
|| !array_key_exists($boundary, $manifest) |
| 270 |
|| !in_array($state, array('active', 'complete'), true)) { |
| 271 |
return ''; |
| 272 |
} |
| 273 |
return $requestId . '|' . $boundary; |
| 274 |
} |
| 275 |
|
| 276 |
/** @param array<mixed, mixed> $record */ |
| 277 |
private static function durableOperationKey(array $record): string { |
| 278 |
if (($record['event'] ?? '') !== 'durable_operation_state') { |
| 279 |
return ''; |
| 280 |
} |
| 281 |
$requestId = is_scalar($record['request_id'] ?? null) |
| 282 |
? (string)$record['request_id'] : ''; |
| 283 |
$checkpointId = is_scalar($record['operation_checkpoint_id'] ?? null) |
| 284 |
? (string)$record['operation_checkpoint_id'] : ''; |
| 285 |
return $requestId === '' || $checkpointId === '' |
| 286 |
? '' |
| 287 |
: $requestId . '|' . $checkpointId; |
| 288 |
} |
| 289 |
|
| 290 |
/** @param array<mixed, mixed> $record */ |
| 291 |
private static function isReservedActiveOperation(array $record): bool { |
| 292 |
if (($record['event'] ?? '') !== 'active_operation_breadcrumb' |
| 293 |
|| ($record['state'] ?? '') !== 'active') { |
| 294 |
return false; |
| 295 |
} |
| 296 |
$boundary = is_scalar($record['boundary'] ?? null) |
| 297 |
? (string)$record['boundary'] : ''; |
| 298 |
$manifest = self::activeBoundaryManifest(); |
| 299 |
$requiredFields = $manifest[$boundary]['required_evidence_fields'] ?? array(); |
| 300 |
return $requiredFields !== array() |
| 301 |
&& self::hasNonEmptyScalarKeys($record, $requiredFields); |
| 302 |
} |
| 303 |
|
| 304 |
/** @param array<mixed, mixed> $record */ |
| 305 |
private static function isReservedDurableOperation(array $record): bool { |
| 306 |
if (($record['event'] ?? '') !== 'durable_operation_state' |
| 307 |
|| !in_array($record['operation_state'] ?? '', array('intent', 'armed'), true)) { |
| 308 |
return false; |
| 309 |
} |
| 310 |
$operationEvent = is_scalar($record['operation_event'] ?? null) |
| 311 |
? (string)$record['operation_event'] : ''; |
| 312 |
if ($operationEvent === 'cache_metrics_probe_start') { |
| 313 |
return self::hasNonEmptyScalarKeys( |
| 314 |
$record, |
| 315 |
array('operation_id', 'source', 'phase', 'operation_checkpoint_id') |
| 316 |
); |
| 317 |
} |
| 318 |
if ($operationEvent !== 'active_operation_breadcrumb' |
| 319 |
|| ($record['state'] ?? '') !== 'active') { |
| 320 |
return false; |
| 321 |
} |
| 322 |
$boundary = is_scalar($record['boundary'] ?? null) |
| 323 |
? (string)$record['boundary'] : ''; |
| 324 |
$requiredFields = self::activeBoundaryManifest()[$boundary]['required_evidence_fields'] |
| 325 |
?? array(); |
| 326 |
return $requiredFields !== array() |
| 327 |
&& self::hasNonEmptyScalarKeys($record, $requiredFields) |
| 328 |
&& self::hasNonEmptyScalarKeys($record, array('operation_checkpoint_id')); |
| 329 |
} |
| 330 |
|
| 331 |
/** |
| 332 |
* Missing diagnostics files must degrade to no reserved active evidence |
| 333 |
* instead of breaking the support-request path on a corrupt install. |
| 334 |
* |
| 335 |
* @return array<string, array{ |
| 336 |
* fields: array<int, string>, |
| 337 |
* required_evidence_fields: array<int, string> |
| 338 |
* }> |
| 339 |
*/ |
| 340 |
private static function activeBoundaryManifest(): array { |
| 341 |
return class_exists('ABJ_404_Solution_ActiveOperationBoundaryManifest') |
| 342 |
? ABJ_404_Solution_ActiveOperationBoundaryManifest::boundaries() |
| 343 |
: array(); |
| 344 |
} |
| 345 |
|
| 346 |
/** |
| 347 |
* @param array<mixed, mixed> $record |
| 348 |
* @param array<int, string> $fields |
| 349 |
*/ |
| 350 |
private static function hasNonEmptyScalarKeys(array $record, array $fields): bool { |
| 351 |
foreach ($fields as $field) { |
| 352 |
$value = $record[$field] ?? null; |
| 353 |
if (!is_scalar($value) || (string)$value === '') { |
| 354 |
return false; |
| 355 |
} |
| 356 |
} |
| 357 |
return true; |
| 358 |
} |
| 359 |
} |
| 360 |
|
