| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Joins pre-aggregated wp_abj404_logs_hits rollup data (logsid, logshits, |
| 9 |
* last_used) onto a list of result rows by canonical URL. |
| 10 |
* |
| 11 |
* Real URLs arrive in many shapes (leading slash / no slash, trailing slash, |
| 12 |
* raw / esc_url_raw'd, with or without query/fragment) so a direct equality |
| 13 |
* lookup misses too often. This populator builds a small set of URL variants |
| 14 |
* per row, queries the hits table in chunked IN() batches, then merges |
| 15 |
* by canonical form. |
| 16 |
* |
| 17 |
* Extracted from LogsRepository under M201. Consumed by the LogsRepository |
| 18 |
* facade, which also passes itself (as LogsRepositoryInterface) so that any |
| 19 |
* test-time override of logsHitsTableExists / scheduleHitsTableRebuild on |
| 20 |
* the facade (or a DataAccess fake DAO bridge) propagates into this |
| 21 |
* populator's existence-check and missing-table-rebuild-scheduling paths. |
| 22 |
*/ |
| 23 |
class ABJ_404_Solution_LogsHitsDataPopulator { |
| 24 |
|
| 25 |
/** @var ABJ_404_Solution_DatabaseCore */ |
| 26 |
private $dbCore; |
| 27 |
|
| 28 |
public function __construct(ABJ_404_Solution_DatabaseCore $dbCore) { |
| 29 |
$this->dbCore = $dbCore; |
| 30 |
} |
| 31 |
|
| 32 |
/** |
| 33 |
* Populate logshits / logsid / last_used on each row from the |
| 34 |
* wp_abj404_logs_hits rollup. |
| 35 |
* |
| 36 |
* @param array<int, array<string, mixed>> $rows |
| 37 |
* @param ABJ_404_Solution_LogsRepositoryInterface $repo Source for logsHitsTableExists / scheduleHitsTableRebuild; passing the facade (rather than the raw rollup) ensures subclass overrides propagate. |
| 38 |
* @return array<int, array<string, mixed>> |
| 39 |
*/ |
| 40 |
public function populate($rows, ABJ_404_Solution_LogsRepositoryInterface $repo) { |
| 41 |
if (empty($rows)) { |
| 42 |
return $rows; |
| 43 |
} |
| 44 |
|
| 45 |
$urls = array(); |
| 46 |
foreach ($rows as $row) { |
| 47 |
if ($row['url'] != null && !empty($row['url'])) { |
| 48 |
$variants = $this->buildHitsLookupUrlVariants($row['url']); |
| 49 |
foreach ($variants as $variant) { |
| 50 |
$urls[] = $variant; |
| 51 |
} |
| 52 |
} |
| 53 |
} |
| 54 |
|
| 55 |
if (empty($urls)) { |
| 56 |
return $rows; |
| 57 |
} |
| 58 |
|
| 59 |
$urls = array_values(array_unique($urls)); |
| 60 |
|
| 61 |
if (!$repo->logsHitsTableExists()) { |
| 62 |
$repo->scheduleHitsTableRebuild(); |
| 63 |
return $rows; |
| 64 |
} |
| 65 |
|
| 66 |
$logsHitsTable = $this->dbCore->doTableNameReplacements('{wp_abj404_logs_hits}'); |
| 67 |
$batchSize = 200; |
| 68 |
$logsResults = array(); |
| 69 |
$urlChunks = array_chunk($urls, $batchSize); |
| 70 |
|
| 71 |
foreach ($urlChunks as $urlChunk) { |
| 72 |
$placeholders = implode(',', array_fill(0, count($urlChunk), '%s')); |
| 73 |
$sql = "SELECT requested_url, logsid, last_used, logshits " |
| 74 |
. "FROM {$logsHitsTable} " |
| 75 |
. "WHERE BINARY requested_url IN ($placeholders)"; |
| 76 |
|
| 77 |
$chunkResult = $this->dbCore->queryAndGetResults($sql, array( |
| 78 |
'query_params' => $urlChunk, |
| 79 |
'log_too_slow' => false, |
| 80 |
)); |
| 81 |
|
| 82 |
if (!empty($chunkResult['timed_out']) || |
| 83 |
(isset($chunkResult['last_error']) && $chunkResult['last_error'] != '')) { |
| 84 |
$errRaw = $chunkResult['last_error'] ?? ''; |
| 85 |
$err = is_string($errRaw) ? $errRaw : ''; |
| 86 |
if ($err !== '' && strpos($err, 'logs_hits') !== false) { |
| 87 |
$repo->scheduleHitsTableRebuild(); |
| 88 |
} |
| 89 |
return $rows; |
| 90 |
} |
| 91 |
|
| 92 |
$chunkResults = is_array($chunkResult['rows'] ?? null) ? $chunkResult['rows'] : array(); |
| 93 |
if (!empty($chunkResults)) { |
| 94 |
$logsResults = array_merge($logsResults, $chunkResults); |
| 95 |
} |
| 96 |
} |
| 97 |
|
| 98 |
$logsDataByUrl = array(); |
| 99 |
foreach ($logsResults as $logRow) { |
| 100 |
$canonicalUrl = $this->canonicalizeUrlForHitsMatch($logRow['requested_url'] ?? ''); |
| 101 |
if ($canonicalUrl === '') { |
| 102 |
continue; |
| 103 |
} |
| 104 |
if (!isset($logsDataByUrl[$canonicalUrl])) { |
| 105 |
$logsDataByUrl[$canonicalUrl] = array( |
| 106 |
'logsid' => (int)($logRow['logsid'] ?? 0), |
| 107 |
'logshits' => (int)($logRow['logshits'] ?? 0), |
| 108 |
'last_used' => (int)($logRow['last_used'] ?? 0), |
| 109 |
); |
| 110 |
continue; |
| 111 |
} |
| 112 |
$existing = $logsDataByUrl[$canonicalUrl]; |
| 113 |
$currentLogsid = (int)($logRow['logsid'] ?? 0); |
| 114 |
$existingLogsid = (int)$existing['logsid']; |
| 115 |
$logsDataByUrl[$canonicalUrl]['logsid'] = ($existingLogsid > 0 && $currentLogsid > 0) |
| 116 |
? min($existingLogsid, $currentLogsid) |
| 117 |
: max($existingLogsid, $currentLogsid); |
| 118 |
$logsDataByUrl[$canonicalUrl]['logshits'] = (int)$existing['logshits'] + (int)($logRow['logshits'] ?? 0); |
| 119 |
$logsDataByUrl[$canonicalUrl]['last_used'] = max((int)$existing['last_used'], (int)($logRow['last_used'] ?? 0)); |
| 120 |
} |
| 121 |
|
| 122 |
foreach ($rows as &$row) { |
| 123 |
if ($row['url'] != null && !empty($row['url'])) { |
| 124 |
$canonicalUrl = $this->canonicalizeUrlForHitsMatch($row['url']); |
| 125 |
if (isset($logsDataByUrl[$canonicalUrl])) { |
| 126 |
$logData = $logsDataByUrl[$canonicalUrl]; |
| 127 |
$row['logsid'] = $logData['logsid']; |
| 128 |
$row['logshits'] = $logData['logshits']; |
| 129 |
$row['last_used'] = $logData['last_used']; |
| 130 |
} |
| 131 |
} |
| 132 |
} |
| 133 |
|
| 134 |
return $rows; |
| 135 |
} |
| 136 |
|
| 137 |
/** @param mixed $url @return string */ |
| 138 |
private function canonicalizeUrlForHitsMatch($url): string { |
| 139 |
if (!is_string($url)) { |
| 140 |
return ''; |
| 141 |
} |
| 142 |
$url = trim($url); |
| 143 |
if ($url === '') { |
| 144 |
return ''; |
| 145 |
} |
| 146 |
$fragment = ''; |
| 147 |
$fragmentPos = strpos($url, '#'); |
| 148 |
if ($fragmentPos !== false) { |
| 149 |
$fragment = substr($url, $fragmentPos); |
| 150 |
$url = substr($url, 0, $fragmentPos); |
| 151 |
} |
| 152 |
$query = ''; |
| 153 |
$queryPos = strpos($url, '?'); |
| 154 |
if ($queryPos !== false) { |
| 155 |
$query = substr($url, $queryPos); |
| 156 |
$url = substr($url, 0, $queryPos); |
| 157 |
} |
| 158 |
$path = trim($url, '/'); |
| 159 |
$normalizedPath = ($path === '') ? '/' : '/' . $path; |
| 160 |
return $normalizedPath . $query . $fragment; |
| 161 |
} |
| 162 |
|
| 163 |
/** @param mixed $url @return array<int, string> */ |
| 164 |
private function buildHitsLookupUrlVariants($url) { |
| 165 |
$variants = array(); |
| 166 |
if (!is_string($url)) { |
| 167 |
return $variants; |
| 168 |
} |
| 169 |
$raw = trim($url); |
| 170 |
if ($raw !== '') { |
| 171 |
$variants[] = $raw; |
| 172 |
} |
| 173 |
$canonical = $this->canonicalizeUrlForHitsMatch($url); |
| 174 |
if ($canonical !== '') { |
| 175 |
$variants[] = $canonical; |
| 176 |
$parts = $this->splitCanonicalHitsUrl($canonical); |
| 177 |
$pathPart = $parts['path']; |
| 178 |
$suffixPart = $parts['suffix']; |
| 179 |
$pathVariants = array($pathPart); |
| 180 |
$noLeadingPath = ltrim($pathPart, '/'); |
| 181 |
if ($noLeadingPath !== '') { |
| 182 |
$pathVariants[] = $noLeadingPath; |
| 183 |
} |
| 184 |
if ($pathPart !== '/') { |
| 185 |
if (substr($pathPart, -1) === '/') { |
| 186 |
$toggleTrailingPath = rtrim($pathPart, '/'); |
| 187 |
} else { |
| 188 |
$toggleTrailingPath = $pathPart . '/'; |
| 189 |
} |
| 190 |
$pathVariants[] = $toggleTrailingPath; |
| 191 |
$toggleNoLeadingPath = ltrim($toggleTrailingPath, '/'); |
| 192 |
if ($toggleNoLeadingPath !== '') { |
| 193 |
$pathVariants[] = $toggleNoLeadingPath; |
| 194 |
} |
| 195 |
} |
| 196 |
foreach (array_unique($pathVariants) as $pathVariant) { |
| 197 |
$variants[] = $pathVariant . $suffixPart; |
| 198 |
} |
| 199 |
} |
| 200 |
return array_values(array_unique($variants)); |
| 201 |
} |
| 202 |
|
| 203 |
/** @return array{path: string, suffix: string} */ |
| 204 |
private function splitCanonicalHitsUrl(string $canonicalUrl): array { |
| 205 |
$firstQueryPos = strpos($canonicalUrl, '?'); |
| 206 |
$firstFragmentPos = strpos($canonicalUrl, '#'); |
| 207 |
if ($firstQueryPos === false && $firstFragmentPos === false) { |
| 208 |
return array('path' => $canonicalUrl, 'suffix' => ''); |
| 209 |
} |
| 210 |
if ($firstQueryPos === false) { |
| 211 |
$splitPos = $firstFragmentPos; |
| 212 |
} elseif ($firstFragmentPos === false) { |
| 213 |
$splitPos = $firstQueryPos; |
| 214 |
} else { |
| 215 |
$splitPos = min($firstQueryPos, $firstFragmentPos); |
| 216 |
} |
| 217 |
return array( |
| 218 |
'path' => substr($canonicalUrl, 0, $splitPos), |
| 219 |
'suffix' => substr($canonicalUrl, $splitPos), |
| 220 |
); |
| 221 |
} |
| 222 |
} |
| 223 |
|
