| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
require_once __DIR__ . '/GscConfig.php'; |
| 8 |
require_once __DIR__ . '/GscFetchLock.php'; |
| 9 |
|
| 10 |
/** |
| 11 |
* Owns Google Search Console Search Analytics requests, cache writes, fetch |
| 12 |
* locks, and background refresh scheduling. |
| 13 |
*/ |
| 14 |
class ABJ_404_Solution_GscSearchAnalyticsClient { |
| 15 |
|
| 16 |
/** |
| 17 |
* Recency window scanned by the GSC URL probe (rows from logsv2). |
| 18 |
* Made explicit at the call site so the cap is visible here, not buried in SQL. |
| 19 |
*/ |
| 20 |
const GSC_URL_PROBE_RECENT_LOG_WINDOW = 5000; |
| 21 |
|
| 22 |
/** Max distinct URLs the GSC URL probe pulls per fetch. */ |
| 23 |
const GSC_URL_PROBE_DISTINCT_URL_CAP = 500; |
| 24 |
|
| 25 |
/** @var ABJ_404_Solution_Logging */ |
| 26 |
private $logger; |
| 27 |
|
| 28 |
/** @var ABJ_404_Solution_GscOAuthTokenStore */ |
| 29 |
private $oauthStore; |
| 30 |
|
| 31 |
/** @var ABJ_404_Solution_GscFetchLock */ |
| 32 |
private $fetchLock; |
| 33 |
|
| 34 |
/** @param ABJ_404_Solution_Logging $logger */ |
| 35 |
public function __construct($logger, ABJ_404_Solution_GscOAuthTokenStore $oauthStore) { |
| 36 |
$this->logger = $logger; |
| 37 |
$this->oauthStore = $oauthStore; |
| 38 |
$this->fetchLock = new ABJ_404_Solution_GscFetchLock($logger); |
| 39 |
} |
| 40 |
|
| 41 |
/** |
| 42 |
* Fetch search analytics data for a list of URLs. |
| 43 |
* |
| 44 |
* @param string[] $urls Relative or absolute URLs to query. |
| 45 |
* @param int $days Number of days to look back. |
| 46 |
* @return array<int, array<string, mixed>> |
| 47 |
*/ |
| 48 |
public function getSearchAnalyticsForUrls(array $urls, int $days = 90): array { |
| 49 |
if (!$this->oauthStore->isAuthorized() || empty($urls)) { |
| 50 |
return array(); |
| 51 |
} |
| 52 |
|
| 53 |
$cached = get_transient(ABJ_404_Solution_GscConfig::TRANSIENT_KEY); |
| 54 |
$cachedRows = $this->normalizeRows($cached); |
| 55 |
if ($cachedRows !== false) { |
| 56 |
return $cachedRows; |
| 57 |
} |
| 58 |
|
| 59 |
$fetchResult = $this->doFetchFromApi($urls, $days); |
| 60 |
$allRows = $fetchResult['rows']; |
| 61 |
// allow-cache-empty: empty GSC result sets are valid recent fetches and drive the explicit no-data UI state. |
| 62 |
set_transient(ABJ_404_Solution_GscConfig::TRANSIENT_KEY, $allRows, ABJ_404_Solution_GscConfig::TRANSIENT_TTL); |
| 63 |
update_option(ABJ_404_Solution_GscConfig::LAST_FETCH_OPTION_KEY, abj_clock()->now(), false); |
| 64 |
return $allRows; |
| 65 |
} |
| 66 |
|
| 67 |
/** |
| 68 |
* Fetch GSC data and cache it. Called by cron and background refresh. |
| 69 |
* |
| 70 |
* @return void |
| 71 |
*/ |
| 72 |
public function fetchAndCacheGscData(): void { |
| 73 |
if (!$this->oauthStore->isAuthorized()) { |
| 74 |
return; |
| 75 |
} |
| 76 |
|
| 77 |
if (!$this->fetchLock->claim()) { |
| 78 |
return; |
| 79 |
} |
| 80 |
|
| 81 |
try { |
| 82 |
$urls = $this->getUrlsToQuery(); |
| 83 |
$fetchResult = $this->doFetchFromApi($urls); |
| 84 |
if (!$fetchResult['completed']) { |
| 85 |
return; |
| 86 |
} |
| 87 |
$allRows = $fetchResult['rows']; |
| 88 |
// allow-cache-empty: empty GSC result sets are valid recent fetches and drive the explicit no-data UI state. |
| 89 |
set_transient(ABJ_404_Solution_GscConfig::TRANSIENT_KEY, $allRows, ABJ_404_Solution_GscConfig::TRANSIENT_TTL); |
| 90 |
update_option(ABJ_404_Solution_GscConfig::LAST_FETCH_OPTION_KEY, abj_clock()->now(), false); |
| 91 |
} finally { |
| 92 |
$this->fetchLock->release(); |
| 93 |
} |
| 94 |
} |
| 95 |
|
| 96 |
/** |
| 97 |
* Get the list of 404 URLs to query from the logs table. |
| 98 |
* |
| 99 |
* @return string[] |
| 100 |
*/ |
| 101 |
protected function getUrlsToQuery(): array { |
| 102 |
$logsRepo = abj_service('logs_repository'); |
| 103 |
return $logsRepo->getDistinctLoggedUrls( |
| 104 |
self::GSC_URL_PROBE_RECENT_LOG_WINDOW, |
| 105 |
self::GSC_URL_PROBE_DISTINCT_URL_CAP |
| 106 |
); |
| 107 |
} |
| 108 |
|
| 109 |
/** |
| 110 |
* Return cached GSC data, or false if the cache is empty. |
| 111 |
* |
| 112 |
* @return array<int, array<string, mixed>>|false |
| 113 |
*/ |
| 114 |
public function getCachedData() { |
| 115 |
$cached = get_transient(ABJ_404_Solution_GscConfig::TRANSIENT_KEY); |
| 116 |
return $this->normalizeRows($cached); |
| 117 |
} |
| 118 |
|
| 119 |
/** |
| 120 |
* Whether a background refresh should be triggered. |
| 121 |
* |
| 122 |
* @return bool |
| 123 |
*/ |
| 124 |
public function isRefreshNeeded(): bool { |
| 125 |
$lastFetch = get_option(ABJ_404_Solution_GscConfig::LAST_FETCH_OPTION_KEY, 0); |
| 126 |
$lastFetchTime = is_numeric($lastFetch) ? (int)$lastFetch : 0; |
| 127 |
return (abj_clock()->now() - $lastFetchTime) > ABJ_404_Solution_GscConfig::STALE_THRESHOLD; |
| 128 |
} |
| 129 |
|
| 130 |
/** |
| 131 |
* Schedule an immediate single-event background refresh via WP-Cron. |
| 132 |
* |
| 133 |
* @return void |
| 134 |
*/ |
| 135 |
public function scheduleBackgroundRefresh(): void { |
| 136 |
$this->fetchLock->initializeAtomicLockMigrationState(); |
| 137 |
if ($this->fetchLock->isHeld()) { |
| 138 |
return; |
| 139 |
} |
| 140 |
abj_cron_scheduler()->scheduleSingleIfMissing( |
| 141 |
ABJ_404_Solution_GscConfig::BACKGROUND_REFRESH_HOOK |
| 142 |
); |
| 143 |
} |
| 144 |
|
| 145 |
/** |
| 146 |
* Fetch top 404 URLs that also have GSC search traffic. |
| 147 |
* |
| 148 |
* @param string[] $capturedUrls Array of captured 404 URL strings. |
| 149 |
* @param int $days Number of days for GSC data. |
| 150 |
* @return array<int, array<string, mixed>> |
| 151 |
*/ |
| 152 |
public function getTrafficDataForCaptured404s(array $capturedUrls, int $days = 90): array { |
| 153 |
if (empty($capturedUrls)) { |
| 154 |
return array(); |
| 155 |
} |
| 156 |
$data = $this->getSearchAnalyticsForUrls($capturedUrls, $days); |
| 157 |
return array_values(array_filter($data, function ($row) { |
| 158 |
return isset($row['clicks']) && is_numeric($row['clicks']) && (int)$row['clicks'] > 0; |
| 159 |
})); |
| 160 |
} |
| 161 |
|
| 162 |
/** |
| 163 |
* Query the GSC Search Analytics API for each URL individually. |
| 164 |
* |
| 165 |
* @param string[] $urls Relative or absolute URLs to query. |
| 166 |
* @param int $days Number of days to look back. |
| 167 |
* @return array{rows: array<int, array<string, mixed>>, completed: bool} |
| 168 |
*/ |
| 169 |
private function doFetchFromApi(array $urls, int $days = 90): array { |
| 170 |
$s = $this->oauthStore->getSettings(); |
| 171 |
$token = get_option(ABJ_404_Solution_GscConfig::TOKEN_OPTION_KEY, false); |
| 172 |
$accessToken = $this->tokenAccessToken($token); |
| 173 |
if ($accessToken === '') { |
| 174 |
return array('rows' => array(), 'completed' => true); |
| 175 |
} |
| 176 |
|
| 177 |
$siteUrl = $s['site_url']; |
| 178 |
$now = abj_clock()->now(); |
| 179 |
$endDayIndex = intdiv($now, 86400); |
| 180 |
$startDayIndex = $endDayIndex - $days; |
| 181 |
$endDate = gmdate('Y-m-d', $endDayIndex * 86400); |
| 182 |
$startDate = gmdate('Y-m-d', $startDayIndex * 86400); |
| 183 |
|
| 184 |
$urls = array_slice($urls, 0, 500); |
| 185 |
$allRows = array(); |
| 186 |
|
| 187 |
foreach ($urls as $url) { |
| 188 |
if (!$this->fetchLock->renewIfDue()) { |
| 189 |
return array('rows' => $allRows, 'completed' => false); |
| 190 |
} |
| 191 |
$absoluteUrl = (strpos($url, 'http') === 0) ? $url : rtrim(home_url('/'), '/') . '/' . ltrim($url, '/'); |
| 192 |
$body = array( |
| 193 |
'startDate' => $startDate, |
| 194 |
'endDate' => $endDate, |
| 195 |
'dimensions' => array('page'), |
| 196 |
'dimensionFilterGroups' => array( |
| 197 |
array( |
| 198 |
'filters' => array( |
| 199 |
array( |
| 200 |
'dimension' => 'page', |
| 201 |
'operator' => 'equals', |
| 202 |
'expression' => $absoluteUrl, |
| 203 |
), |
| 204 |
), |
| 205 |
), |
| 206 |
), |
| 207 |
'rowLimit' => 1000, |
| 208 |
); |
| 209 |
|
| 210 |
$encodedSiteUrl = urlencode($siteUrl); |
| 211 |
$response = wp_remote_post( |
| 212 |
ABJ_404_Solution_GscConfig::API_BASE_URL . "/sites/{$encodedSiteUrl}/searchAnalytics/query", |
| 213 |
array( |
| 214 |
'headers' => array( |
| 215 |
'Authorization' => 'Bearer ' . $accessToken, |
| 216 |
'Content-Type' => 'application/json', |
| 217 |
), |
| 218 |
'body' => (string)wp_json_encode($body), |
| 219 |
'timeout' => 20, |
| 220 |
) |
| 221 |
); |
| 222 |
|
| 223 |
if (is_wp_error($response)) { |
| 224 |
$this->logger->warn('GSC API transport error: ' . $response->get_error_message()); |
| 225 |
break; |
| 226 |
} |
| 227 |
|
| 228 |
$httpCode = (int) wp_remote_retrieve_response_code($response); |
| 229 |
if ($httpCode !== 200) { |
| 230 |
$this->logger->warn('GSC API returned HTTP ' . $httpCode . ': ' . wp_remote_retrieve_body($response)); |
| 231 |
break; |
| 232 |
} |
| 233 |
|
| 234 |
$data = json_decode(wp_remote_retrieve_body($response), true); |
| 235 |
if (!is_array($data) || empty($data['rows']) || !is_array($data['rows'])) { |
| 236 |
continue; |
| 237 |
} |
| 238 |
|
| 239 |
foreach ($data['rows'] as $row) { |
| 240 |
if (!is_array($row)) { |
| 241 |
continue; |
| 242 |
} |
| 243 |
$allRows[] = $this->normalizeApiRow($row); |
| 244 |
} |
| 245 |
} |
| 246 |
|
| 247 |
usort($allRows, function ($a, $b) { |
| 248 |
return $b['clicks'] - $a['clicks']; |
| 249 |
}); |
| 250 |
|
| 251 |
return array('rows' => $allRows, 'completed' => true); |
| 252 |
} |
| 253 |
|
| 254 |
/** |
| 255 |
* @param mixed $cached |
| 256 |
* @return array<int, array<string, mixed>>|false |
| 257 |
*/ |
| 258 |
private function normalizeRows($cached) { |
| 259 |
if (!is_array($cached)) { |
| 260 |
return false; |
| 261 |
} |
| 262 |
$rows = array(); |
| 263 |
foreach ($cached as $row) { |
| 264 |
if (is_array($row)) { |
| 265 |
$normalized = array(); |
| 266 |
foreach ($row as $key => $value) { |
| 267 |
if (is_string($key)) { |
| 268 |
$normalized[$key] = $value; |
| 269 |
} |
| 270 |
} |
| 271 |
$rows[] = $normalized; |
| 272 |
} |
| 273 |
} |
| 274 |
return $rows; |
| 275 |
} |
| 276 |
|
| 277 |
/** @param mixed $token */ |
| 278 |
private function tokenAccessToken($token): string { |
| 279 |
if (!is_array($token)) { |
| 280 |
return ''; |
| 281 |
} |
| 282 |
$accessToken = $token['access_token'] ?? ''; |
| 283 |
return is_scalar($accessToken) ? (string)$accessToken : ''; |
| 284 |
} |
| 285 |
|
| 286 |
/** |
| 287 |
* @param array<mixed, mixed> $row |
| 288 |
* @return array{url: string, clicks: int, impressions: int, position: float} |
| 289 |
*/ |
| 290 |
private function normalizeApiRow(array $row): array { |
| 291 |
return array( |
| 292 |
'url' => $this->rowUrl($row), |
| 293 |
'clicks' => $this->rowInt($row, 'clicks'), |
| 294 |
'impressions' => $this->rowInt($row, 'impressions'), |
| 295 |
'position' => round($this->rowFloat($row, 'position'), 1), |
| 296 |
); |
| 297 |
} |
| 298 |
|
| 299 |
/** @param array<mixed, mixed> $row */ |
| 300 |
private function rowUrl(array $row): string { |
| 301 |
$keys = $row['keys'] ?? array(); |
| 302 |
if (!is_array($keys)) { |
| 303 |
return ''; |
| 304 |
} |
| 305 |
$url = $keys[0] ?? ''; |
| 306 |
return is_scalar($url) ? (string)$url : ''; |
| 307 |
} |
| 308 |
|
| 309 |
/** @param array<mixed, mixed> $row */ |
| 310 |
private function rowInt(array $row, string $key): int { |
| 311 |
$value = $row[$key] ?? 0; |
| 312 |
return is_numeric($value) ? (int)$value : 0; |
| 313 |
} |
| 314 |
|
| 315 |
/** @param array<mixed, mixed> $row */ |
| 316 |
private function rowFloat(array $row, string $key): float { |
| 317 |
$value = $row[$key] ?? 0.0; |
| 318 |
return is_numeric($value) ? (float)$value : 0.0; |
| 319 |
} |
| 320 |
} |
| 321 |
|
