添加whatshub工单
This commit is contained in:
root
Regular → Executable
+107
-5
@@ -67,22 +67,36 @@ abstract class AbstractScrmSpider implements ScrmSpiderInterface
|
||||
$apiUrlsToIntercept[] = $countApi;
|
||||
}
|
||||
|
||||
$antiBot = $config['antiBot'] ?? null;
|
||||
|
||||
$initResult = $this->requestNode('/api/auth-and-intercept', [
|
||||
'pageUrl' => $config['pageUrl'],
|
||||
'apiUrls' => $apiUrlsToIntercept,
|
||||
'authActions' => $config['authActions'] ?? [],
|
||||
'antiBot' => $antiBot,
|
||||
]);
|
||||
|
||||
if (empty($initResult['success'])) {
|
||||
$cfCode = (string) ($initResult['code'] ?? '');
|
||||
SplitTicketSyncLogger::log('spider', 'auth-and-intercept failed', [
|
||||
'error' => $initResult['error'] ?? '未知',
|
||||
'error' => $initResult['error'] ?? '未知',
|
||||
'code' => $cfCode !== '' ? $cfCode : null,
|
||||
'challengeType' => $initResult['challengeType'] ?? null,
|
||||
'stage' => $initResult['stage'] ?? null,
|
||||
'cf' => $initResult['cf'] ?? null,
|
||||
]);
|
||||
throw new Exception('初始化失败: ' . ($initResult['error'] ?? '未知'));
|
||||
$errMsg = (string) ($initResult['error'] ?? '未知');
|
||||
if ($cfCode === 'CF_TURNSTILE_FAILED') {
|
||||
throw new Exception('Cloudflare Turnstile 验证失败: ' . ($initResult['stage'] ?? 'timeout'));
|
||||
}
|
||||
throw new Exception('初始化失败: ' . $errMsg);
|
||||
}
|
||||
|
||||
$interceptedApis = $initResult['interceptedApis'];
|
||||
$finalPageUrl = (string) ($initResult['finalPageUrl'] ?? ($config['pageUrl'] ?? ''));
|
||||
SplitTicketSyncLogger::log('spider', 'auth-and-intercept ok', [
|
||||
'intercepted' => array_keys($interceptedApis),
|
||||
'intercepted' => array_keys($interceptedApis),
|
||||
'finalPageUrl' => $finalPageUrl,
|
||||
]);
|
||||
$cookies = $initResult['cookies'];
|
||||
|
||||
@@ -121,6 +135,7 @@ abstract class AbstractScrmSpider implements ScrmSpiderInterface
|
||||
'method' => $config['listMethod'] ?? 'GET',
|
||||
]],
|
||||
'cookies' => $cookies,
|
||||
'antiBot' => $antiBot,
|
||||
], 120);
|
||||
|
||||
if (!empty($fetchResult['success'])) {
|
||||
@@ -138,12 +153,14 @@ abstract class AbstractScrmSpider implements ScrmSpiderInterface
|
||||
$uiResult = $this->requestNode('/api/ui-pagination', [
|
||||
'apiUrl' => $listApi,
|
||||
'pageUrl' => $config['pageUrl'],
|
||||
'finalPageUrl' => $finalPageUrl,
|
||||
'nextBtnSelector' => $uiConfig['nextBtnSelector'] ?? '',
|
||||
'waitMs' => $uiConfig['waitMs'] ?? 2000,
|
||||
'clicksToPerform' => $clicksToPerform,
|
||||
'cookies' => $cookies,
|
||||
'firstPageData' => $firstPageData,
|
||||
'authActions' => $config['authActions'] ?? [],
|
||||
'antiBot' => $antiBot,
|
||||
], 1200);
|
||||
|
||||
if (!empty($uiResult['success']) && !empty($uiResult['data'])) {
|
||||
@@ -170,13 +187,43 @@ abstract class AbstractScrmSpider implements ScrmSpiderInterface
|
||||
* @return array<string, mixed>
|
||||
*/
|
||||
protected function requestNode(string $endpoint, array $payload, int $timeout = 60): array
|
||||
{
|
||||
$maxAttempts = 3;
|
||||
$lastDecoded = [];
|
||||
for ($attempt = 1; $attempt <= $maxAttempts; $attempt++) {
|
||||
$lastDecoded = $this->doRequestNode($endpoint, $payload, $timeout, $attempt);
|
||||
if (!empty($lastDecoded['success'])) {
|
||||
return $lastDecoded;
|
||||
}
|
||||
$error = (string) ($lastDecoded['error'] ?? '');
|
||||
if (!$this->shouldRetryNodeResponse($lastDecoded, $error) || $attempt >= $maxAttempts) {
|
||||
return $lastDecoded;
|
||||
}
|
||||
$delaySeconds = $attempt;
|
||||
SplitTicketSyncLogger::log('node_retry', 'retry ' . $endpoint, [
|
||||
'attempt' => $attempt + 1,
|
||||
'error' => $error,
|
||||
'delay' => $delaySeconds,
|
||||
]);
|
||||
sleep($delaySeconds);
|
||||
}
|
||||
return $lastDecoded;
|
||||
}
|
||||
|
||||
/**
|
||||
* @param array<string, mixed> $payload
|
||||
* @return array<string, mixed>
|
||||
*/
|
||||
private function doRequestNode(string $endpoint, array $payload, int $timeout, int $attempt): array
|
||||
{
|
||||
$url = $this->nodeHost . $endpoint;
|
||||
$started = microtime(true);
|
||||
$logPayload = SplitTicketSyncLogger::isEnabled() ? $payload : self::summarizeNodePayload($payload);
|
||||
SplitTicketSyncLogger::log('node_request', 'POST ' . $endpoint, [
|
||||
'url' => $url,
|
||||
'timeout' => $timeout,
|
||||
'payload' => $payload,
|
||||
'attempt' => $attempt,
|
||||
'payload' => $logPayload,
|
||||
]);
|
||||
|
||||
$ch = curl_init($url);
|
||||
@@ -184,6 +231,7 @@ abstract class AbstractScrmSpider implements ScrmSpiderInterface
|
||||
curl_setopt($ch, CURLOPT_POST, true);
|
||||
curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($payload, JSON_UNESCAPED_UNICODE));
|
||||
curl_setopt($ch, CURLOPT_HTTPHEADER, ['Content-Type: application/json']);
|
||||
curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 5);
|
||||
curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
|
||||
$response = curl_exec($ch);
|
||||
$httpCode = (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
|
||||
@@ -194,6 +242,7 @@ abstract class AbstractScrmSpider implements ScrmSpiderInterface
|
||||
SplitTicketSyncLogger::log('node_response', 'curl error on ' . $endpoint, [
|
||||
'httpCode' => $httpCode,
|
||||
'elapsedMs' => $elapsedMs,
|
||||
'attempt' => $attempt,
|
||||
'error' => $err,
|
||||
]);
|
||||
throw new Exception($err);
|
||||
@@ -204,9 +253,62 @@ abstract class AbstractScrmSpider implements ScrmSpiderInterface
|
||||
SplitTicketSyncLogger::log('node_response', 'POST ' . $endpoint, array_merge([
|
||||
'httpCode' => $httpCode,
|
||||
'elapsedMs' => $elapsedMs,
|
||||
'attempt' => $attempt,
|
||||
'responseSize' => strlen((string) $response),
|
||||
], $summary));
|
||||
return is_array($decoded) ? $decoded : [];
|
||||
if (!is_array($decoded)) {
|
||||
return ['success' => false, 'error' => 'Node 返回非 JSON', 'httpCode' => $httpCode];
|
||||
}
|
||||
if ($httpCode === 503) {
|
||||
$decoded['success'] = false;
|
||||
$decoded['error'] = (string) ($decoded['error'] ?? '服务繁忙');
|
||||
$decoded['httpCode'] = 503;
|
||||
}
|
||||
return $decoded;
|
||||
}
|
||||
|
||||
/**
|
||||
* @param array<string, mixed> $decoded
|
||||
*/
|
||||
private function shouldRetryNodeResponse(array $decoded, string $error): bool
|
||||
{
|
||||
if ((int) ($decoded['httpCode'] ?? 0) === 503) {
|
||||
return true;
|
||||
}
|
||||
$lower = mb_strtolower($error, 'UTF-8');
|
||||
foreach (['排队超时', 'queue_timeout', 'timeout', 'net::', 'navigation', 'connection'] as $needle) {
|
||||
if ($needle !== '' && mb_strpos($lower, mb_strtolower($needle, 'UTF-8')) !== false) {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
return false;
|
||||
}
|
||||
|
||||
/**
|
||||
* @param array<string, mixed> $payload
|
||||
* @return array<string, mixed>
|
||||
*/
|
||||
private static function summarizeNodePayload(array $payload): array
|
||||
{
|
||||
$summary = $payload;
|
||||
if (isset($summary['tasks']) && is_array($summary['tasks'])) {
|
||||
$summary['tasks'] = array_map(static function ($task) {
|
||||
if (!is_array($task)) {
|
||||
return $task;
|
||||
}
|
||||
$copy = $task;
|
||||
if (isset($copy['paramList']) && is_array($copy['paramList'])) {
|
||||
$copy['paramList_count'] = count($copy['paramList']);
|
||||
unset($copy['paramList']);
|
||||
}
|
||||
return $copy;
|
||||
}, $summary['tasks']);
|
||||
}
|
||||
if (isset($summary['cookies']) && is_array($summary['cookies'])) {
|
||||
$summary['cookies_count'] = count($summary['cookies']);
|
||||
unset($summary['cookies']);
|
||||
}
|
||||
return $summary;
|
||||
}
|
||||
|
||||
/**
|
||||
|
||||
Reference in New Issue
Block a user