2026-06-09 03:36:30 +08:00
|
|
|
|
<?php
|
2026-07-01 01:26:27 +08:00
|
|
|
|
// A2c云控 — 测试脚本(短链/长链 + Real Browser 拦截 API,对齐生产 A2cSpider)
|
2026-06-09 03:36:30 +08:00
|
|
|
|
|
|
|
|
|
|
require_once __DIR__ . '/AbstractScrmSpider.class.php';
|
2026-07-01 01:26:27 +08:00
|
|
|
|
require_once __DIR__ . '/SplitPageUrlResolver.php';
|
2026-06-09 03:36:30 +08:00
|
|
|
|
|
|
|
|
|
|
class A2c extends AbstractScrmSpider
|
|
|
|
|
|
{
|
2026-07-01 01:26:27 +08:00
|
|
|
|
const API_LIST = '/api/talk/counter/share/record/list';
|
|
|
|
|
|
|
|
|
|
|
|
const API_DETAILS = '/api/talk/counter/share/detail';
|
|
|
|
|
|
|
|
|
|
|
|
const API_COUNT = '';
|
|
|
|
|
|
|
|
|
|
|
|
const DEFAULT_PER_PAGE_COUNT = 20;
|
|
|
|
|
|
|
|
|
|
|
|
/** @var string */
|
2026-06-09 03:36:30 +08:00
|
|
|
|
private $pageUrl;
|
2026-07-01 01:26:27 +08:00
|
|
|
|
|
|
|
|
|
|
/** @var string */
|
2026-06-09 03:36:30 +08:00
|
|
|
|
private $account;
|
2026-07-01 01:26:27 +08:00
|
|
|
|
|
|
|
|
|
|
/** @var string */
|
2026-06-09 03:36:30 +08:00
|
|
|
|
private $password;
|
2026-07-01 01:26:27 +08:00
|
|
|
|
|
|
|
|
|
|
/** @var UnifiedScrmData */
|
2026-06-09 03:36:30 +08:00
|
|
|
|
private $unifiedData;
|
2026-07-01 01:26:27 +08:00
|
|
|
|
|
|
|
|
|
|
/** @var string HTTP 预解析后的落地 URL */
|
|
|
|
|
|
private $landingUrl = '';
|
|
|
|
|
|
|
2026-06-09 03:36:30 +08:00
|
|
|
|
public function __construct($pageUrl, $account, $password, $nodeHost = 'http://127.0.0.1:3001')
|
|
|
|
|
|
{
|
|
|
|
|
|
parent::__construct($nodeHost);
|
|
|
|
|
|
$this->account = $account;
|
|
|
|
|
|
$this->password = $password;
|
|
|
|
|
|
$this->pageUrl = $pageUrl;
|
2026-07-01 01:26:27 +08:00
|
|
|
|
$this->landingUrl = SplitPageUrlResolver::resolveFinalUrl($pageUrl);
|
2026-06-09 03:36:30 +08:00
|
|
|
|
$this->unifiedData = new UnifiedScrmData();
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
protected function getSpiderConfig()
|
|
|
|
|
|
{
|
2026-07-01 01:26:27 +08:00
|
|
|
|
$antiBot = $this->buildAntiBotConfig();
|
|
|
|
|
|
|
|
|
|
|
|
$this->logDebug('a2c spider config', [
|
|
|
|
|
|
'pageUrl' => $this->pageUrl,
|
|
|
|
|
|
'landingUrl' => $this->landingUrl,
|
|
|
|
|
|
'sessionKey' => $antiBot['sessionKey'] ?? '',
|
|
|
|
|
|
'cfClearanceRequired' => $antiBot['cfClearanceRequired'] ?? null,
|
|
|
|
|
|
]);
|
|
|
|
|
|
|
2026-06-09 03:36:30 +08:00
|
|
|
|
return [
|
2026-07-01 01:26:27 +08:00
|
|
|
|
'pageUrl' => $this->pageUrl,
|
|
|
|
|
|
'listApi' => self::API_LIST,
|
|
|
|
|
|
'detailApi' => self::API_DETAILS,
|
|
|
|
|
|
'listMethod' => 'POST',
|
2026-06-09 03:36:30 +08:00
|
|
|
|
'paginationMode' => self::MODE_UI,
|
2026-07-01 01:26:27 +08:00
|
|
|
|
'authActions' => [
|
|
|
|
|
|
['type' => 'wait', 'ms' => 2000],
|
|
|
|
|
|
],
|
|
|
|
|
|
'antiBot' => $antiBot,
|
2026-06-09 03:36:30 +08:00
|
|
|
|
];
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2026-07-01 01:26:27 +08:00
|
|
|
|
/**
|
|
|
|
|
|
* A2C:业务域无需 cf_clearance;短链仍作 pageUrl,sessionKey 固定业务域
|
|
|
|
|
|
*
|
|
|
|
|
|
* @return array<string, mixed>
|
|
|
|
|
|
*/
|
|
|
|
|
|
private function buildAntiBotConfig()
|
|
|
|
|
|
{
|
|
|
|
|
|
$sessionHost = $this->resolveA2cSessionHost();
|
|
|
|
|
|
|
|
|
|
|
|
return [
|
|
|
|
|
|
'enabled' => true,
|
|
|
|
|
|
'profile' => 'real',
|
|
|
|
|
|
'turnstile' => true,
|
|
|
|
|
|
'solverFallback' => false,
|
|
|
|
|
|
'cfClearanceRequired' => false,
|
|
|
|
|
|
'sessionKey' => 'a2c:' . $sessionHost,
|
|
|
|
|
|
'challengeTimeoutMs' => 60000,
|
|
|
|
|
|
'landingUrlHint' => $this->landingUrl,
|
|
|
|
|
|
'businessHostHint' => $sessionHost,
|
|
|
|
|
|
'postCfReadyUrlContains' => '/visitors/counter/share',
|
|
|
|
|
|
'postCfReadySelector' => '.btn-next',
|
|
|
|
|
|
'postCfReadyTimeoutMs' => 30000,
|
|
|
|
|
|
'postCfSettleMs' => 500,
|
|
|
|
|
|
'postCfSpaWaitMs' => 4000,
|
|
|
|
|
|
];
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
|
* A2C 会话域:HTTP 预解析无法穿透短链 CF 时,回退到业务域 user.a2c.chat
|
|
|
|
|
|
*/
|
|
|
|
|
|
private function resolveA2cSessionHost()
|
|
|
|
|
|
{
|
|
|
|
|
|
$candidates = [];
|
|
|
|
|
|
if ($this->landingUrl !== '') {
|
|
|
|
|
|
$candidates[] = (string) parse_url($this->landingUrl, PHP_URL_HOST);
|
|
|
|
|
|
}
|
|
|
|
|
|
$candidates[] = (string) parse_url($this->pageUrl, PHP_URL_HOST);
|
|
|
|
|
|
|
|
|
|
|
|
foreach ($candidates as $host) {
|
|
|
|
|
|
$host = strtolower(trim($host));
|
|
|
|
|
|
if ($host !== '' && strpos($host, 'a2c.chat') !== false) {
|
|
|
|
|
|
return $host;
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
return 'user.a2c.chat';
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2026-06-09 03:36:30 +08:00
|
|
|
|
protected function extractListTotalPages($listFirstPageData, $countData = null)
|
|
|
|
|
|
{
|
|
|
|
|
|
$default_per_page_count = self::DEFAULT_PER_PAGE_COUNT;
|
|
|
|
|
|
$this->unifiedData->total = $listFirstPageData['data']['total'];
|
2026-07-01 01:26:27 +08:00
|
|
|
|
|
|
|
|
|
|
if ($listFirstPageData['data']['total'] <= $default_per_page_count) {
|
2026-06-09 03:36:30 +08:00
|
|
|
|
return 1;
|
|
|
|
|
|
}
|
2026-07-01 01:26:27 +08:00
|
|
|
|
|
|
|
|
|
|
return (int) ceil($listFirstPageData['data']['total'] / $default_per_page_count);
|
2026-06-09 03:36:30 +08:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
protected function buildListPageParams($page)
|
|
|
|
|
|
{
|
|
|
|
|
|
return ['page' => $page, 'pageSize' => self::DEFAULT_PER_PAGE_COUNT];
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
protected function getUiPaginationConfig()
|
|
|
|
|
|
{
|
|
|
|
|
|
return [
|
2026-07-01 01:26:27 +08:00
|
|
|
|
'nextBtnSelector' => '.btn-next',
|
|
|
|
|
|
'waitMs' => 2000,
|
2026-06-09 03:36:30 +08:00
|
|
|
|
];
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
protected function parseToUnifiedData($detailData, $allListPagesData)
|
|
|
|
|
|
{
|
|
|
|
|
|
$unifiedData = $this->unifiedData;
|
|
|
|
|
|
|
|
|
|
|
|
if ($detailData) {
|
2026-07-01 01:26:27 +08:00
|
|
|
|
$unifiedData->todayNewCount = (int) ($detailData['data']['newFollowersToday'] ?? 0);
|
2026-06-09 03:36:30 +08:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
foreach ($allListPagesData as $pageRaw) {
|
2026-07-01 01:26:27 +08:00
|
|
|
|
$records = $pageRaw['data']['rows'] ?? [];
|
2026-06-09 03:36:30 +08:00
|
|
|
|
foreach ($records as $item) {
|
2026-07-01 01:26:27 +08:00
|
|
|
|
if (!empty($item['account'])) {
|
|
|
|
|
|
$number = $item['account'];
|
|
|
|
|
|
$isOnline = (isset($item['numberStatus']) && (int) $item['numberStatus'] === 1);
|
|
|
|
|
|
$unifiedData->addNumber($number, $isOnline, (int) ($item['newFollowersToday'] ?? 0));
|
2026-06-09 03:36:30 +08:00
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
return $unifiedData;
|
|
|
|
|
|
}
|
2026-07-01 01:26:27 +08:00
|
|
|
|
}
|