完整版V1 加入爬虫功能

2026-06-09 03:36:30 +08:00
parent 34d76cce74
commit a68b83fcbd
69 changed files with 5058 additions and 56 deletions
@@ -0,0 +1,131 @@
+<?php
+// SS云控(Customer)
+
+require_once __DIR__ . '/AbstractScrmSpider.class.php';
+
+// 可以获取每日具体进线数据
+class SsCustomer extends AbstractScrmSpider
+{
+    const API_LIST  = '/sys/share/report/get-customer-analysis-dimension-list';     // 列表API地址
+    const API_DETAILS = '/sys/share/report/get-customer-analysis-statistics';                   // 详情API地址
+    const API_COUNT = '';    // 总数API地址
+    const DEFAULT_PER_PAGE_COUNT = 20;                          // List默认每页显示的数量
+    private $pageUrl;
+    private $account;
+    private $password;
+    private $unifiedData;
+    
+    // 实例化时动态传入账号和密码
+    public function __construct($pageUrl, $account, $password, $nodeHost = 'http://127.0.0.1:3001')
+    {
+        parent::__construct($nodeHost);
+        $this->account = $account;
+        $this->password = $password;
+        $this->pageUrl = $pageUrl;
+        
+        $this->unifiedData = new UnifiedScrmData();
+    }
+
+    protected function getSpiderConfig()
+    {
+        return [
+            'pageUrl' => $this->pageUrl,
+            'apiUrls' => [self::API_LIST, self::API_DETAILS, self::API_COUNT],
+            
+            // 明确指派角色
+            'listApi'   => self::API_LIST,                  // 必须
+            'detailApi' => self::API_DETAILS,               // 选填
+            'countApi'  => self::API_COUNT,
+            
+            'listMethod' => 'POST',
+            
+            'paginationMode' => self::MODE_UI,
+            
+            'authActions' => [
+                ['type' => 'wait', 'ms' => 2000],
+                
+                ['type' => 'type', 'selector' => 'input[type="password"]', 'value' => $this->password],
+                
+                ['type' => 'press', 'key' => 'Enter'],
+                
+                ['type' => 'wait', 'ms' => 3000],
+                
+                [
+                    'type'     => 'vue_click', 
+                    'selector' => 'button[class*="reports-customers__dimension"]',
+                    'text'     => 'social media accounts'
+                ],
+                
+                ['type' => 'wait', 'ms' => 3000],
+            ]
+        ];
+    }
+
+    // 只负责解析 List 的总页数
+    protected function extractListTotalPages($listFirstPageData, $countData = null)
+    {
+        return null;
+        
+        // $default_per_page_count = self::DEFAULT_PER_PAGE_COUNT;
+        
+        // // 场景 A：如果单独的 Count 接口有数据返回
+        // if ($countData) {
+        //     // 假设 count 接口返回的是 {"data": {"total": 312}}
+        //     $totalRecords = $countData['data']['total'];
+        // } else {
+        //     // 场景 B
+        //     $totalRecords = $listFirstPageData['data']['total'];
+        // }
+        
+        // $this->unifiedData->total = $totalRecords;
+        // if($totalRecords <= $default_per_page_count) {
+        //     return 1;
+        // }
+        
+        // return ceil($listFirstPageData['data']['total']/$default_per_page_count);
+    }
+
+    // 只负责组装 List 的翻页参数
+    protected function buildListPageParams($page)
+    {
+        return ['page' => $page, 'pageSize' => self::DEFAULT_PER_PAGE_COUNT];
+    }
+
+    // 提供 List 的下一页按钮信息
+    protected function getUiPaginationConfig()
+    {
+        return [
+            'nextBtnSelector' => '.arco-pagination-item-next', 
+            'waitMs'          => 2000
+        ];
+    }
+
+    // 清爽至极的数据清洗：详情是详情，列表是列表
+    protected function parseToUnifiedData($detailData, $allListPagesData)
+    {
+        $unifiedData = $this->unifiedData;
+
+        // 1. 如果捕获到了详情数据，提取今日新增
+        if ($detailData) {
+            $unifiedData->todayNewCount = (int)($detailData['data']['distinct_contacts_total'] ?? 0);
+        }
+
+        // 2. 循环合并了所有页数的 List 数组
+        foreach ($allListPagesData as $pageRaw) {
+            $records = $pageRaw['data']['list'] ?? []; 
+            foreach ($records as $item) {
+                if(!empty($item['channel_tag'])) {
+                    $number = $item['channel_tag'] ?? null;
+                    $isOnline = true;   // 离线状态的账号同步不到
+                    
+                    $unifiedData->addNumber($number, $isOnline, $item['distinct_contacts_total']);
+                }
+            }
+        }
+        
+        // 🚀 3. 终极统计：所有翻页数据均已入库，此时再统计真实的 Total 总数
+        $unifiedData->total = count($unifiedData->numbers);
+
+        return $unifiedData;
+    }
+}