first commit

scripts/dl_biz.py

@@ -0,0 +1,57 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""业务数据列表页导出深挖：各模块有真实数据的列表页(花名册/工资表/账单/考勤明细等)
+触发导出(类型B: #icon-export / 文字“导出”) -> 异步任务中心 -> 统一收割落盘。
+复用 dl_export 的安全逻辑(防登出授权处理 + 非均匀拟人间隔 + 四类入口)。"""
+import sys, time
+sys.path.insert(0, "scripts")
+import dl_export as X
+
+# (保存路径 full_path, url, node_key)  —— 各模块有真实数据的业务列表页
+DATA = [
+    ("02_人事 / 员工管理 / 员工花名册", "https://2haohr.com/personnel/employee/roster", "biz_roster"),
+    ("02_人事 / 员工管理 / 员工档案库", "https://2haohr.com/personnel/employee/archive", "biz_arch"),
+    ("02_人事 / 员工管理 / 万能人事表格", "https://2haohr.com/personnel/employee/personnel-form", "biz_form"),
+    ("02_人事 / 员工关系 / 入职管理", "https://2haohr.com/personnel/relation/enrollment", "biz_enroll"),
+    ("02_人事 / 员工关系 / 离职管理", "https://2haohr.com/personnel/relation/departure", "biz_depart"),
+    ("02_人事 / 员工关系 / 合同管理", "https://2haohr.com/personnel/relation/contract", "biz_contract"),
+    ("02_人事 / 员工关系 / 奖惩记录", "https://2haohr.com/personnel/relation/award", "biz_award"),
+    ("02_人事 / 员工关系 / 人事异动", "https://2haohr.com/personnel/relation/unusual-action", "biz_unusual"),
+    ("06_薪酬 / 薪酬管理 / 月工资表", "https://2haohr.com/payroll/salary/monthly", "biz_monthly"),
+    ("06_薪酬 / 薪酬管理 / 员工薪资", "https://2haohr.com/payroll/salary/profiles", "biz_profiles"),
+    ("06_薪酬 / 薪酬管理 / 工资发放", "https://2haohr.com/payroll/salary/payoff", "biz_payoff"),
+    ("05_社保公积金 / 常规 / 参保账单", "https://2haohr.com/social-insurance/records", "biz_sirec"),
+    ("05_社保公积金 / 常规 / 账单核对", "https://2haohr.com/social-insurance/bill", "biz_sibill"),
+    ("04_考勤假期 / 考勤统计 / 月考勤表", "https://2haohr.com/attendance/stat/monthly", "biz_kqmonth"),
+    ("04_考勤假期 / 考勤统计 / 明细记录", "https://2haohr.com/attendance/stat/records", "biz_kqrec"),
+    ("04_考勤假期 / 考勤统计 / 每日考勤", "https://2haohr.com/attendance/stat/daily", "biz_kqday"),
+    ("04_考勤假期 / 假期管理 / 休假一览", "https://2haohr.com/attendance/holiday/leaveRecords", "biz_leave"),
+    ("04_考勤假期 / 假期管理 / 假期余额", "https://2haohr.com/attendance/holiday/balance", "biz_balance"),
+    ("03_组织 / 岗位管理 / 员工标签", "https://2haohr.com/org/tag/emp-tags", "biz_tag"),
+    ("07_绩效 / 考核结果 / 绩效考核记录", "https://2haohr.com/performance-evaluation/result/assessment", "biz_perfrec"),
+]
+
+
+def main():
+    X.ensure()
+    X.call("find_tab", {"url": "2haohr.com"})
+    pages = [{"full_path": a, "url": b, "node_key": c, "screenshot_path": None} for a, b, c in DATA]
+    X.log("=== BIZ 业务列表页深挖, 共 %d 页 ===" % len(pages))
+    aborted = False
+    for i, p in enumerate(pages, 1):
+        try:
+            n = X.process_page(p, i, len(pages))
+            if n == -1:
+                X.log("=== 登录态失效, 中止 ===")
+                aborted = True
+                break
+        except Exception as e:
+            X.log("    页面异常: %s" % str(e)[:100])
+        X.human_delay("page")
+    if not aborted:
+        X.log("=== 收割任务中心 stage2=%d ===" % X.harvest_task_center())
+    X.log("=== BIZ DONE 累计文件=%d ===" % len(X.SAVED))
+
+
+if __name__ == "__main__":
+    main()