星枢: PST处理脚本备份 (2026-04-13)

openclaw/xingshu/pst-processing/rules/apply_rules.py

@@ -0,0 +1,101 @@
+#!/usr/bin/env python3
+"""
+PST 邮件清理规则执行脚本 v1.2
+用法: python3 apply_rules.py <csv_file>
+"""
+import csv
+import json
+import sys
+from collections import defaultdict
+
+RULES_FILE = '/Users/weishen/pst-processing/rules/delete_rules.json'
+
+def load_rules():
+    with open(RULES_FILE) as f:
+        return json.load(f)['rules']
+
+def apply_rules(csv_path):
+    rules = load_rules()
+    rows = list(csv.DictReader(open(csv_path, encoding='utf-8')))
+    
+    for r in rows:
+        r['delete_flag'] = 'N'
+    
+    total_deleted = 0
+    
+    for rule in rules:
+        folder_pattern = rule['folder_contains']
+        action = rule['action']
+        
+        folder_rows = [r for r in rows if folder_pattern in r['folder']]
+        if not folder_rows:
+            continue
+        
+        deleted_count = 0
+        kept_count = 0
+        
+        if action == 'keep':
+            kept_count = len(folder_rows)
+        
+        elif action == 'delete_all':
+            for r in folder_rows:
+                r['delete_flag'] = 'Y'
+                deleted_count += 1
+        
+        elif action == 'keep_sample':
+            keep_limit = rule['keep_count']
+            subject_map = defaultdict(list)
+            for r in folder_rows:
+                subj = r['subject'].strip()[:80]
+                subject_map[subj].append(r)
+            sorted_subjects = sorted(subject_map.items(), key=lambda x: -len(x[1]))
+            
+            for subj, emails in sorted_subjects:
+                if kept_count >= keep_limit:
+                    for r in emails:
+                        r['delete_flag'] = 'Y'
+                        deleted_count += 1
+                else:
+                    for i, r in enumerate(emails):
+                        if i == 0:
+                            kept_count += 1
+                        else:
+                            r['delete_flag'] = 'Y'
+                            deleted_count += 1
+        
+        elif action == 'keep_if_attachment':
+            for r in folder_rows:
+                if r['has_attachment'] == 'Y':
+                    kept_count += 1
+                else:
+                    r['delete_flag'] = 'Y'
+                    deleted_count += 1
+        
+        print(f"  {folder_pattern}: {len(folder_rows)} | 保留{kept_count} | 删除{deleted_count}")
+        total_deleted += deleted_count
+    
+    total_kept = len(rows) - total_deleted
+    print(f"📊 总计: 保留{total_kept} | 删除{total_deleted}")
+    
+    fieldnames = list(rows[0].keys())
+    out_marked = csv_path.replace('.csv', '_marked.csv')
+    out_delete = csv_path.replace('.csv', '_delete_list.csv')
+    
+    with open(out_marked, 'w', newline='', encoding='utf-8') as f:
+        writer = csv.DictWriter(f, fieldnames=fieldnames)
+        writer.writeheader()
+        writer.writerows(rows)
+    
+    with open(out_delete, 'w', newline='', encoding='utf-8') as f:
+        writer = csv.DictWriter(f, fieldnames=fieldnames)
+        writer.writeheader()
+        writer.writerows([r for r in rows if r['delete_flag'] == 'Y'])
+    
+    print(f"✅ 输出: {out_marked}")
+    print(f"✅ 删除清单: {out_delete} ({total_deleted} 封)")
+
+if __name__ == '__main__':
+    if len(sys.argv) < 2:
+        print("用法: python3 apply_rules.py <csv_file>")
+        sys.exit(1)
+    apply_rules(sys.argv[1])

openclaw/xingshu/pst-processing/rules/delete_rules.json

@@ -0,0 +1,84 @@
+{
+  "version": "1.2",
+  "description": "PST邮件清理规则",
+  "rules": [
+    {
+      "id": "aws_notification",
+      "folder_contains": "AWS Notification",
+      "action": "keep_sample",
+      "keep_count": 5,
+      "keep_by": "unique_subject",
+      "description": "AWS告警通知，每不同subject保留1封，最多5封"
+    },
+    {
+      "id": "prisma_cloud",
+      "folder_contains": "Prisma Cloud Notifications",
+      "action": "keep_sample",
+      "keep_count": 5,
+      "keep_by": "unique_subject",
+      "description": "Prisma Cloud通知，每不同subject保留1封，最多5封"
+    },
+    {
+      "id": "x4x_tenant_provisioning",
+      "folder_contains": "X4X-Tenant Provisioning",
+      "action": "keep_sample",
+      "keep_count": 5,
+      "keep_by": "unique_subject",
+      "description": "X4X租户配置通知，每不同subject保留1封，最多5封"
+    },
+    {
+      "id": "qualys",
+      "folder_contains": "Qualys",
+      "action": "keep_sample",
+      "keep_count": 5,
+      "keep_by": "unique_subject",
+      "description": "Qualys安全扫描通知，每不同subject保留1封，最多5封"
+    },
+    {
+      "id": "teams_notification",
+      "folder_contains": "Teams Notification",
+      "action": "keep_if_attachment",
+      "description": "Teams会议通知，有附件保留，无附件删除"
+    },
+    {
+      "id": "sma_notification",
+      "folder_contains": "SMA Notficiation",
+      "action": "keep_sample",
+      "keep_count": 10,
+      "keep_by": "unique_subject",
+      "description": "SMA工单通知，每月保留10封不同subject，其余删除"
+    },
+    {
+      "id": "ppm_saas_change",
+      "folder_contains": "PPM SaaS Change",
+      "action": "delete_all",
+      "description": "PPM故障单解决通知，全部删除"
+    },
+    {
+      "id": "cloudhealth",
+      "folder_contains": "CloudHealth",
+      "action": "keep",
+      "description": "CloudHealth成本报告，全部保留"
+    },
+    {
+      "id": "saas_bi_report",
+      "folder_contains": "SaaS BI Report",
+      "action": "keep",
+      "description": "BI数据推送，全部保留"
+    },
+    {
+      "id": "x4x_tenant_decommissioning",
+      "folder_contains": "X4X-Tenant Decommissioning",
+      "action": "delete_all",
+      "description": "租户下线通知，全部删除"
+    },
+    {
+      "id": "x4x_license_renewal",
+      "folder_contains": "X4X-License Renewal",
+      "action": "delete_all",
+      "description": "续期通知，全部删除"
+    }
+  ],
+  "default_action": "keep",
+  "updated": "2026-04-13"
+}