fix: pass pdf_path to normalizer for data provenance

- Add pdfPath to extraction results in parse_session
- Pass pdf_path to normalizer functions in generate_ppt
- Fix 'missing PDF hash' validation error

api/insurance/ppt/routes.py

@@ -138,6 +138,7 @@ def parse_session(session_id):
             result = asyncio.run(orchestrator.extract_plan(filepath, plan_type))
             extractions.append({
                 "pdfName": file_info.get("name", ""),
+                "pdfPath": filepath,
                 "planType": result.plan_type,
                 "status": result.status,
                 "productName": result.product_name,
@@ -149,6 +150,7 @@ def parse_session(session_id):
             logger.error(f"PDF 解析失败 [{file_info.get('name', '')}]: {e}", exc_info=True)
             extractions.append({
                 "pdfName": file_info.get("name", ""),
+                "pdfPath": filepath,
                 "planType": plan_type,
                 "status": "error",
                 "productName": "unknown",
@@ -275,9 +277,11 @@ def generate_ppt(session_id):
 
     # 取第一个成功的提取结果
     ext_data = None
+    pdf_path = None
     for ext in extractions:
         if ext.get("status") in ("success", "cached") and ext.get("data"):
             ext_data = ext["data"]
+            pdf_path = ext.get("pdfPath")
             break
 
     if not ext_data:
@@ -289,13 +293,13 @@ def generate_ppt(session_id):
 
     plan_type = ext_data.get("product_type", "savings")
     if plan_type == "ci":
-        normalized = normalize_ci_plan(ext_data)
+        normalized = normalize_ci_plan(ext_data, pdf_path)
         issues = validate_formal_ci_plan(normalized)
     elif plan_type == "iul":
-        normalized = normalize_iul_plan(ext_data)
+        normalized = normalize_iul_plan(ext_data, pdf_path)
         issues = validate_formal_iul_plan(normalized)
     else:
-        normalized = normalize_savings_plan(ext_data)
+        normalized = normalize_savings_plan(ext_data, pdf_path)
         issues = validate_formal_savings_plan(normalized)
 
     # 检查是否有阻断性错误