YG-Soft/X-Financial
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: 本体字段治理与风险规则模板执行器重构

Browse Source 
- 新增本体字段注册表与字段治理审计脚本
- 重构风险规则模板执行器、DSL 验证与清单分类器
- 完善票据夹服务与差旅请求详情页交互
- 优化趋势图表与总览页数据展示
- 增强报销平台风险分级与模拟公司筛选
- 补充本体字段、风险规则生成与票据夹服务测试覆盖
This commit is contained in:
caoxiaozhu
2026-06-03 15:46:56 +08:00
parent e12b140508
commit 34457f9c3e
81 changed files with 4858 additions and 1073 deletions
Download Patch File Download Diff File Expand all files Collapse all files

146
server/src/app/services/expense_claim_platform_risk.py
View File

@@ -16,6 +16,7 @@ from app.services.expense_rule_runtime import (
)
from app.services.expense_type_keywords import resolve_expense_type_code_from_text
from app.services.expense_claim_platform_risk_flag import build_platform_risk_flag
from app.services.risk_rule_manifest_classifier import is_budget_risk_manifest
from app.services.risk_rule_manifest_normalizer import normalize_risk_rule_manifest
from app.services.risk_rule_template_executor import RiskRuleTemplateExecutor
@@ -23,6 +24,44 @@ from app.services.risk_rule_template_executor import RiskRuleTemplateExecutor
class ExpenseClaimPlatformRiskMixin:
_DEFAULT_RISK_BUSINESS_STAGE = "reimbursement"
_SUPPORTED_RISK_BUSINESS_STAGES = {"expense_application", "reimbursement"}
_CLEAR_TRAVEL_DOCUMENT_TYPES = {
"flight_itinerary",
"train_ticket",
"ship_ticket",
"hotel_invoice",
"taxi_receipt",
"parking_toll_receipt",
}
_CLEAR_TRAVEL_SCENE_CODES = {"travel", "hotel", "transport"}
_GOODS_DESCRIPTION_FIELD_KEYS = {
"goodsname",
"servicename",
"itemname",
"project",
"productname",
"description",
"content",
"expensecontent",
"feeitem",
}
_GOODS_DESCRIPTION_LABEL_TOKENS = (
"商品",
"服务",
"货物",
"项目",
"品名",
"名称",
"费用内容",
"消费内容",
)
_VAGUE_KEYWORD_NEGATION_MARKERS = (
"不含",
"不包含",
"不包括",
"未包含",
"不涉及",
"不属于",
)
def evaluate_platform_risk_rules(
self,
@@ -127,6 +166,8 @@ class ExpenseClaimPlatformRiskMixin:
manifest_code = str(payload.get("rule_code") or rule_code).strip()
if not manifest_code or (code_filter and manifest_code not in code_filter):
continue
if is_budget_risk_manifest(payload):
continue
if payload.get("enabled") is False or not self._risk_manifest_matches_business_stage(
payload,
business_stage=business_stage,
@@ -162,6 +203,8 @@ class ExpenseClaimPlatformRiskMixin:
continue
if code_filter and rule_code not in missing_codes:
continue
if is_budget_risk_manifest(payload):
continue
if payload.get("enabled") is False or not self._risk_manifest_matches_business_stage(
payload,
business_stage=business_stage,
@@ -364,7 +407,7 @@ class ExpenseClaimPlatformRiskMixin:
fallback_message="票据文本中出现作废、红冲或红字发票相关信息，建议退回补充或人工复核。",
)
if evaluator == "vague_goods_description":
return self._evaluate_text_keyword_risk(
return self._evaluate_vague_goods_description_risk(
manifest,
contexts=contexts,
keywords=["详见清单", "服务费", "咨询费", "其他", "办公用品"],
@@ -663,6 +706,107 @@ class ExpenseClaimPlatformRiskMixin:
evidence={"matched_keywords": matched},
)
def _evaluate_vague_goods_description_risk(
self,
manifest: dict[str, Any],
*,
contexts: list[dict[str, Any]],
keywords: list[str],
fallback_message: str,
) -> dict[str, Any] | None:
matched_keywords: list[str] = []
matched_fields: list[dict[str, str]] = []
for context in contexts:
document_info = context.get("document_info") or {}
if self._is_clear_travel_document(document_info):
continue
field_values = self._collect_goods_description_field_values(document_info)
if field_values:
for value in field_values:
hits = self._collect_non_negated_keyword_hits(value, keywords)
for keyword in hits:
if keyword not in matched_keywords:
matched_keywords.append(keyword)
if hits:
matched_fields.append(
{
"item_index": str(context.get("index") or ""),
"value": value[:80],
}
)
continue
fallback_text = f"{context.get('ocr_summary') or ''}\n{context.get('ocr_text') or ''}"
hits = self._collect_non_negated_keyword_hits(fallback_text, keywords)
for keyword in hits:
if keyword not in matched_keywords:
matched_keywords.append(keyword)
if hits:
matched_fields.append(
{
"item_index": str(context.get("index") or ""),
"value": "OCR全文兜底",
}
)
if not matched_keywords:
return None
return self._build_platform_risk_flag(
manifest,
message=fallback_message,
evidence={
"matched_keywords": matched_keywords,
"matched_fields": matched_fields[:5],
},
)
@classmethod
def _is_clear_travel_document(cls, document_info: dict[str, Any]) -> bool:
document_type = str(document_info.get("document_type") or "").strip().lower()
scene_code = str(document_info.get("scene_code") or "").strip().lower()
return (
document_type in cls._CLEAR_TRAVEL_DOCUMENT_TYPES
or scene_code in cls._CLEAR_TRAVEL_SCENE_CODES
)
@classmethod
def _collect_goods_description_field_values(cls, document_info: dict[str, Any]) -> list[str]:
values: list[str] = []
for field in list(document_info.get("fields") or []):
if not isinstance(field, dict):
continue
field_key = str(field.get("key") or "").strip().lower().replace("_", "")
label = str(field.get("label") or "").replace(" ", "")
value = str(field.get("value") or "").strip()
if not value:
continue
if field_key in cls._GOODS_DESCRIPTION_FIELD_KEYS or any(
token in label for token in cls._GOODS_DESCRIPTION_LABEL_TOKENS
):
values.append(value)
return values
@classmethod
def _collect_non_negated_keyword_hits(cls, text: str, keywords: list[str]) -> list[str]:
normalized = str(text or "")
if not normalized:
return []
hits: list[str] = []
for keyword in keywords:
if not keyword:
continue
for match in re.finditer(re.escape(keyword), normalized):
window = normalized[max(0, match.start() - 12): match.end() + 12]
if any(marker in window for marker in cls._VAGUE_KEYWORD_NEGATION_MARKERS):
continue
hits.append(keyword)
break
return hits
def _evaluate_multi_city_reason_required_risk(
self,
manifest: dict[str, Any],