e1e515ecae
后端新增预算本体解析模块和风险规则评分回填服务，优化规则 生成本体对齐和提示词构建，增强费用类型关键词和本体验证， 完善报销查询和审计接口，前端预算中心页面增加对话框和本 体工具函数，重构审计页面元数据和视图模型，补充单元测试。
270 lines
10 KiB
Python
270 lines
10 KiB
Python
from __future__ import annotations
|
|
|
|
import re
|
|
from typing import Any
|
|
|
|
from app.schemas.ontology import OntologyEntity, OntologyMetric
|
|
from app.services.ontology_rules import (
|
|
BUDGET_CONTEXT_TYPES,
|
|
BUDGET_CONTROL_ACTION_KEYWORDS,
|
|
BUDGET_KEYWORDS,
|
|
BUDGET_REQUIRED_SLOT_KEYS,
|
|
BUDGET_STATUS_KEYWORDS,
|
|
BUDGET_SUBJECT_KEYWORDS,
|
|
BUDGET_SUBJECT_LABEL_BY_CODE,
|
|
)
|
|
|
|
|
|
class BudgetOntologyMixin:
|
|
@staticmethod
|
|
def _is_budget_context_value(context_json: dict[str, Any]) -> bool:
|
|
document_type = str(context_json.get("document_type") or "").strip()
|
|
entry_source = str(context_json.get("entry_source") or "").strip()
|
|
session_type = str(context_json.get("session_type") or "").strip()
|
|
conversation_scenario = str(context_json.get("conversation_scenario") or "").strip()
|
|
return (
|
|
document_type in BUDGET_CONTEXT_TYPES
|
|
or entry_source in BUDGET_CONTEXT_TYPES
|
|
or session_type in BUDGET_CONTEXT_TYPES
|
|
or conversation_scenario == "budget"
|
|
)
|
|
|
|
@staticmethod
|
|
def _has_budget_signal(compact_query: str) -> bool:
|
|
return any(keyword in compact_query for keyword in BUDGET_KEYWORDS)
|
|
|
|
@staticmethod
|
|
def _infer_budget_missing_slots(
|
|
entities: list[OntologyEntity],
|
|
context_json: dict[str, Any],
|
|
) -> list[str]:
|
|
entity_types = {item.type for item in entities}
|
|
budget_values = context_json.get("budget_header")
|
|
if not isinstance(budget_values, dict):
|
|
budget_values = {}
|
|
detail_values = context_json.get("budget_details")
|
|
if not isinstance(detail_values, list):
|
|
detail_values = []
|
|
|
|
missing_slots: list[str] = []
|
|
has_budget_period = str(budget_values.get("budget_period") or "").strip()
|
|
has_department = str(budget_values.get("department") or "").strip()
|
|
if "budget_period" not in entity_types and not has_budget_period:
|
|
missing_slots.append("budget_period")
|
|
if "department" not in entity_types and not has_department:
|
|
missing_slots.append("department")
|
|
has_subject = "budget_subject" in entity_types or any(
|
|
str(item.get("budget_subject") or "").strip()
|
|
for item in detail_values
|
|
if isinstance(item, dict)
|
|
)
|
|
if not has_subject:
|
|
missing_slots.append("budget_subject")
|
|
has_amount = "budget_amount" in entity_types or any(
|
|
str(item.get("budget_amount") or "").strip()
|
|
for item in detail_values
|
|
if isinstance(item, dict)
|
|
)
|
|
if not has_amount:
|
|
missing_slots.append("budget_amount")
|
|
return [item for item in BUDGET_REQUIRED_SLOT_KEYS if item in missing_slots]
|
|
|
|
@staticmethod
|
|
def _extract_budget_metrics(compact_query: str) -> list[OntologyMetric]:
|
|
metrics: list[OntologyMetric] = []
|
|
if any(keyword in compact_query for keyword in ("预算金额", "预算总额", "预算额度")):
|
|
metrics.append(OntologyMetric(name="budget_amount", aggregation="sum", unit="CNY"))
|
|
if any(
|
|
keyword in compact_query
|
|
for keyword in ("可用预算", "剩余预算", "可用余额", "剩余可用")
|
|
):
|
|
metrics.append(OntologyMetric(name="available_amount", aggregation="sum", unit="CNY"))
|
|
if any(
|
|
keyword in compact_query
|
|
for keyword in ("已占用", "已预占", "预算占用", "占用金额")
|
|
):
|
|
metrics.append(OntologyMetric(name="reserved_amount", aggregation="sum", unit="CNY"))
|
|
if any(keyword in compact_query for keyword in ("已发生", "已核销", "已消耗", "已使用")):
|
|
metrics.append(OntologyMetric(name="consumed_amount", aggregation="sum", unit="CNY"))
|
|
if any(keyword in compact_query for keyword in ("执行率", "使用率")):
|
|
metrics.append(
|
|
OntologyMetric(name="budget_usage_rate", aggregation="ratio", unit="percent")
|
|
)
|
|
return metrics
|
|
|
|
def _extract_budget_entities(
|
|
self,
|
|
query: str,
|
|
compact_query: str,
|
|
context_json: dict[str, Any],
|
|
) -> list[OntologyEntity]:
|
|
entities: list[OntologyEntity] = []
|
|
|
|
if self._is_budget_context_value(context_json) or self._has_budget_signal(compact_query):
|
|
entities.append(
|
|
self._make_entity(
|
|
"document_type",
|
|
"预算",
|
|
"budget_plan",
|
|
role="target",
|
|
confidence=0.94,
|
|
)
|
|
)
|
|
entities.append(
|
|
self._make_entity(
|
|
"workflow_stage",
|
|
"预算控制",
|
|
"budget_control",
|
|
role="target",
|
|
confidence=0.9,
|
|
)
|
|
)
|
|
|
|
period_pattern = (
|
|
r"(?P<year>20\d{2})\s*年\s*"
|
|
r"(?:(?P<quarter>Q[1-4]|[一二三四]季度)|(?P<month>\d{1,2})\s*月|度)?"
|
|
)
|
|
for match in re.finditer(period_pattern, query, flags=re.IGNORECASE):
|
|
year = match.group("year")
|
|
quarter = match.group("quarter")
|
|
month = match.group("month")
|
|
if quarter:
|
|
quarter_text = quarter.upper() if quarter.upper().startswith("Q") else quarter
|
|
normalized = f"{year}年{quarter_text}"
|
|
elif month:
|
|
normalized = f"{year}年{int(month)}月"
|
|
else:
|
|
normalized = f"{year}年度"
|
|
entities.append(
|
|
self._make_entity(
|
|
"budget_period",
|
|
match.group(0).strip(),
|
|
normalized,
|
|
role="filter",
|
|
confidence=0.88,
|
|
)
|
|
)
|
|
|
|
for code in re.findall(r"CC-\d+", query, flags=re.IGNORECASE):
|
|
entities.append(
|
|
self._make_entity(
|
|
"cost_center",
|
|
code,
|
|
code.upper(),
|
|
role="filter",
|
|
confidence=0.92,
|
|
)
|
|
)
|
|
|
|
for label, normalized in BUDGET_SUBJECT_KEYWORDS.items():
|
|
if label in query:
|
|
subject_label = BUDGET_SUBJECT_LABEL_BY_CODE.get(normalized, label)
|
|
entities.append(
|
|
self._make_entity(
|
|
"budget_subject",
|
|
label,
|
|
normalized,
|
|
role="filter",
|
|
confidence=0.9,
|
|
)
|
|
)
|
|
entities.append(
|
|
self._make_entity(
|
|
"expense_type",
|
|
subject_label,
|
|
normalized,
|
|
role="filter",
|
|
confidence=0.9,
|
|
)
|
|
)
|
|
|
|
for label, normalized in BUDGET_STATUS_KEYWORDS.items():
|
|
if label in query:
|
|
entities.append(
|
|
self._make_entity(
|
|
"budget_status",
|
|
label,
|
|
normalized,
|
|
role="filter",
|
|
confidence=0.86,
|
|
)
|
|
)
|
|
|
|
for label, normalized in BUDGET_CONTROL_ACTION_KEYWORDS.items():
|
|
if label in query:
|
|
entities.append(
|
|
self._make_entity(
|
|
"control_action",
|
|
label,
|
|
normalized,
|
|
role="target",
|
|
confidence=0.84,
|
|
)
|
|
)
|
|
|
|
version_match = re.search(r"V\d+(?:\.\d+){0,2}", query, flags=re.IGNORECASE)
|
|
if version_match:
|
|
version = version_match.group(0).upper()
|
|
entities.append(
|
|
self._make_entity(
|
|
"budget_version",
|
|
version,
|
|
version,
|
|
role="filter",
|
|
confidence=0.86,
|
|
)
|
|
)
|
|
|
|
warning_match = re.search(r"(?:预警线|预警阈值|预算预警)\s*(?P<value>\d{1,3})\s*%", query)
|
|
if warning_match:
|
|
value = f"{warning_match.group('value')}%"
|
|
entities.append(
|
|
self._make_entity(
|
|
"warning_threshold",
|
|
value,
|
|
value,
|
|
role="threshold",
|
|
confidence=0.9,
|
|
)
|
|
)
|
|
|
|
entities.extend(self._extract_budget_amount_entities(query))
|
|
return entities
|
|
|
|
def _extract_budget_amount_entities(self, query: str) -> list[OntologyEntity]:
|
|
entities: list[OntologyEntity] = []
|
|
patterns = (
|
|
(
|
|
"budget_amount",
|
|
r"(?:预算金额|预算额度|预算总额)\s*(?P<value>\d+(?:\.\d+)?)\s*(?P<unit>万元|万|元)?",
|
|
),
|
|
(
|
|
"available_amount",
|
|
r"(?:可用预算|剩余预算|可用余额|剩余可用)\s*(?P<value>\d+(?:\.\d+)?)\s*(?P<unit>万元|万|元)?",
|
|
),
|
|
(
|
|
"reserved_amount",
|
|
r"(?:已占用|已预占|占用金额|预算占用)\s*(?P<value>\d+(?:\.\d+)?)\s*(?P<unit>万元|万|元)?",
|
|
),
|
|
(
|
|
"consumed_amount",
|
|
r"(?:已发生|已核销|已消耗|已使用)\s*(?P<value>\d+(?:\.\d+)?)\s*(?P<unit>万元|万|元)?",
|
|
),
|
|
)
|
|
for entity_type, pattern in patterns:
|
|
for match in re.finditer(pattern, query):
|
|
raw_value = match.group("value")
|
|
unit = match.group("unit")
|
|
amount_value = self._normalize_amount(raw_value, unit)
|
|
display_value = f"{raw_value}{unit or ''}"
|
|
entities.append(
|
|
self._make_entity(
|
|
entity_type,
|
|
display_value,
|
|
str(amount_value),
|
|
role="target",
|
|
confidence=0.9,
|
|
)
|
|
)
|
|
return entities
|