feat:修改nl2sql功能

This commit is contained in:
2026-09-14 10:57:48 +08:00
parent a7f9e182a4
commit 67d5cfc2b8
15 changed files with 206 additions and 54 deletions
+7 -1
View File
@@ -27,6 +27,7 @@ async def execute_readonly_sql(
masks: dict[tuple[str, str], str] | None = None,
max_rows: int | None = None,
timeout_seconds: float | None = None,
parameters: dict[str, object] | None = None,
user_id: int = 0,
connection_id: int | None = None,
kill_query=None,
@@ -49,7 +50,12 @@ async def execute_readonly_sql(
connection_id=connection_id,
)
try:
execution = session.execute(text(validated_sql.sql))
statement = text(validated_sql.sql)
execution = (
session.execute(statement, parameters)
if parameters
else session.execute(statement)
)
result = (
await asyncio.wait_for(execution, timeout_seconds)
if timeout_seconds is not None
+1 -1
View File
@@ -35,7 +35,7 @@ def apply_row_scope(sql: str, permission: dict, data_scope: dict | None) -> str:
continue
values = _values(data_scope, scope["type"])
condition = exp.In(
this=exp.column(scope["column"]),
this=exp.column(scope["column"], table=table.alias_or_name),
expressions=[exp.Literal.number(value) for value in values],
)
statement = statement.where(condition)
+18 -4
View File
@@ -107,18 +107,26 @@ def resolve_semantics(question: str, *, catalog: dict[str, Any] | None = None) -
{
"term": item["term"],
"field": item["fields"][0],
"hint": item.get("metric_hint") or item.get("value_hint", ""),
"hint": item["metric_hint"],
}
# metric_hint(指标口径)与 value_hint(字段取值口径)都要进入
# SQL 生成上下文;只过滤 metric_hint 会让枚举值提示永远丢失。
for item in matched
if "metric_hint" in item or "value_hint" in item
if item.get("metric_hint")
]
value_hints = [
{
"term": item["term"],
"field": item["fields"][0],
"hint": item["value_hint"],
}
for item in matched
if item.get("value_hint")
]
return {
"terms": [item["term"] for item in matched],
"tables": tables,
"fields": fields,
"metrics": metrics,
"value_hints": value_hints,
"relationships": list(catalog.get("relationships", [])),
}
@@ -149,6 +157,11 @@ def build_semantic_context(question: str, schema: dict[str, Any]) -> dict[str, A
for metric in resolved["metrics"]
if any(metric["field"] == field for _, field in schema_fields if _ in tables)
]
value_hints = [
hint
for hint in resolved["value_hints"]
if any(hint["field"] == field for _, field in schema_fields if _ in tables)
]
relationships = [
relation
for relation in resolved["relationships"]
@@ -160,5 +173,6 @@ def build_semantic_context(question: str, schema: dict[str, Any]) -> dict[str, A
"tables": tables,
"fields": fields,
"metrics": metrics,
"value_hints": value_hints,
"relationships": relationships,
}
+9
View File
@@ -83,7 +83,16 @@ def validate_select_sql(
if statement.find(exp.Star):
raise SqlSecurityError("配置字段权限时禁止 SELECT *")
default_table = next(iter(access_tables), None) if len(access_tables) == 1 else None
select_aliases = {
expression.alias
for expression in statement.expressions
if isinstance(expression, exp.Alias) and expression.alias
}
for column in statement.find_all(exp.Column):
# ORDER BY may legally reference an alias defined in SELECT. The
# alias is already covered by the expressions validated below.
if not column.table and column.name in select_aliases and isinstance(column.parent, exp.Ordered):
continue
table_name = aliases.get(column.table, column.table) or default_table
if table_name is None or column.name not in authorized_columns.get(table_name, set()):
raise SqlSecurityError("SQL 访问了未授权字段")