feat: Spanner ADK toolset supports customizable template SQL and parameterized SQL

PiperOrigin-RevId: 813909122
This commit is contained in:
Google Team Member
2025-10-01 14:15:01 -07:00
committed by Copybara-Service
parent a5cf80b952
commit da62700d73
3 changed files with 246 additions and 53 deletions
+129 -3
View File
@@ -16,14 +16,21 @@ import os
from google.adk.agents.llm_agent import LlmAgent
from google.adk.auth.auth_credential import AuthCredentialTypes
from google.adk.tools.google_tool import GoogleTool
from google.adk.tools.spanner.settings import Capabilities
from google.adk.tools.spanner.settings import SpannerToolSettings
from google.adk.tools.spanner.spanner_credentials import SpannerCredentialsConfig
from google.adk.tools.spanner.spanner_toolset import SpannerToolset
import google.adk.tools.spanner.utils as spanner_tool_utils
from google.adk.tools.tool_context import ToolContext
import google.auth
from google.auth.credentials import Credentials
from google.cloud.spanner_v1 import param_types as spanner_param_types
# Define an appropriate credential type
CREDENTIALS_TYPE = AuthCredentialTypes.OAUTH2
# Set to None to use the application default credentials (ADC) for a quick
# development.
CREDENTIALS_TYPE = None
# Define Spanner tool config with read capability set to allowed.
@@ -56,10 +63,115 @@ else:
credentials=application_default_credentials
)
# Example 1: Use tools from the Spanner toolset.
# For example, data exploration agents help the Spanner database developer or
# data engineer of the organization.
spanner_toolset = SpannerToolset(
credentials_config=credentials_config, spanner_tool_settings=tool_settings
credentials_config=credentials_config,
spanner_tool_settings=tool_settings,
# Uncomment to explicitly specify allowed tools.
# tool_filter=["execute_sql", "get_table_schema"],
)
# Replace the following settings with your specific Spanner database for example
# 2 and 3.
# For example, these settings can also be read from a configuration file or
# environment variables.
_SPANNER_PROJECT_ID = "<PROJECT_ID>"
_SPANNER_INSTANCE_ID = "<INSTANCE_ID>"
_SPANNER_DATABASE_ID = "<DATABASE_ID>"
# Example 2: Create a customized Spanner query tool with a template SQL query.
# Note that this approach makes it **more vulnerable to SQL injection**. This
# might be suitable for some specific use cases, and **adding additional checks
# or callbacks** is recommended.
def count_rows_in_table(
table_name: str,
credentials: Credentials,
settings: SpannerToolSettings,
tool_context: ToolContext,
):
"""Counts the total number of rows for a specified table.
Args:
table_name: The name of the table for which to count rows.
Returns:
The total number of rows in the table.
"""
# Example of adding additional checks:
# if table_name not in ["table1", "table2"]:
# raise ValueError("Table name is not allowed.")
sql_template = f"SELECT COUNT(*) FROM {table_name}"
return spanner_tool_utils.execute_sql(
project_id=_SPANNER_PROJECT_ID,
instance_id=_SPANNER_INSTANCE_ID,
database_id=_SPANNER_DATABASE_ID,
query=sql_template,
credentials=credentials,
settings=settings,
tool_context=tool_context,
)
# Example 3: Create a customized Spanner query tool with a template
# parameterized SQL query.
# For example, it could query data that all authenticated users of the system
# have access to. This can also work for searching public knowledge bases, such
# as company policies and FAQs.
def search_hotels(
location_name: str,
credentials: Credentials,
settings: SpannerToolSettings,
tool_context: ToolContext,
):
"""Search hotels for a specific location.
This function takes a geographical location name and returns a list of hotels
in that area, including key details for each.
Args:
location_name (str): The geographical location (e.g., city or town) for the
hotel search.
Example:
{
"location_name": "Seattle"
}
Example:
{
"location_name": "New York"
}
Example:
{
"location_name": "Los Angeles"
}
Returns:
The hotels name, rating and description.
"""
sql_template = """
SELECT name, rating, description FROM hotels
WHERE location_name = @location_name
"""
return spanner_tool_utils.execute_sql(
project_id=_SPANNER_PROJECT_ID,
instance_id=_SPANNER_INSTANCE_ID,
database_id=_SPANNER_DATABASE_ID,
query=sql_template,
credentials=credentials,
settings=settings,
tool_context=tool_context,
params={"location_name": location_name},
params_types={"location_name": spanner_param_types.STRING},
)
# The variable name `root_agent` determines what your root agent is for the
# debug CLI
root_agent = LlmAgent(
@@ -73,5 +185,19 @@ root_agent = LlmAgent(
You are a data agent with access to several Spanner tools.
Make use of those tools to answer the user's questions.
""",
tools=[spanner_toolset],
tools=[
# Use tools from Spanner toolset.
spanner_toolset,
# Or, uncomment to use customized Spanner tools.
# GoogleTool(
# func=count_rows_in_table,
# credentials_config=credentials_config,
# tool_settings=tool_settings,
# ),
# GoogleTool(
# func=search_hotels,
# credentials_config=credentials_config,
# tool_settings=tool_settings,
# ),
],
)
+10 -50
View File
@@ -14,17 +14,12 @@
from __future__ import annotations
import json
from google.auth.credentials import Credentials
from google.cloud.spanner_admin_database_v1.types import DatabaseDialect
from . import client
from . import utils
from ..tool_context import ToolContext
from .settings import SpannerToolSettings
DEFAULT_MAX_EXECUTED_QUERY_RESULT_ROWS = 50
def execute_sql(
project_id: str,
@@ -68,47 +63,12 @@ def execute_sql(
Note:
This is running with Read-Only Transaction for query that only read data.
"""
try:
# Get Spanner client
spanner_client = client.get_spanner_client(
project=project_id, credentials=credentials
)
instance = spanner_client.instance(instance_id)
database = instance.database(database_id)
if database.database_dialect == DatabaseDialect.POSTGRESQL:
return {
"status": "ERROR",
"error_details": "PostgreSQL dialect is not supported.",
}
with database.snapshot() as snapshot:
result_set = snapshot.execute_sql(query)
rows = []
counter = (
settings.max_executed_query_result_rows
if settings and settings.max_executed_query_result_rows > 0
else DEFAULT_MAX_EXECUTED_QUERY_RESULT_ROWS
)
for row in result_set:
try:
# if the json serialization of the row succeeds, use it as is
json.dumps(row)
except:
row = str(row)
rows.append(row)
counter -= 1
if counter <= 0:
break
result = {"status": "SUCCESS", "rows": rows}
if counter <= 0:
result["result_is_likely_truncated"] = True
return result
except Exception as ex:
return {
"status": "ERROR",
"error_details": str(ex),
}
return utils.execute_sql(
project_id,
instance_id,
database_id,
query,
credentials,
settings,
tool_context,
)
+107
View File
@@ -0,0 +1,107 @@
# Copyright 2025 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import annotations
import json
from typing import Optional
from google.auth.credentials import Credentials
from google.cloud.spanner_admin_database_v1.types import DatabaseDialect
from . import client
from ..tool_context import ToolContext
from .settings import SpannerToolSettings
DEFAULT_MAX_EXECUTED_QUERY_RESULT_ROWS = 50
def execute_sql(
project_id: str,
instance_id: str,
database_id: str,
query: str,
credentials: Credentials,
settings: SpannerToolSettings,
tool_context: ToolContext,
params: Optional[dict] = None,
params_types: Optional[dict] = None,
) -> dict:
"""Utility function to run a Spanner Read-Only query in the spanner database and return the result.
Args:
project_id (str): The GCP project id in which the spanner database
resides.
instance_id (str): The instance id of the spanner database.
database_id (str): The database id of the spanner database.
query (str): The Spanner SQL query to be executed.
credentials (Credentials): The credentials to use for the request.
settings (SpannerToolSettings): The settings for the tool.
tool_context (ToolContext): The context for the tool.
params (dict): values for parameter replacement. Keys must match the
names used in ``query``.
params_types (dict): maps explicit types for one or more param values.
Returns:
dict: Dictionary with the result of the query.
If the result contains the key "result_is_likely_truncated" with
value True, it means that there may be additional rows matching the
query not returned in the result.
"""
try:
# Get Spanner client
spanner_client = client.get_spanner_client(
project=project_id, credentials=credentials
)
instance = spanner_client.instance(instance_id)
database = instance.database(database_id)
if database.database_dialect == DatabaseDialect.POSTGRESQL:
return {
"status": "ERROR",
"error_details": "PostgreSQL dialect is not supported.",
}
with database.snapshot() as snapshot:
result_set = snapshot.execute_sql(
sql=query, params=params, param_types=params_types
)
rows = []
counter = (
settings.max_executed_query_result_rows
if settings and settings.max_executed_query_result_rows > 0
else DEFAULT_MAX_EXECUTED_QUERY_RESULT_ROWS
)
for row in result_set:
try:
# if the json serialization of the row succeeds, use it as is
json.dumps(row)
except:
row = str(row)
rows.append(row)
counter -= 1
if counter <= 0:
break
result = {"status": "SUCCESS", "rows": rows}
if counter <= 0:
result["result_is_likely_truncated"] = True
return result
except Exception as ex:
return {
"status": "ERROR",
"error_details": str(ex),
}