"""Cohere SQL agent."""
from __future__ import annotations
from typing import (
TYPE_CHECKING,
Any,
Dict,
Optional,
Sequence,
Union,
)
from langchain_community.agent_toolkits.sql.toolkit import SQLDatabaseToolkit
from langchain_community.tools.sql_database.tool import (
InfoSQLDatabaseTool,
ListSQLDatabaseTool,
)
from langchain_core.messages import BaseMessage
from langchain_core.prompts import BasePromptTemplate
from langchain_core.prompts.chat import (
ChatPromptTemplate,
HumanMessage,
HumanMessagePromptTemplate,
MessagesPlaceholder,
)
from langchain_cohere.chat_models import ChatCohere
from langchain_cohere.sql_agent.prompts import (
SQL_FUNCTIONS_SUFFIX,
SQL_PREAMBLE,
SQL_PREFIX,
)
if TYPE_CHECKING:
from langchain.agents.agent import AgentExecutor
from langchain_community.utilities.sql_database import SQLDatabase
from langchain_core.callbacks import BaseCallbackManager
from langchain_core.language_models import BaseLanguageModel
from langchain_core.tools import BaseTool
from datetime import datetime
from langchain.agents import (
create_tool_calling_agent,
)
from langchain.agents.agent import (
AgentExecutor,
RunnableMultiActionAgent,
)
[docs]def create_sql_agent(
llm: BaseLanguageModel,
toolkit: Optional[SQLDatabaseToolkit] = None,
callback_manager: Optional[BaseCallbackManager] = None,
prefix: Optional[str] = None,
suffix: Optional[str] = None,
top_k: int = 10,
max_iterations: Optional[int] = 15,
max_execution_time: Optional[float] = None,
early_stopping_method: str = "force",
verbose: bool = False,
agent_executor_kwargs: Optional[Dict[str, Any]] = None,
extra_tools: Sequence[BaseTool] = (),
*,
db: Optional[SQLDatabase] = None,
prompt: Optional[BasePromptTemplate] = None,
**kwargs: Any,
) -> AgentExecutor:
"""Construct a SQL agent from an LLM and toolkit or database.
Args:
llm: Language model to use for the agent. If agent_type is "tool-calling" then
llm is expected to support tool calling.
toolkit: SQLDatabaseToolkit for the agent to use. Must provide exactly one of
'toolkit' or 'db'. Specify 'toolkit' if you want to use a different model
for the agent and the toolkit.
callback_manager: DEPRECATED. Pass "callbacks" key into 'agent_executor_kwargs'
instead to pass constructor callbacks to AgentExecutor.
prefix: Prompt prefix string. Must contain variables "top_k" and "dialect".
suffix: Prompt suffix string. Default depends on agent type.
input_variables: DEPRECATED.
top_k: Number of rows to query for by default.
max_iterations: Passed to AgentExecutor init.
max_execution_time: Passed to AgentExecutor init.
early_stopping_method: Passed to AgentExecutor init.
verbose: AgentExecutor verbosity.
agent_executor_kwargs: Arbitrary additional AgentExecutor args.
extra_tools: Additional tools to give to agent on top of the ones that come with
SQLDatabaseToolkit.
db: SQLDatabase from which to create a SQLDatabaseToolkit. Toolkit is created
using 'db' and 'llm'. Must provide exactly one of 'db' or 'toolkit'.
prompt: Complete agent prompt. prompt and {prefix, suffix, format_instructions,
input_variables} are mutually exclusive. Must contain variables "top_k" and "dialect".
Can contain variables "table_info" or "table_names" if the prompt requires them.
**kwargs: Arbitrary additional Agent args.
Returns:
An AgentExecutor with the specified agent_type agent.
Example:
.. code-block:: python
from langchain_cohere import ChatCohere, create_sql_agent
from langchain_community.utilities import SQLDatabase
db = SQLDatabase.from_uri("sqlite:///Chinook.db")
llm = ChatCohere(model="command-r-plus", temperature=0)
agent_executor = create_sql_agent(llm, db=db, verbose=True)
resp = agent_executor.run("Show me the first 5 rows of the 'Album' table.")
print(resp.get("output"))
""" # noqa: E501
if toolkit is None and db is None:
raise ValueError(
"Must provide exactly one of 'toolkit' or 'db'. Received neither."
)
if toolkit and db:
raise ValueError(
"Must provide exactly one of 'toolkit' or 'db'. Received both."
)
toolkit = toolkit or SQLDatabaseToolkit(llm=llm, db=db) # type: ignore[arg-type]
tools = toolkit.get_tools() + list(extra_tools)
if prompt is None:
prefix = prefix or SQL_PREFIX
prefix = prefix.format(dialect=toolkit.dialect, top_k=top_k)
suffix = suffix or SQL_FUNCTIONS_SUFFIX
# .bind params get overwritten by .bind_tools params
if "preamble" in llm.__dict__ and not llm.__dict__.get("preamble"):
preamble = SQL_PREAMBLE.format(
dialect=toolkit.dialect,
top_k=top_k,
current_date=datetime.now().strftime("%A, %B %d, %Y %H:%M:%S"),
)
chat_cohere_args = {k: v for k, v in llm.__dict__.items() if v}
chat_cohere_args["preamble"] = preamble
llm = ChatCohere(**chat_cohere_args)
sys_prompt = suffix
else:
# If llm is passed after .bind/.bind_tools, then preamble cannot be passed
sys_prompt = prefix + "\n\n" + suffix
messages: Sequence[
Union[BaseMessage, HumanMessagePromptTemplate, MessagesPlaceholder]
] = [
HumanMessage(content=sys_prompt),
HumanMessagePromptTemplate.from_template("{input}"),
MessagesPlaceholder(variable_name="agent_scratchpad"),
]
prompt = ChatPromptTemplate.from_messages(messages)
else:
if "top_k" in prompt.input_variables:
prompt = prompt.partial(top_k=str(top_k))
if "dialect" in prompt.input_variables:
prompt = prompt.partial(dialect=toolkit.dialect)
if any(key in prompt.input_variables for key in ["table_info", "table_names"]):
db_context = toolkit.get_context()
if "table_info" in prompt.input_variables:
prompt = prompt.partial(table_info=db_context["table_info"])
tools = [
tool for tool in tools if not isinstance(tool, InfoSQLDatabaseTool)
]
if "table_names" in prompt.input_variables:
prompt = prompt.partial(table_names=db_context["table_names"])
tools = [
tool for tool in tools if not isinstance(tool, ListSQLDatabaseTool)
]
runnable = create_tool_calling_agent(llm, tools, prompt) # type: ignore
agent = RunnableMultiActionAgent( # type: ignore[assignment]
runnable=runnable,
input_keys_arg=["input"],
return_keys_arg=["output"],
**kwargs,
)
return AgentExecutor(
name="Cohere SQL Agent Executor",
agent=agent,
tools=tools,
callback_manager=callback_manager,
verbose=verbose,
max_iterations=max_iterations,
max_execution_time=max_execution_time,
early_stopping_method=early_stopping_method,
**(agent_executor_kwargs or {}),
)
