File size: 13,537 Bytes
b585c7f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 |
"""Agent for working with pandas objects."""
from io import IOBase
from typing import Any, Dict, List, Optional, Sequence, Tuple, Union
from langchain._api import warn_deprecated
from langchain.agents import AgentExecutor, BaseSingleActionAgent
from langchain_experimental.agents.agent_toolkits.pandas.prompt import (
FUNCTIONS_WITH_DF,
FUNCTIONS_WITH_MULTI_DF,
MULTI_DF_PREFIX,
MULTI_DF_PREFIX_FUNCTIONS,
PREFIX,
PREFIX_FUNCTIONS,
SUFFIX_NO_DF,
SUFFIX_WITH_DF,
SUFFIX_WITH_MULTI_DF,
)
from langchain.agents.mrkl.base import ZeroShotAgent
from langchain.agents.mrkl.prompt import FORMAT_INSTRUCTIONS
from langchain.agents.openai_functions_agent.base import OpenAIFunctionsAgent
from langchain.agents.types import AgentType
from langchain.callbacks.base import BaseCallbackManager
from langchain.chains.llm import LLMChain
from langchain.schema import BasePromptTemplate
from langchain.schema.language_model import BaseLanguageModel
from langchain.schema.messages import SystemMessage
from langchain.tools import BaseTool
from langchain_experimental.tools.python.tool import PythonAstREPLTool
def _get_multi_prompt(
dfs: List[Any],
prefix: Optional[str] = None,
suffix: Optional[str] = None,
input_variables: Optional[List[str]] = None,
include_df_in_prompt: Optional[bool] = True,
number_of_head_rows: int = 5,
) -> Tuple[BasePromptTemplate, List[PythonAstREPLTool]]:
num_dfs = len(dfs)
if suffix is not None:
suffix_to_use = suffix
include_dfs_head = True
elif include_df_in_prompt:
suffix_to_use = SUFFIX_WITH_MULTI_DF
include_dfs_head = True
else:
suffix_to_use = SUFFIX_NO_DF
include_dfs_head = False
if input_variables is None:
input_variables = ["input", "agent_scratchpad", "num_dfs"]
if include_dfs_head:
input_variables += ["dfs_head"]
if prefix is None:
prefix = MULTI_DF_PREFIX
df_locals = {}
for i, dataframe in enumerate(dfs):
df_locals[f"df{i + 1}"] = dataframe
tools = [PythonAstREPLTool(locals=df_locals)]
prompt = ZeroShotAgent.create_prompt(
tools, prefix=prefix, suffix=suffix_to_use, input_variables=input_variables
)
partial_prompt = prompt.partial()
if "dfs_head" in input_variables:
dfs_head = "\n\n".join([d.head(number_of_head_rows).to_markdown() for d in dfs])
partial_prompt = partial_prompt.partial(num_dfs=str(num_dfs), dfs_head=dfs_head)
if "num_dfs" in input_variables:
partial_prompt = partial_prompt.partial(num_dfs=str(num_dfs))
return partial_prompt, tools
def _get_single_prompt(
df: Any,
prefix: Optional[str] = None,
suffix: Optional[str] = None,
input_variables: Optional[List[str]] = None,
include_df_in_prompt: Optional[bool] = True,
number_of_head_rows: int = 5,
format_instructions=FORMAT_INSTRUCTIONS,
) -> Tuple[BasePromptTemplate, List[PythonAstREPLTool]]:
if suffix is not None:
suffix_to_use = suffix
include_df_head = True
elif include_df_in_prompt:
suffix_to_use = SUFFIX_WITH_DF
include_df_head = True
else:
suffix_to_use = SUFFIX_NO_DF
include_df_head = False
if input_variables is None:
input_variables = ["input", "agent_scratchpad"]
if include_df_head:
input_variables += ["df_head"]
if prefix is None:
prefix = PREFIX
tools = [PythonAstREPLTool(locals={"df": df})]
prompt = ZeroShotAgent.create_prompt(
tools, prefix=prefix, suffix=suffix_to_use, input_variables=input_variables,
format_instructions=format_instructions,
)
partial_prompt = prompt.partial()
if "df_head" in input_variables:
partial_prompt = partial_prompt.partial(
df_head=str(df.head(number_of_head_rows).to_markdown())
)
return partial_prompt, tools
def _get_prompt_and_tools(
df: Any,
prefix: Optional[str] = None,
suffix: Optional[str] = None,
input_variables: Optional[List[str]] = None,
include_df_in_prompt: Optional[bool] = True,
number_of_head_rows: int = 5,
format_instructions=FORMAT_INSTRUCTIONS,
) -> Tuple[BasePromptTemplate, List[PythonAstREPLTool]]:
try:
import pandas as pd
pd.set_option("display.max_columns", None)
except ImportError:
raise ImportError(
"pandas package not found, please install with `pip install pandas`"
)
if include_df_in_prompt is not None and suffix is not None:
raise ValueError("If suffix is specified, include_df_in_prompt should not be.")
if isinstance(df, list):
for item in df:
if not isinstance(item, pd.DataFrame):
raise ValueError(f"Expected pandas object, got {type(df)}")
return _get_multi_prompt(
df,
prefix=prefix,
suffix=suffix,
input_variables=input_variables,
include_df_in_prompt=include_df_in_prompt,
number_of_head_rows=number_of_head_rows,
)
else:
if not isinstance(df, pd.DataFrame):
raise ValueError(f"Expected pandas object, got {type(df)}")
return _get_single_prompt(
df,
prefix=prefix,
suffix=suffix,
input_variables=input_variables,
include_df_in_prompt=include_df_in_prompt,
number_of_head_rows=number_of_head_rows,
format_instructions=format_instructions,
)
def _get_functions_single_prompt(
df: Any,
prefix: Optional[str] = None,
suffix: Optional[str] = None,
include_df_in_prompt: Optional[bool] = True,
number_of_head_rows: int = 5,
) -> Tuple[BasePromptTemplate, List[PythonAstREPLTool]]:
if suffix is not None:
suffix_to_use = suffix
if include_df_in_prompt:
suffix_to_use = suffix_to_use.format(
df_head=str(df.head(number_of_head_rows).to_markdown())
)
elif include_df_in_prompt:
suffix_to_use = FUNCTIONS_WITH_DF.format(
df_head=str(df.head(number_of_head_rows).to_markdown())
)
else:
suffix_to_use = ""
if prefix is None:
prefix = PREFIX_FUNCTIONS
tools = [PythonAstREPLTool(locals={"df": df})]
system_message = SystemMessage(content=prefix + suffix_to_use)
prompt = OpenAIFunctionsAgent.create_prompt(system_message=system_message)
return prompt, tools
def _get_functions_multi_prompt(
dfs: Any,
prefix: Optional[str] = None,
suffix: Optional[str] = None,
include_df_in_prompt: Optional[bool] = True,
number_of_head_rows: int = 5,
) -> Tuple[BasePromptTemplate, List[PythonAstREPLTool]]:
if suffix is not None:
suffix_to_use = suffix
if include_df_in_prompt:
dfs_head = "\n\n".join(
[d.head(number_of_head_rows).to_markdown() for d in dfs]
)
suffix_to_use = suffix_to_use.format(
dfs_head=dfs_head,
)
elif include_df_in_prompt:
dfs_head = "\n\n".join([d.head(number_of_head_rows).to_markdown() for d in dfs])
suffix_to_use = FUNCTIONS_WITH_MULTI_DF.format(
dfs_head=dfs_head,
)
else:
suffix_to_use = ""
if prefix is None:
prefix = MULTI_DF_PREFIX_FUNCTIONS
prefix = prefix.format(num_dfs=str(len(dfs)))
df_locals = {}
for i, dataframe in enumerate(dfs):
df_locals[f"df{i + 1}"] = dataframe
tools = [PythonAstREPLTool(locals=df_locals)]
system_message = SystemMessage(content=prefix + suffix_to_use)
prompt = OpenAIFunctionsAgent.create_prompt(system_message=system_message)
return prompt, tools
def _get_functions_prompt_and_tools(
df: Any,
prefix: Optional[str] = None,
suffix: Optional[str] = None,
input_variables: Optional[List[str]] = None,
include_df_in_prompt: Optional[bool] = True,
number_of_head_rows: int = 5,
) -> Tuple[BasePromptTemplate, List[PythonAstREPLTool]]:
try:
import pandas as pd
pd.set_option("display.max_columns", None)
except ImportError:
raise ImportError(
"pandas package not found, please install with `pip install pandas`"
)
if input_variables is not None:
raise ValueError("`input_variables` is not supported at the moment.")
if include_df_in_prompt is not None and suffix is not None:
raise ValueError("If suffix is specified, include_df_in_prompt should not be.")
if isinstance(df, list):
for item in df:
if not isinstance(item, pd.DataFrame):
raise ValueError(f"Expected pandas object, got {type(df)}")
return _get_functions_multi_prompt(
df,
prefix=prefix,
suffix=suffix,
include_df_in_prompt=include_df_in_prompt,
number_of_head_rows=number_of_head_rows,
)
else:
if not isinstance(df, pd.DataFrame):
raise ValueError(f"Expected pandas object, got {type(df)}")
return _get_functions_single_prompt(
df,
prefix=prefix,
suffix=suffix,
include_df_in_prompt=include_df_in_prompt,
number_of_head_rows=number_of_head_rows,
)
def create_pandas_dataframe_agent(
llm: BaseLanguageModel,
df: Any,
agent_type: AgentType = AgentType.ZERO_SHOT_REACT_DESCRIPTION,
callback_manager: Optional[BaseCallbackManager] = None,
prefix: Optional[str] = None,
suffix: Optional[str] = None,
input_variables: Optional[List[str]] = None,
verbose: bool = False,
return_intermediate_steps: bool = False,
max_iterations: Optional[int] = 15,
max_execution_time: Optional[float] = None,
early_stopping_method: str = "force",
agent_executor_kwargs: Optional[Dict[str, Any]] = None,
include_df_in_prompt: Optional[bool] = True,
number_of_head_rows: int = 5,
extra_tools: Sequence[BaseTool] = (),
format_instructions="",
**kwargs: Any,
) -> AgentExecutor:
"""Construct a pandas agent from an LLM and dataframe."""
warn_deprecated(
since="0.0.314",
message=(
"On 2023-10-27 this module will be be deprecated from langchain, and "
"will be available from the langchain-experimental package."
"This code is already available in langchain-experimental."
"See https://github.com/langchain-ai/langchain/discussions/11680."
),
pending=True,
)
agent: BaseSingleActionAgent
if agent_type == AgentType.ZERO_SHOT_REACT_DESCRIPTION:
prompt, base_tools = _get_prompt_and_tools(
df,
prefix=prefix,
suffix=suffix,
input_variables=input_variables,
include_df_in_prompt=include_df_in_prompt,
number_of_head_rows=number_of_head_rows,
format_instructions=format_instructions,
)
tools = base_tools + list(extra_tools)
llm_chain = LLMChain(
llm=llm,
prompt=prompt,
callback_manager=callback_manager,
)
tool_names = [tool.name for tool in tools]
agent = ZeroShotAgent(
llm_chain=llm_chain,
allowed_tools=tool_names,
callback_manager=callback_manager,
**kwargs,
)
elif agent_type == AgentType.OPENAI_FUNCTIONS:
_prompt, base_tools = _get_functions_prompt_and_tools(
df,
prefix=prefix,
suffix=suffix,
input_variables=input_variables,
include_df_in_prompt=include_df_in_prompt,
number_of_head_rows=number_of_head_rows,
)
tools = base_tools + list(extra_tools)
agent = OpenAIFunctionsAgent(
llm=llm,
prompt=_prompt,
tools=tools,
callback_manager=callback_manager,
**kwargs,
)
else:
raise ValueError(f"Agent type {agent_type} not supported at the moment.")
return AgentExecutor.from_agent_and_tools(
agent=agent,
tools=tools,
callback_manager=callback_manager,
verbose=verbose,
return_intermediate_steps=return_intermediate_steps,
max_iterations=max_iterations,
max_execution_time=max_execution_time,
early_stopping_method=early_stopping_method,
**(agent_executor_kwargs or {}),
)
def create_csv_agent(
llm: BaseLanguageModel,
path: Union[str, IOBase, List[Union[str, IOBase]]],
pandas_kwargs: Optional[dict] = None,
**kwargs: Any,
) -> AgentExecutor:
"""Create csv agent by loading to a dataframe and using pandas agent."""
try:
import pandas as pd
except ImportError:
raise ImportError(
"pandas package not found, please install with `pip install pandas`"
)
_kwargs = pandas_kwargs or {}
if isinstance(path, (str, IOBase)):
df = pd.read_csv(path, **_kwargs)
elif isinstance(path, list):
df = []
for item in path:
if not isinstance(item, (str, IOBase)):
raise ValueError(f"Expected str or file-like object, got {type(path)}")
df.append(pd.read_csv(item, **_kwargs))
else:
raise ValueError(f"Expected str, list, or file-like object, got {type(path)}")
return create_pandas_dataframe_agent(llm, df, **kwargs)
|