Scrapegraph-ai/scrapegraphai/nodes/generate_answer_node.py

197 lines
7.9 KiB
Python

"""
GenerateAnswerNode Module
"""
from typing import List, Optional
from json.decoder import JSONDecodeError
import time
from langchain.prompts import PromptTemplate
from langchain_core.output_parsers import JsonOutputParser
from langchain_core.runnables import RunnableParallel
from langchain_openai import ChatOpenAI, AzureChatOpenAI
from langchain_aws import ChatBedrock
from langchain_community.chat_models import ChatOllama
from tqdm import tqdm
from .base_node import BaseNode
from ..utils.output_parser import get_pydantic_output_parser
from requests.exceptions import Timeout
from ..prompts import (
TEMPLATE_CHUNKS, TEMPLATE_NO_CHUNKS, TEMPLATE_MERGE,
TEMPLATE_CHUNKS_MD, TEMPLATE_NO_CHUNKS_MD, TEMPLATE_MERGE_MD
)
class GenerateAnswerNode(BaseNode):
"""
Initializes the GenerateAnswerNode class.
Args:
input (str): The input data type for the node.
output (List[str]): The output data type(s) for the node.
node_config (Optional[dict]): Configuration dictionary for the node,
which includes the LLM model, verbosity, schema, and other settings.
Defaults to None.
node_name (str): The name of the node. Defaults to "GenerateAnswer".
Attributes:
llm_model: The language model specified in the node configuration.
verbose (bool): Whether verbose mode is enabled.
force (bool): Whether to force certain behaviors, overriding defaults.
script_creator (bool): Whether the node is in script creation mode.
is_md_scraper (bool): Whether the node is scraping markdown data.
additional_info (Optional[str]): Any additional information to be
included in the prompt templates.
"""
def __init__(
self,
input: str,
output: List[str],
node_config: Optional[dict] = None,
node_name: str = "GenerateAnswer",
):
super().__init__(node_name, "node", input, output, 2, node_config)
self.llm_model = node_config["llm_model"]
if isinstance(node_config["llm_model"], ChatOllama):
self.llm_model.format = "json"
self.verbose = node_config.get("verbose", False)
self.force = node_config.get("force", False)
self.script_creator = node_config.get("script_creator", False)
self.is_md_scraper = node_config.get("is_md_scraper", False)
self.additional_info = node_config.get("additional_info")
self.timeout = node_config.get("timeout", 120)
def invoke_with_timeout(self, chain, inputs, timeout):
"""Helper method to invoke chain with timeout"""
try:
start_time = time.time()
response = chain.invoke(inputs)
if time.time() - start_time > timeout:
raise Timeout(f"Response took longer than {timeout} seconds")
return response
except Timeout as e:
self.logger.error(f"Timeout error: {str(e)}")
raise
except Exception as e:
self.logger.error(f"Error during chain execution: {str(e)}")
raise
def execute(self, state: dict) -> dict:
"""
Executes the GenerateAnswerNode.
Args:
state (dict): The current state of the graph. The input keys will be used
to fetch the correct data from the state.
Returns:
dict: The updated state with the output key containing the generated answer.
"""
self.logger.info(f"--- Executing {self.node_name} Node ---")
input_keys = self.get_input_keys(state)
input_data = [state[key] for key in input_keys]
user_prompt = input_data[0]
doc = input_data[1]
if self.node_config.get("schema", None) is not None:
if isinstance(self.llm_model, ChatOpenAI):
output_parser = get_pydantic_output_parser(self.node_config["schema"])
format_instructions = output_parser.get_format_instructions()
else:
if not isinstance(self.llm_model, ChatBedrock):
output_parser = get_pydantic_output_parser(self.node_config["schema"])
format_instructions = output_parser.get_format_instructions()
else:
output_parser = None
format_instructions = ""
else:
if not isinstance(self.llm_model, ChatBedrock):
output_parser = JsonOutputParser()
format_instructions = output_parser.get_format_instructions()
else:
output_parser = None
format_instructions = ""
if isinstance(self.llm_model, (ChatOpenAI, AzureChatOpenAI)) \
and not self.script_creator \
or self.force \
and not self.script_creator or self.is_md_scraper:
template_no_chunks_prompt = TEMPLATE_NO_CHUNKS_MD
template_chunks_prompt = TEMPLATE_CHUNKS_MD
template_merge_prompt = TEMPLATE_MERGE_MD
else:
template_no_chunks_prompt = TEMPLATE_NO_CHUNKS
template_chunks_prompt = TEMPLATE_CHUNKS
template_merge_prompt = TEMPLATE_MERGE
if self.additional_info is not None:
template_no_chunks_prompt = self.additional_info + template_no_chunks_prompt
template_chunks_prompt = self.additional_info + template_chunks_prompt
template_merge_prompt = self.additional_info + template_merge_prompt
if len(doc) == 1:
prompt = PromptTemplate(
template=template_no_chunks_prompt,
input_variables=["question"],
partial_variables={"context": doc, "format_instructions": format_instructions}
)
chain = prompt | self.llm_model
if output_parser:
chain = chain | output_parser
try:
answer = self.invoke_with_timeout(chain, {"question": user_prompt}, self.timeout)
except Timeout:
state.update({self.output[0]: {"error": "Response timeout exceeded"}})
return state
state.update({self.output[0]: answer})
return state
chains_dict = {}
for i, chunk in enumerate(tqdm(doc, desc="Processing chunks", disable=not self.verbose)):
prompt = PromptTemplate(
template=template_chunks_prompt,
input_variables=["question"],
partial_variables={"context": chunk,
"chunk_id": i + 1,
"format_instructions": format_instructions}
)
chain_name = f"chunk{i+1}"
chains_dict[chain_name] = prompt | self.llm_model
if output_parser:
chains_dict[chain_name] = chains_dict[chain_name] | output_parser
async_runner = RunnableParallel(**chains_dict)
try:
batch_results = self.invoke_with_timeout(
async_runner,
{"question": user_prompt},
self.timeout
)
except Timeout:
state.update({self.output[0]: {"error": "Response timeout exceeded during chunk processing"}})
return state
merge_prompt = PromptTemplate(
template=template_merge_prompt,
input_variables=["context", "question"],
partial_variables={"format_instructions": format_instructions}
)
merge_chain = merge_prompt | self.llm_model
if output_parser:
merge_chain = merge_chain | output_parser
try:
answer = self.invoke_with_timeout(
merge_chain,
{"context": batch_results, "question": user_prompt},
self.timeout
)
except Timeout:
state.update({self.output[0]: {"error": "Response timeout exceeded during merge"}})
return state
state.update({self.output[0]: answer})
return state