ragflow/agent/tools/tavily.py

228 lines
9.5 KiB
Python
Raw Normal View History

#
# Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
#
# 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.
#
import logging
import os
import time
from abc import ABC
from tavily import TavilyClient
from agent.tools.base import ToolParamBase, ToolBase, ToolMeta
from api.utils.api_utils import timeout
class TavilySearchParam(ToolParamBase):
"""
Define the Retrieval component parameters.
"""
def __init__(self):
self.meta:ToolMeta = {
"name": "tavily_search",
"description": """
Tavily is a search engine optimized for LLMs, aimed at efficient, quick and persistent search results.
When searching:
- Start with specific query which should focus on just a single aspect.
- Number of keywords in query should be less than 5.
- Broaden search terms if needed
- Cross-reference information from multiple sources
""",
"parameters": {
"query": {
"type": "string",
"description": "The search keywords to execute with Tavily. The keywords should be the most important words/terms(includes synonyms) from the original request.",
"default": "{sys.query}",
"required": True
},
"topic": {
"type": "string",
"description": "default:general. The category of the search.news is useful for retrieving real-time updates, particularly about politics, sports, and major current events covered by mainstream media sources. general is for broader, more general-purpose searches that may include a wide range of sources.",
"enum": ["general", "news"],
"default": "general",
"required": False,
},
"include_domains": {
"type": "array",
"description": "default:[]. A list of domains only from which the search results can be included.",
"default": [],
"items": {
"type": "string",
"description": "Domain name that must be included, e.g. www.yahoo.com"
},
"required": False
},
"exclude_domains": {
"type": "array",
"description": "default:[]. A list of domains from which the search results can not be included",
"default": [],
"items": {
"type": "string",
"description": "Domain name that must be excluded, e.g. www.yahoo.com"
},
"required": False
},
}
}
super().__init__()
self.api_key = ""
self.search_depth = "basic" # basic/advanced
self.max_results = 6
self.days = 14
self.include_answer = False
self.include_raw_content = False
self.include_images = False
self.include_image_descriptions = False
def check(self):
self.check_valid_value(self.topic, "Tavily topic: should be in 'general/news'", ["general", "news"])
self.check_valid_value(self.search_depth, "Tavily search depth should be in 'basic/advanced'", ["basic", "advanced"])
self.check_positive_integer(self.max_results, "Tavily max result number should be within [1 20]")
self.check_positive_integer(self.days, "Tavily days should be greater than 1")
def get_input_form(self) -> dict[str, dict]:
return {
"query": {
"name": "Query",
"type": "line"
}
}
class TavilySearch(ToolBase, ABC):
component_name = "TavilySearch"
@timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 12)))
def _invoke(self, **kwargs):
if not kwargs.get("query"):
self.set_output("formalized_content", "")
return ""
self.tavily_client = TavilyClient(api_key=self._param.api_key)
last_e = None
for fld in ["search_depth", "topic", "max_results", "days", "include_answer", "include_raw_content", "include_images", "include_image_descriptions", "include_domains", "exclude_domains"]:
if fld not in kwargs:
kwargs[fld] = getattr(self._param, fld)
for _ in range(self._param.max_retries+1):
try:
kwargs["include_images"] = False
kwargs["include_raw_content"] = False
res = self.tavily_client.search(**kwargs)
self._retrieve_chunks(res["results"],
get_title=lambda r: r["title"],
get_url=lambda r: r["url"],
get_content=lambda r: r["raw_content"] if r["raw_content"] else r["content"],
get_score=lambda r: r["score"])
self.set_output("json", res["results"])
return self.output("formalized_content")
except Exception as e:
last_e = e
logging.exception(f"Tavily error: {e}")
time.sleep(self._param.delay_after_error)
if last_e:
self.set_output("_ERROR", str(last_e))
return f"Tavily error: {last_e}"
assert False, self.output()
def thoughts(self) -> str:
return """
Keywords: {}
Looking for the most relevant articles.
""".format(self.get_input().get("query", "-_-!"))
class TavilyExtractParam(ToolParamBase):
"""
Define the Retrieval component parameters.
"""
def __init__(self):
self.meta:ToolMeta = {
"name": "tavily_extract",
"description": "Extract web page content from one or more specified URLs using Tavily Extract.",
"parameters": {
"urls": {
"type": "array",
"description": "The URLs to extract content from.",
"default": "",
"items": {
"type": "string",
"description": "The URL to extract content from, e.g. www.yahoo.com"
},
"required": True
},
"extract_depth": {
"type": "string",
"description": "The depth of the extraction process. advanced extraction retrieves more data, including tables and embedded content, with higher success but may increase latency.basic extraction costs 1 credit per 5 successful URL extractions, while advanced extraction costs 2 credits per 5 successful URL extractions.",
"enum": ["basic", "advanced"],
"default": "basic",
"required": False,
},
"format": {
"type": "string",
"description": "The format of the extracted web page content. markdown returns content in markdown format. text returns plain text and may increase latency.",
"enum": ["markdown", "text"],
"default": "markdown",
"required": False,
}
}
}
super().__init__()
self.api_key = ""
self.extract_depth = "basic" # basic/advanced
self.urls = []
self.format = "markdown"
self.include_images = False
def check(self):
self.check_valid_value(self.extract_depth, "Tavily extract depth should be in 'basic/advanced'", ["basic", "advanced"])
self.check_valid_value(self.format, "Tavily extract format should be in 'markdown/text'", ["markdown", "text"])
def get_input_form(self) -> dict[str, dict]:
return {
"urls": {
"name": "URLs",
"type": "line"
}
}
class TavilyExtract(ToolBase, ABC):
component_name = "TavilyExtract"
@timeout(int(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60)))
def _invoke(self, **kwargs):
self.tavily_client = TavilyClient(api_key=self._param.api_key)
last_e = None
for fld in ["urls", "extract_depth", "format"]:
if fld not in kwargs:
kwargs[fld] = getattr(self._param, fld)
if kwargs.get("urls") and isinstance(kwargs["urls"], str):
kwargs["urls"] = kwargs["urls"].split(",")
for _ in range(self._param.max_retries+1):
try:
kwargs["include_images"] = False
res = self.tavily_client.extract(**kwargs)
self.set_output("json", res["results"])
return self.output("json")
except Exception as e:
last_e = e
logging.exception(f"Tavily error: {e}")
if last_e:
self.set_output("_ERROR", str(last_e))
return f"Tavily error: {last_e}"
assert False, self.output()
def thoughts(self) -> str:
return "Opened {}—pulling out the main text…".format(self.get_input().get("urls", "-_-!"))