feat: 异步重写

perf: 优化网页搜索回答规范
This commit is contained in:
Soulter
2024-03-03 18:54:50 +08:00
parent 53ef3bbf4f
commit 1cd1c8ea0d
12 changed files with 205 additions and 222 deletions
+54 -34
View File
@@ -1,18 +1,19 @@
import requests
import util.general_utils as gu
from bs4 import BeautifulSoup
import traceback
import time
import json
import asyncio
from googlesearch import search, SearchResult
from readability import Document
from bs4 import BeautifulSoup
from openai.types.chat.chat_completion_message_tool_call import Function
from util.function_calling.func_call import (
FuncCall,
FuncCallJsonFormatError,
FuncNotFoundError
)
from openai.types.chat.chat_completion_message_tool_call import Function
import traceback
from googlesearch import search, SearchResult
from model.provider.provider import Provider
import json
from readability import Document
def tidy_text(text: str) -> str:
@@ -53,11 +54,11 @@ def google_web_search(keyword) -> str:
for i in ls:
desc = i.description
try:
gu.log(f"搜索网页: {i.url}", tag="网页搜索", level=gu.LEVEL_INFO)
# gu.log(f"搜索网页: {i.url}", tag="网页搜索", level=gu.LEVEL_INFO)
desc = fetch_website_content(i.url)
except BaseException as e:
print(f"(google) fetch_website_content err: {str(e)}")
gu.log(f"# No.{str(index)}\ntitle: {i.title}\nurl: {i.url}\ncontent: {desc}\n\n", level=gu.LEVEL_DEBUG, max_len=9999)
# gu.log(f"# No.{str(index)}\ntitle: {i.title}\nurl: {i.url}\ncontent: {desc}\n\n", level=gu.LEVEL_DEBUG, max_len=9999)
ret += f"# No.{str(index)}\ntitle: {i.title}\nurl: {i.url}\ncontent: {desc}\n\n"
index += 1
except Exception as e:
@@ -80,7 +81,7 @@ def web_keyword_search_via_bing(keyword) -> str:
try:
response = requests.get(url, headers=headers)
response.encoding = "utf-8"
gu.log(f"bing response: {response.text}", tag="bing", level=gu.LEVEL_DEBUG, max_len=9999)
# gu.log(f"bing response: {response.text}", tag="bing", level=gu.LEVEL_DEBUG, max_len=9999)
soup = BeautifulSoup(response.text, "html.parser")
res = ""
result_cnt = 0
@@ -96,7 +97,7 @@ def web_keyword_search_via_bing(keyword) -> str:
# "link": link,
# })
try:
gu.log(f"搜索网页: {link}", tag="网页搜索", level=gu.LEVEL_INFO)
# gu.log(f"搜索网页: {link}", tag="网页搜索", level=gu.LEVEL_INFO)
desc = fetch_website_content(link)
except BaseException as e:
print(f"(bing) fetch_website_content err: {str(e)}")
@@ -124,11 +125,11 @@ def web_keyword_search_via_bing(keyword) -> str:
if result_cnt == 0: break
return res
except Exception as e:
gu.log(f"bing fetch err: {str(e)}")
# gu.log(f"bing fetch err: {str(e)}")
_cnt += 1
time.sleep(1)
gu.log("fail to fetch bing info, using sougou.")
# gu.log("fail to fetch bing info, using sougou.")
return web_keyword_search_via_sougou(keyword)
def web_keyword_search_via_sougou(keyword) -> str:
@@ -157,7 +158,7 @@ def web_keyword_search_via_sougou(keyword) -> str:
break
except Exception as e:
pass
gu.log(f"sougou parse err: {str(e)}", tag="web_keyword_search_via_sougou", level=gu.LEVEL_ERROR)
# gu.log(f"sougou parse err: {str(e)}", tag="web_keyword_search_via_sougou", level=gu.LEVEL_ERROR)
# 爬取网页内容
_detail_store = []
for i in res:
@@ -173,7 +174,7 @@ def web_keyword_search_via_sougou(keyword) -> str:
return ret
def fetch_website_content(url):
gu.log(f"fetch_website_content: {url}", tag="fetch_website_content", level=gu.LEVEL_DEBUG)
# gu.log(f"fetch_website_content: {url}", tag="fetch_website_content", level=gu.LEVEL_DEBUG)
headers = {
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) \
AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
@@ -187,7 +188,7 @@ def fetch_website_content(url):
ret = tidy_text(soup.get_text())
return ret
def web_search(question, provider: Provider, session_id, official_fc=False):
async def web_search(question, provider: Provider, session_id, official_fc=False):
'''
official_fc: 使用官方 function-calling
'''
@@ -197,7 +198,7 @@ def web_search(question, provider: Provider, session_id, official_fc=False):
"name": "keyword",
"description": "google search query (分词,尽量保留所有信息)"
}],
"通过搜索引擎搜索。如果问题需要在网页上搜索(如天气、新闻或任何需要通过网页获取信息的问题),则调用此函数;如果没有,不要调用此函数。",
"通过搜索引擎搜索。如果问题需要获取近期、实时的消息,在网页上搜索(如天气、新闻或任何需要通过网页获取信息的问题),则调用此函数;如果没有,不要调用此函数。",
web_keyword_search_via_bing
)
new_func_call.add_func("fetch_website_content", [{
@@ -205,16 +206,16 @@ def web_search(question, provider: Provider, session_id, official_fc=False):
"name": "url",
"description": "网址"
}],
"获取网页的内容。如果问题带有合法的网页链接(例如: `帮我总结一下https://github.com的内容`), 就调用此函数。如果没有,不要调用此函数。",
"获取网页的内容。如果问题带有合法的网页链接(例如: `帮我总结一下 https://github.com 的内容`), 就调用此函数。如果没有,不要调用此函数。",
fetch_website_content
)
question1 = f"{question} \n> hint: 最多只能调用1个function, 并且存在不会调用任何function的可能性。"
has_func = False
function_invoked_ret = ""
if official_fc:
func = provider.text_chat(question1, session_id, function_call=new_func_call.get_func())
# we use official function-calling
func = await provider.text_chat(question1, session_id, function_call=new_func_call.get_func())
if isinstance(func, Function):
# arguments='{\n "keyword": "北京今天的天气"\n}', name='google_web_search'
# 执行对应的结果:
func_obj = None
for i in new_func_call.func_list:
@@ -222,49 +223,68 @@ def web_search(question, provider: Provider, session_id, official_fc=False):
func_obj = i["func_obj"]
break
if not func_obj:
gu.log("找不到返回的 func name " + func.name, level=gu.LEVEL_ERROR)
return provider.text_chat(question1, session_id) + "\n(网页搜索失败, 此为默认回复)"
# gu.log("找不到返回的 func name " + func.name, level=gu.LEVEL_ERROR)
return await provider.text_chat(question1, session_id) + "\n(网页搜索失败, 此为默认回复)"
try:
args = json.loads(func.arguments)
function_invoked_ret = func_obj(**args)
# we use to_thread to avoid blocking the event loop
function_invoked_ret = await asyncio.to_thread(func_obj, **args)
has_func = True
except BaseException as e:
traceback.print_exc()
return provider.text_chat(question1, session_id) + "\n(网页搜索失败, 此为默认回复)"
return await provider.text_chat(question1, session_id) + "\n(网页搜索失败, 此为默认回复)"
else:
# now func is a string
return func
else:
# we use our own function-calling
try:
function_invoked_ret, has_func = new_func_call.func_call(question1, new_func_call.func_dump(), is_task=False, is_summary=False)
args = {
'question': question1,
'func_definition': new_func_call.func_dump(),
'is_task': False,
'is_summary': False,
}
function_invoked_ret, has_func = await asyncio.to_thread(new_func_call.func_call, **args)
except BaseException as e:
res = provider.text_chat(question) + "\n(网页搜索失败, 此为默认回复)"
res = await provider.text_chat(question) + "\n(网页搜索失败, 此为默认回复)"
return res
has_func = True
if has_func:
provider.forget(session_id)
await provider.forget(session_id)
question3 = f"""
以下是相关材料,你的任务是:
1. 根据材料对问题`{question}`做切题的总结回答;
2. 发表你对这个问题的看法.
你的任务是:
1. 根据末尾的材料对问题`{question}`做切题的总结(详细);
2. 简单地发表你对这个问题的看法(简略)。
你的总结末尾应当有对材料的引用, 如果有链接, 请在末尾附上引用网页链接。引用格式严格按照 `\n[1] title url \n`。
不要提到任何函数调用的信息。以下是相关材料:
不要提到任何函数调用的信息。
一些回复的消息模板:
模板1:
```
从网上的信息来看,可以知道...我个人认为...你觉得呢?
```
模板2:
```
根据网上的最新信息,可以得知...我觉得...你怎么看?
```
你可以根据这些模板来组织回答,但可以不照搬,要根据问题的内容来回答。
以下是相关材料:
"""
gu.log(f"web_search: {question3}", tag="web_search", level=gu.LEVEL_DEBUG, max_len=99999)
_c = 0
while _c < 3:
try:
print('text chat')
final_ret = provider.text_chat(question3 + "```" + function_invoked_ret + "```", session_id)
final_ret = await provider.text_chat(question3 + "```" + function_invoked_ret + "```", session_id)
return final_ret
except Exception as e:
print(e)
_c += 1
if _c == 3: raise e
if "The message you submitted was too long" in str(e):
provider.forget(session_id)
await provider.forget(session_id)
function_invoked_ret = function_invoked_ret[:int(len(function_invoked_ret) / 2)]
time.sleep(3)
return function_invoked_ret