async-bing-client

中文文档

Features

• Use API similar to the Newbing website
• Support image recognition by providing images as input
• Enable human language-based drawing
• Facilitate persistent storage of session data
• Organize WebSocket communication (WSS) information into custom classes, supporting output categorization for easy developer encapsulation
• High concurrency processing capability
• Rapid retrieval of chat data
• Bulk conversation deletion functionality
• No need for Tun mode proxy
• Support HTTP and SOCKS proxies

Usage Guide

Table of Contents

Step 1: Create bing client

Step 2: Use client

• [1]. Create new conversation

• [2]. Get all cached chat lists

• [3]. Reload cached chat lists

• [4]. Delete specified number of chats (up to 200)

• [5]. Delete specified chat

• [5]. Chat with bing

• [6]. AI drawing (provided by openai's dall-e)

You can click the table of contents to jump to the corresponding content.

Step 1: Create bing client

1. Bing_Client: Parameters:
   • (1).cookie: bing.com cookie, need to use browser extension cookie-editor Export cookie to json in bing chat interface, then pass file path of str or Path, can also directly pass list[dict] formatted cookie
   • (2).proxy: LAN proxy, need to fill in when there is no global proxy or split tunneling on local machine, support http, httpx, socks5

import asyncio

from async_bing_client import Bing_Client

client = Bing_Client(cookie="cookie.json", proxy="socks5://127.0.0.1:7890")

if __name__ == '__main__':
    async def main():
        await client.init()


    asyncio.run(main())

Step 2: Use client

[1]. Create new conversation

# The new conversation created here can be directly used for ask_stream
chat = await client.create_chat()

[2]. Get all cached chat lists

Contains chat history of all latest 200 conversations

chat_list = client.chat_list

[3]. Reload cached chat lists

# Just reinitialize 
await client.init()

[4]. Delete specified number of chats (up to 200)

Parameters:

• (1). count: int = 20: Number to delete
• (2). del_all: bool = False: Whether to delete all (up to 200, because bing can only load latest 200 conversations)

await client.delete_conversation_by_count()

[5]. Delete specified chat

Parameters:

• (1). conversation_id:str: conversation_id of chat to delete (chat key is conversation_id)

await client.delete_conversation('conversation_id')

[5]. Chat with bing

Can pass pictures for image recognition, can also return Image type or md image format including new ai drawings generated by bing

1. Return different generator functions: ask_stream_raw

Parameters:

• (1). question: str: Your question
• (2). image: str | Path | bytes = None: Image you want to recognize (str here can be link of web image, can also be str of path)
• (3). chat: dict = None: Chat window to send to, will create new one if not passed, and yield a NewChat class containing the new chat
• (4). conversation_style: ConversationStyle = ConversationStyle.Creative: Conversation style to use
• (5). personality=None: Preset personality to load (just needs to be normal text, will auto obfuscate), content must be within bing's moral and legal limits, otherwise will keep returning Apology
• (6). locale:str =guess_locale(): Locale to use (usually don't need to specify yourself)

Return types: AsyncGenerator[NewChat | Apology | Notice | SearchResult | Text | SourceAttribution | SuggestRely | Limit | Response]

• Text: bing's text reply. Can access content attribute or use str method to get content. Addition of Text will still return Text
• Image: bing's image reply. url attribute is image link, base64 (Optional) attribute is image base64 content, name is image name
• NewChat: Newly generated chat, can access chat attribute of NewChat to obtain.
• Apology: bing's refusal or error message. Can access content attribute or use str method to obtain content
• Notice: bing's notification message. Can access content attribute or use str method to obtain content
• SearchResult: bing's search results. content attribute stores list[dict] or str. Can use str method to convert to suitable str format
• SourceAttribution: bing's resource link reply, display_name attribute is displayed link name, see_more_url attribute is link, image (optional) attribute is resource preview image
• SuggestRely: bing's suggested reply. Can access content attribute or use str method to obtain content
• Limit: bing's conversation limit content, will return Apology after exceeding limit. num_user_messages attribute is number of messages user has sent, max_num_user_messages is max number of consecutive messages user can send
• Response: Summary of bing's messages in this conversation, content stores dict of all data generated in this conversation. ask_stream function wraps ask_stream_raw example

chat = client.create_chat()
sources = []
suggest_reply = []
images = []
limit = None
async for data in client.ask_stream_raw('question', 'image', chat=chat):
    if isinstance(data, Text):
        yield data.content
    elif isinstance(data, SuggestRely):
        suggest_reply.append(data)
    elif isinstance(data, SourceAttribution):
        sources.append(data)
    elif isinstance(data, Apology):
        yield '\n' + data.content
    elif isinstance(data, Image):
        images.append(data)
    elif isinstance(data, Limit):
        limit = data
    elif isinstance(data, SearchResult) and yield_search:
        yield str(data)
for index, source in enumerate(sources):
    if index == 0:
        yield "\nSee more:  \n"
    yield f"({index + 1}):{source}  \n"
for index, image in enumerate(images):
    if index == 0:
        yield "\nDrew images:  \n"
    yield f"{index + 1}:{image}  \n"
for index, reply in enumerate(suggest_reply):
    if index == 0:
        yield "\nSuggest Replys:  \n"
    yield f"{index + 1}:{reply.content}  \n"
if limit:
    yield f"\n\nLimit:{limit.num_user_messages} of {limit.max_num_user_messages}  "

1. Generator function outputting pure str ask_stream:

async for text in client.ask_stream("hello", chat=chat,
                                    yield_search=False,
                                    conversation_style=ConversationStyle.Balanced):
    print(text, end="")

[6]. AI drawing (provided by openai's dall-e)

(This function will be called automatically in ask_stream, can directly use human language to let bing generate images) Return value: List[Image]

Image: bing's image reply. url attribute is image link, name is image name

images = await client.draw("drawing prompt")