evilpan / gptcli

ChatGPT in command line with OpenAI API (gpt-3.5-turbo/gpt-4/gpt-4-32k)

Take chatGPT into command line.

Setup

1. clone this repo
2. pip3 install -U -r requirements.txt
3. copy demo_config.json to config.json
4. get your OPENAI_API_KEY and put it in config.json

Run

$ ./gptcli.py -h
usage: gptcli.py [-h] [-c CONFIG]

options:
  -h, --help  show this help message and exit
  -c CONFIG   path to your config.json (default: config.json)

Sample config.json:

{
    "api_key": "sk-xxx",
    "api_base": "https://chat.pppan.net/v1",
    "model": "gpt-3.5-turbo",
    "context": 2,
    "stream": true,
    "stream_render": true,
    "showtokens": false,
    "proxy": "socks5://localhost:1080",
    "prompt": [
        { "role": "system", "content": "If your response contains code, show with syntax highlight, for example ```js\ncode\n```" }
    ]
}

• (required) api_key: OpenAI's api key. will read from OPENAI_API_KEY envronment variable if not set
• (optional) api_base: OpenAI's api base url. Can set to a server reverse proxy, for example Azure OpenAI Service or chatgptProxyAPI. By default it's from OPENAI_API_BASE or just https://api.openai.com/v1;
• (optional) api_type: OpenAI's api type, read from env OPENAI_API_TYPE by default;
• (optional) api_version: OpenAI's api version, read from env OPENAI_API_VERSION by default;
• (optional) api_organization: OpenAI's organization info, read from env OPENAI_ORGANIZATION by default;
• (optional) model: OpenAI's chat model, by default it's gpt-3.5-turbo; choices are:
  • gpt-3.5-turbo
  • gpt-4
  • gpt-4-32k
• (optional) context: Chat session context, choices are:
  • 0: no context provided for every chat request, cost least tokens, but AI don't kown what you said before;
  • 1: only use previous user questions as context;
  • 2: use both previous questions and answers as context, would cost more tokens;
• (optional) stream: Output in stream mode;
• (optional) stream_render: Render markdown in stream mode, you can disable it to avoid some UI bugs;
• (optional) showtokens: Print used tokens after every chat;
• (optional) proxy: Use http/https/socks4a/socks5 proxy for requests to api_base;
• (optional) prompt: Customize your prompt. This will appear in every chat request;

Console help (with tab-complete):

gptcli> .help -v

gptcli commands (use '.help -v' for verbose/'.help <topic>' for details):
======================================================================================================
.edit                 Run a text editor and optionally open a file with it
.help                 List available commands or provide detailed help for a specific command
.load                 Load conversation from Markdown/JSON file
.multiline            input multiple lines, end with ctrl-d(Linux/macOS) or ctrl-z(Windows). Cancel
                      with ctrl-c
.prompt               Load different prompts
.quit                 Exit this application
.reset                Reset session, i.e. clear chat history
.save                 Save current conversation to Markdown/JSON file
.set                  Set a settable parameter or show current settings of parameters
.usage                Tokens usage of current session / last N days, or print detail billing info

Run in Docker:

# build
$ docker build -t gptcli:latest .

# run
$ docker run -it --rm -v $PWD/.key:/gptcli/.key gptcli:latest -h

# for host proxy access:
$ docker run --rm -it -v $PWD/config.json:/gptcli/config.json --network host gptcli:latest -c /gptcli/config.json

Feature

• Single Python script
• Session based
• Markdown support with code syntax highlight
• Stream output support
• Proxy support (HTTP/HTTPS/SOCKS4A/SOCKS5)
• Multiline input support (via .multiline command)
• Save and load session from file (Markdown/JSON) (via .save and .load command)
• Print tokens usage in realtime, and tokens usage for last N days, and billing details
• Integrate with llama_index to support chatting with documents

LINK

• https://platform.openai.com/docs/introduction
• https://platform.openai.com/docs/api-reference/completions
• https://platform.openai.com/docs/models/overview
• https://platform.openai.com/account/api-keys

gptcli's Issues

The model: `gpt-4` and `gpt-4-32k` does not exist.

I made absolutely sure my local directory was updated to the latest version. I have GPT Pro, and it works fine with gpt-3.5-turbo, but when the config is set to gpt-4 or gpt-4-32k, it says the model does not exist.

获取输出的时候会多次重复显示。

系统：macOS
电脑：MacBookair M1芯片
在M2 macmini上也会出现这样的情况：

1. 在输出比较长的时候就会出现这种情况，在输出长度变长的时候就会出现这样的情况。

Error communicating with OpenAI: Missing dependencies for SOCKS support.

缺少pysocks依赖，少个依赖就不提pr了，安装依赖后解决报错。

Please make the pyreadline dependency optional (for platforms with no c compiler)

I have trouble installing gptcli because the pyreadline3 dependency gives me error, since in my environment I’m not allowed to compile binaries.
Can you make the pyreadline3 lib optional, or replace it with a pure python lib?
Thanks

有时候输入几个字符然后使用退格建会导致显示问题

windows10不可用

输入python gptcli.py -p https://gpt.pawan.krd/backend-api/conversation，报错：

(chatgpt) G:\[Notes]\chatgpt-demo\gptcli>python gptcli.py -p https://gpt.pawan.krd/backend-api/conversation
Loading key from G:[Notes]\chatgpt-demo\gptcli\.key
Using proxy: https://gpt.pawan.krd/backend-api/conversation
Attach response in prompt: False
Stream mode: True
�[1;33mInput:�[0m hello?
C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\connector.py:899: RuntimeWarning: An HTTPS request is being
sent through an HTTPS proxy. This support for TLS in TLS is known to be disabled in the stdlib asyncio. This is why
you'll probably see an error in the log below.

It is possible to enable it via monkeypatching under Python 3.7 or higher. For more details, see:
* https://bugs.python.org/issue37179
* https://github.com/python/cpython/pull/28073

You can temporarily patch this as follows:
* https://docs.aiohttp.org/en/stable/client_advanced.html#proxy-support
* https://github.com/aio-libs/aiohttp/discussions/6044

  _, proto = await self._create_proxy_connection(req, traces, timeout)
RuntimeWarning: Enable tracemalloc to get the object allocation traceback

Traceback (most recent call last):
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\openai\api_requestor.py", line 587, in arequest_raw
    result = await session.request(**request_kwargs)
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\client.py", line 536, in _request
    conn = await self._connector.connect(
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\connector.py", line 540, in connect
    proto = await self._create_connection(req, traces, timeout)
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\connector.py", line 899, in _create_connection
    _, proto = await self._create_proxy_connection(req, traces, timeout)
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\connector.py", line 1288, in _create_proxy_connection
    raise ClientHttpProxyError(
aiohttp.client_exceptions.ClientHttpProxyError: 400, message='Bad Request', url=URL('https://gpt.pawan.krd/backend-api/conversation')

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "G:\[Notes]\chatgpt-demo\gptcli\gptcli.py", line 161, in <module>
    answer = asyncio.run(query_openai_stream(data))
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\runners.py", line 44, in run
    return loop.run_until_complete(main)
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\base_events.py", line 641, in run_until_complete
    return future.result()
  File "G:\[Notes]\chatgpt-demo\gptcli\gptcli.py", line 52, in query_openai_stream
    async for part in await openai.ChatCompletion.acreate(
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\openai\api_resources\chat_completion.py", line 45, in acreate
    return await super().acreate(*args, **kwargs)
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\openai\api_resources\abstract\engine_api_resource.py", line 217, in acreate
    response, _, api_key = await requestor.arequest(
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\openai\api_requestor.py", line 300, in arequest
    result = await self.arequest_raw(
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\openai\api_requestor.py", line 604, in arequest_raw
    raise error.APIConnectionError("Error communicating with OpenAI") from e
openai.error.APIConnectionError: Error communicating with OpenAI
Exception ignored in: <function _ProactorBasePipeTransport.__del__ at 0x000002AC05F29A20>
Traceback (most recent call last):
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\proactor_events.py", line 116, in __del__
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\proactor_events.py", line 108, in close
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\base_events.py", line 745, in call_soon
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\base_events.py", line 510, in _check_closed
RuntimeError: Event loop is closed

(chatgpt) G:\[Notes]\chatgpt-demo\gptcli>

输入python gptcli.py -r，报错：

(chatgpt) G:\[Notes]\chatgpt-demo\gptcli>python gptcli.py -r
Loading key from G:[Notes]\chatgpt-demo\gptcli\.key
Attach response in prompt: True
Stream mode: True
�[1;33mInput:�[0m
�[1;33mInput:�[0m Hello?

Traceback (most recent call last):
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\connector.py", line 980, in _wrap_create_connection
    return await self._loop.create_connection(*args, **kwargs)  # type: ignore[return-value]  # noqa
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\base_events.py", line 1055, in create_connection
    raise exceptions[0]
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\base_events.py", line 1040, in create_connection
    sock = await self._connect_sock(
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\base_events.py", line 954, in _connect_sock
    await self.sock_connect(sock, address)
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\proactor_events.py", line 704, in sock_connect
    return await self._proactor.connect(sock, address)
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\windows_events.py", line 812, in _poll
    value = callback(transferred, key, ov)
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\windows_events.py", line 599, in finish_connect
    ov.getresult()
OSError: [WinError 121] 信号灯超时时间已到

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\openai\api_requestor.py", line 587, in arequest_raw
    result = await session.request(**request_kwargs)
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\client.py", line 536, in _request
    conn = await self._connector.connect(
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\connector.py", line 540, in connect
    proto = await self._create_connection(req, traces, timeout)
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\connector.py", line 901, in _create_connection
    _, proto = await self._create_direct_connection(req, traces, timeout)
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\connector.py", line 1206, in _create_direct_connection
    raise last_exc
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\connector.py", line 1175, in _create_direct_connection
    transp, proto = await self._wrap_create_connection(
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\aiohttp\connector.py", line 988, in _wrap_create_connection
    raise client_error(req.connection_key, exc) from exc
aiohttp.client_exceptions.ClientConnectorError: Cannot connect to host api.openai.com:443 ssl:default [信号灯超时时间已 到]

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "G:\[Notes]\chatgpt-demo\gptcli\gptcli.py", line 161, in <module>
    answer = asyncio.run(query_openai_stream(data))
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\runners.py", line 44, in run
    return loop.run_until_complete(main)
  File "C:\Users\li\.conda\envs\chatgpt\lib\asyncio\base_events.py", line 641, in run_until_complete
    return future.result()
  File "G:\[Notes]\chatgpt-demo\gptcli\gptcli.py", line 52, in query_openai_stream
    async for part in await openai.ChatCompletion.acreate(
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\openai\api_resources\chat_completion.py", line 45, in acreate
    return await super().acreate(*args, **kwargs)
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\openai\api_resources\abstract\engine_api_resource.py", line 217, in acreate
    response, _, api_key = await requestor.arequest(
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\openai\api_requestor.py", line 300, in arequest
    result = await self.arequest_raw(
  File "C:\Users\li\.conda\envs\chatgpt\lib\site-packages\openai\api_requestor.py", line 604, in arequest_raw
    raise error.APIConnectionError("Error communicating with OpenAI") from e
openai.error.APIConnectionError: Error communicating with OpenAI

难不成要挂代理才能用？😂

.help不展示Console help

环境：mac
终端：iterm2, zsh

ChatGPT UI App

Base gptcli. ChatGPT UI in window & mac with open api (gpt-3.5/gpt-4) and generate chat files. https://github.com/xhunmon/GPT-UI.

Feature request: Add support for ChatGPT plugins

https://openai.com/blog/chatgpt-plugins

Saving session to file

Hi!

Is it possible to save the current conversation session to some file using logging or f.write? That would be cool, that's what's missing in chat.openai.com

something like

filename = (date)+generate_random_string(10)
    with open(f"{filename}.txt", "w") as f:
    while true:
        f.write(prompt)
        f.write(message)

response displayed word by word, line by line.

如何在容器内配置宿主机的代理地址，现在配置还是报错

Can't copy-paste code without getting indent/line breaks errors and tons of blank spaces messing around the code

When presenting ChatGPT a piece of code to refactor it, GPT answers back the refactored code inside a black colored block. It looks fine but If i were to copy the code and paste it in a text editor or IDE, it will look something like this:

 if __name__ ==
 "__main__":                                                                                                                                                                                                      
          main()   

Not only it breaks it down in three rows when it should be only one two but also it will be filled with blank spaces everywhere. Editing/cleaning it takes a lot of time.

EDIT:
Using Gnome terminal version 0.54.2 +GNUTLS

如何一次输入多行文字？

比如我想问一段代码的含义，中间包含了几个回车，当前的程序检测到一次回车就会当做一个问题进行询问。

Not working on windows 11

I am trying to run gptcli on windows 11.

Here is the error I am facing

(venv) PS D:\experiments\gptcli-main\gptcli-main> python .\gptcli.py -c .\config.json
Loading config from: .\config.json
openai.api_key=sk-mEkw*****
openai.api_base=https://chat.pppan.net/v1
openai.api_type=open_ai
openai.api_version=None
openai.api_organization=None
Proxy: socks5://localhost:1080
Context level: ContextLevel.FULL
Stream mode: True
gptcli> hello
Traceback (most recent call last):
  File "D:\experiments\gptcli-main\gptcli-main\gptcli.py", line 252, in query_openai_stream
    response = openai.ChatCompletion.create(
  File "D:\experiments\gptcli-main\gptcli-main\venv\lib\site-packages\openai\lib\_old_api.py", line 39, in __call__
    raise APIRemovedInV1(symbol=self._symbol)
openai.lib._old_api.APIRemovedInV1:

You tried to access openai.ChatCompletion, but this is no longer supported in openai>=1.0.0 - see the README at https://github.com/openai/openai-python for the API.

You can run `openai migrate` to automatically upgrade your codebase to use the 1.0.0 interface.

Alternatively, you can pin your installation to the old version, e.g. `pip install openai==0.28`

A detailed migration guide is available here: https://github.com/openai/openai-python/discussions/742


During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "D:\experiments\gptcli-main\gptcli-main\gptcli.py", line 420, in <module>
    main()
  File "D:\experiments\gptcli-main\gptcli-main\gptcli.py", line 417, in main
    app.cmdloop()
  File "D:\experiments\gptcli-main\gptcli-main\venv\lib\site-packages\cmd2\cmd2.py", line 5285, in cmdloop
    self._cmdloop()
  File "D:\experiments\gptcli-main\gptcli-main\venv\lib\site-packages\cmd2\cmd2.py", line 3142, in _cmdloop
    stop = self.onecmd_plus_hooks(line)
  File "D:\experiments\gptcli-main\gptcli-main\gptcli.py", line 131, in onecmd_plus_hooks
    self.handle_input(line)
  File "D:\experiments\gptcli-main\gptcli-main\gptcli.py", line 163, in handle_input
    answer = self.query_openai_stream(self.messages)
  File "D:\experiments\gptcli-main\gptcli-main\gptcli.py", line 272, in query_openai_stream
    except openai.error.OpenAIError as e:
AttributeError: module 'openai' has no attribute 'error'

Let me know if any more information is required from my side.

Feature request: Estimate token use/cost

A low priority idea.
There is a slight delay between the use of tokens and when it shows up on the usage page: https://platform.openai.com/account/usage

It would be good to know the tokens required to send a query and/or the tokens used for a query.

Some ideas:

• If a flag (confirmtokens) is set to true, after entering an input the number of tokens will be displayed and a further confirmation - like enter - is required to move on.
• If a flag (showtokens) is set to true, the token usage / cost of an input will be displayed with the output such as [Tokens used: XX Est. Cost $0.0002]
• The values should adjust based on which model is selected

OpenAI do provide some information on the way token usage should be estimated, or can be called from the API:
https://platform.openai.com/docs/guides/chat/introduction#:~:text=it%27s%20more%20difficult%20to%20count%20how%20many%20tokens

Can't immediately see pricing information via the API, however they are published here: https://openai.com/pricing

使用.save 保存到文件时，中文乱码

Can I use it with Azure openai service

refer document:

https://learn.microsoft.com/en-us/azure/cognitive-services/openai/chatgpt-quickstart?pivots=programming-language-python&tabs=command-line

the ENDPOINT and API-KEY

Variable name	Value
ENDPOINT	This value can be found in the Keys & Endpoint section when examining your resource from the Azure portal. Alternatively, you can find the value in the Azure AI Studio > Playground > Code View. An example endpoint is: https://docs-test-001.openai.azure.com/.
API-KEY	This value can be found in the Keys & Endpoint section when examining your resource from the Azure portal. You can use either KEY1 or KEY2.

demo Code

#Note: The openai-python library support for Azure OpenAI is in preview.
import os
import openai
openai.api_type = "azure"
openai.api_base = os.getenv("AZURE_OPENAI_ENDPOINT") 
openai.api_version = "2023-05-15"
openai.api_key = os.getenv("AZURE_OPENAI_KEY")

response = openai.ChatCompletion.create(
    engine="gpt-35-turbo", # engine = "deployment_name".
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Does Azure OpenAI support customer managed keys?"},
        {"role": "assistant", "content": "Yes, customer managed keys are supported by Azure OpenAI."},
        {"role": "user", "content": "Do other Azure Cognitive Services support this too?"}
    ]
)

print(response)
print(response['choices'][0]['message']['content'])

Answer with panic

cannot import name 'MarkdownIt' from 'rich.markdown'

Hello, 我是windows 10, python 3.7.8.

Traceback (most recent call last):
File "gptcli.py", line 10, in
from rich.markdown import Markdown, MarkdownIt
ImportError: cannot import name 'MarkdownIt' from 'rich.markdown'