Error on input --a b

[Jasha10]

I have a reproducible error when the bot is given input string --a b.

$ gpt
Hi! I'm here to help. Type :q or Ctrl-D to exit, :c or Ctrl-C and Enter to clear
the conversation, :r or Ctrl-R to re-generate the last response. To enter
multi-line mode, enter a backslash \ followed by a new line. Exit the multi-line
mode by pressing ESC and then Enter (Meta+Enter). Try :? for help.
> --a b
Invalid argument: a. Allowed arguments: ['model', 'temperature', 'top_p']
Invalid argument: a. Allowed arguments: ['model', 'temperature', 'top_p']
NoneType: None

$ gpt --version
gpt-cli v0.1.3

Here is the output from running `$ gpt --log_file log.txt --log_level DEBUG`:

$ cat log.txt
2023-07-17 18:26:33,531 - gptcli - INFO - Starting a new chat session. Assistant config: {'messages': [], 'temperature': 0.0, 'model': 'gpt-4'}
2023-07-17 18:26:33,539 - gptcli-session - INFO - Chat started
2023-07-17 18:26:33,539 - asyncio - DEBUG - Using selector: EpollSelector
2023-07-17 18:26:35,314 - gptcli-session - ERROR - Invalid argument: a. Allowed arguments: ['model', 'temperature', 'top_p']
NoneType: None

For context, this error came up when I copy/pasted the following rustc error message into the cli using multiline> mode:

 15 | ) -> impl Iterator<Item = AdaptedRecord> {
    |      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ `()` is not an iterator
    |
    = help: the trait `Iterator` is not implemented for `()`

 For more information about this error, try `rustc --explain E0277`

This resulted in Invalid argument: explain. Allowed arguments: ...

[Jasha10]

It looks like the root cause is that the parse_args function matches the regex pattern --(\w+)(?:\s+|=)([^\s]+) and treats it specially.

@kharvd, I'd like to submit a PR that allows the user to disable this regex matching behavior. Would that be acceptable to you?

[kharvd]

Thanks for a detailed report. If you'd like to add a config option to disable this (and maybe even make it disabled by default), I'm on board - just want to make sure the option is there, because I use --model a lot to run the same prompt with a different model.

One alternative idea is to allow switching the model and other args via : commands now that we have :q and friends: e.g. :model gpt-4 or :temperature 0.0

[Jasha10]

One alternative idea is to allow switching the model and other args via : commands

Let me make sure I understand what you have in mind: if you write :model gpt-4, would this change the model for all subsequent messages?
I believe the current behavior is that Message ... --model gpt-4 changes the model to gpt-4 only for the current message, and subsequent messages will revert back to the model that was configured at the start of the chat session.

[wfjt]

Was hit this as well by including a -----BEGIN AGE ENCRYPTED FILE-----\n... string in multiline mode, got an error about -BEGIN being an invalid parameter. Quite annoying having to escape 20 times for the header and footer. I tried gpt -- but that made no difference, same error.