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

[evilpan]

Seems that this is caused by rich text, you may try copy and paste as plaintext to see if it solves the problem.

[evilpan]

I've update the code to add .save command to save conversation to file, by copying the content in the json file you may get the original results. I am also planing to add options to save/load conversation file as text file (markdown), and it would be more friendly for human reading.

[roperi]

Seems that this is caused by rich text, you may try copy and paste as plaintext to see if it solves the problem.

I just tried it and looks fine but still has tons of blank spaces surrounding the code.

Also, paste it as plaintext is not available for some editors. I had to open it in Writer Document in Libre Office to be able to paste it as Unformatted Text (and blank spaces are still everywhere).

The idea of saving it is great though! Thanks, @evilpan . You are not so evil after all ;)

[roperi]

I just git pulled but i don't get how to use .save.

This is what I do:

~/workspace/gptcli$ ./gptcli.py 
Loading config from /home/user/workspace/gptcli/config.json
Response in prompt: True
Stream mode: True
Input: 

I don't see the gptclip> as shown in README . Then in the input prompt I type .save and get back:

I'm sorry, I cannot perform file operations such as saving a file. However, I can still help you with any questions or information you need.

[roperi]

Also tried this:

./gptcli.py .save
usage: gptcli.py [-h] [-c CONFIG]
gptcli.py: error: unrecognized arguments: .save

[evilpan]

Seems you are not using the latest code, are you sure the git pull command returns successfully?

[roperi]

@evilpan

Sorry, i thought I was git pulling but didn't notice I was getting an error back, so I wasn't really pulling anything.

Anyway, I fixed the error and git pulled your latest commits but now it fails with FileNotFoundError: [Errno 2] No such file or directory /home/user/gptcli/config.json. I see the default config file is config.json in code, but in the repo you have it named as demo_config.json. So it fails when starting it up. So i had to start it with ./gptcli.py -c demo_config.json (which i didn't want to as i wanted to run on defaults without any arg flag).

Now, since i deleted the OpenAi key from the config (so it can it read from the environment) it fails with :

OpenAIError: You didn't provide an API key...

[roperi]

Anyway! I can save the chat!

[
  {
    "role": "user",
    "content": "write a python script that prints 'hello, world!' and the time."
  },
  {
    "role": "assistant",
    "content": "Here's a Python script that prints 'Hello, World!' and the current time:\n\n```python\nimport datetime\n\ncurrent_time = datetime.datetime.now().strftime(\"%Y-%m-%d %H:%M:%S\")\nprint(\"Hello, World!\")\nprint(\"Current Time:\", current_time)\n```\n\nJust copy and paste the code above into a Python file, save it, and run the file to see the output."
  }
]

Now it's a matter of just parsing the json file. I'm gonna ask it to write a python script to parse it and get the python script from it.

Oh wait. :/

[evilpan]

Hi! Now we can save/load from both JSON and Markdown file, using the following command:

.save [-m {json, md}] file
.load [-m {json, md}] file

And we use the md (markdown) format by default now.

See: 0777a45

[roperi]

Hi! Now we can save/load from both JSON and Markdown file, using the following command:

.save [-m {json, md}] file
.load [-m {json, md}] file

And we use the md (markdown) format by default now.

See: 0777a45

Ah, amazing! Now the tool is super useful. Thanks so much!