NiceGUI-primer

This project provides the plain text primer version of NiceGUI's documentation for usage with AI or as a portable offline file.

Why?

• NiceGUI is new compared to the cut-off training dates of many AI LLM's, so they are not aware of recent updates and will often make incorrect assumptions when coding.
• While dynamically created websites are impressive and useful for humans, a plain text file is smaller, faster, and easier for AIs to ingest and interpret.

---

A Typical Usage

The NiceGUI Primer is available for use as a GPT via OpenAI's GPT store. While it's offered by me as a free to use tool; you may have to set up a ChatGPT account (maybe $'s). But the NiceGUI Primer GPT is not just limited to chatting about and coding NiceGUI, since it is backed by GPT-4, it can be used for any typical chat activities as well as the built-in ability to code in Python with an awareness of other PyPi's and libraries.

It turns out that other GPT's exist regarding NiceGUI, see: Exploring GPTs.

---

Further information here:

• https://nicegui.io/
• https://nicegui.io/documentation
• https://github.com/zauberzeug/nicegui/
• https://github.com/zauberzeug/nicegui/blob/main/CONTRIBUTING.md#documentation

---

TL;DR

1. open a command line window:

   • clone the nicegui repo
   • cd nicegui
   • do env/pip stuff (tested with miniconda and python 3.10)
   • python main.py ... to launch the website server at http://127.0.0.1:8080/

2. open another command line window:

   • clone this repo
   • cd NiceGUI-primer
   • do env/pip stuff (tested with miniconda and python 3.10)
   • then run:
     • python 1_get_all_urls.py ... makes a list of urls to be scraped
     • python 2_scrape_all_urls.py ... scrape list of urls into a text file

3. use the final text file with AI's or for other purposes

OR ... just grab the included doc's text file and be done.

⚠️ WARNING:
If you want your NiceGUI documentation text file to be up-to-date, you will have to re-run the scraping.

---

1. Install NiceGUI's documentation website for scraping

To avoid disturbing the nicegui.io online server, let's run the nicegui documentation website at:
http://localhost:8080/ by doing the following:

By the way, this was a very thoughtful option for the NiceGUI project to offer, so why not use it.

Clone the github repo or download the zip file:

git clone nicegui

cd nicegui

Setup a virtual environment using conda or something:

conda create -n nicegui python=3.10

conda activate nicegui

Perhaps not all of the following are actually required, but this works:

python3 -m pip install -e .

pip install itsdangerous prometheus_client isort docutils pandas plotly pyecharts matplotlib requests dnspython

Sometimes these were needed too:

pip install pytest
pip install selenium

Now launch main.py in the root directory:

python -B main.py

Now, your default web browser should appear with the launch of the website server at http://127.0.0.1:8080/.

---

2. Install NiceGUI-primer

Clone the github repo or download the zip file:

git clone https://github.com/cleesmith/NiceGUI-primer.git

cd NiceGUI-primer

Setup a virtual environment using conda or something:

Also, other python versions may work, but I used 3.10.14:

conda create -n NiceGUI-primer python=3.10

conda activate NiceGUI-primer

---

3. Scrape NiceGUI documentation

Collect a list of the scrapable urls using Playwright with the Chromium option

cd NiceGUI-primer

pip install playwright

python -B 1_get_all_urls.py

---

4. Scrape using the list of urls into a text file

This process also uses Playwright with the Chromium option.

cd NiceGUI-primer

python -B 2_scrape_all_urls.py

---

🤓 NOTE:
While it would be simple to combine steps 3. and 4. above into a single script, I preferred having the sanity check of a text file with the list of known URLs to be scraped. So I could double check the list before scraping.

---

5. Scrape using the list of urls into a PDF file

Again, this process also uses Playwright with the Chromium option.

⚠️ WARNING:
While this code works to create a PDF file, the result is just too big to be useful with AI's as they exist in August 2024. But your mileage may vary.

So here's how:

cd NiceGUI-primer

python -B 3_scrape_to_pdf.py

---

Thoughts

First and foremost, the state of AI in August 2024 can make one wonder about the "I" (intelligence) in AI,
and this tiny project really seems like overkill just to help the "I" understand.

The online NiceGUI documentation is great for humans, but not so great for AI.

And the scraping process is affected by:

1. repeated links due to the navigation, which is helpful to humans; see #1 and #2 in the screenshot below
2. so many anchor links, such as: https://nicegui.io/documentation/section_text_elements#label i.e. it's the #label thing, and there are a lot of them in the doc's; see #3 in the screenshot below
3. the example display/output or live demo shown by executing the example code; see #4 in the screenshot below

This example code, regardless of how it's executed, impacts page load time.

After digging around in the source code, there are ways to turn off #1 and #2 (the nav bars) but there was no easy way found to stop #3 (anchor links) and #4 (code live demos or example display/output).

Perhaps these changes are not possible, or just too much to ask of any project, but a
command line arg such as: --nonhuman
would certainly help with the scraping of the doc's to be feed to AI's
in removing all 4 of the troublesome spots, like so:

cd nicegui
python -B main.py --nonhuman

A screenshot of a typical NiceGUI documentation page with annotation of the troublesome spots:

---

Exiting cleanly

See ~/nicegui/main.py:

NOTE: do not reload on fly.io (see https://github.com/zauberzeug/nicegui/discussions/1720#discussioncomment-7288741)
ui.run(uvicorn_reload_includes='*.py, *.css, *.html', reload=not on_fly, reconnect_timeout=10.0)

Sure it executes just fine, but to exit requires a ctrl-c and then one sees:

(nicegui) booboobear:~$ python -B main.py   
NiceGUI ready to go on http://localhost:8080, and http://192.168.0.8:8080
^C%                                                                                                 
(nicegui) booboobear:~$ /opt/miniconda3/envs/nicegui/lib/python3.10/multiprocessing/resource_tracker.py:224: 
UserWarning: resource_tracker: There appear to be 6 leaked semaphore objects to clean up at shutdown
warnings.warn('resource_tracker: There appear to be %d '

So let's add a tiny bit of code to catch the ctrl-c:

if __name__ == "__main__":
    try:
        # just say no to reload
        ui.run(reload=False, reconnect_timeout=10.0, show_welcome_message=False, title="NiceGUI documentation")
    except KeyboardInterrupt:
        print(f"\nShutting down NiceGUI documentation server . . .", end=" ")
        app.shutdown() # is not required but feels logical and says it all
    finally:
        print("catch you later!")

And now, a clean exit:

cd nicegui
python -B main.py
^C
Shutting down NiceGUI documentation server . . . catch you later!

---

fin

---