posit-dev / py-shinylive Goto Github PK

Export the commands that power the CLI so they can be called in Python scripts.

I'm working on a toy application to show how a shinylive app could make API calls. It's turning out to be harder than I thought.
It seems like networking in general does not work very well with pyodide (and thus shinylive). I'm still learning here, but the issue looks to be accessing sockets.

A few posts (e.g., here) suggest that you should be able to do this with javascript. I see the tags reference in support, but I'm a bit lost in how to implement this in practice.

My toy app is calling a published API that was built on the palmerpenguins data set. Here is the code that works as a standalone shiny app but not with shinylive:

from shiny import *
import urllib.request
import json

app_ui = ui.page_fluid(
    ui.layout_sidebar(
    ui.panel_sidebar(
    ui.input_select("species", "Penguin Species",
                    {"Gentoo":"Gentoo", "Chinstrap":"Chinstrap", "Adelie":"Adelie"}),
    ui.input_action_button("go", "Predict", width="100%")),
    ui.panel_main(ui.h2(ui.output_text("txt"))))
)

def server(input, output, session):
    @output
    @render.text
    @reactive.event(input.go)
    def txt():
       url = "http://penguin.eastus.azurecontainer.io:8000/predict"
       payload = [{"species":input. Species(),
                   "bill_length_mm":40.5,
                   "bill_depth_mm":18.9,
                   "flipper_length_mm":180,
                   "body_mass_g":3950}]
       headers = {"Content-Type": "application/json"}
       data = json.dumps(payload).encode("utf-8")
       headers = {k: v.encode("utf-8") for k, v in headers. Items()}
       request = urllib.request.Request(url, data, headers)
       response = urllib.request.urlopen(request)
       data = response. Read()
       response_data = json.loads(data. Decode("utf-8"))
       prediction = response_data[0]
       class_pred = prediction[".pred_class"]
       return f"The {input.species()} 🐧 is predicted to be {class_pred}." 
  
app = App(app_ui, server)

Is there a known workaround with the python route that I'm unable to find? If not, any suggestions on how to insert a javascript component that can take the dynamic inputs from the UI to be used for the prediction?

Request:

Hmm, shinylive appears to scan every .py file in the directory to figure out what pyodide needs to install, even scripts that aren't used at runtime. Like sometimes I'll have a helper script that builds a database (in this case, using sqlalchemy) but the actual shiny app only needs built-in sqlite3.

I hacked around this by making a stripped-down branch and exporting that, but it could be nice if either the build process honored requirements.txt (in a minimalistic sense) or if somehow it only scanned app.py and what it imports? or some other solution to ignore certain files when figuring this out.

My reply:

... correct. py-shinylive currently scans every that app.json provides. (Which is every file.) Link

It seems fair to me that if a requirements.txt is found, then only those packages are inspected.

Putting Jinja2 in a requirements.txt file, as suggested e.g. here, doesn't seem to be enough for the shinylive CLI tool to actually bundle the Jinja2 wheel in the output dir. But if I explicitly import jinja2 in app.py, the wheel gets included and rendering tables works.

Is this intentional? If so, it should probably be documented?

As an aside, thank you for bringing Shiny to Python in great style!

When I deploy my project with shinylive I get problems with the character handling. This issue was fixed in shiny with the last version. But the webassembly still can't handle it.

Running it with shiny

Running as webassembly

I couldn't understand from documentation how can I make a local file accessible to the application, so that it would be possible to read_csv and do things like that?
Should I copy it to some location in shinylive folder? An what path should I provide for the script to work?

Some files in project directory can be accessed by the py-shiny app which works fine, but shinylive app gives file_not_found_error if I build a shinylive app from exactly the same project.

Hi there,

I was trying out shinylive on a virtualenv on 3.9.14 and could not get it to work:

pip install shiny

pip install shinylive

shiny create myapp

shinylive export myapp site

Creating site/
Copying base Shinylive files from /Users/XX/Library/Caches/shinylive/shinylive-0.0.8/ to site/
Copying imported packages from /Users/XX/Library/Caches/shinylive/shinylive-0.0.8/shinylive/pyodide/ to site/shinylive/pyodide/
Traceback (most recent call last):
  File "/usr/local/bin/shinylive", line 8, in <module>
    sys.exit(main())
  File "/usr/local/lib/python3.9/site-packages/click/core.py", line 1130, in __call__
    return self.main(*args, **kwargs)
  File "/usr/local/lib/python3.9/site-packages/click/core.py", line 1055, in main
    rv = self.invoke(ctx)
  File "/usr/local/lib/python3.9/site-packages/click/core.py", line 1657, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "/usr/local/lib/python3.9/site-packages/click/core.py", line 1404, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/usr/local/lib/python3.9/site-packages/click/core.py", line 760, in invoke
    return __callback(*args, **kwargs)
  File "/usr/local/lib/python3.9/site-packages/shinylive/_main.py", line 79, in export
    _export.export(
  File "/usr/local/lib/python3.9/site-packages/shinylive/_export.py", line 101, in export
    copy_fn(src_path, dest_path)
  File "/usr/local/lib/python3.9/site-packages/shinylive/_utils.py", line 93, in copy_fn
    verbose_print(f"Skipping {dst}", file=sys.stderr)
TypeError: verbose_print() got an unexpected keyword argument 'file'

Any idea what might be going on? Really looking forward to playing around with this!