Textual in WASM?

[showipintbri]

Will & Team,
What is the current level of Textual functionality in WASM Python? (pyscript or pyodide)

I'd love to be able to use Textual to build 1 single interface for both terminal and web based environments instead of building 1 for terminal interaction and 1 for web based interaction.

Is anyone already using Textual in WASM?

[davep]

Someone was experimenting with pyscript late on last year, and someone posted last month that they'd got their app running in pyodide.

[pmp-p]

Hi, I have only found a few things that would prevent Textual from running on a python wasm runtime "out of the box"

First is that textual.worker.Worker will not enforce thread = False even though
( not hasattr(sys, "_emscripten_info") or sys._emscripten_info.pthreads )
also if threading is not available and worker callable is not async, the WorkerError("Unsupported attempt to run a thread worker") should be turned into a warning/console.warn and attempt sync execution.

The annoying two print in texual/app.py when using pilot mode ( easiest way i've found to start the textual app as a normal async task without any tty check ) but that's only a problem if select does not work properly with stdin ( pyodide maybe ?).

It may also have some trouble when app code is coming from stdin and not catching OSError here

The monkey patch I used to run codebrowser.py on pygbag

import asyncio
import select
import os

import textual.events
import textual.driver
import textual.drivers.headless_driver
import textual._xterm_parser

class HeadlessDriver(textual.driver.Driver):
    def send_event(self, event: textual.events.Event) -> None:
        self._loop.create_task(self._app._post_message(event))

    def _get_terminal_size(self) -> tuple[int, int]:
        return tuple(map(int, __import__("os").get_terminal_size()))

    def write(self, data: str) -> None:
        sys.__stdout__.write(data)

    def start_application_mode(self) -> None:
        loop = asyncio.get_running_loop()
        loop.create_task(self.input_handler())

    def disable_input(self) -> None:
        ...

    def stop_application_mode(self) -> None:
        ...

    async def input_handler(self):
        import sys, os
        from platform import window
        import textual._xterm_parser

        STDIN = sys.stdin.fileno()

        def more_data() -> bool:
            return select.select([STDIN], [], [], 0)[0]

        parser = textual._xterm_parser.XTermParser(more_data, debug=self._debug)
        while True:
            await asyncio.sleep(0)
            if select.select([STDIN], [], [], 0)[0]:
                try:
                    payload = os.read(STDIN, 1024)
                    if payload:
                        for event in parser.feed(payload.decode('utf-8')):
                            self.process_event(event)
                except OSError:
                    continue


textual.drivers.headless_driver.HeadlessDriver = HeadlessDriver

There is surely a better way to integrate python wasm but i only spent an hour on this.

probably some font width / xterm.js specific problems here with textual-paint