hr_visualize/main.py

from typing import (
    Annotated,
    AsyncGenerator,
    Final,
    Generator,
    List,
    Literal,
    Optional,
    Tuple,
    TypeVar,
    TypedDict,
    Any,
    cast,
)

from loguru import logger
import numpy as np
import plotly.graph_objects as go
import streamlit as st
import anyio
from anyio.abc import TaskGroup, UDPSocket
from anyio import create_memory_object_stream, create_udp_socket
from anyio.streams.memory import MemoryObjectSendStream, MemoryObjectReceiveStream
from threading import Thread
from time import sleep
from pydantic import BaseModel, computed_field
from datetime import datetime
import awkward as ak
from awkward import Array as AwkwardArray, Record as AwkwardRecord
from app.model import AlgoReport
from collections import deque


class AppHistory(TypedDict):
    timescape: deque[datetime]
    hr_data: deque[float]
    hr_conf: deque[int]  # in %
    accel_x_data: deque[int]
    accel_y_data: deque[int]
    accel_z_data: deque[int]


# https://handmadesoftware.medium.com/streamlit-asyncio-and-mongodb-f85f77aea825
class AppState(TypedDict):
    worker_thread: Thread
    message_queue: MemoryObjectReceiveStream[bytes]
    task_group: TaskGroup
    history: AppHistory


UDP_SERVER_HOST: Final[str] = "localhost"
UDP_SERVER_PORT: Final[int] = 50_000
MAX_LENGTH = 600
NDArray = np.ndarray

T = TypeVar("T")


def unwrap(value: Optional[T]) -> T:
    if value is None:
        raise ValueError("Value is None")
    return value


@st.cache_resource
def resource(params: Any = None):
    set_ev = anyio.Event()
    tx, rx = create_memory_object_stream[bytes]()
    tg: Optional[TaskGroup] = None

    async def poll_task():
        nonlocal set_ev
        nonlocal tg
        tg = anyio.create_task_group()
        set_ev.set()
        async with tg:
            async with await create_udp_socket(
                local_host=UDP_SERVER_HOST, local_port=UDP_SERVER_PORT, reuse_port=True
            ) as udp:
                async for packet, _ in udp:
                    await tx.send(packet)

    tr = Thread(target=anyio.run, args=(poll_task,))
    tr.start()
    while not set_ev.is_set():
        sleep(0.01)
    logger.info("Poll task initialized")
    state: AppState = {
        "worker_thread": tr,
        "message_queue": rx,
        "task_group": unwrap(tg),
        "history": {
            "timescape": deque(maxlen=MAX_LENGTH),
            "hr_data": deque(maxlen=MAX_LENGTH),
            "hr_conf": deque(maxlen=MAX_LENGTH),
            "accel_x_data": deque(maxlen=MAX_LENGTH),
            "accel_y_data": deque(maxlen=MAX_LENGTH),
            "accel_z_data": deque(maxlen=MAX_LENGTH),
        },
    }
    logger.info("Resource created")
    return state


def main():
    state = resource()
    history = state["history"]

    def on_export():
        file_name = f"history_{datetime.now().strftime('%Y%m%d_%H%M%S')}.parquet"
        logger.info(f"Exporting to {file_name}")
        rec = ak.Record(history)
        ak.to_parquet(rec, file_name)

    def on_clear():
        nonlocal history
        logger.info("Clearing history")
        history["timescape"].clear()
        history["hr_data"].clear()
        history["hr_conf"].clear()
        history["accel_x_data"].clear()
        history["accel_y_data"].clear()
        history["accel_z_data"].clear()

    # https://docs.streamlit.io/develop/api-reference/layout
    st.title("MAX-BAND Visualizer")
    with st.container(border=True):
        c1, c2 = st.columns(2)
        with c1:
            st.button(
                "Export",
                help="Export the current data to a parquet file",
                on_click=on_export,
            )
        with c2:
            st.button(
                "Clear",
                help="Clear the current data",
                on_click=on_clear,
            )

    placeholder = st.empty()

    while True:
        try:
            message = state["message_queue"].receive_nowait()
        except anyio.WouldBlock:
            continue
        with placeholder.container():
            report = AlgoReport.unmarshal(message)
            history["timescape"].append(datetime.now())
            history["hr_data"].append(report.data.hr_f)
            history["hr_conf"].append(report.data.hr_conf)
            history["accel_x_data"].append(report.accel_x)
            history["accel_y_data"].append(report.accel_y)
            history["accel_z_data"].append(report.accel_z)

            # with st.container():
            #     c1, c2 = st.columns(2)
            #     with c1:
            #         c1.write(f"HR: {report.data.hr_f}")
            #     with c2:
            #         c2.write(f"HR Confidence: {report.data.hr_conf}")

            fig_hr, fig_accel = st.tabs(["Heart Rate", "Accelerometer"])

            with fig_hr:
                fig_hr.plotly_chart(
                    go.Figure(
                        data=[
                            go.Scatter(
                                x=list(history["timescape"]),
                                y=list(history["hr_data"]),
                                mode="lines",
                                name="HR",
                            ),
                            go.Scatter(
                                x=list(history["timescape"]),
                                y=list(history["hr_conf"]),
                                mode="lines",
                                name="HR Confidence",
                            ),
                        ]
                    )
                )
            with fig_accel:
                fig_accel.plotly_chart(
                    go.Figure(
                        data=[
                            go.Scatter(
                                x=list(history["timescape"]),
                                y=list(history["accel_x_data"]),
                                mode="lines",
                                name="x",
                            ),
                            go.Scatter(
                                x=list(history["timescape"]),
                                y=list(history["accel_y_data"]),
                                mode="lines",
                                name="y",
                            ),
                            go.Scatter(
                                x=list(history["timescape"]),
                                y=list(history["accel_z_data"]),
                                mode="lines",
                                name="z",
                            ),
                        ]
                    )
                )


if __name__ == "__main__":
    main()