使用FloatSlider手动调整标记位置时的异常问题排查

我目前正尝试通过ipywidgets根据用户输入为数据分配标记。定义偏移量（offset）、步长（Step length）、休息时长（rest length）等初始用户输入步骤已正常运行，但添加手动调整各阶段Start和End标记位置的功能后，点击interact运行时，标记位置未改变甚至消失。我已获取各Start/End标记的索引，想请教这一现象的原因？

• 滑块与交互函数未绑定：手动创建的调整标记滑块没有传递给interact_manual，导致更新函数无法获取滑块最新值，无法触发标记位置更新。
• Output组件被错误清空：update_marker_positions中调用output.clear_output(wait=True)会将之前显示的滑块一并清除，丢失调整控件的同时，图表也无法继续接收输入。
• 标记更新逻辑有遗漏：循环中的条件判断会在参数缺失时跳过标记添加，而由于滑块未绑定，参数始终缺失，最终导致标记全部消失。

from ipywidgets import Output, VBox, HBox, FloatSlider, Button, interact_manual
import plotly.graph_objs as go
import pandas as pd
import numpy as np

output = Output()
markers = {}  # 全局存储标记位置
np.random.seed(0)
df = pd.DataFrame({
    'timeinsec': np.linspace(0, 1000, 100),
    "data": np.random.rand(100)
})

def set_markers(df, x_col, y_col, offset, step_length, rest_length, num_intervals, last_step_length, start_velocity, velocity_increment):
    with output:
        output.clear_output(wait=True)
        
        fig = go.Figure()
        fig.add_trace(go.Scatter(x=df[x_col], y=df[y_col], mode='markers', name='original', 
                                marker=dict(color='red', size=8, line=dict(width=1, color='DarkSlateGrey'))))
        
        velocities = [start_velocity + i * velocity_increment for i in range(num_intervals)]
        
        # 初始化标记和图表形状
        for i in range(num_intervals):
            start = offset + i * (step_length + rest_length)
            end = start + (last_step_length if i == num_intervals - 1 else step_length)
            markers[f"Start{i+1}"] = start
            markers[f"End{i+1}"] = end
            
            # 添加初始标记线
            fig.add_shape(type="line", x0=start, y0=df[y_col].min(), x1=start, y1=df[y_col].max(), 
                         line=dict(color="blue", width=2))
            fig.add_shape(type="line", x0=end, y0=df[y_col].min(), x1=end, y1=df[y_col].max(), 
                         line=dict(color="green", width=2))
            
            # 分配速度值
            df.loc[(df[x_col] >= start) & (df[x_col] <= end), 'v'] = velocities[i]
        
        # 处理最后阶段的速度比例
        if num_intervals > 1:
            stage_per = last_step_length / step_length
            new_v = velocities[-2] + velocity_increment * stage_per
            df.loc[(df[x_col] >= markers[f"Start{num_intervals}"]) & (df[x_col] <= markers[f"End{num_intervals}"]), 'v'] = new_v
            df.v.fillna(0, inplace=True)
        
        fig.update_layout(title='Assign Stage:', xaxis_title=x_col, yaxis_title=y_col)
        fig.show()

    # 创建调整滑块（放在Output外部，避免被清空）
    sliders_column_1 = VBox()
    sliders_column_2 = VBox()
    slider_args = {}

    for i in range(num_intervals):
        start_slider = FloatSlider(value=markers[f"Start{i+1}"], min=0, max=df[x_col].max(), description=f'Start {i+1}:')
        end_slider = FloatSlider(value=markers[f"End{i+1}"], min=0, max=df[x_col].max(), description=f'End {i+1}:')
        sliders_column_1.children += (start_slider,)
        sliders_column_2.children += (end_slider,)
        slider_args[f"Start{i+1}"] = start_slider
        slider_args[f"End{i+1}"] = end_slider

    sliders_box = HBox([sliders_column_1, sliders_column_2])
    display(sliders_box)

    # 更新标记位置的函数
    def update_marker_positions(**kwargs):
        with output:
            output.clear_output(wait=True)
            # 重新构建所有标记形状
            shapes = []
            for i in range(num_intervals):
                start_pos = kwargs[f"Start{i+1}"]
                end_pos = kwargs[f"End{i+1}"]
                markers[f"Start{i+1}"] = start_pos
                markers[f"End{i+1}"] = end_pos
                
                shapes.append(dict(type="line", x0=start_pos, y0=df[y_col].min(), x1=start_pos, y1=df[y_col].max(), 
                                  line=dict(color="blue", width=2)))
                shapes.append(dict(type="line", x0=end_pos, y0=df[y_col].min(), x1=end_pos, y1=df[y_col].max(), 
                                  line=dict(color="green", width=2)))
            
            # 更新图表并重新显示
            fig.update_layout(shapes=shapes)
            fig.show()
            
            # 重新分配速度值
            df['v'] = 0
            velocities = [start_velocity + i * velocity_increment for i in range(num_intervals)]
            for i in range(num_intervals):
                df.loc[(df[x_col] >= markers[f"Start{i+1}"]) & (df[x_col] <= markers[f"End{i+1}"]), 'v'] = velocities[i]
            if num_intervals > 1:
                stage_per = last_step_length / step_length
                new_v = velocities[-2] + velocity_increment * stage_per
                df.loc[(df[x_col] >= markers[f"Start{num_intervals}"]) & (df[x_col] <= markers[f"End{num_intervals}"]), 'v'] = new_v

    # 将滑块绑定到交互函数
    interact_manual(update_marker_positions, **slider_args)

def interactive_line_plot(df, x_col, y_col):
    from ipywidgets import IntSlider, FloatText
    offset = IntSlider(value=10, min=0, max=df[x_col].max(), step=1, description='Offset:')
    step_length = IntSlider(value=240, min=1, max=500, step=1, description='Step Length:')
    rest_length = IntSlider(value=30, min=1, max=500, step=1, description='Rest Length:')
    num_intervals = IntSlider(value=5, min=1, max=10, step=1, description='Intervals:')
    last_step_length = IntSlider(value=240, min=1, max=500, step=1, description='t Last Step:')
    start_velocity = FloatText(value=8.0, description='Start:')
    velocity_increment = FloatText(value=1.0, description='Increment:')
    
    update_plot_button = Button(description="Update Plot")
    update_plot_button.on_click(lambda b: set_markers(df, x_col, y_col, offset.value, step_length.value, rest_length.value, num_intervals.value, last_step_length.value, start_velocity.value, velocity_increment.value))
    
    control_box = VBox([offset, step_length, rest_length, num_intervals, last_step_length, start_velocity, velocity_increment, update_plot_button])
    display(HBox([control_box, output]))

interactive_line_plot(df, "timeinsec", "data")

1. 滑块与交互函数绑定：创建滑块后存入slider_args字典，传递给interact_manual，确保滑块值能正确传入更新函数。
2. 分离滑块与Output：将滑块的display放在Output组件外部，避免更新图表时被clear_output清除。
3. 简化更新逻辑：移除不必要的条件判断，确保所有标记都会被重新绘制，同时同步更新DataFrame中的速度值。

内容的提问来源于stack exchange，提问作者Hansson