PyCharm调试模式下seaborn、matplotlib代码报错但运行模式正常

运行环境 #

• PyCharm 2021.2.3 社区版
• Python 解释器 3.10.0
• matplotlib 3.5.0
• seaborn 0.11.2
• numpy 1.21.4
• pandas 1.3.4
• PySimpleGUI 4.55.1

问题现象 #

以下脚本在普通运行模式下执行正常，但调试模式下会抛出异常：

import numpy as np
import pandas as pd
import PySimpleGUI as sg
import seaborn as sns

from matplotlib.backends.backend_tkagg import FigureCanvasTkAgg
from matplotlib.figure import Figure


def init_data():
    r1 = np.random.rand(5, 4)
    columns = [f"var{i}" for i in range(1, 5)]
    df = pd.DataFrame(r1, columns=columns)
    df.insert(0, 'year', range(2021, 2026))
    df.insert(1, 'scenario', 'test1')
    ldf = pd.melt(df, id_vars=['year', 'scenario'], value_vars=columns, var_name='percentile', value_name='carbon')
    return ldf


def draw_figure(canvas, figure):
    figure_canvas_agg = FigureCanvasTkAgg(figure, canvas)
    figure_canvas_agg.draw()
    figure_canvas_agg.get_tk_widget().pack(side='top', fill='both', expand=1)
    return figure_canvas_agg


# define the window layout
layout = [[sg.Text('Plot test')],
          [sg.Canvas(key='-CANVAS-')],
          [sg.Button('Ok')]]

# create the form and show it without the plot
window = sg.Window('Testing seaborn in PySimpleGUI', layout, finalize=True,
                   element_justification='center', font='Helvetica 18')

figure = Figure()
ax = figure.subplots()
sns.lineplot(x='year', y='carbon', hue='percentile', data=init_data(), ax=ax)

# add the plot to the window
fig_canvas_agg = draw_figure(window['-CANVAS-'].TKCanvas, figure)

event, values = window.read()

window.close()

报错信息 #

Traceback (most recent call last):
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\numpy\core\getlimits.py", line 384, in __new__
    dtype = numeric.dtype(dtype)
TypeError: 'NoneType' object is not callable
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
  File "C:\Program Files\JetBrains\PyCharm Community Edition 2021.2.3\plugins\python-ce\helpers\pydev\pydevd.py", line 1483, in _exec
    pydev_imports.execfile(file, globals, locals)  # execute the script
  File "C:\Program Files\JetBrains\PyCharm Community Edition 2021.2.3\plugins\python-ce\helpers\pydev\_pydev_imps\_pydev_execfile.py", line 18, in execfile
    exec(compile(contents+"\n", file, 'exec'), glob, loc)
  File "D:/OneDrive/OneDrive - Louise Pryor & Co Ltd/Actuarial/Carbon/Carbon/seaborntest.py", line 39, in <module>
    sns.lineplot(x='year', y='carbon', hue='percentile', data=init_data(), ax=ax)
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\seaborn\_decorators.py", line 46, in inner_f
    return f(**kwargs)
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\seaborn\relational.py", line 710, in lineplot
    p.plot(ax, kwargs)
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\seaborn\relational.py", line 557, in plot
    self._add_axis_labels(ax)
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\seaborn\_core.py", line 1194, in _add_axis_labels
    x_visible = any(t.get_visible() for t in ax.get_xticklabels())
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\axes\_base.py", line 75, in wrapper
    return get_method(self)(*args, **kwargs)
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\axis.py", line 1249, in get_ticklabels
    return self.get_majorticklabels()
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\axis.py", line 1201, in get_majorticklabels
    ticks = self.get_major_ticks()
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\axis.py", line 1371, in get_major_ticks
    numticks = len(self.get_majorticklocs())
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\axis.py", line 1277, in get_majorticklocs
    return self.major.locator()
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\ticker.py", line 2113, in __call__
    vmin, vmax = self.axis.get_view_interval()
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\axis.py", line 1987, in getter
    return getattr(getattr(self.axes, lim_name), attr_name)
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\axes\_base.py", line 781, in viewLim
    self._unstale_viewLim()
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\axes\_base.py", line 776, in _unstale_viewLim
    self.autoscale_view(**{f"scale{name}": scale
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\axes\_base.py", line 2932, in autoscale_view
    handle_single_axis(
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\axes\_base.py", line 2895, in handle_single_axis
    x0, x1 = locator.nonsingular(x0, x1)
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\ticker.py", line 1654, in nonsingular
    return mtransforms.nonsingular(v0, v1, expander=.05)
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\matplotlib\transforms.py", line 2880, in nonsingular
    if maxabsvalue < (1e6 / tiny) * np.finfo(float).tiny:
  File "C:\Users\drlou\AppData\Local\Programs\Python\Python310\lib\site-packages\numpy\core\getlimits.py", line 387, in __new__
    dtype = numeric.dtype(type(dtype))
TypeError: 'NoneType' object is not callable

该问题是PyCharm 2021版本调试器与Python 3.10、对应版本numpy的兼容冲突导致的，调试器会意外将numpy.numeric.dtype覆盖为None，触发调用异常。可按以下优先级尝试解决：

1. 关闭PyCharm调试器的Gevent兼容选项：打开路径 File > Settings > Build, Execution, Deployment > Python Debugger，取消勾选Gevent compatible选项，重启调试即可生效，这是该问题最常用的解决方式。
2. 升级PyCharm版本：PyCharm 2022.1及以上版本已经修复了Python 3.10的调试兼容问题，升级到最新社区版即可彻底解决冲突。
3. 降级Python版本到3.9.x：如果暂时无法升级PyCharm，将Python解释器替换为3.9版本，搭配适配的numpy 1.21.x版本也可以规避该问题。
4. 临时补丁：如果以上方案都无法执行，可在脚本开头导入numpy后手动重置dtype引用：

import numpy as np
np.numeric.dtype = np.dtype

该操作会强制修正被调试器覆盖的dtype引用，保证后续代码调用正常。

内容的提问来源于stack exchange，提问作者lpryor