Fedora 36下使用pandas绘图出现报错问题求助

问题背景 #

我在RHEL8及更早版本使用该工具一直正常，但同事在Fedora 36（安装了新版本pandas/matplotlib）运行时出现报错。

环境信息 #

matplotlib         3.6.1     /home/loberman/.local/lib/python3.10/site-packages pip
pandas             1.5.1     /home/loberman/.local/lib/python3.10/site-packages pip

出错代码 #

cpu_mem_data = read_csv(fname_mem, sep=' ', index_col=1, parse_dates=True)
cpumain = cpu_mem_data[['CPU-Wait%','CPU-User%','CPU-Nice%','CPU-Sys%',]]

cpumain.plot(grid='on',subplots='True',style=['r','g','b','y'],sharex=True,kind='area',title=plot_title+' cpu-view')

这段代码已经稳定运行近10年，从未出现过问题。

报错信息 #

File "/home/loberman/collectl-parser/./parse-collectl-python3.py", line 264, in <module>
    cpumain.plot(grid='on',subplots='True',style=['r','g','b','y'],sharex=True,kind='area',title=plot_title+' cpu-view')
  File "/home/loberman/.local/lib/python3.10/site-packages/pandas/plotting/_core.py", line 1000, in __call__
    return plot_backend.plot(data, kind=kind, **kwargs)
  File "/home/loberman/.local/lib/python3.10/site-packages/pandas/plotting/_matplotlib/__init__.py", line 70, in plot
    plot_obj = PLOT_CLASSES[kind](data, **kwargs)
  File "/home/loberman/.local/lib/python3.10/site-packages/pandas/plotting/_matplotlib/core.py", line 1530, in __init__
    LinePlot.__init__(self, data, **kwargs)
  File "/home/loberman/.local/lib/python3.10/site-packages/pandas/plotting/_matplotlib/core.py", line 1337, in __init__
    MPLPlot.__init__(self, data, **kwargs)
  File "/home/loberman/.local/lib/python3.10/site-packages/pandas/plotting/_matplotlib/core.py", line 187, in __init__
    self.subplots = self._validate_subplots_kwarg(subplots)
  File "/home/loberman/.local/lib/python3.10/site-packages/pandas/plotting/_matplotlib/core.py", line 352, in _validate_subplots_kwarg
    raise ValueError(
ValueError: When subplots is an iterable, each entry should be a list/tuple of column names.

临时解决方法 #

将pandas降级至1.4版本后，问题恢复正常，降级操作日志如下：

Defaulting to user installation because normal site-packages is not writeable
Collecting pandas==1.4
  Downloading pandas-1.4.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (11.7 MB)
     |████████████████████████████████| 11.7 MB 2.7 MB/s            
Requirement already satisfied: python-dateutil>=2.8.1 in /usr/lib/python3.10/site-packages (from pandas==1.4) (2.8.1)
Requirement already satisfied: pytz>=2020.1 in /home/loberman/.local/lib/python3.10/site-packages (from pandas==1.4) (2022.5)
Requirement already satisfied: numpy>=1.21.0 in /home/loberman/.local/lib/python3.10/site-packages (from pandas==1.4) (1.23.4)
Requirement already satisfied: six>=1.5 in /usr/lib/python3.10/site-packages (from python-dateutil>=2.8.1->pandas==1.4) (1.16.0)
Installing collected packages: pandas
  Attempting uninstall: pandas
    Found existing installation: pandas 1.5.1
    Uninstalling pandas-1.5.1:
      Successfully uninstalled pandas-1.5.1
Successfully installed pandas-1.4.0

问题根源与永久修复 #

pandas 1.5.x对subplots参数的校验逻辑做了严格调整：旧版本中字符串格式的'True'会被自动解析为布尔值True，但新版本会把字符串视为可迭代对象，触发了针对迭代类型参数的校验规则，从而抛出错误。

不需要降级pandas，只需要把代码中的subplots='True'修改为布尔值subplots=True即可彻底解决问题，修改后的代码如下：

cpumain.plot(grid='on',subplots=True,style=['r','g','b','y'],sharex=True,kind='area',title=plot_title+' cpu-view')

内容的提问来源于stack exchange，提问作者Laurence Oberman