Jupyter Notebook安装库时触发pexpect模块AttributeError错误求助

问题重现 #

执行以下安装命令时触发错误：

!pip install tensorflow==2.4.1 tensorflow-gpu==2.4.1 opencv-python mediapip sklearn matplotlib

报错信息：

AttributeError                            Traceback (most recent call last)
Cell In[1], line 1
----> 1 get_ipython().system('pip install tensorflow==2.4.1 tensorflow-gpu==2.4.1 opencv-python mediapip sklearn matplotlib')

File /lib/python3.11/site-packages/IPython/core/interactiveshell.py:2590, in InteractiveShell.system_piped(self, cmd)
   2585     raise OSError("Background processes not supported.")
   2587 # we explicitly do NOT return the subprocess status code, because
   2588 # a non-None value would trigger :func:`sys.displayhook` calls.
   2589 # Instead, we store the exit_code in user_ns.
-> 2590 self.user_ns['_exit_code'] = system(self.var_expand(cmd, depth=1))

File /lib/python3.11/site-packages/IPython/utils/_process_posix.py:129, in ProcessHandler.system(self, cmd)
    125 enc = DEFAULT_ENCODING
    127 # Patterns to match on the output, for pexpect.  We read input and
    128 # allow either a short timeout or EOF
-> 129 patterns = [pexpect.TIMEOUT, pexpect.EOF]
    130 # the index of the EOF pattern in the list.
    131 # even though we know it's 1, this call means we don't have to worry if
    132 # we change the above list, and forget to change this value:
    133 EOF_index = patterns.index(pexpect.EOF)

AttributeError: module 'pexpect' has no attribute 'TIMEOUT

原因分析 #

这个错误是IPython版本与pexpect版本不兼容导致的：IPython的_process_posix.py代码中引用了pexpect.TIMEOUT和pexpect.EOF，但新版本pexpect（>=4.8.0）已将这些属性移至pexpect.exceptions模块，旧版IPython未适配该变更。此外，你使用的Python 3.11与TensorFlow 2.4.1本身存在兼容性问题（TensorFlow 2.4.1不支持Python 3.11），会间接加剧依赖冲突。

解决方案 #

方案1：修复pexpect兼容性 #

先降级pexpect到适配IPython的版本，再重新执行安装命令：

!pip install pexpect==4.7.0
!pip install tensorflow==2.4.1 tensorflow-gpu==2.4.1 opencv-python mediapipe scikit-learn matplotlib

注意：TensorFlow 2.4.1仅支持Python 3.7-3.9，若坚持使用该版本，需先切换到对应Python环境。

方案2：更换TensorFlow版本适配Python 3.11 #

放弃TensorFlow 2.4.1，安装支持Python 3.11的版本（如TensorFlow 2.13+）：

!pip install tensorflow==2.13.0 opencv-python mediapipe scikit-learn matplotlib

修正了原命令中的拼写错误：mediapip改为mediapipe，建议使用官方包名scikit-learn替代别名sklearn。

方案3：绕过IPython的system方法 #

直接用Python的subprocess模块执行安装，避开pexpect的兼容性问题：

import subprocess
subprocess.run(["pip", "install", "tensorflow==2.4.1", "tensorflow-gpu==2.4.1", "opencv-python", "mediapipe", "scikit-learn", "matplotlib"])

额外注意事项 #

• TensorFlow 2.4.1需要搭配CUDA 11.0和cuDNN 8.0才能启用GPU支持；TensorFlow 2.10之后已无需单独安装tensorflow-gpu，GPU支持整合在tensorflow包中。
• 优先推荐匹配Python版本与TensorFlow版本，避免不必要的依赖冲突。

内容的提问来源于stack exchange，提问作者Essalution44