如何在 MacOSX 上使用 pyinstaller 打包 tensorflow？

Question

我正在分发一个包含 tensorflow 的 python 程序。在 OSX，python 2.7 上，加载库和运行 tensorflow 没有错误。当使用 pyinstaller 创建可执行文件时，它会因 PBR 错误而失败。

我们可以将此问题简化为一个简单的（希望可重现）错误。

这是我们的程序。让我们加载 tensorflow 并让它打印出一些路径信息。

tf_check.py

import tensorflow
print(tensorflow.__file__)

使用 pyinstaller 运行。

Bens-MacBook-Pro:tests ben$ pyinstaller tf_check.py 
66 INFO: PyInstaller: 3.3
66 INFO: Python: 2.7.10
78 INFO: Platform: Darwin-16.7.0-x86_64-i386-64bit
79 INFO: wrote /Users/ben/Documents/DeepMeerkat/tests/tf_check.spec
87 INFO: UPX is not available.

构建没有问题。尝试运行dist/tf_check/下的tf_check

产量

Last login: Thu Sep 28 12:54:34 on ttys000
Bens-MacBook-Pro:~ ben$ /Users/ben/Documents/DeepMeerkat/tests/dist/tf_check/tf_check ; exit;
Traceback (most recent call last):
  File "tf_check.py", line 1, in <module>
  File "/Library/Python/2.7/site-packages/PyInstaller-3.3-py2.7.egg/PyInstaller/loader/pyimod03_importers.py", line 396, in load_module
    exec(bytecode, module.__dict__)
  File "tensorflow/__init__.py", line 24, in <module>
  File "/Library/Python/2.7/site-packages/PyInstaller-3.3-py2.7.egg/PyInstaller/loader/pyimod03_importers.py", line 396, in load_module
    exec(bytecode, module.__dict__)
  File "tensorflow/python/__init__.py", line 107, in <module>
  File "/Library/Python/2.7/site-packages/PyInstaller-3.3-py2.7.egg/PyInstaller/loader/pyimod03_importers.py", line 396, in load_module
    exec(bytecode, module.__dict__)
  File "tensorflow/python/platform/test.py", line 60, in <module>
  File "/Library/Python/2.7/site-packages/PyInstaller-3.3-py2.7.egg/PyInstaller/loader/pyimod03_importers.py", line 396, in load_module
    exec(bytecode, module.__dict__)
  File "mock/__init__.py", line 2, in <module>
  File "/Library/Python/2.7/site-packages/PyInstaller-3.3-py2.7.egg/PyInstaller/loader/pyimod03_importers.py", line 396, in load_module
    exec(bytecode, module.__dict__)
  File "mock/mock.py", line 71, in <module>
  File "pbr/version.py", line 461, in semantic_version
  File "pbr/version.py", line 448, in _get_version_from_pkg_resources
  File "pbr/packaging.py", line 755, in get_version
Exception: Versioning for this project requires either an sdist tarball, or access to an upstream git repository. It's also possible that there is a mismatch between the package name in setup.cfg and the argument given to pbr.version.VersionInfo. Project name mock was given, but was not able to be found.
[1916] Failed to execute script tf_check
logout
Saving session...
...copying shared history...
...saving history...truncating history files...
...completed.
Deleting expired sessions...9 completed.

[Process completed]

一些系统信息

Python 2.7.10 (default, Feb  7 2017, 00:08:15) 
[GCC 4.2.1 Compatible Apple LLVM 8.0.0 (clang-800.0.34)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow
>>> tensorflow.__version__
'1.3.0'

相同的代码在 Windows 上运行没有问题。同样，仅在使用 pyinstaller 分发时。

谁能展示在 OSX 上打包 tensorflow 的示例？

编辑#1

关注Tensorflow CUDA GTX 1070 import error

• 我尝试在干净的虚拟环境中安装，只需 pip install tensorflow 和 pip 安装 pyinstaller，没有改善。

• 我尝试在构建 pyinstaller 之前导出 PBR 版本 tf_check.py，没有改进。

(env) Bens-MacBook-Pro:tests ben$ echo $PBR_VERSION 3.1.1
(env) Bens-MacBook-Pro:tests ben$ open dist/tf_check/tf_check

Answer 1

在导入之前放置在 tensorflow 脚本的顶部。

import os

#export pbr version for tensorflow user
os.environ["PBR_VERSION"]='3.1.1'

或者 pbr -v 给你的任何版本