为什么在文件中而不是在Python控制台中执行sys.path.append时会出现ImportError？

Question

I'm trying to run a script that requires a package in another directory.

This works:

• execfile("../test.py") when starting python in package parent directory
• sys.path.append("package parent") while in the python interpreter and then calling execfile("test.py")

This doesn't work (gives an ImportError ):

• python ../test.py when in package parent directory
• python test.py when in script directory

The test.py file contains the same sys.path.append instruction I used in the interpreter. Is there a difference between running it in the file or in the interpreter?

---

My directory structure:

• test.py
• package
  • stuff.py

The error message I get:

from package.stuff import SomeClass
ImportError: No module named package.stuff

Answer 1

A hint why the behavior from the interactive shell is different from script behavior can be found in the docs for sys.path :

As initialized upon program startup, the first item of this list, path[0] , is the directory containing the script that was used to invoke the Python interpreter. If the script directory is not available (eg if the interpreter is invoked interactively or if the script is read from standard input), path[0] is the empty string, which directs Python to search modules in the current directory first. Notice that the script directory is inserted before the entries inserted as a result of PYTHONPATH .

This should explain why executing from an interactive shell works. However, without further information it can only be guessed why the script fails.

Setting PYTHONPATH is essentially the same as updating sys.path . On bash this would be:

PYTHONPATH=/path/to/package1:/path/to/package2 python test.py

Check out this post and the docs for details.