Import problem

[Johny]

I have this directory structure

C:
\A
__init__.py
amodule.py

\B
__init__.py
bmodule.py

\D
__init__.py
dmodule.py

and I want to import bmodule.py
C:\>cd \

C:\>python
Python 2.5 (r25:51908, Sep 19 2006, 09:52:17) [MSC v.1310 32 bit
(Intel)] on win
32
Type "help", "copyright", "credits" or "license" for more information.
>>> from A.B import bmodule
I am bmodule
>>>
C:\>

so far so good. Now I would like to import bmodule but if the current
directory is \D subdirectory.

C:> cd \A\B\D
C:\A\B\D>
C:\A\B\D>python
Python 2.5 (r25:51908, Sep 19 2006, 09:52:17) [MSC v.1310 32 bit
(Intel)] on win
32
Type "help", "copyright", "credits" or "license" for more information.
>>> import sys
>>> sys.path.append('C:\\A')
>>> from A.B import bmodule
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: No module named A.B

C:\>

so I can not import a module from the parent directory? Or where did I
make an error?
Thanks for help

L.

[Steven D'Aprano]

On Sat, 06 Mar 2010 03:53:53 -0800, Johny wrote:

>>>> import sys
>>>> sys.path.append('C:\\A')
>>>> from A.B import bmodule
> Traceback (most recent call last):
> File "<stdin>", line 1, in <module>
> ImportError: No module named A.B

The current directory is irrelevant, except that it is automatically
added to the PYTHONPATH. That's why you can import A.B when the current
directory is C.

You are trying to import module B from package A *inside* directory C:\A,
but there is no such package A inside C:\A. You need to add C to the
path, and then it should work.

--
Steven

[Jean-Michel Pichavant]

try

import sys
sys.path.append('C:\\')
from A.B import bmodule

JM

[News123]

Hi Steven,

Steven D'Aprano wrote:
> On Sat, 06 Mar 2010 03:53:53 -0800, Johny wrote:
>
>>>>> import sys
>>>>> sys.path.append('C:\\A')
>>>>> from A.B import bmodule
>> Traceback (most recent call last):
>> File "<stdin>", line 1, in <module>
>> ImportError: No module named A.B
>
> The current directory is irrelevant, except that it is automatically
> added to the PYTHONPATH. That's why you can import A.B when the current
> directory is C.

Minor currection:

It doesn't seem to be the current directory, but the directory, where
the script is located in, which is auto-appended to the pythonpath

Please see following example:

$ python -V
Python 2.6.4

$ mkdir A

$ touch A/__init__

$ # create A/blla.py an A/blo.py

$ cat A/bla.py
print "I am bla"
import A.blo
print "and I found blo",dir(A.blo)
$ cat A/blo.py
avar = 3
print "I am blo"

$ python A/bla.py
I am bla

Traceback (most recent call last):

File "A/bla.py", line 2, in <module>
import A.blo
ImportError: No module named A.blo

However:
$ cat alternative_bla.py
import sys
sys.path.append(".")
print "I am bla"
import A.blo
print "and I found blo",dir(A.blo)

$ python A/alternativ_bla.py
I am bla
I am blo
and I found blo ['__builtins__', '__doc__', '__file__', '__name__',
'__package__', 'avar']

bye N

[News123]

is there any 'automatic' way of finding the top level
directory?basically the 'top level directory is the first directory
going upwards, that doesn't contain a __init__.py file.

of course you could do this 'manually' by
doing:

# assume, that this module is A.amodule
import sys
import os

# I'd love to have a similiar automatic construct
if __name__ == "__main__":
level = 1 # or function locating how far to go up before
# finding a dir, whcih does not contain a __init__.py
mydir = os.path.split(__file__)[0]
topdir = os.path.join( mydir,*(("..",)*level))
abstop = os.path.abspath(topdir)
sys.path.append(abstop)

## now you can import with the normal module paths

import A.blo
print "and I found blo",dir(A.blo)

bye N

[News123]

[Jean-Michel Pichavant]

what if some user has an __init__.py file the top level directory of
your package ?

> of course you could do this 'manually' by
> doing:
>
> # assume, that this module is A.amodule
> import sys
> import os
>
> # I'd love to have a similiar automatic construct
> if __name__ == "__main__":
> level = 1 # or function locating how far to go up before
> # finding a dir, whcih does not contain a __init__.py
> mydir = os.path.split(__file__)[0]
> topdir = os.path.join( mydir,*(("..",)*level))
> abstop = os.path.abspath(topdir)
> sys.path.append(abstop)
>
> ## now you can import with the normal module paths
>
> import A.blo
> print "and I found blo",dir(A.blo)
>
>
> bye N
>
>
>
>

You don't want to do that and you don't need it neither. That's what the
env variable PYTHONPATH is for. set it correctly, install your package
inside and everything works just fine (+standard). With a linux OS it
easy to create smb links to point to any working directory. It should be
possible on windows as well.

If your package is meant to be destributed, you may use setup.py

JM

[News123]

Hi JM,

Is there any other usage of __init.py__ than indicating a module directory?
I wasn't aware of it, but you're right I did not investigte in depth and
users can of course do whatever they like.

>> of course you could do this 'manually' by
>> doing:
>>
>> # assume, that this module is A.amodule
>> import sys
>> import os
>>
>> # I'd love to have a similiar automatic construct
>> if __name__ == "__main__":
>> level = 1 # or function locating how far to go up before
>> # finding a dir, whcih does not contain a __init__.py
>> mydir = os.path.split(__file__)[0]
>> topdir = os.path.join( mydir,*(("..",)*level))
>> abstop = os.path.abspath(topdir)
>> sys.path.append(abstop)
>>
>> ## now you can import with the normal module paths
>>
>> import A.blo
>> print "and I found blo",dir(A.blo)
>>
>>

> You don't want to do that and you don't need it neither. That's what the
> env variable PYTHONPATH is for. set it correctly, install your package
> inside and everything works just fine (+standard). With a linux OS it
> easy to create smb links to point to any working directory. It should be
> possible on windows as well.

I like your idea with the symlinks.
However not sure how to do it with windows.
I assume default shortcuts won't do.

>
> If your package is meant to be destributed, you may use setup.py
>
>

Well,

It's nice if a user just unpacks a zip file and can click on any script
with the .py suffix in the tree.

(Its nice for example for tutorials / demos )

It's also nice if he can later on just delete the unpacked directory and
there will be no trace left in the registry or in the python base dir.

This is why I'm interested in solutions without setup.py or changing
environment variables.

bye

N

[News123]

Hi JM,

Is there any other usage of __init.py__ than indicating a module directory?

I wasn't aware of it, but you're right I did not investigte in depth and
users can of course do whatever they like.

>> of course you could do this 'manually' by

>> doing:
>>
>> # assume, that this module is A.amodule
>> import sys
>> import os
>>
>> # I'd love to have a similiar automatic construct
>> if __name__ == "__main__":
>> level = 1 # or function locating how far to go up before
>> # finding a dir, whcih does not contain a __init__.py
>> mydir = os.path.split(__file__)[0]
>> topdir = os.path.join( mydir,*(("..",)*level))
>> abstop = os.path.abspath(topdir)
>> sys.path.append(abstop)
>>
>> ## now you can import with the normal module paths
>>
>> import A.blo
>> print "and I found blo",dir(A.blo)
>>
>>

> You don't want to do that and you don't need it neither. That's what the
> env variable PYTHONPATH is for. set it correctly, install your package
> inside and everything works just fine (+standard). With a linux OS it
> easy to create smb links to point to any working directory. It should be
> possible on windows as well.

I like your idea with the symlinks.
However not sure how to do it with windows.
I assume default shortcuts won't do.
>

> If your package is meant to be destributed, you may use setup.py
>
>