[BUG] dash doesn't get imported when a file named "org.py", "dash.py", or "test.py" with specific content is present in the current directory // "AttributeError: module 'dash' has no attribute 'Dash'"

[cannatag]

Describe your context

dash (1.9.1)
dash-core-components (1.8.1)
dash-html-components (1.0.2)
dash-renderer (1.2.4)
dash-table (4.6.1)

Describe the bug

If a file named org.py is present in the current directory with the following content:

import dash_core_components as dcc

then dash doesn't import and I get the following message:

>>> import dash
Dash was not successfully imported. Make sure you don't have a file named
'dash.py' in your current directory.

Expected behavior
dash should import without any error.

Additional info

• The org.py is never imported
• If I rename the file to a different name dash get imported without any problem.
• The problem is shown also with ``import dash_html_components as html```
• The problem is shown either on Windows and in Linux
• Tested with python3.4, python3.6, python3.8

Steps to replicate the problem on Linux

$ mkdir mytest
$ cd mytest
$ echo "import dash_core_components as dcc" > org.py
$ python3 -m venv venv
$ . venv/bin/activate
(venv) $ pip install dash
(venv) $ python
Python 3.4.6 (default, Mar 01 2017, 16:52:22) [GCC] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import dash
Dash was not successfully imported. Make sure you don't have a file named
'dash.py' in your current directory.
(venv) $

if I rename the file the import works:

(venv) $ mv org.py othername.py
(venv) $ python
Python 3.4.6 (default, Mar 01 2017, 16:52:22) [GCC] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import dash
>>>

[Charlenator]

I had the exact same issue with a file named "test.py"

[alexcjohnson]

OK this one is very weird... it appears to come from the pickle module:

# Jython has PyStringMap; it's a dict subclass with string keys
try:
    from org.python.core import PyStringMap
except ImportError:
    PyStringMap = None

Which you can see if you raise an error inside org.py - it is trying to import it, to find org.python.core

That seems like a terribly hacky bit of code to have living in the Python core, but we'll have to live with that part. This becomes a problem when org.py tries to import something that itself needs dash (in this case dash_core_components) and finds a partially-initialized module. I suppose we could put something unique very early in the dash __init__.py (something like __plotly_dash__ = True) and short-circuit the "Make sure you don't have a file named 'dash.py' in your current directory." errors if we find it.

I don't see a problem with test.py but if we make a generic fix like ^^ it should cover any related oddities.

[karolprzyb]

Exact same issue with test.py
For some reason downgrading to dash 1.12.0 or lower also fixed the issue. I cannot for the life of me find why though.

[alexcjohnson]

@meltbox360 can you try raising an exception at the top of test.py and post the traceback you see?

[karolprzyb]

Traceback (most recent call last):
  File "C:/Users/xxx/PycharmProjects/plotGPS_OSM/Legacy/test6.py", line 14, in <module>
    import dash
  File "C:\Users\xxx\Anaconda3\envs\OSMNX and Dash\lib\site-packages\dash\__init__.py", line 1, in <module>
    from .dash import Dash, no_update  # noqa: F401
  File "C:\Users\xxx\Anaconda3\envs\OSMNX and Dash\lib\site-packages\dash\dash.py", line 15, in <module>
    from future.moves.urllib.parse import urlparse
  File "C:\Users\kprzybys\Anaconda3\envs\OSMNX and Dash\lib\site-packages\future\moves\__init__.py", line 8, in <module>
    import_top_level_modules()
  File "C:\Users\xxx\Anaconda3\envs\OSMNX and Dash\lib\site-packages\future\standard_library\__init__.py", line 810, in import_top_level_modules
    with exclude_local_folder_imports(*TOP_LEVEL_MODULES):
  File "C:\Users\xxx\Anaconda3\envs\OSMNX and Dash\lib\site-packages\future\standard_library\__init__.py", line 781, in __enter__
    module = __import__(m, level=0)
  File "C:\Users\xxx\PycharmProjects\plotGPS_OSM\Legacy\test.py", line 1, in <module>
    raise Exception("TEST")
Exception: TEST

Process finished with exit code 1

[karolprzyb]

Good call. I think test is a reserved module name in python see standard_library __init__.py ~line 800 probably

TOP_LEVEL_MODULES = ['builtins',
                     'copyreg',
                     'html',
                     'http',
                     'queue',
                     'reprlib',
                     'socketserver',
                     'test',
                     'tkinter',
                     'winreg',
                     'xmlrpc',
                     '_dummy_thread',
                     '_markupbase',
                     '_thread',
                    ]

Still have no idea why this works with 1.12.0 and before though.
I also have no idea if what I said above is true. Relatively new to python and haven't learned it too well yet.

[alexcjohnson]

Thanks @meltbox360 that's perfect. Haha with exclude_local_folder_imports(*TOP_LEVEL_MODULES) obviously didn't accomplish its goal, as test is in your local folder 🤔 It's also puzzling to me why I don't see the same issue with test via future, maybe exclude_local_folder_imports does the right thing on Mac but not on Windows? Anyway normally it should be fine to make a module named test. There does seem to be a built-in test module but the docs strongly discourage its use so I have no idea why future would reference it. But this is indeed the same issue as org so should the same fix should work.

[amirdora]

rename file test.py to something else

for it was caused by my file name which i named "test.py".

[chriddyp]

For folks googling, I believe this error also appears in the form "AttributeError: module 'dash' has no attribute 'Dash'"