Develop Reference

Import error even though file is clearly in path

I am running a script and I am getting the error:
Traceback (most recent call last):
File "common/tensorflow/run_tf_benchmark.py", line 30, in <module>
from common.base_benchmark_util import BaseBenchmarkUtil
ModuleNotFoundError: No module named 'common'
I am running on an ec2 instance with the following AMI: Deep Learning AMI (Ubuntu 16.04) Version 25.0 (ami-025c308193ac1a136) and I am also working in the tensorflow_p36 anaconda environment that comes from the AMI. My Ubuntu directory is structured as so:
home
--ubuntu
--benchmark_models
--benchmark
--common
--__init__.py
--base_benchmark_util.py
--tensorflow
--run_tf_benchmark.py
--__init__.py
Note, there are other files in the directory, but these are the relevant python files.
Currently, the imports at the top of run_tf_benchmark.py are
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import sys
sys.path.append('/home/ubuntu/benchmark_models/benchmarks/common')
print(sys.path)
from argparse import ArgumentParser
from common.base_benchmark_util import BaseBenchmarkUtil
and the printed sys.path is
['/home/ubuntu/benchmark_models/benchmarks/common/tensorflow', '/home/ubuntu/src/cntk/bindings/python',
'/home/ubuntu/benchmark_models/models/image_recognition/tensorflow/mobilenet_v2', '/home/ubuntu/models',
'/home/ubuntu/models/research', '/home/ubuntu/models/research/slim',
'/home/ubuntu/anaconda3/lib/python36.zip', '/home/ubuntu/anaconda3/lib/python3.6',
'/home/ubuntu/anaconda3/lib/python3.6/lib-dynload', '/home/ubuntu/anaconda3/lib/python3.6/site-packages',
'/home/ubuntu/benchmark_models/benchmarks/common']
I can clearly see common in the python path, but the import error still happens. What am I doing wrong?

Just remove 'common' while importing because you are already in 'common' directory
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import sys
sys.path.append('/home/ubuntu/benchmark_models/benchmarks/common')
print(sys.path)
from argparse import ArgumentParser
from base_benchmark_util import BaseBenchmarkUtil

Alright, I solved the issue. The issue was that I added /home/ubuntu/benchmark_models/benchmarks/common, not /home/ubuntu/benchmark_models/benchmarks/. By import benchmarks, it also allowed all of my other code to import common as well.

how to make a file act as a python module

I want to make a file that is consisted of 3 python programs.
but, when I want to access one of the there files from one of them, it cant find the folder.
I made a init python file in it so python can recognize it as a module
my folder struct:
dlgo/
__init__.py
goboard_slow.py
gotypes.py
my goboard_slow:
from dlgo.gotypes import player
error:
Traceback (most recent call last):
File "C:\Users\asus\Desktop\dlgo\goboard_slow.py", line 2, in <module>
from dlgo.gotypes import player
ImportError: No module named 'dlgo'

Access as below:
from dlgo.gotypes import players
See here more info on "Guido's decision" on imports in python 3 and complete example on how to import in python 3.

maybe try from (filename) import (functionname)

Tl;dr:
from gotypes.py import player
when you specify path to a file, interpreter starts looking for it inside same folder, unless you give it path from main dirrectory like '/' or 'C:\'

You can import a py file with the following statement:
# Other import
import os
import sys
if './dlgo' not in sys.path:
sys.path.insert(0, './dlgo')
from dlgo.gotypes import player
NOTE:
For IDE like PyCharm, you can specify the import path using the Project Structure setting tab (CTRL+ALT+S)
Helpful stack overflow questions [maybe off topic]:
What is the right way to create project structure in pycharm?
Manage import with PyCharm documentation:
https://www.jetbrains.com/help/pycharm/configuring-project-structure.html

How to import a function from a module in the same folder? [duplicate]

This question already has answers here:
import function from a file in the same folder
(4 answers)
Closed last month.
I am trying to separate my script into several files with functions, so I moved some functions into separate files and want to import them into one main file. The structure is:
core/
main.py
posts_run.py
posts_run.py has two functions, get_all_posts and retrieve_posts, so I try import get_all_posts with:
from posts_run import get_all_posts
Python 3.5 gives the error:
ImportError: cannot import name 'get_all_posts'
Main.py contains following rows of code:
import vk
from configs import client_id, login, password
session = vk.AuthSession(scope='wall,friends,photos,status,groups,offline,messages', app_id=client_id, user_login=login,
user_password=password)
api = vk.API(session)
Then i need to import api to functions, so I have ability to get API calls to vk.
Full stack trace
Traceback (most recent call last):
File "E:/gited/vkscrap/core/main.py", line 26, in <module>
from posts_run import get_all_posts
File "E:\gited\vkscrap\core\posts_run.py", line 7, in <module>
from main import api, absolute_url, fullname
File "E:\gited\vkscrap\core\main.py", line 26, in <module>
from posts_run import get_all_posts
ImportError: cannot import name 'get_all_posts'
api - is a api = vk.API(session) in main.py.
absolute_url and fullname are also stored in main.py.
I am using PyCharm 2016.1 on Windows 7, Python 3.5 x64 in virtualenv.
How can I import this function?

You need to add __init__.py in your core folder. You getting this error because python does not recognise your folder as python package
After that do
from .posts_run import get_all_posts
# ^ here do relative import
# or
from core.posts_run import get_all_posts
# because your package named 'core' and importing looks in root folder

MyFile.py:
def myfunc():
return 12
start python interpreter:
>>> from MyFile import myFunc
>>> myFunc()
12
Alternatively:
>>> import MyFile
>>> MyFile.myFunc()
12
Does this not work on your machine?

Python doesn't find the module to import because it is executed from another directory.
Open a terminal and cd into the script's folder, then execute python from there.
Run this code in your script to print from where python is being executed from:
import os
print(os.getcwd())
EDIT:
This is a demonstration of what I mean
Put the code above in a test.py file located at C:\folder\test.py
open a terminal and type
python3 C:\folder\test.py
This will output the base directory of python executable
now type
cd C:\folder
python3 test.py
This will output C:\folder\. So if you have other modules in folder importing them should not be a problem
I usually write a bash/batch script to cd into the directory and start my programs. This allows to have zero-impact on host machines

A cheat solution can be found from this question (question is Why use sys.path.append(path) instead of sys.path.insert(1, path)? ). Essentially you do the following
import sys
sys.path.insert(1, directory_path_your_code_is_in)
import file_name_without_dot_py_at_end
This will get round that as you are running it in PyCharm 2016.1, it might be in a different current directory to what you are expecting...

python shell windows - import error

I'm trying to import module from local path in Python2.7.10 Shell on Windows
I add local path to sys.path by:
import sys
sys.path.append('C:\download')
next I try to import by:
from download.program01 import *
but I've got this error:
Traceback (most recent call last):
File "<pyshell#5>", line 1, in <module>
from download.program01 import *
ImportError: No module named download.program01
On Linux this code works fine.
Does someone know what is wrong?

If download is in your pythonpath, then you should import program01 directly.
Also, please don't import *; it makes things very hard to debug. Just do import program01.

put a file __init__.py in your download folder so that python knows it is a module and do sys.path.append('C:') instead.
If you want to keep just using path and not create a module file (the __init___.py) then just keep your code like that but import doing
import program01

From imports doesn't work on linux

I have probably path inaccuracies but can't track it. I have a following statement:
from serial import SerialException
# from Utilities.UnitsFormat import UnitsFormat
from Loggers.MainLogger import NetworkLogger as Logger
import Utilities.Serial.SerialHandle.SerialHandle as SH
It works well on Windows buy fails to find path in Debian Linux:
> root#debian-armhf:/# cd usr/CROW/ATE/Drivers/PSU_PR_V2/
root#debian-armhf:/usr/CROW/ATE/Drivers/PSU_PR_V2# python PSU_PR.py
Traceback (most recent call last):
File "PSU_PR.py", line 79, in <module>
from Loggers.MainLogger import NetworkLogger as Logger
ImportError: No module named Loggers.MainLogger
root#debian-armhf:/usr/CROW/ATE/Drivers/PSU_PR_V2#
How can I resolve this while I remain simple with path management and support both operating systems?
I have tried this with no help:
import os
if os.name == 'nt':
sys.path.append("C:\CROW\ATE")
else:
sys.path.append("usr/CROW/ATE")

Assuming that the file you are manipulating the path from is located some where under the CROW/ATE directory. You could use this code to get a system independent path to that directory like so:
import re
import os
def get_project_dir():
return re.findall(''.join(['.*', os.path.join('CROW', 'ATE')]), os.path.abspath(__file__))[0]
and then do
sys.path.append(get_project_dir())
*I wasn't able to test this on windows but this should be close.