I am running a Flask server in Docker and cannot import create_app from app.py into my integration tests. I've tried a variety of naming approaches but Python is unable to find app.py.
The directory structure is as follows
/server
/test/integration/test.py
app.py
test.py has this import
from app import create_app
I tried relative imports as well but there was a "parent" error. I then played around with empty __init__.py files in an attempt to use relative imports. That didn't work. I am not sure why this is so involved to do really. What is the solution for this?
In test.py add:
# some_file.py
import sys
# caution: path[0] is reserved for script path (or '' in REPL)
sys.path.insert(1, '/path/to/app/folder')
import app
from app import create_app
This add the directory of the source file to the system so Python will know to go there when searching the things to import. This is true when both files are local on your computer. If there is a server you need to see how you modify the in-server Python environment to know the folder of the app.py file.
See: Importing files from different folder
You can try with relative imports
from ..app import create_app
Related
apologies if I'm going about this wrong, but I am quite new to python and can't quite figure out what the problem is! I have a simple Flask app that I'm trying to write tests using pytest for. The file structure is like this:
│── app.py
|── tests
│ ├── tests_app.py
I originally had app.py and tests_app.py in the same level and it all worked fine with tests passing, but now I have put tests in their own folder I can no longer import app without error.
I have tried the following in tests_app.py:
from app import app #this is what worked fine when app and tests were in the same folder
from ..app import app
from .. import app
and the linting error is 'Attempted relative import beyond top-level package' and when I run pytest it says 'ImportError: attempted relative import with no known parent package'.
Many thanks in advance!
p.s. I am using Python 3.8.5
You need to have app on the module search path. A quick fix would be to add the path to app.py to the PYTHONPATH environment variable:
export PYTHONPATH=/path/to/dir # where app.py is
If app.py is on the python path, then import app will work (barring any other import problems such as circular imports).
An effective and reproducible way to do this is to install your application. You can do this in develop mode, which means the code runs from its current directory. This is an understated, important part of developing a Python application, but to do it you will need to write code to package your application.
I am working on some python project (2.7) and I have issue with imports. When I run main.py it start scripts from tests folder etc. and save output to logs and everything works fine.
/root------------
-logs
-staticCfg
-config.py
-tests
-systemTest
-scrypt1.py
-scrypt2.py
-userTest
-uScrypt1.py
main.py
My static variables (email, name etc.) are located in config.py. I need to import config.py in scrypt1.py or scrypt2.py. I tryed adding __init__.py to tests, systemTest and staticCfg folder but I always get an error.
In my scrypt1.py:
import staticCfg as cfg
...
or
from staticCfg import *
...
I get the error:
ImportError: No module named staticCfg
The import mechanism of Python can be a bit tricky.
You can refer to the documentation for more information: Python Import Mechanism
When you use absolute imports (your import does not start with a .) as you do, the import path will start from your main script (the one you launch). In your case, it's scrypt1.py. So starting from this location, python can't find the package staticCfg.
For me, the simplest solution is to create a main script in your root directory and call scrypt1.py from there (imported using from tests.systemTet import scrypt1.py). In this case, the base package will be your root folder and you will have access to the package staticCfg from all your script files as you wanted to do.
you may add root folder to PYTHONPATH.
I have read numerous SO questions and blogs. I am trying to structure my flask application. The current structure of my application is the following:
application
run_server.py
/config
__init__.py
production.py
staging.py
development.py
/app
__init__.py
/site
__init__.py
views.py
Now, inside app/__init__.py I want to access the config based on my environment( dev, staging, production).
from ..config import config
I am getting this error:
ValueError: Attempted relative import beyond toplevel package
I have tried using -m switch.
I have also tried to set PYTHONPATH as my root directory to tell interpreter what is top level package.
I think I am missing some fundamental in relative imports.
Try using absolute import. IMHO it makes things easier to grok
from __future__ import absolute_import
from application.config import production
This is absolute because you are specifying the exact path you are importing from which reduces ambiguity.
Also, you are missing __init__.py in the application folder
If you are running your application through run_server.py then there is no need of relative import in app/__init__.py. You can simply say,
from config import <production/staging/development>
This is because, when your interpreter interprets run_server.py, at the line say, from app import <something>, it will fetch app/__init__.py content and try to execute them at toplevel i.e. from directory application.
Now, assume you are running from ..config import config at toplevel. Obviously, it will throw an error.
Assume you have configin application/config/__init__.py.
You also need __init__.py under application directory, if not, then the application/app is your top level package, you can not access application/config.
On my development (Win7) machine, my app runs fine. The folder structure looks like this (the script being run is run.py below):
package1/
__init__.py
run.py
..
Inside the app, there are some modules which do stuff like
import app from package1
Which works fine.
However, when I try deploying to a linux machine, and run that exact same file, I get an error:
ImportError: No module named package1
I looked into the sys.path of both machines when this script is being run. The first two lines of the windows machine are these:
C:\\Users\\USERNAME\\IdeaProjects\\PROJ_NAME\\package1
C:\\Users\\USERNAME\\IdeaProjects\\PROJ_NAME
Whereas the linux one only has this:
/home/username/webapps/PROJ_NAME/package1
I recognize the problem is that the second line is missing. But why is it missing? What am I missing? Did I build the folder structure wrong?
This is a crappy solution, and I don't like it, but it works.
I added this to run.py
# Fixing the python path
import sys
import os
file_location = os.path.dirname(os.path.abspath(__file__))
project_dir = os.path.abspath(os.path.join(file_location, os.pardir))
if not project_dir in sys.path:
sys.path.insert(0, project_dir)
In my home directory, I have a folder called local. In it, there are files called __init__.py and local_settings.py. My django app is in a completely different directory. When the app is NOT running in DEBUG mode, I want it to load the local_settings.py file. How can this be acheived? I read the below:
Import a module from a relative path
Importing files from different folder in Python
http://docs.python.org/tutorial/modules.html
Basically, those tutorials are allowing to import from another directory, but what about a completely different working tree? I don't want to keep doing .., .., .. etc. Is there a way to goto the home directory?
I tried the following code:
import os, sys
os.chdir(os.path.join(os.getenv("HOME"), 'local'))
from local_settings import *
But i keep seeing errors in my apache error.log for it...
os.chdir just affects the current working directory, which has nothing whatsoever to do with where Python imports modules from.
What you need to do is to add the the local directory to the Pythonpath. You can either do this from the shell by modifying PYTHONPATH, or from inside Python by modifying sys.path:
import sys
import os
sys.path.append(os.path.expanduser("~/local"))
import local_settings
In response to your concerns about source control, you can just set the source control to ignore that file, and/or have a symlink installed as part of your deploy script to link the file on the os into another. I do both , though not in django. but it's a trivial task.
your deploy layout could look like this:
/current ( symlink to /releases/v3 )
/settings/local_settings.py
/releases/v1
/releases/v2
/releases/v3
and a task runs as part of your deploy:
cd /current
ln -s /settings/local_settings.py local_settings.py
if you're deploying with fab or capistrano, it's a few lines of configuration. i'm sure you could do this with puppet/chef simply too.