Importing .py files in Google Colab

问题

Is there any way to upload my code in .py files and import them in colab code cells?

The other way I found is to create a local Jupyter notebook then upload it to Colab, is it the only way?

回答1:

You can save it first, then import it.

from google.colab import files
src = list(files.upload().values())[0]
open('mylib.py','wb').write(src)
import mylib

Update (nov 2018): Now you can upload easily by

click at [>] to open the left pane
choose file tab
click [upload] and choose your [mylib.py]
import mylib

Update (oct 2019): If you don't want to upload every time, you can store it in S3 and mount it to Colab, as shown in this gist

回答2:

You can upload local files to google colab by using upload() function in google.colab.files
If you have files on github, then clone the repo using !git clone https://github.com/username/repo_name.git. Then just like in jupyter notebook load it using the magic function %load %load filename.py.

回答3:

In case anyone else is interested to know how to import files/packages from gdrive inside a google colab. The following procedure worked for me:

1) Mount your google drive in google colab:

from google.colab import drive
drive.mount('/content/gdrive/')

2) Append the directory to your python path using sys:

import sys
sys.path.append('/content/gdrive/mypythondirectory')

Now you should be able to import stuff from that directory!

回答4:

Based on the answer by Korakot Chaovavanich, I created the function below to download all files needed within a Colab instance.

from google.colab import files
def getLocalFiles():
    _files = files.upload()
    if len(_files) >0:
       for k,v in _files.items():
         open(k,'wb').write(v)
getLocalFiles()

You can then use the usual 'import' statement to import your local files in Colab. I hope this helps

回答5:

Try this way:

I have a package named plant_seedlings. The package is stored in google drive. What I should do is to copy this package in /usr/local/lib/python3.6/dist-packages/.

!cp /content/drive/ai/plant_seedlings.tar.gz /usr/local/lib/python3.6/dist-packages/

!cd /usr/local/lib/python3.6/dist-packages/ && tar -xzf plant_seedlings.tar.gz

!cd /content

!python -m plant_seedlings

回答6:

You can upload those .py files to Google drive and allow Colab to use to them:

!mkdir -p drive
!google-drive-ocamlfuse drive

All your files and folders in root folder will be in drive.

回答7:

I face the same problem. After reading numerous posts, I would like to introduce the following solution I finally chose over many other methods (e.g. use urllib, httpimport, clone from GitHub, package the modules for installation, etc). The solution utilizes Google Drive API (official doc) for proper authorization.

Advantages:

Easy and safe (no need for code to handle file operation exceptions and/or additional authorization)
Module files safeguarded by Google account credentials (no one else can view/take/edit them)
You control what to upload/access (you can change/revoke access anytime on a file-by-file basis)
Everything in one place (no need to rely upon or manage another file hosting service)
Freedom to rename/relocate module files (not path-based and won't break your/other's notebook code)

Steps:

Save your .py module file to Google Drive - you should have that since you're already using Colab
Right click on it, "Get shareable link", copy the part after "id=" - the file id assigned by Google Drive
Add and run the following code snippets to your Colab notebook:

!pip install pydrive                             # Package to use Google Drive API - not installed in Colab VM by default
from pydrive.auth import GoogleAuth
from pydrive.drive import GoogleDrive

from google.colab import auth                    # Other necessary packages
from oauth2client.client import GoogleCredentials

auth.authenticate_user()                         # Follow prompt in the authorization process
gauth = GoogleAuth()
gauth.credentials = GoogleCredentials.get_application_default()

drive = GoogleDrive(gauth)
your_module = drive.CreateFile({"id": "your_module_file_id"})   # "your_module_file_id" is the part after "id=" in the shareable link
your_module.GetContentFile("your_module_file_name.py")          # Save the .py module file to Colab VM
import your_module_file_name                                    # Ready to import. Don't include".py" part, of course :)

Side note

Last but not least, I should credit the original contributor of this approach. That post might have some typo in the code as it triggered an error when I tried it. After more reading and troubleshooting my code snippets above worked (as of today on Colab VM OS: Linux 4.14.79).

回答8:

It's Jun 2019. Make sure in the Python package's __init__.py all related files are imported in order. Push the code to Git or use this code.

for e.g,

from .Boxes import *
from .Circles import *
from .Rectangles import *
...

Don't use Package name in __init__.py file for importing the files.

in Google colab,

! rm -rf SorghumHeadDetection
! git clone https://github.com/user/amazing-repo-name/

来源：https://stackoverflow.com/questions/48905127/importing-py-files-in-google-colab

标签

ipython

jupyter-notebook

google-colaboratory