DEV Community

guzmanojero
guzmanojero

Posted on • Originally published at stackoverflow.com

Running Jupyter via command line on Windows

#python #jupyter #windows

I had the exact same problem and it was driving me crazy. Other answers provide a solution, but they don't explain why you and I are having this problem.

I will try to explain why this is happening and then provide some solutions.

You can go to the end to see the TL;DR.

1)What's going on? Why is this error happening?

I'll try to make a step-by-step answer so everything is explained clearly.
If you think it's too basic at the beginning, go to the end of this "article".

I'll first start with common things like running the python shell from the terminal or running pip. You'll see why you can do that from the terminal and we'll end up on why and how you can run the jupyter notebook from the terminal as well.

Ready? Let's start!

Have you ever wondered why you can type python in the terminal (command prompt) and suddenly start the Python interpreter?

    Microsoft Windows [Version 10.0.18363.1440]
    (c) 2019 Microsoft Corporation. All rights reserved.

    C:\Users\YOUR-USERNAME>python
    Python 3.9.1 (tags/v3.9.1:1e5d33e, Dec  7 2020, 17:08:21) [MSC v.1927 64 bit (AMD64)] on win32
    Type "help", "copyright", "credits" or "license" for more information.
    >>>
Enter fullscreen mode Exit fullscreen mode

You probably already know (but maybe don't) that this is because Python was added to the Windows PATH environment variable. You probably did it at installation time or afterwards.

But, what is this PATH environment variable?

It basically allows you to run any executables, that are located inside
the paths specified in the variable, at the command prompt without
having to give the full path to the executable.

You can check the content of that PATH variable with:

    >>> import sys
    >>> for path in sys.path:
            print(path)
Enter fullscreen mode Exit fullscreen mode 
    C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39\python39.zip
    C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39\DLLs
    C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39\lib
    C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39
    C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39\lib\site-packages
    ... (some other paths were taken out for clarity)
Enter fullscreen mode Exit fullscreen mode

You can see this folder: C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39. This is the place where Python version 3.9 is installed.

Let's check its content:

    <DIR>   DLLs
    <DIR>   Doc
    <DIR>   etc
    <DIR>   include
    <DIR>   Lib
    <DIR>   libs
    <DIR>   Scripts
    <DIR>   share
    <DIR>   tcl
    <DIR>   Tools
            LICENSE.txt
            NEWS.txt
            python.exe
            python3.dll
            python39.dll
            pythonw.exe
            vcruntime140.dll
            vcruntime140_1.dll
Enter fullscreen mode Exit fullscreen mode

Voilà! We have the python.exe file (an executable). We have a Python executable file in the PATH, that's why you can start the Python interpreter from the terminal with just typing python. If this wasn't the case you would have to type the full path to the executable file in the terminal:

    C:\Users\YOUR-USERNAME> C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39\python)
Enter fullscreen mode Exit fullscreen mode

Instead of just:

    C:\Users\YOUR-USERNAME> python
Enter fullscreen mode Exit fullscreen mode

And what about when you use pip?

It's the same principle. You can run pip from the terminal because there is a pip executable file in the PATH variable.

If you go to C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39\Scripts\ (which is in the PATH showed above) you'll see many executables files. One of them is pip. Actually I have three versions: pip, pip3.9 and pip3.

The Scripts folder allows exectuable files to be run from the terminal. Like pip or other libraries that you intend to run directly from the terminal. The Scripts folder:

...is not intended for you, it's for scripts that are installed as
components of modules that you install. For example, pip is a module,
but it also has a wrapper script by the same name, pip, which will be
installed in that directory.

If you put something there and it is properly in your PATH, then it
should be executable

That wrapper script would be the pip executable file. When this executable file is run, it locates the pip folder in the Python installation folder and runs pip.

But you could also run pip directly from the installation folder (C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39\Lib\site-packages), without needing the executable pip file.

But, how can you do it?

I'm glad you ask. There is a Python way to run modules as the main module (without the need to import it).

    python -m pip
Enter fullscreen mode Exit fullscreen mode

When you run a module directly its name becomes __main__. What -m does is:

Search sys.path for the named module and execute its contents as the __main__ module.

What is __main__?

'__main__' is the name of the scope in which top-level code executes.

A module’s __name__ is set equal to '__main__' when read from standard
input, a script, or from an interactive prompt.
...

I guess that the pip executable does something similar, or at least, has the same effect: to start pip.

2)What does this have to do with the Jupyter Notebook?!

Think of the Jupyter Notebook as the same as pip. If you want to run jupyter in the terminal, you need an executable that it's on the PATH.

We have already seen that the executables of modules like pip or jupyter are located here C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39\Scripts\.

If I check the content of the folder I see this:

    easy_install-3.9.exe
    easy_install.exe
    f2py.exe
    jsonschema.exe
    jupyter-bundlerextension.exe
    jupyter-console.exe
    jupyter-nbconvert.exe
    jupyter-nbextension.exe
    jupyter-notebook.exe
    jupyter-qtconsole.exe
    jupyter-serverextension.exe
    jupyter-trust.exe
    pip.exe
    pip3.9.exe
    pip3.exe
Enter fullscreen mode Exit fullscreen mode

I see the already mentioned pip, pip3.9 and pip3. But I don't see jupyter (the word "jupyter" alone).

If I type jupyter in the terminal I get the error that started all:

'jupyter' is not recognized as an internal or external command, operable program or batch file.
Enter fullscreen mode Exit fullscreen mode

Finally we've reached an answer to your question!!!

'jupyter' is not recognized as a command because there is no executable file in the Scripts folder called jupyter.

So, let's try a different executable. What about jupyter-notebook?

BINGO! The notebook is running!

    Serving notebooks from local directory:
    C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39\Scripts
    Jupyter Notebook 6.3.0 is running at:
    http://localhost:8888/?token=... (edited)
    or http://127.0.0.1:8888/?token=... (edited)
    Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
Enter fullscreen mode Exit fullscreen mode

I don't know why I don't have a jupyter executable called 'jupyter'. The official documentation says to use jupyter notebook on the terminal, but it seems that in some cases it doesn't work. And I think it has to do with what I mentioned above: there is no jupyter exectuable in the Scripts folder.

If you remember, I told you that you can run pip as the main module using python -m pip.

It happens that you can do the same with jupyter.We just need to know how to call it. As in with pip, we have to check the folder where 3rd party libraries are installed: C:\Users\YOUR-USERNAME\AppData\Local\Programs\Python\Python39\Lib\site-packages.

You'll see jupyter_console, but this just creates an interactive notebook in the terminal, not exactly what you were looking for. You're also going to find folders ending with .dist.info, like jupyter_console-6.4.0.dist-info. This is just metadata of the Wheel Binary Package builder. You'll also see a folder like jupyterlab_pygments, but that's for JupyterLab. We want to use the classic Jupyter notebook.

What we want is to run notebook. How do we know this?

You'll see in the folder site-packages the folder (package) notebook. Inside there is a file called __main__.py:

    #__main__.py
    if __name__ == '__main__':
        from notebook import notebookapp as app
        app.launch_new_instance()
Enter fullscreen mode Exit fullscreen mode

It's calling notebookapp.py which is a "A tornado based Jupyter notebook server." Yes, this is what we need.

We can see that launch_new_instance in the notebookapp calls launch_instance(), which "launches an instance of a Jupyter Application".

Perfect! We are in the correct folder. To run the jupyter notebook from the Python interactive shell we have to run the notebook package with:

    python -m notebook
Enter fullscreen mode Exit fullscreen mode

3)*** SUMMARY: SOLUTION ***

tl;dr:

I have explained and showed why this error is happening.

Now let's summarize the solutions:

  1. To know the name of the jupyter executable (in the Scripts folder), so you can run directly from the terminal (Command Prompt) as: 
        jupyter notebook
Enter fullscreen mode Exit fullscreen mode

or as:

        jupyter-notebook
Enter fullscreen mode Exit fullscreen mode

Or whatever name you have.

  1. Run the notebook as the main module from Python: 
        python -m notebook
Enter fullscreen mode Exit fullscreen mode

I hope this helps you as much as it helped me. I'm open to your comments and suggestions.

NOTE: Originally posted in StackOverflow
https://stackoverflow.com/a/66938692/6710903

Top comments (0)

Read next

kumarkalyan profile image

Building a 🧾 PDF Invoice Generator with 📨 Twilio SendGrid, 🐍 Python, Flask, and 🖌️ Bootstrap

Kumar Kalyan -

namimai profile image

Python Error Types Explained: Troubleshooting for Beginners

Namima Islam -

radchenko profile image

Practical Experience: Integrating Over 50 Neural Networks Into One Open-Source Project

Vladislav Radchenko -

craftedwithintent profile image

Enhancing Observability in Machine Learning with OpenTelemetry: InsightfulAI Update

Philip Thomas -