Develop Reference

Unable to install pypiwin32 library

Tried to install the pip install pypiwin32 in Google Colab for reading outlook emails. But the installation keeps on getting failed.
Tried by downgrading the python version to 3.9 as well, but didn't worked.
Any suggestions for fixing the issue?

Are you running it actually on Windows? If no, it pyiwin32 only is supported for Windows
Try using pip install --no-cache-dir pyiwin32

Tried installing !pip install pypiwin32 in Jupyter notebook, it works like a charm.
But didn't worked in colab.

PyTorch Installation - PiP - Erorr

When trying to install PyTorch, by entering the command:
pip3 install torch==1.10.2+cu113 torchvision==0.11.3+cu113 torchaudio===0.10.2+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html
I get the following Error
Error - "The Package index page used does not have a proper HTML doctype declaration."
it also says to contact the owner of the package index, to get this fixed. https://github.com/pypa/pip/issues/10825
Any solutions for this? I want to install PyTorch
EDIT: I downloaded the latest version of pip, pip 22.0.3

Seems like it's an ongoing issue with pip==22.0.{0,1,2,3} (confirm your pip version using pip3 --version or pip --version). They started a migration process to remove an HTML parser 1, but it seems like PyTorch didn't notice and now they're trying to solve it (this is the GitHub issue where they're tracking the progress on this matter).
At the moment, it seems like you have these options:
Try to use the pip install command by adding this flag: --use-deprecated=html5lib. For example:
pip3 install --use-deprecated=html5lib torch==1.10.2+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html
Downgrade pip to a "non-22.0.*" version. It seems like the recommended way is to use a version that comes with your Python version 2:
python -m pip uninstall pip && python -m ensurepip
Then, use the pip install ... command normally.
Install PyTorch using a different technique.

Cannot install tensorflow:

trying to install tensorflow with conda but this error shows up when i try to pip install it:
Unable to create process using 'C:\Users\myusername\.conda\envs\tf\python.exe "C:\Users\myusername\.conda\envs\tf\Scripts\pip-script.py" install tensorflow'
can anyone help?

PIP Install: Cannot combine --user and --target

My goal is to install a package to a specific directory on my machine so I can package it up to be used with AWS Lambda.
Here is what I have tried:
pip install snowflake-connector-python -t .
pip install --system --target=C:\Users\path2folder --install-option=--install-scripts=C:\Users\path2folder --upgrade snowflake-connector-python
Both of these options have returned the following error message:
ERROR: Can not combine '--user' and '--target'
In order for the AWS Lambda function to work, I need to have my dependencies installed in a specific directory to create a .zip file for deployment. I have searched through Google and StackOverflow, but have not seen a thread that has answered this issue.
Update: This does not seem to be a problem on Mac. The issue described is on Windows 10.

We encountered the same issue when running pip install --target ./py_pkg -r requirements.txt --upgrade with Microsoft store version of Python 3.9.
Adding --no-user to the end of it seems solves the issue. Maybe you can try that in your command and let us know if this solution works?
pip install --target ./py_pkg -r requirements.txt --upgrade --no-user

We had the same issue just in a Python course: The error comes up if Python is installed as an app from the Microsoft app store. In our case it was resolved after re-installing Python by downloading and using the installation package directly from the Python website.

I got a similar error recently. Adding my solution so that it might help someone facing the error due to the same reason.
I was facing an issue where all my pip installed packages were going to an older python brew installation folder.
As part of debugging, I was trying to install awscli-local package to user site-package using:
pip install --user awscli-local
Then I got:
ERROR: cannot combine --user and --target
In my case, it was due to the changes in pip config I had set some time back for some other reason.
I had set the 'target' config globally - removing which removed this error and my actual issue I was debugging for.
Check the following if solutions given above doesn't resolve your issue:
try the command:
pip config edit --editor <your_text_editor>
For me:
pip config edit --editor sublime
This will open the current config file where you can check if there's any conflicting configuration like the 'target' set in my case.

Adding --no-user at the end does work, and by far the easiest solution.

Cannot install tflearn in python 3

I have tensorflow 1.11.9 for python 3 and am trying to install tflearn.
I tried installation using:
"pip3 install git+https://github.com/tflearn/tflearn.git"
"pip3 install tflearn"
"pip install git+https://github.com/tflearn/tflearn.git"
"pip install git+https://github.com/tflearn/tflearn.git"
Nothing seems to work. pip install is installing a tflearn version for python 2 (I have python2.7 and 3.4 on Ubuntu 14.04), whereas pip3 is claiming that installation is successful, but I get an error saying
"No module named 'tensorflow.contrib.rnn.python.ops.core_rnn'".
Can anyone suggest what am I doing wrong? I would be happy to provide any further information that might be required. Thanks in advance.

I think your tensorflow version is probably a bit low. Current one is 1.4

I figured out that I was installing tensorflow using
sudo pip3 install --upgrade tfBinaryURL,
where, tfBinaryURL identifies the proper URL of the TensorFlow Python package. Doing the tensorflow installation using
pip3 install tensorflow
and then installling tflearn using
pip3 install git+github.com/tflearn/tflearn.git
solves my problem.