Conda version pip install -r requirements.txt --target ./lib

Question

What is the conda version of this?

pip install -r requirements.txt --target ./lib

I've found these commands:

while read requirement; do conda install --yes $requirement; done < requirements.txt

But it doesn't tell how to specify --target ./lib

Solution 1

You can run conda install --file requirements.txt instead of the loop, but there is no target directory in conda install. conda install installs a list of packages into a specified conda environment.

Solution 2

To create an environment named py37 with python 3.7, using the channel conda-forge and a list of packages:

conda create -y --name py37 python=3.7
conda install --force-reinstall -y -q --name py37 -c conda-forge --file requirements.txt
conda activate py37
...
conda deactivate

Flags explained:

• -y: Yes, do not ask for confirmation.
• --force-reinstall: Install the package even if it already exists.
• -q: Quiet, do not display progress bar.
• -c: Channels, additional channels to search for packages. These are URLs searched in the order

Alternatively you can create an environment.yml file instead of requirements.txt:

name: py37
channels:
  - conda-forge
dependencies:
  - python=3.7
  - numpy=1.9.*
  - pandas

Use these commands to create and activate the conda environment based on the specifications in the Yaml file:

conda env create --file environment.yml
conda activate py37

Use this command to list the environments you have:

conda info --envs

Use this command to remove the environment:

conda env remove --name py37

New! The ansible-role dockpack.base_conda can manage conda environments on Linux, Mac and Windows, and can be used to create a docker image with custom conda environments.

Solution 3

You can always try this:

/home/user/anaconda3/bin/pip install -r requirements.txt

This simply uses the pip installed in the conda environment. If pip is not preinstalled in your environment you can always run the following command

conda install pip

Solution 4

A quick search on the conda official docs will help you to find what each flag does.

So far:

• -y: Do not ask for confirmation.
• -f: I think it should be --file, so it read package versions from the given file.
• -q: Do not display progress bar.
• -c: Additional channel to search for packages. These are URLs searched in the order

Solution 5

would this work?

cat requirements.txt | while read x; do conda install "$x" -p ./lib ;done

or

conda install --file requirements.txt -p ./lib

View more solutions

Comments

[Charlie Parker]

did you try conda install --file requirements.txt?

[Ataxias]

explaining the flags would be useful

[ionox0]

Is there any reason why after the 2nd command above python3 is removed from the environment and python2 is the only one remaining?

[bbaassssiiee]

conda manages python environments, conda deactivate resets your shell, conda activate py37 sets your PATH.

[Dr_Zaszuś]

When I do this on my requirements.txt specifying versions of packages, I get InvalidVersionSpec: Invalid version '3.0.': empty version component

[Remi Cuingnet]

There us no target directory for in conda install. However specifying a location for a virtual environment is possible with the --prefix optional argument (doc) and target environment specifications can be given with the --name or --prefix optional arguments (doc).

[Hrvoje]

It could be usefull if you add where to put yaml file and how to install environment from it. I guess environment doesn't simply come to existence if you have yaml file.

[Tms91]

Just pip install -r requirements.txt works great for me.

[EfratBlaier]

best solution for me: cat requirements.txt | while read x; do conda install -y "$x" ;done

[Nikolay Fominyh]

python -m pip install -r requirements.txt with activated conda env

[Charlie Parker]

what do you mean by conda install installs a list of packages into a specified conda environment.? For me it usually installs it for whatever conda env is active at the moment.

[phd]

@CharlieParker The OP wanted to install to a different directory; pip can do this but conda probably cannot.

[William Torkington]

@Ataxias was very right, the flag explanations are very helpful. +1!

[john.da.costa]

i like that one, it could possibly run slower, but more reliable.

[arranjdavis]

Same as above, once I activated the conda environment, I just did pip3 install -r requirements.txt.