condascan

CLI tool to scan through existing conda environments. It helps you manage your conda environments by:

• Finding environments that satisfy a requirements.txt or environment.yml file. You can then clone them, instead of creating a new environment from scratch.
• Checking if a command is available in some environments, as well as getting the result of executing the command.
• Compare different environments to see which packages are unique or shared. Then, you can delete environments that are not needed anymore.

Installation

You can install condascan using pip:

pip install condascan

Additionally, conda has to be installed on your system, either using miniconda, anaconda, miniforge, mambaforge, etc. Make sure that you can execute the following command:

conda --version

Usage

Search by Requirements

Using the have command, you can scan through your conda environments and find those that satisfy a given requirement. The bash command is as follows:

condascan have <requirement>

<requirement> can be one of the following:

• A string enclosed in quotes, specifying space-separated package names in PEP 440 format, e.g., "numpy pandas", "numpy==1.2.* pandas>=2.1.0".
• A path to a requirements.txt file, generated by pip freeze
• A path to a environment.yml file, generated by conda env export

Check Command Availability

To see if a command is available in any of your conda environments, use the can-execute command:

condascan can-execute <command>

<command> can be one of the following:

• A string enclosed in quotes, specifying the command to run, e.g., "nvcc --version"
• A path to .txt file containing list of commands, one per line.

Note: To determine if a command can be executed in a given environment, condascan will actually run the command inside each environment. This means any side effects (e.g., creating files, modifying state, or triggering installations) will occur if the command succeeds. Make sure the commands you're testing are safe and have predictable behavior across environments.

Compare Environments

To compare two or more environments, use the compare command:

condascan compare <envs>

can be one of the following:

• A string enclosed in quotes, specifying space-separated installed environment names, e.g., "base env1 env2"
• A path to a .txt file containing list of installed environment names, one per line.

Caching

To speed up execution, condascan caches the results of previous runs. The cache is stored in ~/.cache/condascan. If in-between executing condascan you modify your conda environments or you want to run without cache, you can do so by adding --no-cache flag. For example:

condascan have "numpy pandas" --no-cache

Formatting Output

--verbose, --limit, and --first Flags

Note: These flags are only applicable for have and can-execute commands

To get the detailed output of condascan, you can add --verbose flag. For example:

condascan can-execute "nvcc --version" --verbose

To limit the number of environments displayed in the output, you can use the --limit flag. Note that this is only applicable for have and can-execute commands. For example:

condascan have "numpy pandas" --limit 5

To find the first environment that satisfies the requirement, you can use the --first flag. Note that in this case, condascan will only scan the environments until it finds the first one that satisfies the requirement, and then it will stop scanning further. This can significantly speed up the search if you only need one environment that meets the requirement. For example:

condascan have "numpy pandas" --first

--pip Flag

Note: These flags are only applicable for compare command

By default, the compare command will compare every installed package from any channel in the environments. If you want to compare only packages installed from pip, you can use the --pip flag. For example:

condascan compare "env1 env2" --pip