I have a saved file with the output of conda list of my previously created environment. It is of the format as below
# packages in environment at /home/*****/intelpython3:
#
# Name Version Build Channel
_tflow_select 2.1.0 gpu anaconda
absl-py 0.8.0 py36_0 anaconda
affine 2.3.0 pypi_0 pypi
asn1crypto 0.24.0 py36_3 intel
astor 0.8.0 py36_0 anaconda
atomicwrites 1.3.0 pypi_0 pypi
attrs 19.3.0 pypi_0 pypi
audioread 2.1.6 py36_0 <unknown>
awscli 1.16.292 pypi_0 pypi
backcall 0.1.0 py36_2 <unknown>
backports 1.0 py36_9 <unknown>
bayesian-optimization 1.0.1 pypi_0 pypi
bleach 2.1.3 py36_2 <unknown>
Can anybody help me with some python/unix script or some other way to convert this to a yaml (environment.yml) file which can be used with conda to create a new environment
To convert your exported file to a yaml spec for conda, you can just join the Name and Version columns with and equal sign. Pandas is a good tool for this (I assumed your current environment list is named 'environment.txt'):
import pandas as pd
df = pd.read_table('environment.txt', sep='\\s+', skiprows=3, header=None,
names=['Name', 'Version', 'Build', 'Channel'])
env = df.Name + '=' + df.Version
env.to_csv('environment.yml', header=False, index=False)
The build number won't be included, but conda doesn't really need those. You will have to specify the channels yourself.
To create a yaml file of your current environment, you need the --export flag and a pipe to a file.
conda list -n intelpython3 --export > environment.yml
To use the file you can run:
conda env create -n <environment name> --file environment.yml
Newer versions of conda can also just use:
conda create -n <environment name> --file environment.yml
but there were some issues on older versions, at least on Windows.