More robust run_notebook.sh

source/_ext/run_notebook.sh

@@ -48,15 +48,14 @@ shopt -s failglob
 
 # Make sure we're in the same directory as this script,
 # the environment file, and the notebook
-cd "${0%/*}"
+cd "$(dirname "$0")"
 
 ################################     CONFIG     ################################
 
-# Choose a prefix for the new environment
-# Choosing a prefix in the working directory means that deleting the working
-# directory will clean up the environment. Note that it will not clean up the
-# micromamba cache, which may be substantial.
-PREFIX='./env'
+# Choose a name for the new environment
+# Note that deleting this will not clean up the micromamba cache, which may be
+# substantial.
+ENVNAME='__openff_examples__'
 
 # The filename of the Conda environment file to run the notebook in
 ENV_FILE='environment.yaml'
@@ -94,19 +93,20 @@ fi
 # script is running)
 eval "$(./micromamba shell hook --shell=bash)"
 
-# Create the new environment, or install packages into it from the YAML file if
-# it already exists
-if [[ -d $PREFIX/conda-meta/ ]]; then
-    micromamba update --file $ENV_FILE --prefix "$PREFIX"
+# Create the new environment, or install and update packages from the YAML file
+# if it already exists
+if micromamba env list | grep "^\s*${ENVNAME}\s*"; then
+    micromamba install --file "$ENV_FILE" --name "$ENVNAME" --yes
+    micromamba update --file "$ENV_FILE" --name "$ENVNAME" --yes
 else
-    micromamba create --file $ENV_FILE --prefix "$PREFIX"
+    micromamba create --file "$ENV_FILE" --name "$ENVNAME" --yes
 fi
 
 # Unset environment variables that might confuse Python
 unset PYTHONPATH
 unset PYTHONHOME
 
 # Activate the new environment
-micromamba activate "$PREFIX"
+micromamba activate "$ENVNAME"
 # Open the notebook in the new environment
 jupyter lab $NOTEBOOK