osh_dir_std 0.8.4

#!/usr/bin/env bash
# SPDX-FileCopyrightText: 2022 Robin Vobruba <hoijui.quaero@gmail.com>
# SPDX-License-Identifier: AGPL-3.0-or-later
#
# See the output of "$0 -h" for details.

# Exit immediately on each error and unset variable;
# see: https://vaneyckt.io/posts/safer_bash_scripts_with_set_euxo_pipefail/
set -Eeuo pipefail
#set -Eeu

APP_NAME="definition (CSV) to File-System+READMEs converter"
CACHE_ROOT="$HOME/.cache/$(basename "$0")"

# parameters

function print_help() {

	script_name="$(basename "$0")"
	echo -e ""
	echo -e "$APP_NAME - Given a definition.csv (as found in mod/*/definition.csv'),"
	echo -e "creates the actual file-system representation of it,"
	echo -e "and a README.md with the description text in each directory,"
	echo -e "respectively it puts the description text as the content of the sample files."
	echo -e ""
	echo -e "Usage:"
	echo -e "\t$script_name [OPTION...] <definition-csv-file> [output-dir]"
	echo -e "Options:"
	echo -e "\t-h, --help"
	echo -e "\t\tPrint this usage help and exit"
	echo -e "Examples:"
	echo -e "\t$script_name mod/unixish/definition.csv /tmp/dir-std-unixish"
	echo -e ""
}

# Process command line arguments
while [[ $# -gt 0 ]]
do
	arg="$1"
	shift # $2 -> $1, $3 -> $2, ...

	case "$arg" in
		-h|--help)
			print_help
			exit 0
			;;
		-*) # unknown switch
			>&2 echo "ERROR: Unknown switch '$arg'"
			exit 1
			;;
		*) # non-switch
			POSITIONAL+=("$arg") # save it in an array for later
			;;
	esac
done
set -- "${POSITIONAL[@]}" # restore positional parameters

if ! which curl > /dev/null
then
	>&2 echo 'ERROR: Tool "curl" not found; Please install it with "sudo apt install curl"'
	exit 1
fi

if ! which csvcut > /dev/null
then
	>&2 echo 'ERROR: Tool "csvcut" not found; Please install it with "sudo pip install csvkit"'
	exit 1
fi

if ! which jq > /dev/null
then
	>&2 echo 'ERROR: Tool "jq" not found; Please install it with "sudo apt install jq"'
	exit 1
fi

if [ -z "${1:-}" ]
then
	>&2 echo 'ERROR: Please supply path to input file (definition.csv).'
	exit 1
fi
input_file="$1"
out_root="${2:-$(pwd)}"

echo
echo "Input file: '$input_file'"
echo "Output dir: '$out_root'"
echo

# Dirs only
for row in $(csvgrep \
	-c Path \
	--regex '/$' \
	"$input_file" \
	| csvcut \
		-c Path,Description,"Sample Content" \
		| csvjson \
		| jq -r '.[] | @base64')
do
	function _col() {
		echo "$row" | base64 --decode | jq -r "{\"$1\"}[]"
	}
	printf '.'
	path="$(_col "Path")"
	desc="$(_col "Description")"
	cont="$(_col "Sample Content")"
	out_path="$out_root/$path"
	mkdir -p "$out_path"
	rdm="$out_path/README.md"
	if ! [ "$cont" = "null" ]
	then
		>&2 echo "ERROR: 'Sample Content' of dirs should always be empty/none/null, but is not for '$out_path'!"
		exit 1
	fi
	if ! [ -f "$rdm" ]
	then
		{
			echo '# README `'"$path"'`'
			echo
			echo "$desc"
			echo
		} > "$rdm"
	fi
done
echo

# Files only
mkdir -p "$CACHE_ROOT"
for row in $(csvgrep \
	-c Path \
	--regex '/$' \
	-i \
	"$input_file" \
	| csvcut \
		-c Path,Description,"Sample Content" \
		| csvjson \
		| jq -r '.[] | @base64')
do
	function _col() {
		echo "$row" | base64 --decode | jq -r "{\"$1\"}[]"
	}
	printf '.'
	path="$(_col "Path")"
	desc="$(_col "Description")"
	cont="$(_col "Sample Content")"
	if [ "$path" = "null" ]
	then
		continue
	fi
	out_file="$out_root/$path"
	mkdir -p "$(dirname "$out_file")"
	if [[ $cont == http* ]]
	then
		url_hash=$(echo "$cont" | sha1sum)
		cont_cache="$CACHE_ROOT/$url_hash"
		if ! [ -e "$cont_cache" ]
		then
			curl "$cont" --output "$cont_cache"
		fi
		cp "$cont_cache" "$out_file"
	else
		if [ "$cont" = "null" ]
		then
			cont=""
		fi
		echo "$cont" > "$out_file"
	fi
done
echo

echo "done."