visualize

#!/usr/bin/env sh
# SPDX-FileCopyrightText: 2020 - 2023 Robin Vobruba <hoijui.quaero@gmail.com>
# SPDX-License-Identifier: Unlicense

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

script_path="$(readlink -f "$0")"
script_dir="$(dirname "$script_path")"
script_name="$(basename "$script_path")"

root_dir="$script_dir"
build_dir="$root_dir/build"
publish_dir="$root_dir/public"
tsdc_tools_dl_url="https://osegermany.gitlab.io/oh-tsdc-tools/tsdc-tools-latest-jar-with-dependencies.jar"
tsdc_tools="$build_dir/tsdc-tools.jar"
vrdf_dl_url="https://gitlab.com/OSEGermany/rdf-visualizer/raw/master/visualize-rdf"
vrdf="$build_dir/visualize-rdf"
force=false

_install_requirements() {
	mkdir -p "$build_dir"

	# Installs wget, JDK, Python
	if ! ( which wget > /dev/null && which pip > /dev/null && which javac > /dev/null )
	then
		apt-get update
		apt-get install -y -qq wget python3-pip default-jdk > /dev/null
	fi

	# Installs `visualize-rdf`
	mkdir -p "$(dirname "$vrdf")"
	wget "$vrdf_dl_url" \
		--output-document="$vrdf"
	chmod +x "$vrdf"
	"$vrdf" --install-requirements

	# Installs TsDC tools
	mkdir -p "$(dirname "$tsdc_tools")"
	wget "$tsdc_tools_dl_url" \
		--output-document="$tsdc_tools"
}

_print_help() {

	echo "Create differently layouted visual (graph-style) representations"
	echo "of all the RDF files from CWD, recursively."
	echo "This generates multiple SVG, PNG and a single DOT file"
	echo "for each TTL file."
	echo "It also creates a structures Markdown file with the RDFs content."
	echo "In the end, it also creates an index.html file"
	echo "that links to all the generated files."
	echo
	echo "Usage:"
	echo "  $script_name [OPTIONS]"
	echo "Options:"
	echo "  -h, --help"
	echo "     Show this help message and exit"
	echo "  -s, --install-requirements"
	echo "     Installs all the requirements for this script to run successfully and exit"
	echo "  -f, --force"
	echo "     Overwrite index.html if it already exists."
	echo "     NOTE: All the other generated files are overwritten"
	echo "           even without this flag."
	echo "Examples:"
	echo "  $script_name"
	echo "  $script_name --help"
}

# read command-line args
i=1
while [ "$i" -lt "$#" ]
do
	arg="$(eval "echo \$$i")"

	case "$arg" in
		-h|--help)
			shift "$i"
			_print_help
			exit 0
			;;
		-s|--install-requirements)
			shift "$i"
			_install_requirements
			exit 0
			;;
		-f|--force)
			shift "$i"
			force=true
			;;
		*) # non-/unknown option
			i=$((i + 1))
			;;
	esac
done

if [ ! -f "$vrdf" ]
then
	_install_requirements
fi

_clean_dot_file() {
	dot_file="$1"

	layout_engine=fdp
	dot_file_clean="${dot_file%.*}-clean.dot"
	echo "Cleaning '$dot_file' -> '$dot_file_clean' ..."
	{
		grep -v 'wikidata' \
			| grep -v ':base' \
			| grep -v 'ObjectProperty' \
			| grep -v 'DatatypeProperty' \
			| grep -v 'Class' \
			| grep -v 'Ontology'
	} < "$dot_file" > "$dot_file_clean"
	svg_file="${dot_file_clean%.dot}.svg"
	png_file="${dot_file_clean%.dot}.png"
	echo "Generating SVG and PNG from '$dot_file_clean' ..."
	dot -Glayout="$layout_engine" -Grankdir=LR -Tsvg -o "$svg_file" "$dot_file_clean"
	dot -Glayout="$layout_engine" -Grankdir=LR -Tpng -o "$png_file" "$dot_file_clean"
}

mkdir -p "$publish_dir"

echo
find . -name "*.ttl" \
	| grep -v "^\./$(basename "$build_dir")/" \
	| grep -v "^\./$(basename "$publish_dir")/" \
	| while read -r ttl_file
do
	printf "\n\nProcessing input file '%s' ...\n" "$ttl_file"

	base_output_dir="$publish_dir/$(dirname "$ttl_file")"
	mkdir -p "$base_output_dir"
	base_name="$(basename "${ttl_file%.*}")"

	cp "$ttl_file" "$base_output_dir/"

	# Create graphs
	for layout in fdp dot
	do
		base_output_file="$base_output_dir/${layout}_$base_name"

		# Generates PNG, SVG & DOT files
		"$vrdf" \
			--do-not-open \
			--layout "$layout" \
			"$ttl_file" \
			"$base_output_file"

		# Generates a cleaned-up DOT file and SVG and PNG from it
		dot_file="${base_output_file}.dot"
		_clean_dot_file "$dot_file"
	done

	# Create structured Markdown
	java \
		-classpath "$tsdc_tools" \
		de.opensourceecology.tsdc.tools.rdf2md.MainKt \
		--input "$ttl_file" \
		--output "${base_output_file}.md"
#		--structure-property "http://purl.org/oseg/din/ontologies/tsdc/0.1/specification#isBelow" \
done

index_file="$publish_dir/index.html"
if ! [ -e "$index_file" ] || $force
then
{
	cd "$publish_dir"

	# Create HTML that links to all the files generated above
	# plus the sources
	echo "<html>"
	echo "<head>"
	echo "<title>Sources & Generated output</title>"
	echo "</head>"
	echo "<body>"
	echo "<h1>Sources & Generated output</h1>"

	echo "<h3>Turtle/RDF sources</h3>"
	echo "<ul>"
	for file in *.ttl
	do
		echo "<li><a href=\"$file\">$file</a></li>"
	done
	echo "</ul>"

	echo "<h3>Markdown files laying out the structure</h3>"
	echo "<ul>"
	for file in *.md
	do
		echo "<li><a href=\"$file\">$file</a></li>"
	done
	echo "</ul>"

	echo "<h3>GraphViz/DOT graps representing the structure</h3>"
	echo "<ul>"
	for file in *.dot
	do
		echo "<li><a href=\"$file\">$file</a></li>"
	done
	echo "</ul>"

	echo "<h3>PNG & SVG renders of the above graphs</h3>"
	echo "<ul>"
	for file in *.svg
	do
		file_png="${file%.svg}.png"
		echo "<li>"
		echo "<a href=\"$file\">$file - <img src=\"$file\" alt=\"$file\" width=\"640\"/></a> - "
		echo "<a href=\"$file_png\">$file_png - <img src=\"$file_png\" alt=\"$file\" width=\"640\"/></a>"
		echo "</li>"
	done
	echo "</ul>"

	echo "</body>"
	echo "</html>"
} > "$index_file"
else
	>&2 echo "WARNING: Index file not written, because it already exists, and --force is not given."
fi