rnc_to_analysis.sh

#!/bin/bash

BASE_DIR="$(dirname $0)/.."
SCRIPT_DIR="$(dirname $0)"
DEP_DIR="$BASE_DIR/dependencies"

TRANG="java -jar $DEP_DIR/Stylesheets/lib/trang.jar"


#######################################
# variables
#######################################

input="$BASE_DIR/schema/resource/eoa-tei-strict.rnc"
output_dir="$BASE_DIR/schema/legacy/generated/analysis"
verbosity="2"
debug_modules=()

#######################################
# functions
#######################################

function print_help {
	echo "analyze a schema in relaxng compact syntax"
	echo
	echo "usage: $0 [OPTIONS]"
	echo
	echo "OPTIONS:"
	echo "  -h | --help: print this help"
	echo "  -i|--input INPUT. default: '$input'"
	echo "  -o|--output-dir OUTPUT_DIR. default: '$output_dir'"
	echo "  -v|--verbosity (0: only errors, 1: warnings, 2: info, 3: debug). default: $verbosity"
	echo "  -d|--debug-module VAL (multiple options accumulate): send debug messages for the specific module"
}

#######################################
# parse command line arguments
#######################################

while [[ $# > 0 ]]; do
	key="$1"
	case $key in
		-h | --help)
			print_help
			exit 0
			;;
		-i | --input)
			shift
			input="$1"
			shift
			;;
		-o | --output-dir)
			shift
			output_dir="$1"
			shift
			;;
		-v | --verbosity)
			shift
			verbosity="$1"
			shift
			;;
		-d | --debug-module)
			shift
			debug_modules+=("$1")
			shift
			;;
		-* )
			echo "wrong syntax!"
			print_help
			exit 1
			;;
		*)
			break
			;;
	esac
done

set -e

mkdir -p "$output_dir"

input_basename=$(basename $input)
rng_file=$output_dir/0_input.rng


echo "rnc -> rng: '$input' -> '$rng_file'"

$SCRIPT_DIR/utils/run_with_cache.sh \
	--output-file "$rng_file" \
	--input-file "$input" \
	--cache 1 \
	--print-cmd \
	-- \
	"$TRANG" '$INPUT' '$OUTPUT' \
|| exit 1

rng_simplified=${rng_file%.*}_simplified7-22.rng
rng_simplified_arg=${rng_file%.*}_simplified
echo "rng -> simplified rng: '$rng_file' -> '$rng_simplified'"

$SCRIPT_DIR/utils/run_with_cache.sh \
	--output-file "$rng_simplified" \
	--input-file "$rng_file" \
	--cache 1 \
	--print-cmd \
	-- \
	xsltproc \
	--stringparam out-name "$rng_simplified_arg" \
	stylesheets/rnc_simplification.xsl \
	'$INPUT' \
|| exit 1

echo "simplified rng -> analysis: '$rng_simplified' -> '$output_dir/{2_flattened.xml,3_serialized.txt}'"

debug_modules_options="debug_modules=${debug_modules[@]}"

# echo "debug_modules_options: $debug_modules_options"

$SCRIPT_DIR/utils/run_with_cache.sh \
	--output-file "$output_dir/5_structured_advanced.xml" \
	--input-file "$rng_simplified" \
	--cache 1 \
	--print-cmd \
	-- \
	$SCRIPT_DIR/saxon.sh \
	"$BASE_DIR/stylesheets/rng_to_odd/rng_to_analysis.xsl" '-s:$INPUT' '-o:$OUTPUT' \
	"output=." \
	"verbosity=$verbosity" \
	"$debug_modules_options" \
|| exit 1

echo "SOURCE: $input" > $output_dir/info.cfg