update_output.sh

#!/bin/bash

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

INTERMEDIATE_DIR="$BASE_DIR/schema/intermediate"
INTERMEDIATE_OUTPUT_0=( "cleaned" "statistics" )
INTERMEDIATE_OUTPUT_1=( "cleaned" "statistics" )
INTERMEDIATE_OUTPUT_2=( "cleaned" "statistics" "rnc" "rng" "schema" "schematron" )

# break script on first error
# set -e
# set -o pipefail

if ! which realpath; then
	if which 'grealpath'; then
		alias realpath=grealpath
	else
		echo "realpath not found (also 'grealpath' is not available)."
		echo "On mac, please install 'coreutils', e.g. using homebrew"
		exit 1
	fi
fi

##########################################
# stage 0: p5subset.xml:
##########################################

# ODD -> * (schemata)
function im_output () {
	echo "intermediate output...:"

	local dir="$1"
	shift
	local local_src="$1"
	shift
	local dump_these_resources=( "$@" )
	echo "dir: $dir"
	echo "local_src: $local_src"
	if [[ " ${dump_these_resources[@]} " =~ " cleaned " ]]; then
		echo "ODD -> illiterate ODD"
		$SCRIPT_DIR/utils/for_each_input.sh \
			--input-dir "$dir/odd_compiled" \
			--output-dir "$dir/generated/odd_cleaned" \
			--print-cmd \
			--log-file "$dir/logs/clean_odd.log" \
			-- \
			$SCRIPT_DIR/saxon.sh \
			"$BASE_DIR/stylesheets/clean_odd.xsl" '-s:$INPUT' '-o:$OUTPUT' \
		| sed 's/^/  /' \
		|| exit 1
	fi

	if [[ " ${dump_these_resources[@]} " =~ " statistics " ]]; then
		echo "illiterate ODD -> statistics"
		$SCRIPT_DIR/utils/for_each_input.sh \
			--input-dir "$dir/generated/odd_cleaned" \
			--output-dir "$dir/generated/statistics" \
			--suffix '.md' \
			--print-cmd \
			--log-file "$dir/logs/statistics.log" \
			-- \
			$SCRIPT_DIR/saxon.sh \
			"$BASE_DIR/stylesheets/statistics.xsl" '-s:$INPUT' '-o:$OUTPUT' \
		| sed 's/^/  /' \
		|| exit 1

		# echo "htmlstatistics.xsl..."
		which pandoc &> /dev/null
		if [[ "$?" != "0" ]]; then
			echo "WARNING: pandoc not found. No HTML statistics generated."
		else
			$SCRIPT_DIR/utils/for_each_input.sh \
				--input-dir "$dir/generated/statistics" \
				--output-dir "$dir/generated/statistics_html" \
				--suffix '.html' \
				--print-cmd \
				--log-file "$dir/logs/statistics_html.log" \
				-- \
				"pandoc" -o '$OUTPUT' '$INPUT' \
			| sed 's/^/  /' \
			|| exit 1
		fi
	fi

	if [[ " ${dump_these_resources[@]} " =~ " rnc " ]]; then
		echo "ODD -> RELAX-NG Compact..."
		local local_src_opt=""
		if [[ "$local_src" != "" ]]; then
			local_src_opt="--localsource='$local_src'"
		fi
		$SCRIPT_DIR/utils/for_each_input.sh \
			--input-dir "$dir/odd" \
			--output-dir "$dir/generated/rnc" \
			--suffix '.rnc' \
			--log-file "$dir/logs/teitornc.log" \
			--print-cmd \
			-- \
			"$DEP_DIR/Stylesheets/bin/teitornc" \
			"$local_src_opt" \
			'$INPUT' '$OUTPUT' \
		| sed 's/^/  /' \
		|| exit 1
	fi

	if [[ " ${dump_these_resources[@]} " =~ " rng " ]]; then
		echo "ODD -> RELAX-NG..."
		local local_src_opt=""
		if [[ "$local_src" != "" ]]; then
			local_src_opt="--localsource='$local_src'"
		fi
		$SCRIPT_DIR/utils/for_each_input.sh \
			--input-dir "$dir/odd" \
			--output-dir "$dir/generated/rng" \
			--suffix '.rng' \
			--log-file "$dir/logs/teitorelaxng.log" \
			--print-cmd \
			-- \
			"$DEP_DIR/Stylesheets/bin/teitorelaxng" \
			"$local_src_opt" \
			'$INPUT' '$OUTPUT' \
		| sed 's/^/  /' \
		|| exit 1
	fi

	if [[ " ${dump_these_resources[@]} " =~ " schema " ]]; then
		echo "ODD -> XML Schema..."
		local local_src_opt=""
		if [[ "$local_src" != "" ]]; then
			local_src_opt="--localsource='$local_src'"
		fi
		$SCRIPT_DIR/utils/for_each_input.sh \
			--input-dir "$dir/odd" \
			--output-dir "$dir/generated/xsd" \
			--suffix '.xsd' \
			--log-file "$dir/logs/teitoxsd.log" \
			--print-cmd \
			-- \
			"$DEP_DIR/Stylesheets/bin/teitoxsd" \
			"$local_src_opt" \
			'$INPUT' '$OUTPUT' \
		| sed 's/^/  /' \
		|| exit 1
	fi

	if [[ " ${dump_these_resources[@]} " =~ " schematron " ]]; then
		echo "ODD -> Schematron"
		local local_src_opt=""
		if [[ "$local_src" != "" ]]; then
			local_src_opt="--localsource='$local_src'"
		fi
		$SCRIPT_DIR/utils/for_each_input.sh \
			--input-dir "$dir/odd" \
			--output-dir "$dir/generated/schematron" \
			--suffix '.xml' \
			--log-file "$dir/logs/teitoschematron.log" \
			--print-cmd \
			-- \
			"$DEP_DIR/Stylesheets/bin/teitoschematron" \
			"$local_src_opt" \
			'$INPUT' '$OUTPUT' \
		| sed 's/^/  /' \
		|| exit 1

		echo "Schematron -> xslt 2"
		$SCRIPT_DIR/utils/for_each_input.sh \
			--input-dir "$dir/generated/schematron" \
			--output-dir "$dir/generated/schematron" \
			--filter '*.xml' \
			--suffix '.xsl' \
			--log-file "$dir/logs/schematrontoxslt2.log" \
			--print-cmd \
			-- \
			$SCRIPT_DIR/saxon.sh \
				"$BASE_DIR/stylesheets/schematron/iso_svrl_for_xslt2.xsl" \
				'-s:$INPUT' '-o:$OUTPUT' \
		| sed 's/^/  /' \
		|| exit 1
	fi
}

function stage0 () {

	echo "##########################################"
	echo "# stage 0: the TEI guidelines"
	echo "##########################################"

	output_dir="$INTERMEDIATE_DIR/s0"
	for x in "${INTERMEDIATE_OUTPUT_0[@]}"; do
		if [[ "$x" != 'cleaned' ]] && [[ "$x" != 'statistics' ]]; then
			echo "ERROR: in stage 0: cannot create intermediate output"
			echo "of type '$x', because we don't have the uncompiled ODD"
			echo "only possible values: cleaned|statistics"
			exit 1
		fi
	done
	im_output "$output_dir" "" "${INTERMEDIATE_OUTPUT_0[@]}"
}

stage0 || exit 1

##########################################

echo "##########################################"
echo "# stage 1: generate ODD from corpus"
echo "##########################################"

mkdir -pv "$INTERMEDIATE_DIR/s1/logs"

echo "create ODD from corpus from example files in '"$(realpath "$BASE_DIR/schema/resource?select=*.xml")"'"

$SCRIPT_DIR/utils/run_with_cache.sh \
	--output-file "$INTERMEDIATE_DIR/s1/odd/tmp.odd" \
	--cache 1 \
	--log-file "$INTERMEDIATE_DIR/s1/logs/oddbyexample.log" \
	--print-cmd \
	-- \
	"$SCRIPT_DIR/saxon.sh" \
		"$DEP_DIR/Stylesheets/tools/oddbyexample.xsl" \
		'-o:$OUTPUT' \
		'-it:main' \
		corpusList=$(realpath "$BASE_DIR/schema/resource?select=*.xml") \
		defaultSource=$(realpath "$BASE_DIR/schema/tei/odd/p5subset.xml") \
		debug=true \
		verbose=true \
| sed 's/^/  /' \
|| exit 1

echo "compile odd..."
$SCRIPT_DIR/utils/for_each_input.sh \
	--input-dir "$INTERMEDIATE_DIR/s1/odd" \
	--output-dir "$INTERMEDIATE_DIR/s1/odd_compiled" \
	--suffix '.xml' \
	--log-file "$INTERMEDIATE_DIR/s1/logs/teitoodd.log" \
	--print-cmd \
	-- \
	"$DEP_DIR/Stylesheets/bin/teitoodd" \
	--localsource="$INTERMEDIATE_DIR/s0/odd_compiled/p5subset.xml" \
	'$INPUT' '$OUTPUT' \
| sed 's/^/  /' \
|| exit 1

im_output "$INTERMEDIATE_DIR/s1" "$INTERMEDIATE_DIR/s0/odd_compiled/p5subset.xml" "${INTERMEDIATE_OUTPUT_1[@]}"


##########################################

echo "##########################################"
echo "# stage 2: autogenerate ODD from RELAX-NG schema"
echo "##########################################"

echo "prepare: rnc -> analyis..."
mkdir -p $INTERMEDIATE_DIR/s2/logs

$SCRIPT_DIR/rnc_to_analysis.sh \
	--input "$BASE_DIR/schema/resource/eoa-tei-strict.rnc" \
	--output-dir "$INTERMEDIATE_DIR/s2/prepare/rnc_analysis" \
	--verbosity 3 \
	>$INTERMEDIATE_DIR/s2/logs/rnc_to_analysis.log \
	2>&1 \
	|| exit 1

echo "analyis -> odd..."
$SCRIPT_DIR/rnc_analysis_to_odd.sh \
	--verbosity 3 \
	--input-dir "$INTERMEDIATE_DIR/s2/prepare/rnc_analysis" \
	--output-dir "$INTERMEDIATE_DIR/s2/odd" \
	--modules 'core,tei,header,namesdates,figures,transcr,linking,textstructure' \
	>$INTERMEDIATE_DIR/s2/logs/rnc_analysis_to_odd.log \
	2>&1 \
	|| exit 1

echo "compile odd..."
$SCRIPT_DIR/utils/for_each_input.sh \
	--input-dir "$INTERMEDIATE_DIR/s2/odd" \
	--output-dir "$INTERMEDIATE_DIR/s2/odd_compiled" \
	--suffix '.xml' \
	--log-file "$INTERMEDIATE_DIR/s2/logs/teitoodd.log" \
	--print-cmd \
	-- \
	"$DEP_DIR/Stylesheets/bin/teitoodd" \
	--localsource="$INTERMEDIATE_DIR/s1/odd_compiled/tmp.xml" \
	'$INPUT' '$OUTPUT' \
	| sed 's/^/  /' \
	|| exit 1

im_output "$INTERMEDIATE_DIR/s2" "$INTERMEDIATE_DIR/s1/odd_compiled/tmp.xml" "${INTERMEDIATE_OUTPUT_2[@]}"

echo "done"