Add or modify Makefile recipes to enable regeneration of *.rdf files …

…and retain expected *.ttl files content `rdfpipe` utility is now used to regenerate *.rdf files what results in the duplicated imports being discarded. Changes were made in Makefile file and affects generation of *.rdf artefacts (owl-core, owl-restrictions and shacl Makefile recipes) as well as conversion of these to TTL format (convert-rdf-to-turtle). The latter were needed to keep the same TTL output when using the new *.rdf files. Scope of the changes: * New generic convert-between-serialization-formats Makefile recipe * The existing owl-core, owl-restrictions and shacl Makefile recipes now include an RDF regeneration step * The existing convert-rdf-to-turtle Makefile recipe now uses the same namespaces that are used for XSLT transformation (test/ePO-default-config/namespaces.xml) to provide rdfpipe with prefixes so the output makes use of compact URIs. * As the reported issue occurs for EPO v4.2.0 which introduces some new prefixes, these prefixes have also been added to test/ePO-default-config/namespaces.xml so it would be possible to reproduce the fix.
OP-TED · Sep 18, 2024 · abde78b · abde78b
1 parent ca63046
commit abde78b
Show file tree

Hide file tree

Showing 3 changed files with 89 additions and 7 deletions.
diff --git a/Makefile b/Makefile
@@ -33,6 +33,9 @@ TURTLE_FILELIST=$(shell ls ${ONTOLOGY_FOLDER_PATH}/*.ttl)
 # Widoco variables
 WIDOCO_RDF_INPUT_FILE_PATH?=test/reasoning-investigation/model-2020-12-16/ePO_restrictions.rdf
 WIDOCO_OUTPUT_FOLDER_PATH?=output/widoco
+NAMESPACES_AS_RDFPIPE_ARGS=$(shell ${MODEL2OWL_FOLDER}/scripts/get_namespaces.sh)
+RDF_XML_MIME_TYPE:='application/rdf+xml'
+TURTLE_MIME_TYPE:='turtle'
 
 # download saxon library 	
 get-saxon:
@@ -110,17 +113,37 @@ generate-convention-SVRL-report:
 #Example how to run transformation commands :
 # make owl-core XMI_INPUT_FILE_PATH=/home/mypc/work/model2owl/eNotice_CM.xml OUTPUT_FOLDER_PATH=./my-folder
 owl-core:
-	@java -jar ${SAXON} -s:${XMI_INPUT_FILE_PATH} -xsl:${MODEL2OWL_FOLDER}/src/owl-core.xsl -o:${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}.rdf
+	@java -jar ${SAXON} -s:${XMI_INPUT_FILE_PATH} -xsl:${MODEL2OWL_FOLDER}/src/owl-core.xsl \
+		-o:${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}.tmp.rdf
+	@make convert-between-serialization-formats INPUT_FORMAT=${RDF_XML_MIME_TYPE} \
+		OUTPUT_FORMAT=${RDF_XML_MIME_TYPE} \
+		FILE_PATH=${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}.tmp.rdf \
+		OUTPUT_FILE_PATH=${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}.rdf
 	@echo Output owl core file:
 	@ls -lh ${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}.rdf
+	@rm -f ${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}.tmp.rdf
+
 owl-restrictions:
-	@java -jar ${SAXON} -s:${XMI_INPUT_FILE_PATH} -xsl:${MODEL2OWL_FOLDER}/src/owl-restrictions.xsl -o:${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_restrictions.rdf
+	@java -jar ${SAXON} -s:${XMI_INPUT_FILE_PATH} -xsl:${MODEL2OWL_FOLDER}/src/owl-restrictions.xsl \
+		-o:${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_restrictions.tmp.rdf
+	@make convert-between-serialization-formats INPUT_FORMAT=${RDF_XML_MIME_TYPE} \
+		OUTPUT_FORMAT=${RDF_XML_MIME_TYPE} \
+		FILE_PATH=${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_restrictions.tmp.rdf \
+		OUTPUT_FILE_PATH=${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_restrictions.rdf
 	@echo Output owl restrictions file:
 	@ls -lh ${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_restrictions.rdf
+	@rm -f ${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_restrictions.tmp.rdf
+
 shacl:
-	@java -jar ${SAXON} -s:${XMI_INPUT_FILE_PATH} -xsl:${MODEL2OWL_FOLDER}/src/shacl-shapes.xsl -o:${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_shapes.rdf
+	@java -jar ${SAXON} -s:${XMI_INPUT_FILE_PATH} -xsl:${MODEL2OWL_FOLDER}/src/shacl-shapes.xsl \
+		-o:${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_shapes.tmp.rdf
+	@make convert-between-serialization-formats INPUT_FORMAT=${RDF_XML_MIME_TYPE} \
+		OUTPUT_FORMAT=${RDF_XML_MIME_TYPE} \
+		FILE_PATH=${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_shapes.tmp.rdf \
+		OUTPUT_FILE_PATH=${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_shapes.rdf
 	@echo Output shacl file location:
 	@ls -lh ${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_shapes.rdf
+	@rm -f ${OUTPUT_FOLDER_PATH}/${XMI_INPUT_FILENAME_WITHOUT_EXTENSION}_shapes.tmp.rdf
 
 
 # Combine xmi UML files
@@ -144,7 +167,12 @@ convert-rdf-to-turtle:
 	@for FILE_PATH in ${RDF_FILELIST}; do \
 		echo Converting $${FILE_PATH} into Turtle; \
 		source model2owl-venv/bin/activate; \
-		rdfpipe -i application/rdf+xml -o  turtle $${FILE_PATH} > $${FILE_PATH%.*}.ttl; \
+		make convert-between-serialization-formats \
+			INPUT_FORMAT=${RDF_XML_MIME_TYPE} \
+			OUTPUT_FORMAT=${TURTLE_MIME_TYPE}  \
+			FILE_PATH=$${FILE_PATH}  \
+			OUTPUT_FILE_PATH=$${FILE_PATH%.*}.ttl \
+			USE_NAMESPACES=1; \
 		echo Input in RDF/XML format;  \
 		echo $${FILE_PATH};  \
 		echo " ==> Output in Turtle format";  \
@@ -182,6 +210,35 @@ convert-rdf-to-rdf:
 		echo " ==> Output in RDF/XML format";  \
 		ls -lh $${FILE_PATH%.*}.rdf;  \
 	done
+
+# A generic recipe for converting RDF data from one serialization format to 
+# another. It can also be used to regenerate a file using the same format.
+# 
+# Arguments:
+# 	FILE_PATH: Input RDF file in any allowed serialization format
+# 	OUTPUT_FILE_PATH: Path for the output file
+# 	INPUT_FORMAT: a MIME type of the given input RDF file
+# 	OUTPUT_FORMAT: a MIME type of any of the valid RDF serializations
+# 	USE_NAMESPACES: optional; if non-empty then namespaces (from the namespaces.xml file).
+#					This can be used if the input (FILE_PATH) doesn't include
+# 					namespaces we want to be applied (e.g. to have compact
+# 					instead of full URIs in the output file).
+#					
+# Supported MIME types: https://rdflib.readthedocs.io/en/7.0.0/plugin_serializers.html
+# 	
+# Example:
+# make convert-between-serialization-formats
+#   INPUT_FORMAT='application/rdf+xml'
+#   OUTPUT_FORMAT='application/rdf+xml' 
+#   FILE_PATH=output/ePO_core.tmp.rdf 
+#   OUTPUT_FILE_PATH=output/ePO_core.rdf
+# 	USE_NAMESPACES=1
+convert-between-serialization-formats:
+	@source model2owl-venv/bin/activate; \
+	rdfpipe -i ${INPUT_FORMAT} -o ${OUTPUT_FORMAT} \
+		$(if $(USE_NAMESPACES),${NAMESPACES_AS_RDFPIPE_ARGS}) \
+		${FILE_PATH} > ${OUTPUT_FILE_PATH}
+
 # make validate-rdf-file FILE_TO_VALIDATE_PATH=./output/eFulfilment.rdf
 validate-rdf-file:
 	@$(JENA_RIOT_TOOL) --validate $(FILE_TO_VALIDATE_PATH)

diff --git a/scripts/get_namespaces.sh b/scripts/get_namespaces.sh
@@ -0,0 +1,20 @@
+#!/bin/bash
+# 
+# Gets namespaces from namespaces.xml file and prepares argument
+# list from them to be used with `rdfpipe` tool.
+# Uses Saxon installed in the project main directory.
+
+PROJECT_DIR=$(dirname $(dirname $(realpath ${BASH_SOURCE[0]})))
+NAMESPACES_DIR=${PROJECT_DIR}/test/ePO-default-config
+
+cd ${NAMESPACES_DIR}
+namespaces=$(
+    java -cp ../../saxon/saxon.jar net.sf.saxon.Query -s:namespaces.xml \
+		-qs:'for $x in /*:prefixes/*:prefix return concat(string($x/@name), "=", string($x/@value))' \
+		\!method=text
+)
+ns_args=$( \
+	echo "$namespaces" | tr ' ' '\n' | awk '{printf("--ns='\''%s'\'' ", $0)}'
+)
+
+echo "$ns_args"
diff --git a/test/ePO-default-config/namespaces.xml b/test/ePO-default-config/namespaces.xml
@@ -12,26 +12,31 @@
     <prefix name="dct" value="http://purl.org/dc/terms/" importURI="http://purl.org/dc/terms/"/>
     <prefix name="org" value="http://www.w3.org/ns/org#"/>
     <prefix name="skos" value="http://www.w3.org/2004/02/skos/core#" importURI="http://www.w3.org/2004/02/skos/core#"/>
+    <prefix name="eli" value="http://data.europa.eu/eli/ontology#"/>
 
     <prefix name="epo" value="http://data.europa.eu/a4g/ontology#"/>
+    <prefix name="epo-acc" value="http://data.europa.eu/a4g/ontology#"/>
     <prefix name="epo-cat" value="http://data.europa.eu/a4g/ontology#"/>
     <prefix name="epo-con" value="http://data.europa.eu/a4g/ontology#"/>
-    <prefix name="epo-ord" value="http://data.europa.eu/a4g/ontology#"/>
-    <prefix name="epo-not" value="http://data.europa.eu/a4g/ontology#"/>
     <prefix name="epo-ful" value="http://data.europa.eu/a4g/ontology#"/>
+    <prefix name="epo-inv" value="http://data.europa.eu/a4g/ontology#"/>
+    <prefix name="epo-not" value="http://data.europa.eu/a4g/ontology#"/>
+    <prefix name="epo-ord" value="http://data.europa.eu/a4g/ontology#"/>
+    <prefix name="epo-sub" value="http://data.europa.eu/a4g/ontology#"/>
+    <prefix name="core-shape" value="http://data.europa.eu/a4g/data-shape#"/>
 
     <prefix name="nuts" value="http://data.europa.eu/nuts/"/>
     <prefix name="reg2015" value="http://data.europa.eu/a4g/extension/ontology#"/>
     <prefix name="espd" value="https://github.com/ESPD/ESPD-EDM/"/>
     <prefix name="at-voc" value="http://publications.europa.eu/resource/authority/"/>
     <prefix name="time" value="http://www.w3.org/2006/time#"/>
     <prefix name="locn" value="http://www.w3.org/ns/locn#"/>
+    <prefix name="cccev" value="http://data.europa.eu/m8g/" importURI="https://data.europa.eu/m8g"/>
     <prefix name="cv" value="http://data.europa.eu/m8g/" importURI="https://data.europa.eu/m8g"/>
     <prefix name="core" value="http://data.europa.eu/m8g/"/>
     <prefix name="cpv" value="http://data.europa.eu/m8g/"/>
     <prefix name="cpsv" value="http://data.europa.eu/m8g/"/>
     <prefix name="cpov" value="http://data.europa.eu/m8g/"/>
-    <prefix name="cccev" value="http://data.europa.eu/m8g/" importURI="https://data.europa.eu/m8g"/>
     <prefix name="geosparql" value="http://www.opengis.net/ont/geosparql#"/>
     <prefix name="dul" value="http://www.loa-cnr.it/ontologies/DUL.owl#"/>    
     <prefix name="person" value="http://www.w3.org/ns/person#"/>