Merge pull request #9 from kbss-cvut/add-docker-compose-for-internal-…

…auth

[New] Add example docker-compose setup for internal authentication

deploy/internal-auth/.env

@@ -0,0 +1,8 @@
+# In this setting it is assumed there is no apapche/nginx in front of the application
+RECORD_SET_NAME=iauth-example
+PORT=1235
+RECORD_MANAGER_SERVER_URL=http://localhost:1235/record-manager/record-manager-server
+FORMGEN_SERVICE_URL=http://s-pipes-engine:8080/s-pipes/service?_pId=clone-form
+RECORD_MANAGER_APP_TITLE=Record manager
+RECORD_MANAGER_BASENAME=/record-manager
+LANGUAGE=en

deploy/internal-auth/db-server/Dockerfile

@@ -0,0 +1,22 @@
+FROM ontotext/graphdb:10.2.0
+
+# Override parent entrypoint
+ENTRYPOINT []
+
+ENV GRAPHDB_HOME=/opt/graphdb/home
+ENV GRAPHDB_INSTALL_DIR=/opt/graphdb/dist
+
+WORKDIR ${GRAPHDB_HOME}
+
+# Install libs related to RDF processing
+RUN apk add py3-rdflib
+RUN apk add perl-uri
+
+# Copy initialization data and repository config
+COPY init-data /root/graphdb-import
+COPY init-config /repo-config
+COPY bin/* ${GRAPHDB_INSTALL_DIR}/bin/
+
+EXPOSE 7200
+
+CMD ${GRAPHDB_INSTALL_DIR}/bin/repo-init.sh /repo-config ${GRAPHDB_HOME} & ${GRAPHDB_INSTALL_DIR}/bin/graphdb -Dgraphdb.home=${GRAPHDB_HOME} -Dgraphdb.logback=${GRAPHDB_INSTALL_DIR}/conf/logback.xml

deploy/internal-auth/db-server/bin/get-rdf-subject-by-type.py

@@ -0,0 +1,75 @@
+#!/usr/bin/python3
+
+import sys
+from rdflib import Graph, URIRef
+
+def log(message):
+    print("ERROR: " + message, file=sys.stderr)
+
+def check_params():
+    if len(sys.argv) != 3:
+        log(f"""Illegal number of parameters.
+
+Script returns single subject of triple matching the pattern '?result a <type-uri>' 
+from the file specified by <rdf-file-path>.
+
+Usage: {sys.argv[0]} <rdf-file-path> <type-uri>
+
+Example: {sys.argv[0]} "./init-data/forms/example-1.ttl" "http://onto.fel.cvut.cz/ontologies/form/form-template"
+""")
+        sys.exit(1)
+
+def check_only_one_instance(results, rdf_type):
+    if len(results) == 0:
+        log(f"No instance found for the specified {rdf_type}.")
+        sys.exit(2)
+    elif len(results) > 1:
+        error_message = f"Multiple instances found for the type {rdf_type}. Triple that match pattern '?s a <{rdf_type}>' are:\n"
+        for row in results:
+            subject = row[0]
+            error_message += f"  {subject.n3()} a <{rdf_type}> .\n"
+        log(error_message)
+        sys.exit(3)
+
+
+def load_rdf_graph(file_path):
+    # Load RDF file into an RDFLib graph
+    g = Graph()
+
+    # Explicitly specify the format based on the file extension
+    if file_path.endswith(".ttl"):
+        g.parse(file_path, format="turtle")
+    elif file_path.endswith(".rdf"):
+        g.parse(file_path, format="xml")
+    else:
+        log(f"Unsupported RDF file format of {file_path}.")
+        sys.exit(1)
+
+    return g
+
+def main():
+    check_params()
+
+    file_path = sys.argv[1]
+    rdf_type = URIRef(sys.argv[2])
+
+    g = load_rdf_graph(file_path)
+
+    # Query for subjects with the specified RDF type
+    query = f"""
+        SELECT ?subject
+        WHERE {{
+            ?subject a <{rdf_type}>.
+        }}
+    """
+
+    results = g.query(query)
+
+    check_only_one_instance(results, rdf_type)
+
+    for row in results:
+        subject = row[0]
+        print(subject.n3())
+
+if __name__ == "__main__":
+    main()

deploy/internal-auth/db-server/bin/get-value-of-rdf-property.py

@@ -0,0 +1,73 @@
+#!/usr/bin/python3
+
+import sys       
+from rdflib import Graph, URIRef
+
+def log(message):
+    print("ERROR: " + message, file=sys.stderr)
+
+def check_params():
+    if len(sys.argv) != 3:
+        log(f"""Illegal number of parameters.
+
+Script returns single value of <rdf-property-uri> from file specified by <rdf-file-path>.
+
+Usage: {sys.argv[0]} <rdf-file-path> <rdf-property-uri>
+
+Example: {sys.argv[0]} "./init-config/repo-config.ttl" "http://www.openrdf.org/config/repository#repositoryID"
+""")
+        sys.exit(1)
+
+
+def check_property_has_single_value(results, rdf_property):
+    if len(results) == 0:
+        log(f"No values found for the specified property {rdf_property}.")
+        sys.exit(2)
+    elif len(results) > 1:
+        error_message = f"Multiple values found for the property {rdf_property}. Triple that match pattern '?s <{rdf_property}> ?o' are:\n"
+        for row in results:
+            subject, value = row
+            error_message += f"  {subject} {rdf_property} {value} .\n"
+        log(error_message)
+        sys.exit(3)
+
+def load_rdf_graph(file_path):
+    # Load RDF file into an RDFLib graph
+    g = Graph()
+
+    # Explicitly specify the format based on the file extension
+    if file_path.endswith(".ttl"):
+        g.parse(file_path, format="turtle")
+    elif file_path.endswith(".rdf"):
+        g.parse(file_path, format="xml")
+    else:
+        log(f"Unsupported RDF file format of {file_path}.")
+        sys.exit(1)
+    return g
+
+def main():
+    check_params()
+
+    file_path = sys.argv[1]
+    rdf_property = URIRef(sys.argv[2])
+
+    g = load_rdf_graph(file_path)
+
+    # Query for subjects with the specified property
+    query = f"""
+        SELECT ?subject ?value                      
+        WHERE {{
+            ?subject <{rdf_property}> ?value.
+        }}      
+    """                                          
+    results = g.query(query)
+
+    check_property_has_single_value(results, rdf_property)
+
+    for row in results:
+        subject, value = row
+        print(f"{value}")
+
+if __name__ == "__main__":                                  
+    main()
+

deploy/internal-auth/db-server/bin/rdf4j-deploy-context.sh

@@ -0,0 +1,99 @@
+#!/bin/bash
+
+
+### DEFAULTS ###
+APPEND=true
+CONTENT_TYPE='text/turtle'
+#CONTENT_TYPE='application/trig'
+#CONTENT_TYPE='application/rdf+xml'
+#CONTENT_TYPE='text/x-nquads'
+#CONTENT_TYPE='application/ld+json'
+
+function print_usage() {
+	  echo "Rdf4j server deploy script."
+          echo "Parameters: "
+          echo -e "\t-R -- replace content of GRAPH_IRI (appends by default)"
+          echo -e "\t-C CONTENT_TYPE -- content type of input FILES, i.e. text/turtle (default), application/rdf+xml ..."
+          echo "Usage: "
+          echo -e "\t$0 -R -C <CONTENT_TYPE> -s <RDF4J_SERVER> -r <REPO_ID> -c <GRAPH_IRI> <FILES>"
+          echo -e "\t$0 -R -C <CONTENT_TYPE> -u <RDF4J_REPOSITORY_URL> -c <GRAPH_IRI> <FILES>"
+          echo "Examples: "
+          echo -e "\tEXAMPLE-1 (append context): $0 -s http://onto.mondis.cz/openrdf-RDF4J -r ontomind_owlim -c http://onto.mondis.cz/resource/mdr-1.0-SNAPSHOT-temporary mdr.owl" 
+          echo -e "\tEXAMPLE-2 (replace context): $0 -R -C 'text/turtle' -s http://onto.fel.cvut.cz/rdf4j-server -r test -c http://vfn.cz/ontologies/fertility-saving-study study-model.ttl"
+	  echo -e "\tEXAMPLE-3 (replace repository): $0 -R -C 'text/x-nquads' -s http://onto.fel.cvut.cz/rdf4j-server -r test fss-study-formgen.ng"
+          echo -e "\tEXAMPLE-4 (use of repository url): $0 -R -C 'text/turtle' -u http://onto.fel.cvut.cz/rdf4j-server/repositories/test -c http://vfn.cz/ontologies/fertility-saving-study study-model.ttl"
+}
+
+if [ "$#" -eq 0 ]; then
+	print_usage
+        exit
+fi
+
+
+while getopts "h:s:r:u:c:RC:" arg; do
+      case $arg in
+        h)
+	  print_usage
+          exit 0
+          ;;
+        s)
+          RDF4J_SERVER=$OPTARG
+          ;;
+        r)
+          REPOSITORY=$OPTARG
+          ;;
+        u)
+          REPOSITORY_URL=$OPTARG
+          ;;
+        c)
+          GRAPH=$(perl -MURI::Escape -e 'print uri_escape($ARGV[0]);' "<$OPTARG>")
+          ;;
+	R)
+	  APPEND=false 
+	  ;;
+	C)
+	  CONTENT_TYPE=$OPTARG
+	  ;;
+      esac
+done
+
+shift $(($OPTIND-1))
+FILES=$@
+
+REPOSITORY_URL=${REPOSITORY_URL:-$RDF4J_SERVER/repositories/$REPOSITORY}
+
+echo "INFO: *** DEPLOYING ***"
+[ ! -z "$REPOSITORY_URL" ] && echo "INFO:    - repository url: $REPOSITORY_URL"
+[ ! -z "$RDF4J_SERVER" ] && echo "INFO:  destination server:  $RDF4J_SERVER"
+[ ! -z "$REPOSITORY" ] && echo "INFO:    - repository $REPOSITORY"
+echo "INFO:    - graph $GRAPH"
+echo "INFO:    - files $FILES"
+
+if [ "$APPEND" = false ] ; then
+   echo "INFO: *** CLEARING THE GRAPH ***"
+   TEMP_FILE=`mktemp`
+   QUERY_PARAMS="context=$GRAPH"
+   if [ ! "$GRAPH" ]; then QUERY_PARAMS= ;  fi
+   curl $REPOSITORY_URL/statements?$QUERY_PARAMS -v -X DELETE &> $TEMP_FILE
+   cat $TEMP_FILE | grep "HTTP/1.1 204" &>/dev/null && echo 'INFO:  clearing graph was sucessfull'  
+   cat $TEMP_FILE | grep "HTTP/1.1 204" &>/dev/null || ( echo 'ERROR:  clearing graph failed. Output of the process : '; cat $TEMP_FILE )
+fi 
+
+echo "INFO: *** SENDING DATA ***"
+for FILE in $FILES
+do
+   echo INFO: " -- sending $FILE";
+   TEMP_FILE=`mktemp`
+   QUERY_PARAMS="context=$GRAPH"
+   if [ ! "$GRAPH" ]; then QUERY_PARAMS= ;  fi
+
+   curl -X POST -H "Content-Type: $CONTENT_TYPE" --data-binary "@$FILE" -o - -v "$REPOSITORY_URL/statements?$QUERY_PARAMS" 2> $TEMP_FILE
+   cat $TEMP_FILE | grep "HTTP/1.1 204" &>/dev/null && echo 'INFO:  sending data was sucessfull'  
+   cat $TEMP_FILE | grep "HTTP/1.1 204" &>/dev/null || ( echo 'ERROR:  sending data failed. Output of the process : '; cat $TEMP_FILE )
+done;
+
+echo "INFO: *** CHECKING ***"
+TEMP_FILE=`mktemp`
+GRAPH_SIZE=`curl $REPOSITORY_URL/size 2> $TEMP_FILE`
+echo "INFO:  size of the new graph is $GRAPH_SIZE"
+

deploy/internal-auth/db-server/bin/repo-init.sh

@@ -0,0 +1,48 @@
+#!/bin/sh
+
+#
+# Initializes Record Manager GraphDB repositories if it does not already exist
+#
+
+SOURCE_DIR=$1
+GRAPHDB_HOME=$2
+
+SCRIPT_DIR="`dirname $0`"
+
+echo "INFO: Running initializer for Record Manager repositories ..."
+
+# Wait for GraphDB to start up
+echo "INFO: Waiting for GraphDB to start up..."
+sleep 15s
+
+ls ${SOURCE_DIR}/*-config.ttl | while read REPO_CONFIG_FILE; do 
+
+	REPO_NAME=`$SCRIPT_DIR/get-value-of-rdf-property.py $REPO_CONFIG_FILE 'http://www.openrdf.org/config/repository#repositoryID'`
+
+	if [ -z "$REPO_NAME" ]; then
+    		echo "ERROR: Could not parse repository name from file $REPO_CONFIG_FILE" 
+    		exit 1
+	fi
+
+	if [ ! -d ${GRAPHDB_HOME}/data/repositories/${REPO_NAME} ] || [ -z "$(ls -A ${GRAPHDB_HOME})/data/repositories/${REPO_NAME}" ]; then
+		echo "INFO: Initializing repository $REPO_NAME..."
+
+	    # Create repository based on configuration
+	    echo "INFO: Creating repository $REPO_NAME..."
+	    curl -X POST --header "Content-Type: multipart/form-data" -F "config=@${REPO_CONFIG_FILE}" "http://localhost:7200/rest/repositories"
+	    echo "INFO: Repository $REPO_NAME successfully initialized."
+	else
+	    echo "INFO: Repository $REPO_NAME already exists. Skipping initialization..."
+	fi
+done
+
+
+DATA_DIR=/root/graphdb-import
+
+ls ${DATA_DIR}/forms/*.ttl | while read DATA_FILE; do
+	REPO_NAME="record-manager-formgen"
+	CONTEXT=`$SCRIPT_DIR/get-rdf-subject-by-type.py $DATA_FILE 'http://onto.fel.cvut.cz/ontologies/form/form-template' | sed 's/[<>]//g'`
+
+	echo "INFO: Deploying form templates ${DATA_FILE} into ${CONTEXT}."
+	$SCRIPT_DIR/rdf4j-deploy-context.sh -R -C 'text/turtle' -s http://localhost:7200 -r ${REPO_NAME} -c ${CONTEXT} ${DATA_FILE}
+done