GH-5215 Support apostrophe in local name for SPARQL 1.1 and Turtle/Trig/etc. (#5216)

Author: hmottestad

core/model/src/main/java/org/eclipse/rdf4j/model/util/URIUtil.java

@@ -228,6 +228,19 @@ public static boolean isValidLocalName(String name) {
 
 		for (int i = 1; i < name.length(); i++) {
 			if (!isNameChar(name.charAt(i))) {
+
+				// PLX
+				if (name.charAt(i) == '%') {
+					continue;
+				} else if (name.charAt(i) == '\\') {
+					if (i + 1 < name.length() && isPN_LOCAL_ESC(name.substring(i, i + 2))) {
+						i++;
+						continue;
+					} else {
+						return false;
+					}
+				}
+
 				return false;
 			}
 
@@ -341,7 +354,7 @@ private static boolean isNameStartChar(int codePoint) {
 	 * @return <code>true</code> if the supplied code point represents a valid name char, <code>false</code> otherwise.
 	 */
 	private static boolean isNameChar(int codePoint) {
-		return isPN_CHARS(codePoint) || codePoint == '.' || codePoint == ':' | codePoint == '\\' || codePoint == '%';
+		return isPN_CHARS(codePoint) || codePoint == '.' || codePoint == ':';
 	}
 
 	/**

core/model/src/test/java/org/eclipse/rdf4j/model/util/URIUtilTest.java

@@ -108,5 +108,6 @@ public void isValidLocalName() {
 		assertFalse(URIUtil.isValidLocalName("foo\tbar"));
 		assertFalse(URIUtil.isValidLocalName("foo\nbar"));
 		assertFalse(URIUtil.isValidLocalName("*foobar"));
+		assertTrue(URIUtil.isValidLocalName("fo\\'obar"));
 	}
 }

core/queryparser/sparql/src/main/java/org/eclipse/rdf4j/query/parser/sparql/ast/SyntaxTreeBuilderTokenManager.java

@@ -2296,7 +2296,7 @@ private int jjMoveNfa_0_curCharLessThan64(int startsAt, int i, int kind) {
 				}
 				break;
 			case 82:
-				if ((0xa800ff7e00000000L & l) != 0L) {
+				if ((0xa800fffa00000000L & l) != 0L) {
 					jjCheckNAddStates(44, 47);
 				}
 				break;
@@ -2326,12 +2326,12 @@ private int jjMoveNfa_0_curCharLessThan64(int startsAt, int i, int kind) {
 				}
 				break;
 			case 88:
-				if ((0xa800ff7e00000000L & l) != 0L && kind > 146) {
+				if ((0xa800fffa00000000L & l) != 0L && kind > 146) {
 					kind = 146;
 				}
 				break;
 			case 90:
-				if ((0xa800ff7e00000000L & l) == 0L) {
+				if ((0xa800fffa00000000L & l) == 0L) {
 					break;
 				}
 				if (kind > 146) {

core/queryparser/sparql/src/main/java/org/eclipse/rdf4j/query/parser/sparql/ast/sparql.jjt

@@ -345,7 +345,7 @@ TOKEN:
 |	<#PN_LOCAL: (<PN_CHARS_U> | ":" | <NUM> | <PLX> ) ( (<PN_CHARS> | "." | ":" | <PLX>)* ( <PN_CHARS> | ":" | <PLX>) )?>
 |	<#PLX: <PERCENT> | <PN_LOCAL_ESC>>
 |	<#PERCENT: "%" <HEX> <HEX>>
-|	<#PN_LOCAL_ESC: "\\" [ "_", "~", ".", "-", "!", "$", "&", "\"", "(", ")", "*", "+", ",", ";", "=", "/", "?", "#", "@", "%" ]>
+|	<#PN_LOCAL_ESC: "\\" [ "_", "~", ".", "-", "!", "$", "&", "'", "(", ")", "*", "+", ",", ";", "=", "/", "?", "#", "@", "%" ]>
 |	<#VARNAME: (<PN_CHARS_U> | <NUM> ) (<VAR_CHAR>)*>
 |	<TRIPLE_OPEN: "<<">
 |	<TRIPLE_CLOSE: ">>">

core/queryparser/sparql/src/test/java/org/eclipse/rdf4j/query/parser/sparql/SPARQLParserTest.java

@@ -976,6 +976,32 @@ public void testGroupByProjectionHandling_function() {
 		verifySerializable(parsedQuery.getTupleExpr());
 	}
 
+	@Test
+	public void testApostrophe() {
+		String query = "PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n" +
+				"select * where { \n" +
+				"  rdf:Test\\'s ?p1 ?o1\n" +
+				"}";
+
+		HashSet<Namespace> customPrefixes = new HashSet<>();
+		SPARQLParser parser = new SPARQLParser(customPrefixes);
+
+		parser.parseQuery(query, null);
+	}
+
+	@Test
+	public void testApostropheInsertData() {
+		String query = "PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n" +
+				"INSERT DATA { \n" +
+				"  rdf:Test\\'s a <http://example.com/Test> .\n" +
+				"}";
+
+		HashSet<Namespace> customPrefixes = new HashSet<>();
+		SPARQLParser parser = new SPARQLParser(customPrefixes);
+
+		parser.parseUpdate(query, null);
+	}
+
 	private AggregateFunctionFactory buildDummyFactory() {
 		return new AggregateFunctionFactory() {
 			@Override

core/rio/trig/src/test/resources/testcases/trig/trig-syntax-minimal-whitespace-01.trig

@@ -19,3 +19,5 @@ d:<http://example/d/>.
 _:s{:s :p :o .:s :p"Alice".:s :p _:o.}
 :{: : :}{: : :}:{: : :}
 :{():()}{():[]}:{[]:[]}
+{b:a\'s a :Foo}
+

core/rio/turtle/src/main/java/org/eclipse/rdf4j/rio/turtle/TurtleUtil.java

@@ -13,6 +13,7 @@
 import java.util.Arrays;
 
 import org.eclipse.rdf4j.common.text.ASCIIUtil;
+import org.eclipse.rdf4j.model.util.URIUtil;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -366,41 +367,7 @@ public static boolean isPN_LOCAL_ESC(String name) {
 	}
 
 	public static boolean isPN_LOCAL(String name) {
-		// Empty names are legal
-		if (name.isEmpty()) {
-			return true;
-		}
-
-		if (!isPN_CHARS_U(name.charAt(0)) && name.charAt(0) != ':' && !ASCIIUtil.isNumber(name.charAt(0))
-				&& !isPLX_START(name)) {
-			logger.debug("PN_LOCAL was not valid (start characters invalid) i=" + 0 + " nextchar="
-					+ name.charAt(0) + " name=" + name);
-			return false;
-		}
-
-		if (!isNameStartChar(name.charAt(0))) {
-			logger.debug("name was not valid (start character invalid) i=" + 0 + " nextchar=" + name.charAt(0)
-					+ " name=" + name);
-			return false;
-		}
-
-		for (int i = 1; i < name.length(); i++) {
-			if (!isNameChar(name.charAt(i))) {
-				logger.debug("name was not valid (intermediate character invalid) i=" + i + " nextchar="
-						+ name.charAt(i) + " name=" + name);
-				return false;
-			}
-
-			// Check if the percent encoding was less than two characters from the
-			// end of the prefix, in which case it is invalid
-			if (name.charAt(i) == '%' && (name.length() - i) < 3) {
-				logger.debug("name was not valid (short percent escape) i=" + i + " nextchar=" + name.charAt(i)
-						+ " name=" + name);
-				return false;
-			}
-		}
-
-		return true;
+		return URIUtil.isValidLocalName(name);
 	}
 
 	// public static boolean isLegalName(String name) {

core/rio/turtle/src/test/java/org/eclipse/rdf4j/rio/turtle/TurtleUtilTest.java

@@ -229,6 +229,7 @@ public final void testIsPN_LOCAL() {
 		assertFalse(TurtleUtil.isPN_LOCAL("foo\tbar"));
 		assertFalse(TurtleUtil.isPN_LOCAL("foo\nbar"));
 		assertFalse(TurtleUtil.isPN_LOCAL("*foobar"));
+		assertTrue(TurtleUtil.isPN_LOCAL("foo\\'bar"));
 	}
 
 	/**

core/rio/turtle/src/test/resources/test-newlines.ttl

@@ -5,7 +5,8 @@
 ex:a a ex:Foo .
 ex:b a ex:Foo .
 ex:b ex:value '''^M'''.
-ex:c 
+ex:c
     :foo ex:Foo .
+rdf:Test\'s a ex:Foo .
 ex:a ex:value 1.0 .
 ex:a ex:value 3.0 .