apacheGH-2887: Tolerate non-NFC IRIs (IRIProviderJenaIRI)

afs · Dec 9, 2024 · 35cacd0 · 35cacd0
1 parent 3d34ac9
commit 35cacd0
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 8 deletions.
diff --git a/jena-core/src/main/java/org/apache/jena/irix/IRIProviderJenaIRI.java b/jena-core/src/main/java/org/apache/jena/irix/IRIProviderJenaIRI.java
@@ -249,23 +249,24 @@ private static IRI exceptions(IRI iri, String iriStr) {
             // Global settings below; this section is for conditional filtering.
             // See also Checker.iriViolations for WARN filtering.
             switch(code) {
-                case Violation.PROHIBITED_COMPONENT_PRESENT:
+                case Violation.PROHIBITED_COMPONENT_PRESENT->{
                     // Allow "u:p@" when non-strict.
                     // Jena3 compatibility.
                     if ( isHTTP(iri) && ! STRICT_HTTP && v.getComponent() == IRIComponents.USER )
                         continue;
-                    break;
-                case Violation.SCHEME_PATTERN_MATCH_FAILED:
+                }
+                case Violation.SCHEME_PATTERN_MATCH_FAILED->{
                     if ( isURN(iri) && ! STRICT_URN )
                         continue;
                     if ( isFILE(iri) )
                         continue;
-                    break;
-                case Violation.REQUIRED_COMPONENT_MISSING:
+                }
+                case Violation.REQUIRED_COMPONENT_MISSING->{
                     // jena-iri handling of "file:" URIs is only for (an interpretation of) RFC 1738.
                     // RFC8089 allows relative file URIs and a wider use of characters.
                     if ( isFILE(iri) )
                         continue;
+                }
             }
             // Signal first error.
             String msg = v.getShortMessage();

diff --git a/jena-core/src/main/java/org/apache/jena/irix/SetupJenaIRI.java b/jena-core/src/main/java/org/apache/jena/irix/SetupJenaIRI.java
@@ -99,9 +99,11 @@ public static IRIFactory iriCheckerFactory() {
         setErrorWarning(iriCheckerFactory, ViolationCodes.NON_INITIAL_DOT_SEGMENT, false, false);
 
         // == Character related.
-        //setErrorWarning(iriFactoryInst, ViolationCodes.NOT_NFC,  false, false);
+        // Causes confusion! And this is only advice in RDF Concepts.
+        setErrorWarning(iriCheckerFactory, ViolationCodes.NOT_NFC,  false, false);
         // NFKC is not mentioned in RDF 1.1. Switch off.
         setErrorWarning(iriCheckerFactory, ViolationCodes.NOT_NFKC, false, false);
+        // The MAYBE ViolationCodes are never generated.
 
         // ** Applies to various unicode blocks.
 

diff --git a/jena-core/src/test/java/org/apache/jena/irix/TestIRIxJenaSystem.java b/jena-core/src/test/java/org/apache/jena/irix/TestIRIxJenaSystem.java
@@ -33,8 +33,8 @@
 
 /**
  * Test of parsing and schema violations.
- * <p>s
- * This is the test suite that compares result with jena-iri.
+ * <p>
+ * This is the test suite that compares results with jena-iri.
  * See also {@link TestIRIxSyntaxRFC3986} for RDF 3986 syntax only parsing.
  */
 @FixMethodOrder(MethodSorters.NAME_ASCENDING)
@@ -90,6 +90,9 @@ public TestIRIxJenaSystem(String name, IRIProvider provider) {
 
     @Test public void parse_18() { good("/z/a:b"); }
 
+    // This character is from a report on users@jena.
+    @Test public void parse_nfc() { good("http://host/ή"); }
+
     // ---- bad
 
     // Leading ':'