apacheGH-2060: Fix for IRI isAbsolute (RFC 3986 vs RFC 2396)

AKSW · Nov 2, 2023 · c47e4c2 · c47e4c2
1 parent efb85bd
commit c47e4c2
Show file tree

Hide file tree

Showing 8 changed files with 212 additions and 206 deletions.
diff --git a/jena-core/src/main/java/org/apache/jena/ext/xerces/util/URI.java b/jena-core/src/main/java/org/apache/jena/ext/xerces/util/URI.java
diff --git a/jena-core/src/main/java/org/apache/jena/irix/IRIx.java b/jena-core/src/main/java/org/apache/jena/irix/IRIx.java
@@ -84,6 +84,9 @@ protected IRIx(String string) {
      * <p>
      * Note that a URI can be both "not absolute" and "not relative", e.g. {@code http://example/path#fragment}.
      * <p>
+     * Beware that <a href="https://datatracker.ietf.org/doc/html/rfc2396#section-3.1">RFC 2396 section 3.1</a> has a
+     * different definition, where the scheme is required but a fragment may be present.
+     * <p>
      * See {@linkplain #isReference()} for testing whether a URI is suitable for use in RDF.
      */
     public abstract boolean isAbsolute();

diff --git a/jena-core/src/main/java/org/apache/jena/rdfxml/xmlinput0/impl/XMLBaselessContext.java b/jena-core/src/main/java/org/apache/jena/rdfxml/xmlinput0/impl/XMLBaselessContext.java
@@ -100,7 +100,8 @@ void baseUsed(XMLHandler forErrors, Taint taintMe, String relUri,
     void checkBaseUse(XMLHandler forErrors, Taint taintMe, String relUri, IRI rslt) throws SAXParseException {
 
         String resolvedURI = rslt.toString();
-        if (relUri.equals(resolvedURI) && rslt.isAbsolute())
+        // "absolute" is not the opposite of "relative" in RFC 3986 (it is in RFC 2396).
+        if (relUri.equals(resolvedURI) && !rslt.isRelative())
             return;
 
         forErrors.warning(taintMe, errno, errmsg + ": <" + relUri + ">");

diff --git a/jena-core/src/main/java/org/apache/jena/rdfxml/xmlinput0/impl/XMLHandler.java b/jena-core/src/main/java/org/apache/jena/rdfxml/xmlinput0/impl/XMLHandler.java
@@ -386,7 +386,7 @@ private void checkNamespaceURI(String uri) throws SAXParseException {
         if (uri.length() != 0)
              {
                 IRI u = iriFactory().create(uri);
-                if (!u.isAbsolute()) {
+                if (u.isRelative()) {
                     warning(null,
                             WARN_RELATIVE_NAMESPACE_URI_DEPRECATED,
                             "The namespace URI: <"

diff --git a/jena-core/src/main/java/org/apache/jena/rdfxml/xmlinput1/impl/XMLBaselessContext.java b/jena-core/src/main/java/org/apache/jena/rdfxml/xmlinput1/impl/XMLBaselessContext.java
@@ -86,7 +86,8 @@ void baseUsed(XMLHandler forErrors, Taint taintMe, String relUri,
     void checkBaseUse(XMLHandler forErrors, Taint taintMe, String relUri, IRIx rslt) throws SAXParseException {
 
         String resolvedURI = rslt.toString();
-        if (relUri.equals(resolvedURI) && rslt.isAbsolute())
+        // "absolute" is not the opposite of "relative" in RFC 3986 (it is in RFC 2396).
+        if (relUri.equals(resolvedURI) && !rslt.isRelative())
             return;
 
         forErrors.warning(taintMe, errno, errmsg + ": <" + relUri + ">");

diff --git a/jena-core/src/main/java/org/apache/jena/rdfxml/xmlinput1/impl/XMLHandler.java b/jena-core/src/main/java/org/apache/jena/rdfxml/xmlinput1/impl/XMLHandler.java
@@ -400,7 +400,7 @@ private void checkNamespaceURI(String uri) throws SAXParseException {
                 } catch (IRIException ex) {
                     u = IRIx.createAny(uri);
                 }
-                if (!u.isAbsolute()) {
+                if (u.isRelative()) {
                     warning(null,
                             WARN_RELATIVE_NAMESPACE_URI_DEPRECATED,
                             "The namespace URI: <"

diff --git a/jena-core/src/test/java/org/apache/jena/irix/TestIRIx.java b/jena-core/src/test/java/org/apache/jena/irix/TestIRIx.java
@@ -34,7 +34,7 @@ public TestIRIx(String name, IRIProvider provider) {
         super(name, provider);
     }
 
-    // ---- RFC 3986 Grammar
+    // ---- RFC 3986 Grammar : misc parsing.
 
     @Test public void uri_01()      { test("http://example/abc"); }
 
@@ -49,6 +49,7 @@ public TestIRIx(String name, IRIProvider provider) {
     @Test public void uri_06()      { test("http://1.2.3.4/abc"); }
 
     // ---- Compliance with HTTP RFC7230. https://tools.ietf.org/html/rfc7230#section-2.7
+
     @Test(expected=IRIException.class)
     public void http_01() { parse("http:"); }
 
@@ -135,6 +136,18 @@ public TestIRIx(String name, IRIProvider provider) {
 
     @Test public void reference_08() { reference("file:///a:/~jena/file", true); }
 
+    @Test public void reference_09() { reference("http://example/abcd#frag", true); }
+
+    // -- isAbsolute, isRelative : These are not opposites in RFC 3986. (String, isAbsolute, isRelative)
+
+    @Test public void abs_rel_01()   { test_abs_rel("http://example/abc", true, false); }
+
+    @Test public void abs_rel_02()   { test_abs_rel("abc", false, true); }
+
+    @Test public void abs_rel_03()   { test_abs_rel("http://example/abc#def", false, false); }
+
+    @Test public void abs_rel_04()   { test_abs_rel("abc#def", false, true); }
+
     // -- Resolving
     @Test public void resolve_http_01() { resolve("http://example/", "path", "http://example/path"); }
 
@@ -202,6 +215,13 @@ private void resolve(String baseUriStr, String otherStr, String expected) {
         assertEquals("Base=<"+baseUriStr+"> Rel=<"+otherStr+">", expected, iriStr);
     }
 
+    // Create - is it suitable for an RDF reference?
+    private void test_abs_rel(String uriStr, boolean isAbsolute, boolean isRelative) {
+        IRIx iri = IRIx.create(uriStr);
+        assertEquals("Absolute test: IRI = "+uriStr, isAbsolute, iri.isAbsolute());
+        assertEquals("Relative test: IRI = "+uriStr, isRelative, iri.isRelative());
+    }
+
     // Create - is it suitable for an RDF reference?
     private void reference(String uriStr) {
         IRIx iri = IRIx.create(uriStr);

diff --git a/jena-iri/src/main/java/org/apache/jena/iri/impl/AbsIRIImpl.java b/jena-iri/src/main/java/org/apache/jena/iri/impl/AbsIRIImpl.java
@@ -208,7 +208,10 @@ private void createExceptions(long m) {
 
     @Override
     public boolean isAbsolute() {
-        return has(SCHEME);
+        // Definition from RFC3986 section 4.3
+        return has(SCHEME) && ! has(FRAGMENT);
+        // Definition from RFC2396 section 3.1
+        //return has(SCHEME);
     }
 
     abstract boolean has(int component);
@@ -218,29 +221,6 @@ public boolean isRelative() {
         return !has(SCHEME);
     }
 
-    /*
-     * public boolean isRDFURIReference() { return !hasException(RDF); }
-     *
-     * public boolean isIRI() { return !hasException(IRI); }
-     *
-     * public boolean isURIinASCII() { return !hasException(URI); }
-     */
-    // public boolean isVeryBad() {
-    // return false;
-    // }
-    // public boolean isXSanyURI() {
-    // return !hasException(XMLSchema);
-    // }
-    /*
-    public boolean hasException(int conformance) {
-        return hasExceptionMask(getFactory().recsToMask(conformance));
-    }
-
-    public Iterator exceptions(int conformance) {
-        return exceptionsMask(getFactory().recsToMask(conformance));
-    }
-    */
-
     @Override
     public boolean hasViolation(boolean includeWarnings) {
         return hasExceptionMask(getSchemeSpec().getMask(includeWarnings));