GH-2979: Test for surrogates after escape processing

apache · Feb 24, 2025 · 55694c2 · 55694c2
1 parent 2c1f02f
commit 55694c2
Show file tree

Hide file tree

Showing 9 changed files with 65 additions and 14 deletions.
diff --git a/jena-arq/Grammar/Final/README.txt b/jena-arq/Grammar/Final/README.txt
@@ -1 +1,16 @@
 Final-for-spec-publication versions of the grammar.
+
+sparql_10-final.jj       - SPARQL 1.0 "sparql_10.jj" ("main.jj" after cpp)
+
+sparql_11-final.jj	     - SPARQL 1.1 "sparql_11.jj" ("main.jj" after cpp)
+
+sparql-main-11.jj        - SPARQL 1.1 "main.jj" (com.hp)
+
+
+sparql_11-dev-final.jj   - End SPARQL 1.1 development.  (org.apache.jena.graph "main.jj" at SPARQLParser11)
+sparql_11-dev-final.txt  - jjdoc
+tokens_11.txt            - Tokens file.
+
+-- Coming soon.
+sparql_12-final.jj       - SPARQL 1.2 "sparql_11.jj" ("main.jj" after cpp)
+sparql-main-12.jj        - SPARQL 1.2 "main.jj"
diff --git a/jena-arq/Grammar/arq.jj b/jena-arq/Grammar/arq.jj
@@ -1910,8 +1910,8 @@ String String() : { Token t ; String lex ; }
   | t = <STRING_LITERAL_LONG1> { lex = stripQuotes3(t.image) ; }
   | t = <STRING_LITERAL_LONG2> { lex = stripQuotes3(t.image) ; }
   )
-    { checkString(lex, t.beginLine, t.beginColumn) ;
-      lex = unescapeStr(lex, t.beginLine, t.beginColumn) ;
+    { lex = unescapeStr(lex, t.beginLine, t.beginColumn) ;
+      checkString(lex, t.beginLine, t.beginColumn) ;
       return lex ;
     }
 }

diff --git a/jena-arq/Grammar/main.jj b/jena-arq/Grammar/main.jj
@@ -2621,8 +2621,7 @@ String String() : { Token t ; String lex ; }
   | t = <STRING_LITERAL_LONG1> { lex = stripQuotes3(t.image) ; }
   | t = <STRING_LITERAL_LONG2> { lex = stripQuotes3(t.image) ; }
   )
-    { checkString(lex, t.beginLine, t.beginColumn) ;
-      lex = unescapeStr(lex, t.beginLine, t.beginColumn) ;
+    { lex = prepareLexicalForm(lex, t.beginLine, t.beginColumn) ;
       return lex ;
     }
 }

diff --git a/jena-arq/Grammar/sparql_12.jj b/jena-arq/Grammar/sparql_12.jj
@@ -1625,8 +1625,8 @@ String String() : { Token t ; String lex ; }
   | t = <STRING_LITERAL_LONG1> { lex = stripQuotes3(t.image) ; }
   | t = <STRING_LITERAL_LONG2> { lex = stripQuotes3(t.image) ; }
   )
-    { checkString(lex, t.beginLine, t.beginColumn) ;
-      lex = unescapeStr(lex, t.beginLine, t.beginColumn) ;
+    { lex = unescapeStr(lex, t.beginLine, t.beginColumn) ;
+      checkString(lex, t.beginLine, t.beginColumn) ;
       return lex ;
     }
 }

diff --git a/jena-arq/src/main/java/org/apache/jena/sparql/lang/QueryParserBase.java b/jena-arq/src/main/java/org/apache/jena/sparql/lang/QueryParserBase.java
@@ -209,10 +209,15 @@ protected Node stripSign(Node node) {
     }
 
     protected void checkString(String string, int line, int column) {
+        // Checks for bare surrogate pairs.
         for ( int i = 0 ; i < string.length() ; i++ ) {
             // Not "codePointAt" which does surrogate processing.
             char ch = string.charAt(i);
-            // Check surrogate pairs are pairs.
+
+            // Check surrogate pairs are in pairs. Pairs are high-low.
+            if ( Character.isLowSurrogate(ch) )
+                throw new QueryParseException("Bad surrogate pair (low surrogate without high surrogate)", line, column);
+
             if ( Character.isHighSurrogate(ch) ) {
                 i++;
                 if ( i == string.length() )
@@ -221,8 +226,6 @@ protected void checkString(String string, int line, int column) {
                 if ( !Character.isLowSurrogate(ch1) ) {
                     throw new QueryParseException("Bad surrogate pair (high surrogate not followed by low surrogate)", line, column);
                 }
-            } else if ( Character.isLowSurrogate(ch) ) {
-                throw new QueryParseException("Bad surrogate pair (low surrogate without high surrogate)", line, column);
             }
         }
     }
@@ -345,7 +348,6 @@ protected static String stripChars(String s, int n) {
 
     protected Var createVariable(String s, int line, int column) {
         s = s.substring(1); // Drop the marker
-
         // This is done by the parser input stream nowadays.
         // s = unescapeCodePoint(s, line, column);
         // Check \ u did not put in any illegals.
@@ -361,6 +363,8 @@ protected Node createTripleTerm(Node s, Node p, Node o, int line, int column) {
     protected String resolveQuotedIRI(String iriStr, int line, int column) {
         iriStr = stripQuotes(iriStr);
         iriStr = unescapeUnicode(iriStr, line, column);
+        // Check for Unicode surrogates
+        checkString(iriStr, line, column);
         return resolveIRI(iriStr, line, column);
     }
 

diff --git a/jena-arq/src/main/java/org/apache/jena/sparql/lang/arq/ARQParser.java b/jena-arq/src/main/java/org/apache/jena/sparql/lang/arq/ARQParser.java
@@ -7442,8 +7442,8 @@ final public Node BooleanLiteral() throws ParseException {
       jj_consume_token(-1);
       throw new ParseException();
     }
-checkString(lex, t.beginLine, t.beginColumn) ;
-      lex = unescapeStr(lex, t.beginLine, t.beginColumn) ;
+lex = unescapeStr(lex, t.beginLine, t.beginColumn) ;
+      checkString(lex, t.beginLine, t.beginColumn) ;
       {if ("" != null) return lex ;}
     throw new Error("Missing return statement in function");
 }

diff --git a/jena-arq/src/main/java/org/apache/jena/sparql/lang/sparql_12/SPARQLParser12.java b/jena-arq/src/main/java/org/apache/jena/sparql/lang/sparql_12/SPARQLParser12.java
@@ -5837,8 +5837,8 @@ final public Node BooleanLiteral() throws ParseException {
       jj_consume_token(-1);
       throw new ParseException();
     }
-checkString(lex, t.beginLine, t.beginColumn) ;
-      lex = unescapeStr(lex, t.beginLine, t.beginColumn) ;
+lex = unescapeStr(lex, t.beginLine, t.beginColumn) ;
+      checkString(lex, t.beginLine, t.beginColumn) ;
       {if ("" != null) return lex ;}
     throw new Error("Missing return statement in function");
 }

diff --git a/jena-arq/src/test/java/org/apache/jena/sparql/syntax/TestQueryParser.java b/jena-arq/src/test/java/org/apache/jena/sparql/syntax/TestQueryParser.java
@@ -19,6 +19,7 @@
 package org.apache.jena.sparql.syntax;
 
 import static org.junit.jupiter.api.Assertions.assertThrows;
+import static org.junit.jupiter.api.Assertions.assertTrue;
 
 import org.junit.jupiter.api.Test;
 
@@ -40,6 +41,36 @@ private static void silent(Runnable action) {
         LogCtl.withLevel(loggerSPARQL, "fatal", action);
     }
 
+    // Single backslash so a Java string escape, raw surrogate in the string.
+    @Test
+    public void syntax_unicode_raw_surrogate_uri() {
+        QueryParseException ex = assertThrows(QueryParseException.class,  ()->testParse("SELECT * { <http://example/\uD800> ?p ?o}"));
+        assertTrue(ex.getMessage().contains("surrogate"));
+    }
+
+    @Test
+    public void syntax_unicode_raw_surrogate_string() {
+        QueryParseException ex = assertThrows(QueryParseException.class,  ()->testParse("SELECT * { ?s ?p '\uD800' }"));
+        assertTrue(ex.getMessage().contains("surrogate"));
+    }
+
+    // Double backslash so the query string has an escape in it.
+    @Test
+    public void syntax_unicode_escaped_surrogate_uri() {
+        QueryParseException ex = assertThrows(QueryParseException.class,  ()->testParse("SELECT * { <http://example/\\uD800> ?p ?o}"));
+        assertTrue(ex.getMessage().contains("surrogate"));
+    }
+
+    @Test
+    public void syntax_unicode_escaped_surrogate_strings() {
+        QueryParseException ex = assertThrows(QueryParseException.class,  ()->testParse("SELECT * { ?s ?p '\\uD800'}"));
+        assertTrue(ex.getMessage().contains("surrogate"));
+    }
+
+    private static void testParse(String string) {
+        QueryFactory.create(string);
+    }
+
     @Test public void syntax_uri_brackets_1() {
         testParseIRIs("<http://example/#[]>");
     }

diff --git a/jena-cmds/src/test/java/arq/rdftests.java b/jena-cmds/src/test/java/arq/rdftests.java
@@ -81,6 +81,8 @@ public class rdftests extends CmdGeneral
         RIOT.getContext().set(RIOT.symTurtleDirectiveStyle, "sparql");
     }
 
+    // Test runners are in jena-arq, package org.apache.jena.arq.junit.runners
+
     public static void main(String...argv) {
         try { new rdftests(argv).mainRun(); }
         catch (TerminationException ex) {