Fix bug: Incorrect parsing of Identifier - "[Customers].[City].&[San …

…Francisco]&[CA]&[USA].&[cust1234]". (Contributed by Antony.)

src/org/olap4j/impl/IdentifierParser.java

@@ -291,10 +291,11 @@ public static int parseMember(
  break;
  case '&':
  state = BEFORE_SEG;
- if (syntax == Builder.Syntax.FIRST_KEY) {
- syntax = Builder.Syntax.NEXT_KEY;
- } else {
+ // Roll the syntax - NAME=>FIRST_KEY=>NEXT_KEY=>NEXT_KEY...
+ if (syntax == Builder.Syntax.NAME) {
  syntax = Builder.Syntax.FIRST_KEY;
+ } else {
+ syntax = Builder.Syntax.NEXT_KEY;
  }
  ++i;
  break;

testsrc/org/olap4j/impl/Olap4jUtilTest.java

@@ -469,20 +469,72 @@ public void testParseIdentifierAdvanced() {
  assertEquals("City", s1.getName());
  assertEquals(Quoting.QUOTED, s1.getQuoting());
  assertTrue(segments.get(2) instanceof KeySegment);
- KeySegment s2 =
- (KeySegment) segments.get(2);
+ final KeySegment s2 = (KeySegment) segments.get(2);
  assertEquals(3, s2.getKeyParts().size());
  final NameSegment s2k0 = s2.getKeyParts().get(0);
  assertEquals("San Francisco", s2k0.getName());
  assertEquals(Quoting.QUOTED, s2k0.getQuoting());
  final NameSegment s2k1 = s2.getKeyParts().get(1);
  assertEquals("CA", s2k1.getName());
+ assertEquals(Quoting.UNQUOTED, s2k1.getQuoting());
+ final NameSegment s2k2 = s2.getKeyParts().get(2);
+ assertEquals("USA", s2k2.getName());
+ assertEquals(Quoting.UNQUOTED, s2k2.getQuoting());
+ final KeySegment s3 = (KeySegment) segments.get(3);
+ assertNull(s3.getName());
+ assertEquals(1, s3.getKeyParts().size());
+ final NameSegment s3k0 = s3.getKeyParts().get(0);
+ assertEquals("cust1234", s3k0.getName());
+ assertEquals(Quoting.QUOTED, s3k0.getQuoting());
+ }
+
+ /**
+ * Advanced test for the {@link IdentifierNode#parseIdentifier}
+ * method. There was a bug that QUOTED sub-segments were skipped
+ * after 2 sub-segments and new segments were created. This test
+ * verifies that any number of QUOTED sub-segments can be under a
+ * segment.
+ */
+ public void testParseIdentifierAdvanced2() {
+ List<IdentifierSegment> segments;
+
+ // detailed example, per javadoc
+ //
+ // A more complex example illustrates a compound key. The identifier
+ // [Customers].[City].&[San Francisco]&[CA]&[USA].&[cust1234]
+ // contains four segments as follows:
+ //
+ // * Segment #0 is QUOTED, name "Customers"
+ // * Segment #1 is QUOTED, name "City"
+ // * Segment #2 is a KEY. It has 3 sub-segments:
+ // ** Sub-segment #0 is QUOTED, name "San Francisco"
+ // ** Sub-segment #1 is QUOTED, name "CA"
+ // ** Sub-segment #2 is QUOTED, name "USA"
+ // * Segment #3 is a KEY. It has 1 sub-segment:
+ // ** Sub-segment #0 is QUOTED, name "cust1234"</li>
+ segments =
+ IdentifierParser.parseIdentifier(
+ "[Customers].[City].&[San Francisco]&[CA]&[USA].&[cust1234]");
+ assertEquals(4, segments.size());
+ final IdentifierSegment s0 = segments.get(0);
+ assertEquals("Customers", s0.getName());
+ assertEquals(Quoting.QUOTED, s0.getQuoting());
+ final IdentifierSegment s1 = segments.get(1);
+ assertEquals("City", s1.getName());
+ assertEquals(Quoting.QUOTED, s1.getQuoting());
+ assertTrue(segments.get(2) instanceof KeySegment);
+ final KeySegment s2 = (KeySegment) segments.get(2);
+ assertEquals(3, s2.getKeyParts().size());
+ final NameSegment s2k0 = s2.getKeyParts().get(0);
+ assertEquals("San Francisco", s2k0.getName());
  assertEquals(Quoting.QUOTED, s2k0.getQuoting());
- final NameSegment s2k2 = s2.getKeyParts().get(0);
- assertEquals("San Francisco", s2k2.getName());
+ final NameSegment s2k1 = s2.getKeyParts().get(1);
+ assertEquals("CA", s2k1.getName());
+ assertEquals(Quoting.QUOTED, s2k1.getQuoting());
+ final NameSegment s2k2 = s2.getKeyParts().get(2);
+ assertEquals("USA", s2k2.getName());
  assertEquals(Quoting.QUOTED, s2k2.getQuoting());
- KeySegment s3 =
- (KeySegment) segments.get(3);
+ final KeySegment s3 = (KeySegment) segments.get(3);
  assertNull(s3.getName());
  assertEquals(1, s3.getKeyParts().size());
  final NameSegment s3k0 = s3.getKeyParts().get(0);