MXParser tokenization fails when PI is before first tag (fixes #7) (#12)

Author: gnodet

src/main/java/org/codehaus/plexus/util/xml/pull/MXParser.java

@@ -1822,13 +1822,9 @@ else if ( ch == '\uFFFD' )
                 {
                     // check if it is 'xml'
                     // deal with XMLDecl
-                    boolean isXMLDecl = parsePI();
+                    parsePI();
                     if ( tokenize )
                     {
-                        if ( isXMLDecl )
-                        {
-                            return eventType = START_DOCUMENT;
-                        }
                         return eventType = PROCESSING_INSTRUCTION;
                     }
                 }
@@ -3111,7 +3107,7 @@ else if ( ch == '\n' )
         }
     }
 
-    private boolean parsePI()
+    private void parsePI()
         throws XmlPullParserException, IOException
     {
         // implements XML 1.0 Section 2.6 Processing Instructions
@@ -3213,7 +3209,7 @@ else if ( ch == '<' )
                                 final int off = piTargetStart + 3;
                                 final int len = pos - 2 - off;
                                 xmlDeclContent = newString( buf, off, len );
-                                return false;
+                                return;
                             }
                         }
                     }
@@ -3291,7 +3287,6 @@ else if ( ch == '\n' )
                 --pcEnd;
             }
         }
-        return true;
     }
 
     // protected final static char[] VERSION = {'v','e','r','s','i','o','n'};

src/test/java/org/codehaus/plexus/util/xml/pull/MXParserTest.java

@@ -1622,4 +1622,46 @@ public void testUnicode() throws IOException {
             fail( "should not raise exception: " + e );
         }
     }
+
+    @Test
+    public void testProcessingInstructionTokenizeBeforeFirstTag()
+        throws Exception
+    {
+        String input = "<?a?><test>nnn</test>";
+
+        MXParser parser = new MXParser();
+        parser.setInput( new StringReader( input ) );
+
+        assertEquals( XmlPullParser.START_DOCUMENT, parser.getEventType() );
+        assertEquals( XmlPullParser.PROCESSING_INSTRUCTION, parser.nextToken() );
+        assertEquals( "a", parser.getText() );
+        assertEquals( XmlPullParser.START_TAG, parser.nextToken() );
+        assertEquals( "test", parser.getName() );
+        assertEquals( XmlPullParser.TEXT, parser.nextToken() );
+        assertEquals( "nnn", parser.getText() );
+        assertEquals( XmlPullParser.END_TAG, parser.nextToken() );
+        assertEquals( XmlPullParser.END_DOCUMENT, parser.nextToken() );
+    }
+
+    @Test
+    public void testProcessingInstructionTokenizeAfterXMLDeclAndBeforeFirstTag()
+        throws Exception
+    {
+        String input = "<?xml version=\"1.0\" encoding=\"UTF-8\"?><?a?><test>nnn</test>";
+
+        MXParser parser = new MXParser();
+        parser.setInput( new StringReader( input ) );
+
+        assertEquals( XmlPullParser.START_DOCUMENT, parser.getEventType() );
+        assertEquals( XmlPullParser.PROCESSING_INSTRUCTION, parser.nextToken() );
+        assertEquals( "xml version=\"1.0\" encoding=\"UTF-8\"", parser.getText() );
+        assertEquals( XmlPullParser.PROCESSING_INSTRUCTION, parser.nextToken() );
+        assertEquals( "a", parser.getText() );
+        assertEquals( XmlPullParser.START_TAG, parser.nextToken() );
+        assertEquals( "test", parser.getName() );
+        assertEquals( XmlPullParser.TEXT, parser.nextToken() );
+        assertEquals( "nnn", parser.getText() );
+        assertEquals( XmlPullParser.END_TAG, parser.nextToken() );
+        assertEquals( XmlPullParser.END_DOCUMENT, parser.nextToken() );
+    }
 }