Fix XMP metadata parser to be able to parse not finished ASCII control characters

DEVSIX-9085

Author: vitali-pr

commons/src/sharpenconfig/java/com/itextpdf/commons/SharpenConfigMapping.java

@@ -83,8 +83,6 @@ public void applyMappingConfiguration(MappingConfigurator configurator) {
         configurator.mapType("java.io.PrintWriter", "iText.Commons.Utils.FormattingStreamWriter");
         configurator.mapMethod("java.nio.charset.Charset.forName", "iText.Commons.Utils.EncodingUtil.GetEncoding", false);
         configurator.mapField("java.nio.charset.StandardCharsets.ISO_8859_1", "iText.Commons.Utils.EncodingUtil.ISO_8859_1");
-        configurator.mapType("java.io.PushbackReader", "iText.Commons.Utils.PushbackReader");
-        configurator.mapType("java.io.FilterReader", "iText.Commons.Utils.FilterReader");
         configurator.mapMethod("java.lang.String.valueOf(char[])", "iText.Commons.Utils.JavaUtil.GetStringForChars", false);
         configurator.mapMethod("java.lang.String.valueOf(char[],int,int)", "iText.Commons.Utils.JavaUtil.GetStringForChars", false);
         configurator.mapMethod("java.lang.String.String(byte[])", "iText.Commons.Utils.JavaUtil.getStringForBytes", false);

kernel/src/main/java/com/itextpdf/kernel/xmp/impl/FixASCIIControlsReader.java

@@ -31,14 +31,12 @@
 package com.itextpdf.kernel.xmp.impl;
 
 import java.io.IOException;
-import java.io.PushbackReader;
 import java.io.Reader;
 
-
 /**
  * @since   22.08.2006
  */
-public class FixASCIIControlsReader extends PushbackReader
+public class FixASCIIControlsReader extends Reader
 {
 	/** */
 	private static final int STATE_START = 0;
@@ -52,43 +50,41 @@ public class FixASCIIControlsReader extends PushbackReader
 	private static final int STATE_DIG1 = 4;
 	/** */
 	private static final int STATE_ERROR = 5;
-	/** */
-	private static final int BUFFER_SIZE = 8;
 	/** the state of the automaton */
 	private int state = STATE_START;
 	/** the result of the escaping sequence */
 	private int control = 0;
 	/** count the digits of the sequence */
-	private int digits = 0; 
-	
+	private int digits = 0;
+
+	private Reader in;
+
 	/**
-	 * The look-ahead size is 6 at maximum («)
-	 * @see java.io.PushbackReader#PushbackReader(java.io.Reader, int)
+	 * A wrapper xmp reader to handle control characters («)
+	 *
 	 * @param input a Reader
 	 */
 	public FixASCIIControlsReader(Reader input)
 	{
-		super(input, BUFFER_SIZE);
+		in = input;
 	}
 
-
 	/**
 	 * @see java.io.Reader#read(char[], int, int)
 	 */
 	public int read(char[] cbuf, int off, int len) throws IOException
 	{
-		int readAhead = 0;
 		int read = 0;
 		int pos = off;
-		char[] readAheadBuffer = new char[BUFFER_SIZE];
+		char[] readAheadBuffer = new char[1];
 
 		boolean available = true;
-		while (available  &&  read < len)
+		while (available && read < len)
 		{
-			available = super.read(readAheadBuffer, readAhead, 1) == 1;
+			available = in.read(readAheadBuffer, 0, 1) == 1;
 			if (available)
 			{
-				char c = processChar(readAheadBuffer[readAhead]);
+				char c = processChar(readAheadBuffer[0]);
 				if (state == STATE_START)
 				{
 					// replace control chars with space
@@ -97,34 +93,26 @@ public int read(char[] cbuf, int off, int len) throws IOException
 						c = ' ';
 					}	
 					cbuf[pos++] = c;
-					readAhead = 0;
 					read++;
 				}
 				else if (state == STATE_ERROR)
 				{
-					unread(readAheadBuffer, 0, readAhead + 1);
-					readAhead = 0;
+					// It's broken ASCII character sequence, let's just skip them
+					// If we try to preserve them, SAX parser will throw later on anyway
 				}
-				else
-				{
-					readAhead++;
-				}
-			}
-			else if (readAhead > 0)
-			{
-				// handles case when file ends within excaped sequence
-				unread(readAheadBuffer, 0, readAhead);
-				state = STATE_ERROR;
-				readAhead = 0;
-				available = true;
 			}
 		}
 
-		
-		return read > 0  ||  available ? read : XMPUtilsImpl.eofReadBytesValue();
+		return read > 0 || available ? read : XMPUtilsImpl.eofReadBytesValue();
 	}
-	
-	
+
+    /**
+     * {@inheritDoc}
+     */
+    public void close() throws IOException {
+        in.close();
+    }
+
 	/**
 	 * Processes numeric escaped chars to find out if they are a control character.
 	 * @param ch a char

kernel/src/test/java/com/itextpdf/kernel/pdf/XMPMetadataTest.java

@@ -51,8 +51,8 @@ This file is part of the iText (R) project.
 @Tag("IntegrationTest")
 public class XMPMetadataTest extends ExtendedITextTest{
 
-    public static final String SOURCE_FOLDER = "./src/test/resources/com/itextpdf/kernel/pdf/XmpWriterTest/";
-    public static final String DESTINATION_FOLDER = TestUtil.getOutputPath() + "/kernel/pdf/XmpWriterTest/";
+    public static final String SOURCE_FOLDER = "./src/test/resources/com/itextpdf/kernel/pdf/XMPMetadataTest/";
+    public static final String DESTINATION_FOLDER = TestUtil.getOutputPath() + "/kernel/pdf/XMPMetadataTest/";
 
     @BeforeAll
     public static void beforeClass() {
@@ -419,4 +419,41 @@ public void listParsingTest() {
         Assertions.assertThrows(XMPException.class,
                 () -> XMPMetaFactory.parseFromBuffer(xmp.getBytes(StandardCharsets.UTF_8)));
     }
+
+    @Test
+    public void readDocumentWithControlCharactersInXMPMetadata() throws IOException {
+        String src = SOURCE_FOLDER + "docWithControlCharactersInXmp.pdf";
+        try (PdfDocument document = new PdfDocument(new PdfReader(src),
+                new PdfWriter(new ByteArrayOutputStream()), new StampingProperties())) {
+            Assertions.assertEquals(PdfConformance.PDF_A_3A, document.getConformance());
+        }
+    }
+
+    @Test
+    public void readDocumentWithBrokenControlCharactersInXMPMetadata() throws IOException {
+        String src = SOURCE_FOLDER + "docWithBrokenControlCharactersInXmp.pdf";
+        try (PdfDocument document = new PdfDocument(new PdfReader(src),
+                new PdfWriter(new ByteArrayOutputStream()), new StampingProperties())) {
+            Assertions.assertEquals(PdfConformance.PDF_A_3A, document.getConformance());
+        }
+    }
+
+    @Test
+    public void readDocumentWithInvalidConformance() throws IOException {
+        String src = SOURCE_FOLDER + "docWithInvalidConformance.pdf";
+        try (PdfDocument document = new PdfDocument(new PdfReader(src),
+                new PdfWriter(new ByteArrayOutputStream()), new StampingProperties())) {
+            Assertions.assertEquals(PdfConformance.PDF_NONE_CONFORMANCE, document.getConformance());
+        }
+    }
+
+    @LogMessages(messages = {@LogMessage(messageTemplate = IoLogMessageConstant.EXCEPTION_WHILE_UPDATING_XMPMETADATA)})
+    @Test
+    public void readDocumentWithInvalidXMPMetadata() throws IOException {
+        String src = SOURCE_FOLDER + "docWithInvalidMetadata.pdf";
+        try (PdfDocument document = new PdfDocument(new PdfReader(src),
+                new PdfWriter(new ByteArrayOutputStream()), new StampingProperties())) {
+            Assertions.assertEquals(PdfConformance.PDF_NONE_CONFORMANCE, document.getConformance());
+        }
+    }
 }

kernel/src/test/resources/com/itextpdf/kernel/pdf/XMPMetadataTest/docWithBrokenControlCharactersInXmp.pdf

@@ -142,34 +142,6 @@ public void saveAndReadDocumentWithCanonicalXmpMetadata() throws IOException, XM
         }
     }
 
-    @Test
-    public void readDocumentWithControlCharactersInXMPMetadata() throws IOException {
-        String src = sourceFolder + "pdfs/docWithControlCharactersInXmp.pdf";
-        try (PdfADocument document = new PdfADocument(new PdfReader(src),
-                new PdfWriter(new java.io.ByteArrayOutputStream()), new StampingProperties())) {
-            Assertions.assertEquals(PdfConformance.PDF_A_3A, document.getConformance());
-        }
-    }
-
-    @Test
-    public void readDocumentWithInvalidConformance() throws IOException {
-        String src = sourceFolder + "pdfs/docWithInvalidConformance.pdf";
-        try (PdfDocument document = new PdfDocument(new PdfReader(src),
-                new PdfWriter(new java.io.ByteArrayOutputStream()), new StampingProperties())) {
-            Assertions.assertEquals(PdfConformance.PDF_NONE_CONFORMANCE, document.getConformance());
-        }
-    }
-
-    @LogMessages(messages = {@LogMessage(messageTemplate = IoLogMessageConstant.EXCEPTION_WHILE_UPDATING_XMPMETADATA)})
-    @Test
-    public void readDocumentWithInvalidXMPMetadata() throws IOException {
-        String src = sourceFolder + "pdfs/docWithInvalidMetadata.pdf";
-        try (PdfDocument document = new PdfDocument(new PdfReader(src),
-                new PdfWriter(new java.io.ByteArrayOutputStream()), new StampingProperties())) {
-            Assertions.assertEquals(PdfConformance.PDF_NONE_CONFORMANCE, document.getConformance());
-        }
-    }
-
     @Test
     public void testPdfUAExtensionMetadata() throws IOException {
 

pdfa/src/test/resources/com/itextpdf/pdfa/pdfs/docWithControlCharactersInXmp.pdf renamed to kernel/src/test/resources/com/itextpdf/kernel/pdf/XMPMetadataTest/docWithControlCharactersInXmp.pdf

@@ -503,6 +503,12 @@
                 <excludes>
                   <!-- Contains interfaces for bouncycastle wrappers which can be changes in any time -->
                   <exclude>com.itextpdf.commons.bouncycastle</exclude>
+                  <!--
+                   Temporary exclude to be able to cleanup code.
+                   This class is not intended to be used by our users directly.
+                   TODO: DEVSIX-9096 - remove after 9.3.0 release
+                  -->
+                  <exclude>com.itextpdf.kernel.xmp.impl.FixASCIIControlsReader</exclude>
                 </excludes>
                 <excludeModules>
                   <excludeModule>bouncy-castle-adapter</excludeModule>