From e12937d97f7827f597d6433f673c402603dac4ed Mon Sep 17 00:00:00 2001
From: samalloing <sam.alloing@gmail.com>
Date: Wed, 29 Apr 2020 10:14:25 +0200
Subject: [PATCH] proposition to fix #277, the value of Destination is printed
 instead of the java object and the error PDF_HUL_149 instead of PDF_HUL_122

---
 .../hul/ois/jhove/module/PdfModule.java       |  8 ++++--
 .../hul/ois/jhove/module/pdf/Literal.java     | 28 ++++++++++---------
 2 files changed, 20 insertions(+), 16 deletions(-)

diff --git a/jhove-modules/pdf-hul/src/main/java/edu/harvard/hul/ois/jhove/module/PdfModule.java b/jhove-modules/pdf-hul/src/main/java/edu/harvard/hul/ois/jhove/module/PdfModule.java
index d05095e25..3d98b3091 100644
--- a/jhove-modules/pdf-hul/src/main/java/edu/harvard/hul/ois/jhove/module/PdfModule.java
+++ b/jhove-modules/pdf-hul/src/main/java/edu/harvard/hul/ois/jhove/module/PdfModule.java
@@ -53,6 +53,7 @@
 import edu.harvard.hul.ois.jhove.IdentifierType;
 import edu.harvard.hul.ois.jhove.InfoMessage;
 import edu.harvard.hul.ois.jhove.InternalSignature;
+import edu.harvard.hul.ois.jhove.Message;
 import edu.harvard.hul.ois.jhove.Module;
 import edu.harvard.hul.ois.jhove.ModuleBase;
 import edu.harvard.hul.ois.jhove.NisoImageMetadata;
@@ -3350,7 +3351,7 @@ protected void addDestination(PdfObject itemObj, String propName,
 				// Encryption messes up name trees
 				if (!_encrypted) {
 					int pageObjNum = resolveIndirectDest(
-							dest.getIndirectDest());
+							dest.getIndirectDest(), info);
 					if (pageObjNum == -1) {
 						// The scope of the reference is outside this
 						// file, so we just report it as such.
@@ -4042,7 +4043,7 @@ protected Property buildOutlineItemProperty(PdfDictionary dict,
 				Destination dest = new Destination(destObj, this, false);
 				if (dest.isIndirect()) {
 					itemList.add(new Property(PROP_NAME_DESTINATION,
-							PropertyType.STRING, dest.getIndirectDest()));
+							PropertyType.STRING, dest.getIndirectDest().getStringValue()));
 				} else {
 					int pageObjNum = dest.getPageDestObjNumber();
 					Integer destPg = _pageSeqMap.get(new Integer(pageObjNum));
@@ -4157,7 +4158,7 @@ protected boolean doOutlineStuff(RepInfo info) {
 	 * We return the page sequence number for the referenced page.
 	 * If we can't find a match for the reference, we return -1.
 	 */
-	protected int resolveIndirectDest(PdfSimpleObject key) throws PdfException {
+	protected int resolveIndirectDest(PdfSimpleObject key, RepInfo info) throws PdfException {
 		if (key == null) {
 			throw new IllegalArgumentException("Argument key can not be null");
 		}
@@ -4173,6 +4174,7 @@ protected int resolveIndirectDest(PdfSimpleObject key) throws PdfException {
 					key.getStringValue());
 			JhoveMessage message = JhoveMessages.getMessageInstance(
 					MessageConstants.PDF_HUL_149.getId(), mess);
+			info.setMessage(new ErrorMessage(message));
 			throw new PdfInvalidException(message); // PDF-HUL-149
 			// OR if this is not considered invalid
 			// return -1;
diff --git a/jhove-modules/pdf-hul/src/main/java/edu/harvard/hul/ois/jhove/module/pdf/Literal.java b/jhove-modules/pdf-hul/src/main/java/edu/harvard/hul/ois/jhove/module/pdf/Literal.java
index fe8d38b63..4adceb003 100644
--- a/jhove-modules/pdf-hul/src/main/java/edu/harvard/hul/ois/jhove/module/pdf/Literal.java
+++ b/jhove-modules/pdf-hul/src/main/java/edu/harvard/hul/ois/jhove/module/pdf/Literal.java
@@ -152,7 +152,7 @@ public long processLiteral (Tokenizer tok) throws IOException
         _rawBytes = new Vector<> (32);
         _state = State.LITERAL;
 
-	long offset = 0;
+       long offset = 0;
         for (;;) {
             ch = tok.readChar ();
             // If we get -1, then we've hit an EOF without proper termination of
@@ -162,7 +162,6 @@ public long processLiteral (Tokenizer tok) throws IOException
             }
             offset++;
             _rawBytes.add (ch);
-
             if (_state == State.LITERAL) {
                 // We are still in a state of flux, determining the encoding
                 if (ch == FE) {
@@ -280,17 +279,18 @@ else if (_state == (State.LITERAL_UTF16_2)) {
                     readUTFLanguageCode (tok);
                 }
                 else {
-		    /* It turns out that a backslash may be double-byte,
-		     * rather than the assumed single.byte.  The following
-		     * allows for this. Suggested by Justin Litman, Library
-		     * of Congress, 2006-03-17.
-		     */
-		    if (utfch == BACKSLASH) {
-			utfch = readBackslashSequence (false, tok);
-			if (utfch == 0) {
-			    continue;   /* Invalid character, ignore. */
-			}
-		    }
+                    /* It turns out that a backslash may be double-byte,
+                     * rather than the assumed single.byte.  The following
+                     * allows for this. Suggested by Justin Litman, Library
+                     * of Congress, 2006-03-17.
+                     */
+				    if (utfch == BACKSLASH) {
+					utfch = readBackslashSequence (false, tok);
+						if (utfch == 0) {
+							_state = State.LITERAL_UTF16_2; // skip the wrong char and reset to previous state
+						    continue;   /* Invalid character, ignore. */
+						}
+				    }
                     buffer.append ((char) utfch);
                 }
             }
@@ -631,6 +631,8 @@ private int readBackslashSequence (boolean utf16, Tokenizer tok)
                 return LF;
             case 0X72:   // r
                 return CR;
+            case 0xd: // this is an error for CR
+            	return 0;
             case 0X74:   // t
                 return HT;
             case 0X68:   // h