Merge branch 'integration' into refact/core-apps-warnings

jhove-modules/src/main/java/edu/harvard/hul/ois/jhove/module/pdf/AProfile.java

@@ -52,12 +52,12 @@ public final class AProfile extends PdfProfile
 
     /* The following are the annotation types which are considered
        non-text annotations. */
-    private String[] nonTextAnnotTypes = {
-        "Link", "Line", "Square", "Circle",
-        "Polygon", "Polyline", "Stamp", "Caret",
-        "Ink", "Popup", "Widget", "Screen",
-        "PrinterMark", "TrapNet"
-    };
+//    private String[] nonTextAnnotTypes = {
+//        "Link", "Line", "Square", "Circle",
+//        "Polygon", "Polyline", "Stamp", "Caret",
+//        "Ink", "Popup", "Widget", "Screen",
+//        "PrinterMark", "TrapNet"
+//    };
 
     private String[] excludedActions = {
         "Launch", "Sound", "Movie", "ResetForm",
@@ -100,6 +100,7 @@ public void setTaggedProfile (TaggedProfile tpr)
      * be called.
      *
      */
+    @Override
     public boolean satisfiesThisProfile ()
     {
         // Assume level A compliance.
@@ -234,7 +235,7 @@ private boolean fontsOK ()
         Iterator<Map<Integer, PdfObject>> iter = lst.listIterator ();
         try {
             while (iter.hasNext ()) {
-                Map<Integer, PdfObject> fmap = (Map<Integer, PdfObject>) iter.next ();
+                Map<Integer, PdfObject> fmap = iter.next ();
                 Iterator<PdfObject> iter1 = fmap.values ().iterator ();
                 while (iter1.hasNext ()) {
                     PdfDictionary font = (PdfDictionary) iter1.next ();
@@ -318,7 +319,7 @@ private boolean fontOK (PdfDictionary font)
 	     *
 	     * PdfStream toUni = (PdfStream) font.get ("ToUnicode");
 	     */
-            PdfObject toUni = (PdfObject) font.get ("ToUnicode");
+            PdfObject toUni = font.get ("ToUnicode");
             if (toUni == null) {
                 _levelA = false;
             }
@@ -368,7 +369,7 @@ private boolean type0FontsOK ()
                 // While Adobe warns that this may change in a
                 // previous version, we require here that the
                 // first descendant be a CIDFont, and ignore any others.
-                Vector subfonts = descendants.getContent ();
+                Vector<PdfObject> subfonts = descendants.getContent ();
 
 		/*
 		 * Fix contributed by FCLA, 2007-05-30, to permit the
@@ -380,7 +381,7 @@ private boolean type0FontsOK ()
 		 * subfont = (PdfDictionary)
 		 *                  _module.resolveIndirectObject (subfont);
 		 */
-                PdfObject objFont = (PdfObject) subfonts.elementAt (0);
+                PdfObject objFont = subfonts.elementAt (0);
                 PdfDictionary subfont = (PdfDictionary)
                         _module.resolveIndirectObject (objFont);
                 PdfSimpleObject subtype = 

jhove-modules/src/main/java/edu/harvard/hul/ois/jhove/module/pdf/AProfileLevelA.java

@@ -42,6 +42,7 @@ public AProfileLevelA(PdfModule module) {
      * been called on the profile assigned by <code>setAProfile</code>.
      *
      */
+    @Override
     public boolean satisfiesThisProfile() {
         return _aProfile.satisfiesLevelA();
     }

jhove-modules/src/main/java/edu/harvard/hul/ois/jhove/module/pdf/Destination.java

@@ -6,6 +6,8 @@
 package edu.harvard.hul.ois.jhove.module.pdf;
 
 import edu.harvard.hul.ois.jhove.module.PdfModule;
+
+import java.io.IOException;
 import java.util.*;
 
 /** 
@@ -42,7 +44,7 @@ public final class Destination
      ******************************************************************/
 
     /* Flag indicating destination is indirect. */
-    private boolean _indirect;
+    private boolean _indirect = false;
 
     /* Name of indirect destination. */
     private PdfSimpleObject _indirectDest;
@@ -61,36 +63,35 @@ public final class Destination
      *  @param  named        Flag indicating whether this object came
      *                       from a named destination.
      */
-    public Destination (PdfObject destObj, PdfModule module, boolean named)
+    public Destination (final PdfObject destObj, final PdfModule module, final boolean named)
                          throws PdfException
     {
+        if (!named && destObj instanceof PdfSimpleObject) {
+            _indirect = true;
+            _indirectDest = (PdfSimpleObject) destObj;
+            return;
+        }
+        PdfArray destArray = null;
         try {
-            if (!named && destObj instanceof PdfSimpleObject) {
-                _indirect = true;
-                _indirectDest = (PdfSimpleObject) destObj;
-            }
-            else if (destObj instanceof PdfArray) {
+            if (destObj instanceof PdfArray) {
+                destArray =  (PdfArray) destObj;
                 // We extract only the page reference, not the view.
-                _indirect = false;
-                Vector v = ((PdfArray) destObj).getContent ();
-                _pageDest = (PdfDictionary) module.resolveIndirectObject
-                        ((PdfObject) v.elementAt (0));
-            }
-            else if (named && destObj instanceof PdfDictionary) {
-                PdfArray destArray = (PdfArray)
+                _pageDest = findDirectDest(module, destArray);
+            } else if (named && destObj instanceof PdfDictionary) {
+                destArray = (PdfArray)
                         ((PdfDictionary) destObj).get ("D");
                 // The D entry is just like the array above.
-                _indirect = false;
-                Vector v = destArray.getContent ();
-                _pageDest = (PdfDictionary) module.resolveIndirectObject
-                        ((PdfObject) v.elementAt (0));
-            }
-            else {
-                throw new Exception ("");
+                _pageDest = findDirectDest(module, destArray);
+            } else {
+                throw new PdfInvalidException (MessageConstants.ERR_DEST_OBJ_INVALID); // PDF-HUL-1
             }
         }
-        catch (Exception e) {
-            throw new PdfInvalidException (MessageConstants.ERR_DEST_OBJ_INVALID);
+        catch (ClassCastException e) {
+            throw new PdfInvalidException (MessageConstants.ERR_DEST_OBJ_INVALID); // PDF-HUL-2
+        }
+        catch (IOException e) {
+            throw new PdfInvalidException (String.format(MessageConstants.ERR_DEST_IOEXCEP_READING, // PDF-HUL-3
+                    Integer.valueOf(destArray._objNumber)));
         }
     }
 
@@ -131,4 +132,9 @@ public int getPageDestObjNumber () throws NullPointerException
     {
         return _pageDest.getObjNumber ();
     }
+
+    private static PdfDictionary findDirectDest(final PdfModule module, final PdfArray destObj)
+            throws PdfException, IOException {
+        return (PdfDictionary) module.resolveIndirectObject(destObj.getContent().elementAt(0));
+    }
 }

jhove-modules/src/main/java/edu/harvard/hul/ois/jhove/module/pdf/DocNode.java

@@ -39,13 +39,14 @@ public abstract class DocNode
      */
     public DocNode (PdfModule module,
                 PageTreeNode parent, 
-                PdfDictionary dict)
+                PdfDictionary dict) throws PdfMalformedException
     {
+         if (dict == null) {
+             throw new PdfMalformedException (MessageConstants.ERR_DOC_NODE_DICT_MISSING); // PDF-HUL-4
+         }
         _module = module;
         _parent = parent;
         _dict = dict;
-        // Debug code
-        PdfIndirectObj parentRef = (PdfIndirectObj) _dict.get ("Parent");
     }
 
     /**
@@ -99,19 +100,13 @@ public PdfDictionary getDict ()
      */
     public PdfDictionary getResources () throws PdfException
     {
-        if (_dict == null) {
-            throw new PdfMalformedException (MessageConstants.ERR_DOC_NODE_DICT_MISSING);
-        }
+        PdfObject resdict = _dict.get ("Resources");
         try {
-            PdfObject resdict = _dict.get ("Resources");
             resdict = _module.resolveIndirectObject (resdict);
             return (PdfDictionary) resdict;
         }
-        catch (ClassCastException e) {
-            throw new PdfInvalidException (MessageConstants.ERR_RESOURCES_ENTRY_INVALID);
-        }
-        catch (IOException f) {
-            throw new PdfInvalidException (MessageConstants.ERR_RESOURCES_ENTRY_INVALID);
+        catch (ClassCastException | IOException f) {
+            throw new PdfInvalidException(MessageConstants.ERR_RESOURCES_ENTRY_INVALID); // PDF-HUL-5
         }
     }
 
@@ -126,18 +121,18 @@ public PdfDictionary getResources () throws PdfException
     public PdfDictionary getFontResources () throws PdfException
     {
         PdfDictionary resdict = getResources ();
-        if (resdict != null) {
-            try {
-                PdfObject fontdict = (PdfObject) resdict.get("Font");
-                fontdict = _module.resolveIndirectObject (fontdict);
-                return (PdfDictionary) fontdict;
-            }
-            catch (Exception e) {
-                throw new PdfMalformedException 
-                        (MessageConstants.ERR_RESOURCES_FONT_ENTRY_INVALID);
-            }
+        if (resdict == null) {
+            return null;
+        }
+        PdfObject fontdict = resdict.get("Font");
+        try {
+            fontdict = _module.resolveIndirectObject (fontdict);
+            return (PdfDictionary) fontdict;
+        }
+        catch (ClassCastException | IOException e) {
+            throw new PdfMalformedException
+                    (MessageConstants.ERR_RESOURCES_FONT_ENTRY_INVALID); // PDF-HUL-6
         }
-        return null;
     }
 
     /**
@@ -148,17 +143,18 @@ public PdfDictionary getFontResources () throws PdfException
      */
     public PdfArray getMediaBox () throws PdfInvalidException
     {
+        PdfArray mbox = null;
         try {
-	    PdfArray mbox = (PdfArray) get ("MediaBox", true);
-            if (mbox.toRectangle () != null) {
-                return mbox;
-            }
-            // There's a MediaBox, but it's not a rectangle
-            throw new PdfInvalidException (MessageConstants.ERR_PAGE_TREE_MEDIA_BOX_MALFORMED);
+            mbox = (PdfArray) get ("MediaBox", true);
         }
-        catch (Exception e) {
-            throw new PdfInvalidException (MessageConstants.ERR_PAGE_TREE_MEDIA_BOX_MALFORMED);
+        catch (ClassCastException e) {
+            throw new PdfInvalidException (MessageConstants.ERR_PAGE_TREE_MEDIA_BOX_MALFORMED); // PDF-HUL-7
+        }
+        if (mbox != null && mbox.toRectangle () == null) {
+            // There's a MediaBox, but it's not a rectangle
+            throw new PdfInvalidException (MessageConstants.ERR_PAGE_TREE_MEDIA_BOX_MALFORMED); // PDF-HUL-8
         }
+        return mbox;
     }
 
     /**

jhove-modules/src/main/java/edu/harvard/hul/ois/jhove/module/pdf/FileSpecification.java

@@ -21,11 +21,10 @@
  *    <LI>The Macintosh file specification string
  *  </UL>
  */
-public class FileSpecification 
+public enum FileSpecification 
 {
-    String _specString;
-    PdfObject _sourceObject;
-
+    INSTANCE;
+    private static final String[] dictKeys = {"F", "Unix", "DOS", "Mac"};
     /**
      *  Constructor.
      * 
@@ -37,50 +36,25 @@ public class FileSpecification
      *               then the first of the keys F, Unix, DOS, and Mac
      *               to be found is used.
      */
-    public FileSpecification (PdfObject obj) throws PdfException
+    public static String getFileSpecString (PdfObject obj) throws PdfInvalidException
     {
+        if (obj instanceof PdfSimpleObject) {
+            return ((PdfSimpleObject) obj).getStringValue ();
+        }
         try {
-            _sourceObject = obj;
             if (obj instanceof PdfDictionary) {
                 PdfDictionary dictObj = (PdfDictionary) obj;
-                PdfSimpleObject pathObj;
-                pathObj = (PdfSimpleObject) dictObj.get ("F");
-                if (pathObj == null) {
-                    pathObj = (PdfSimpleObject) dictObj.get ("Unix");
-                }
-                if (pathObj == null) {
-                    pathObj = (PdfSimpleObject) dictObj.get ("DOS");
-                }
-                if (pathObj == null) {
-                    pathObj = (PdfSimpleObject) dictObj.get ("Mac");
-                }
-                if (pathObj != null) {
-                    _specString = pathObj.getStringValue ();
+                for (final String dictKey : dictKeys) {
+                    PdfSimpleObject pathObj = (PdfSimpleObject) dictObj.get(dictKey);
+                    if (pathObj != null) {
+                        return pathObj.getStringValue();
+                    }
                 }
             }
-            else if (obj instanceof PdfSimpleObject) {
-                _specString = ((PdfSimpleObject) obj).getStringValue ();
-            }
         }
         catch (ClassCastException e) {
-            throw new PdfInvalidException (MessageConstants.ERR_FILE_SPEC_INVALID);
+            throw new PdfInvalidException(MessageConstants.ERR_FILE_SPEC_INVALID); // PDF-HUL-9
         }
-    }
-
-    /**
-     *  Returns the file specification as a string.  
-     */
-    public String getSpecString ()
-    {
-        return _specString;
-    }
-
-
-    /**
-     *   Returns the PdfObject from which the file specification was created.
-     */
-    public PdfObject getSourceObject ()
-    {
-	return _sourceObject;
+        return null;
     }
 }

jhove-modules/src/main/java/edu/harvard/hul/ois/jhove/module/pdf/FileTokenizer.java

@@ -50,13 +50,14 @@ private void initFileBuffer ()
 
     /** Gets the current position in the file.  This method is
      *  aware of buffering. */
-    public long getFilePos () throws IOException
+    public long getFilePos ()
     {
         return _fileBufferPositionOffset + _fileBufferOffset;
     }
 
 
     /** Gets a character from the file, using a buffer. */
+    @Override
     public int readChar () throws IOException
     {
         if (_fileBufferOffset >= _fileBufferBytes) {
@@ -73,14 +74,15 @@ public int readChar () throws IOException
             }
             _fileBufferOffset = 0;
         }
-        return (int) (_fileBuffer[_fileBufferOffset++] & 0XFF);
+        return (_fileBuffer[_fileBufferOffset++] & 0XFF);
     }
 
     /**
      *  Set the Tokenizer to a new position in the file.
      *
      *  @param  offset  The offset in bytes from the start of the file.
      */
+    @Override
     public void seek (long offset)
         throws IOException
     {
@@ -102,6 +104,7 @@ public void seek (long offset)
     /**
      *   Back up a byte so it will be read again.
      */
+    @Override
     public void backupChar ()
     {
         _fileBufferOffset--;
@@ -111,7 +114,8 @@ public void backupChar ()
      *  so some of the initialization of a stream object 
      *  goes here.
      */
-    protected void initStream (Stream token) throws IOException
+    @Override
+    protected void initStream (Stream token)
     {
         token.setOffset (getFilePos ());
     }
@@ -121,7 +125,8 @@ protected void initStream (Stream token) throws IOException
      *  Only the file-based tokenizer can do this, which is why this
      *  overrides the Tokenizer method. 
      */
-    protected void setStreamOffset (Stream token) throws IOException
+    @Override
+    protected void setStreamOffset (Stream token)
     {
         if (token.getOffset() < 0) {
             token.setOffset (getFilePos ());