Fix bug #45290 - Support odd files where the POIFS header block comes after the data blocks, and is on the data blocks list

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@713447 13f79535-47bb-0310-9956-ffa450edef68

7 files changed, 55 insertions, 27 deletions

diff --git a/src/java/org/apache/poi/poifs/eventfilesystem/POIFSReader.java b/src/java/org/apache/poi/poifs/eventfilesystem/POIFSReader.java
index 73911e6b0e..6a1423ea48 100644
--- a/src/java/org/apache/poi/poifs/eventfilesystem/POIFSReader.java
+++ b/src/java/org/apache/poi/poifs/eventfilesystem/POIFSReader.java
@@ -245,13 +245,13 @@ public class POIFSReader
                     {
                         document =
                             new POIFSDocument(name, small_blocks
-                                .fetchBlocks(startBlock), size);
+                                .fetchBlocks(startBlock, -1), size);
                     }
                     else
                     {
                         document =
                             new POIFSDocument(name, big_blocks
-                                .fetchBlocks(startBlock), size);
+                                .fetchBlocks(startBlock, -1), size);
                     }
                     while (listeners.hasNext())
                     {
@@ -270,11 +270,11 @@ public class POIFSReader
                     // consume the document's data and discard it
                     if (property.shouldUseSmallBlocks())
                     {
-                        small_blocks.fetchBlocks(startBlock);
+                        small_blocks.fetchBlocks(startBlock, -1);
                     }
                     else
                     {
-                        big_blocks.fetchBlocks(startBlock);
+                        big_blocks.fetchBlocks(startBlock, -1);
                     }
                 }
             }
diff --git a/src/java/org/apache/poi/poifs/filesystem/POIFSFileSystem.java b/src/java/org/apache/poi/poifs/filesystem/POIFSFileSystem.java
index 92348ceaff..6ea37ea387 100644
--- a/src/java/org/apache/poi/poifs/filesystem/POIFSFileSystem.java
+++ b/src/java/org/apache/poi/poifs/filesystem/POIFSFileSystem.java
@@ -173,11 +173,16 @@ public class POIFSFileSystem
                               data_blocks);
 
         // init documents
-        processProperties(SmallBlockTableReader
-            .getSmallDocumentBlocks(data_blocks, properties
-                .getRoot(), header_block_reader
-                    .getSBATStart()), data_blocks, properties.getRoot()
-                        .getChildren(), null);
+        processProperties(
+        		SmallBlockTableReader.getSmallDocumentBlocks(
+        				data_blocks, properties.getRoot(), 
+        				header_block_reader.getSBATStart()
+        		), 
+        		data_blocks, 
+        		properties.getRoot().getChildren(), 
+        		null,
+        		header_block_reader.getPropertyStart()
+        );
     }
     /**
      * @param stream the stream to be closed
@@ -491,7 +496,8 @@ public class POIFSFileSystem
     private void processProperties(final BlockList small_blocks,
                                    final BlockList big_blocks,
                                    final Iterator properties,
-                                   final DirectoryNode dir)
+                                   final DirectoryNode dir,
+                                   final int headerPropertiesStartAt)
         throws IOException
     {
         while (properties.hasNext())
@@ -511,7 +517,8 @@ public class POIFSFileSystem
 
                 processProperties(
                     small_blocks, big_blocks,
-                    (( DirectoryProperty ) property).getChildren(), new_dir);
+                    (( DirectoryProperty ) property).getChildren(), 
+                    new_dir, headerPropertiesStartAt);
             }
             else
             {
@@ -522,14 +529,15 @@ public class POIFSFileSystem
                 if (property.shouldUseSmallBlocks())
                 {
                     document =
-                        new POIFSDocument(name, small_blocks
-                            .fetchBlocks(startBlock), size);
+                        new POIFSDocument(name, 
+                                          small_blocks.fetchBlocks(startBlock, headerPropertiesStartAt), 
+                                          size);
                 }
                 else
                 {
                     document =
                         new POIFSDocument(name,
-                                          big_blocks.fetchBlocks(startBlock),
+                                          big_blocks.fetchBlocks(startBlock, headerPropertiesStartAt),
                                           size);
                 }
                 parent.createDocument(document);
diff --git a/src/java/org/apache/poi/poifs/property/PropertyTable.java b/src/java/org/apache/poi/poifs/property/PropertyTable.java
index 11c56e886d..00b306b792 100644
--- a/src/java/org/apache/poi/poifs/property/PropertyTable.java
+++ b/src/java/org/apache/poi/poifs/property/PropertyTable.java
@@ -78,7 +78,7 @@ public class PropertyTable
         _blocks      = null;
         _properties  =
             PropertyFactory
-                .convertToProperties(blockList.fetchBlocks(startBlock));
+                .convertToProperties(blockList.fetchBlocks(startBlock, -1));
         populatePropertyTree(( DirectoryProperty ) _properties.get(0));
     }
 
diff --git a/src/java/org/apache/poi/poifs/storage/BlockAllocationTableReader.java b/src/java/org/apache/poi/poifs/storage/BlockAllocationTableReader.java
index b167d2fca1..1016db9c46 100644
--- a/src/java/org/apache/poi/poifs/storage/BlockAllocationTableReader.java
+++ b/src/java/org/apache/poi/poifs/storage/BlockAllocationTableReader.java
@@ -179,17 +179,34 @@ public class BlockAllocationTableReader
      */
 
     ListManagedBlock [] fetchBlocks(final int startBlock,
+                                    final int headerPropertiesStartBlock,
                                     final BlockList blockList)
         throws IOException
     {
         List blocks       = new ArrayList();
         int  currentBlock = startBlock;
-
-        while (currentBlock != POIFSConstants.END_OF_CHAIN)
-        {
-            blocks.add(blockList.remove(currentBlock));
-            currentBlock = _entries.get(currentBlock);
+        boolean firstPass = true;
+        
+        // Process the chain from the start to the end
+        // Normally we have header, data, end
+        // Sometimes we have data, header, end
+        // For those cases, stop at the header, not the end
+        while (currentBlock != POIFSConstants.END_OF_CHAIN) {
+        	try {
+        		blocks.add(blockList.remove(currentBlock));
+                currentBlock = _entries.get(currentBlock);
+        	} catch(IOException e) {
+        		if(currentBlock == headerPropertiesStartBlock) {
+        			// Special case where things are in the wrong order
+        			System.err.println("Warning, header block comes after data blocks in POIFS block listing");
+        			currentBlock = POIFSConstants.END_OF_CHAIN;
+        		} else {
+        			// Ripple up
+        			throw e;
+        		}
+        	}
         }
+        
         return ( ListManagedBlock [] ) blocks
             .toArray(new ListManagedBlock[ 0 ]);
     }
diff --git a/src/java/org/apache/poi/poifs/storage/BlockList.java b/src/java/org/apache/poi/poifs/storage/BlockList.java
index 2352d3a2c1..a5eb7df217 100644
--- a/src/java/org/apache/poi/poifs/storage/BlockList.java
+++ b/src/java/org/apache/poi/poifs/storage/BlockList.java
@@ -59,13 +59,14 @@ public interface BlockList
      * blocks are removed from the list.
      *
      * @param startBlock the index of the first block in the stream
+     * @param headerPropertiesStartBlock the index of the first header block in the stream
      *
      * @return the stream as an array of correctly ordered blocks
      *
      * @exception IOException if blocks are missing
      */
 
-    public ListManagedBlock [] fetchBlocks(final int startBlock)
+    public ListManagedBlock [] fetchBlocks(final int startBlock, final int headerPropertiesStartBlock)
         throws IOException;
 
     /**
diff --git a/src/java/org/apache/poi/poifs/storage/BlockListImpl.java b/src/java/org/apache/poi/poifs/storage/BlockListImpl.java
index 7e44fda3f3..f315b5d515 100644
--- a/src/java/org/apache/poi/poifs/storage/BlockListImpl.java
+++ b/src/java/org/apache/poi/poifs/storage/BlockListImpl.java
@@ -94,8 +94,10 @@ class BlockListImpl
             result = _blocks[ index ];
             if (result == null)
             {
-                throw new IOException("block[ " + index
-                                      + " ] already removed");
+                throw new IOException(
+                		"block[ " + index + " ] already removed - " +
+                		"does your POIFS have circular or duplicate block references?"
+                );
             }
             _blocks[ index ] = null;
         }
@@ -119,7 +121,7 @@ class BlockListImpl
      * @exception IOException if blocks are missing
      */
 
-    public ListManagedBlock [] fetchBlocks(final int startBlock)
+    public ListManagedBlock [] fetchBlocks(final int startBlock, final int headerPropertiesStartBlock)
         throws IOException
     {
         if (_bat == null)
@@ -127,7 +129,7 @@ class BlockListImpl
             throw new IOException(
                 "Improperly initialized list: no block allocation table provided");
         }
-        return _bat.fetchBlocks(startBlock, this);
+        return _bat.fetchBlocks(startBlock, headerPropertiesStartBlock, this);
     }
 
     /**
diff --git a/src/java/org/apache/poi/poifs/storage/SmallBlockTableReader.java b/src/java/org/apache/poi/poifs/storage/SmallBlockTableReader.java
index e70774aa16..4da79e71f1 100644
--- a/src/java/org/apache/poi/poifs/storage/SmallBlockTableReader.java
+++ b/src/java/org/apache/poi/poifs/storage/SmallBlockTableReader.java
@@ -56,9 +56,9 @@ public class SmallBlockTableReader
     {
         BlockList list =
             new SmallDocumentBlockList(SmallDocumentBlock
-                .extract(blockList.fetchBlocks(root.getStartBlock())));
+                .extract(blockList.fetchBlocks(root.getStartBlock(), -1)));
 
-        new BlockAllocationTableReader(blockList.fetchBlocks(sbatStart),
+        new BlockAllocationTableReader(blockList.fetchBlocks(sbatStart, -1),
                                        list);
         return list;
     }