Help for bug #44840 - Improved handling of HSSFObjectData, especially for entries with data held not in POIFS

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@659575 13f79535-47bb-0310-9956-ffa450edef68

src/documentation/content/xdocs/changes.xml

@@ -37,6 +37,7 @@
 
 		<!-- Don't forget to update status.xml too! -->
         <release version="3.1-final" date="2008-06-??">
+           <action dev="POI-DEVELOPERS" type="add">44840 - Improved handling of HSSFObjectData, especially for entries with data held not in POIFS</action>
            <action dev="POI-DEVELOPERS" type="add">45043 - Support for getting excel cell comments when extracting text</action>
            <action dev="POI-DEVELOPERS" type="add">Extend the support for specifying a policy to HSSF on missing / blank cells when fetching, to be able to specify the policy at the HSSFWorkbook level</action>
            <action dev="POI-DEVELOPERS" type="fix">45025 - improved FormulaParser parse error messages</action>

src/documentation/content/xdocs/status.xml

@@ -34,6 +34,7 @@
 	<!-- Don't forget to update changes.xml too! -->
     <changes>
         <release version="3.1-final" date="2008-06-??">
+           <action dev="POI-DEVELOPERS" type="add">44840 - Improved handling of HSSFObjectData, especially for entries with data held not in POIFS</action>
            <action dev="POI-DEVELOPERS" type="add">45043 - Support for getting excel cell comments when extracting text</action>
            <action dev="POI-DEVELOPERS" type="add">Extend the support for specifying a policy to HSSF on missing / blank cells when fetching, to be able to specify the policy at the HSSFWorkbook level</action>
            <action dev="POI-DEVELOPERS" type="fix">45025 - improved FormulaParser parse error messages</action>

src/java/org/apache/poi/hssf/record/EmbeddedObjectRefSubRecord.java

@@ -108,7 +108,10 @@ public class EmbeddedObjectRefSubRecord
             in.readByte(); // discard
         }
 
-        field_6_stream_id              = in.readInt();
+        // Fetch the stream ID
+        field_6_stream_id = in.readInt();
+        
+        // Store what's left
         remainingBytes = in.readRemainder();
     }
 

src/java/org/apache/poi/hssf/usermodel/HSSFObjectData.java

@@ -55,36 +55,72 @@ public class HSSFObjectData
         this.record = record;
         this.poifs = poifs;
     }
+    
+    /**
+     * Returns the OLE2 Class Name of the object
+     */
+    public String getOLE2ClassName() {
+    	EmbeddedObjectRefSubRecord subRecord = findObjectRecord();
+    	return subRecord.field_5_ole_classname;
+    }
 
     /**
-     * Gets the object data.
+     * Gets the object data. Only call for ones that have
+     *  data though. See {@link #hasDirectoryEntry()}
      *
      * @return the object data as an OLE2 directory.
      * @throws IOException if there was an error reading the data.
      */
-    public DirectoryEntry getDirectory() throws IOException
-    {
-        Iterator subRecordIter = record.getSubRecords().iterator();
-        while (subRecordIter.hasNext())
-        {
-            Object subRecord = subRecordIter.next();
-            if (subRecord instanceof EmbeddedObjectRefSubRecord)
-            {
-                int streamId = ((EmbeddedObjectRefSubRecord) subRecord).getStreamId();
-                String streamName = "MBD" + HexDump.toHex(streamId);
+    public DirectoryEntry getDirectory() throws IOException {
+    	EmbeddedObjectRefSubRecord subRecord = findObjectRecord();
 
-                Entry entry = poifs.getRoot().getEntry(streamName);
-                if (entry instanceof DirectoryEntry)
-                {
-                    return (DirectoryEntry) entry;
-                }
-                else
-                {
-                    throw new IOException("Stream " + streamName + " was not an OLE2 directory");
-                }
+    	int streamId = ((EmbeddedObjectRefSubRecord) subRecord).getStreamId();
+        String streamName = "MBD" + HexDump.toHex(streamId);
+
+        Entry entry = poifs.getRoot().getEntry(streamName);
+        if (entry instanceof DirectoryEntry) {
+            return (DirectoryEntry) entry;
+        } else {
+            throw new IOException("Stream " + streamName + " was not an OLE2 directory");
+        }
+    }
+    
+    /**
+     * Returns the data portion, for an ObjectData
+     *  that doesn't have an associated POIFS Directory
+     *  Entry
+     */
+    public byte[] getObjectData() {
+    	EmbeddedObjectRefSubRecord subRecord = findObjectRecord();
+    	return subRecord.remainingBytes;
+    }
+    
+    /**
+     * Does this ObjectData have an associated POIFS 
+     *  Directory Entry?
+     * (Not all do, those that don't have a data portion)
+     */
+    public boolean hasDirectoryEntry() {
+    	EmbeddedObjectRefSubRecord subRecord = findObjectRecord();
+    	
+    	// Field 6 tells you
+    	return (subRecord.field_6_stream_id != 0);
+    }
+    
+    /**
+     * Finds the EmbeddedObjectRefSubRecord, or throws an 
+     *  Exception if there wasn't one
+     */
+    protected EmbeddedObjectRefSubRecord findObjectRecord() {
+        Iterator subRecordIter = record.getSubRecords().iterator();
+        
+        while (subRecordIter.hasNext()) {
+            Object subRecord = subRecordIter.next();
+            if (subRecord instanceof EmbeddedObjectRefSubRecord) {
+            	return (EmbeddedObjectRefSubRecord)subRecord;
             }
         }
-
+        
         throw new IllegalStateException("Object data does not contain a reference to an embedded object OLE2 directory");
     }
 }

src/testcases/org/apache/poi/hssf/usermodel/TestBugs.java

@@ -18,14 +18,17 @@
 package org.apache.poi.hssf.usermodel;
 
 import java.io.File;
+import java.io.FileNotFoundException;
 import java.io.FileOutputStream;
 import java.io.IOException;
 import java.util.Iterator;
+import java.util.List;
 
 import junit.framework.AssertionFailedError;
 import junit.framework.TestCase;
 
 import org.apache.poi.hssf.HSSFTestDataSamples;
+import org.apache.poi.hssf.record.EmbeddedObjectRefSubRecord;
 import org.apache.poi.hssf.util.Region;
 import org.apache.poi.util.TempFile;
 
@@ -950,4 +953,40 @@ public final class TestBugs extends TestCase {
         writeOutAndReadBack(wb);
         assertTrue("no errors writing sample xls", true);
     }
+    
+    /**
+     * Problems with extracting check boxes from
+     *  HSSFObjectData
+     * @throws Exception
+     */
+    public void test44840() throws Exception {
+        HSSFWorkbook wb = openSample("WithCheckBoxes.xls");
+
+        // Take a look at the embeded objects
+        List objects = wb.getAllEmbeddedObjects();
+        assertEquals(1, objects.size());
+        
+        HSSFObjectData obj = (HSSFObjectData)objects.get(0);
+        assertNotNull(obj);
+        
+        // Peek inside the underlying record
+        EmbeddedObjectRefSubRecord rec = obj.findObjectRecord();
+        assertNotNull(rec);
+        
+        assertEquals(32, rec.field_1_stream_id_offset);
+        assertEquals(0, rec.field_6_stream_id); // WRONG!
+        assertEquals("Forms.CheckBox.1", rec.field_5_ole_classname);
+        assertEquals(12, rec.remainingBytes.length);
+        
+        // Doesn't have a directory
+        assertFalse(obj.hasDirectoryEntry());
+        assertNotNull(obj.getObjectData());
+        assertEquals(12, obj.getObjectData().length);
+        assertEquals("Forms.CheckBox.1", obj.getOLE2ClassName());
+        
+        try {
+        	obj.getDirectory();
+        	fail();
+        } catch(FileNotFoundException e) {}
+    }
 }