Improve how POIFS works with directory entries, and update HWPFDocument to support reading an embeded word document

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@646870 13f79535-47bb-0310-9956-ffa450edef68

build.xml

@@ -521,6 +521,8 @@ under the License.
         file="${main.src.test}/org/apache/poi/hwpf/data"/> 
       <sysproperty key="HPSF.testdata.path"
         file="${main.src.test}/org/apache/poi/hpsf/data"/> 
+      <sysproperty key="POIFS.testdata.path"
+        file="${main.src.test}/org/apache/poi/poifs/data"/> 
       <sysproperty key="java.awt.headless" value="true"/>
       <formatter type="plain"/>
       <formatter type="xml"/>
@@ -556,6 +558,8 @@ under the License.
         file="${main.src.test}/org/apache/poi/hpsf/data"/>
       <sysproperty key="HWPF.testdata.path"
         file="${scratchpad.src.test}/org/apache/poi/hwpf/data"/>
+      <sysproperty key="POIFS.testdata.path"
+        file="${main.src.test}/org/apache/poi/poifs/data"/> 
       <sysproperty key="java.awt.headless" value="true"/>
       <formatter type="plain" usefile="no"/>
       <batchtest todir="${main.reports.test}">
@@ -585,6 +589,7 @@ under the License.
             <sysproperty key="HWPF.testdata.path" file="${scratchpad.src.test}/org/apache/poi/hwpf/data"/>
             <sysproperty key="HSMF.testdata.path" file="${scratchpad.src.test}/org/apache/poi/hsmf/data"/>
             <sysproperty key="HDGF.testdata.path" file="${scratchpad.src.test}/org/apache/poi/hdgf/data"/>
+            <sysproperty key="POIFS.testdata.path" file="${main.src.test}/org/apache/poi/poifs/data"/> 
                 <sysproperty key="java.awt.headless" value="true"/>
             <formatter type="plain" usefile="no"/>
             <formatter type="xml"/>
@@ -601,6 +606,7 @@ under the License.
             <classpath refid="test.classpath"/>
             <sysproperty key="HSSF.testdata.path" file="${main.src.test}/org/apache/poi/hssf/data"/>
             <sysproperty key="HPSF.testdata.path" file="${main.src.test}/org/apache/poi/hpsf/data"/>
+            <sysproperty key="POIFS.testdata.path" file="${main.src.test}/org/apache/poi/poifs/data"/> 
             <sysproperty key="java.awt.headless" value="true"/>
             <formatter type="plain" usefile="no"/>
             <test name="${testcase}"/>
@@ -639,6 +645,7 @@ under the License.
             <sysproperty key="HSLF.testdata.path" file="${scratchpad.src.test}/org/apache/poi/hslf/data"/>
             <sysproperty key="HSMF.testdata.path" file="${scratchpad.src.test}/org/apache/poi/hsmf/data"/>
             <sysproperty key="HDGF.testdata.path" file="${scratchpad.src.test}/org/apache/poi/hdgf/data"/>
+            <sysproperty key="POIFS.testdata.path" file="${main.src.test}/org/apache/poi/poifs/data"/> 
             <sysproperty key="java.awt.headless" value="true"/>
             <formatter type="plain"/>
             <formatter type="xml"/>
@@ -673,6 +680,7 @@ under the License.
                 <sysproperty key="HSLF.testdata.path" file="${scratchpad.src.test}/org/apache/poi/hslf/data"/>
                 <sysproperty key="HSMF.testdata.path" file="${scratchpad.src.test}/org/apache/poi/hsmf/data"/>
                 <sysproperty key="HDGF.testdata.path" file="${scratchpad.src.test}/org/apache/poi/hdgf/data"/>
+                <sysproperty key="POIFS.testdata.path" file="${main.src.test}/org/apache/poi/poifs/data"/> 
 	            <sysproperty key="java.awt.headless" value="true"/>
 	                <sysproperty key="java.awt.headless" value="true"/>
 	            <formatter type="plain" usefile="no"/>

src/documentation/content/xdocs/changes.xml

@@ -37,6 +37,7 @@
 
 		<!-- Don't forget to update status.xml too! -->
         <release version="3.0.3-beta1" date="2008-04-??">
+           <action dev="POI-DEVELOPERS" type="add">Improve how POIFS works with directory entries, and update HWPFDocument to support reading an embeded word document</action>
            <action dev="POI-DEVELOPERS" type="add">Initial support for getting and changing chart and series titles</action>
            <action dev="POI-DEVELOPERS" type="add">Implement a proxy HSSFListener which tracks the format records, and lets you lookup the format string for a given cell. Convert the xls to csv example to use it</action>
            <action dev="POI-DEVELOPERS" type="fix">44792 - fixed encode/decode problems in ExternalNameRecord and CRNRecord.</action>

src/documentation/content/xdocs/status.xml

@@ -34,6 +34,7 @@
 	<!-- Don't forget to update changes.xml too! -->
     <changes>
         <release version="3.0.3-beta1" date="2008-04-??">
+           <action dev="POI-DEVELOPERS" type="add">Improve how POIFS works with directory entries, and update HWPFDocument to support reading an embeded word document</action>
            <action dev="POI-DEVELOPERS" type="add">Initial support for getting and changing chart and series titles</action>
            <action dev="POI-DEVELOPERS" type="add">Implement a proxy HSSFListener which tracks the format records, and lets you lookup the format string for a given cell. Convert the xls to csv example to use it</action>
            <action dev="POI-DEVELOPERS" type="fix">44792 - fixed encode/decode problems in ExternalNameRecord and CRNRecord.</action>

src/java/org/apache/poi/POIDocument.java

@@ -29,6 +29,7 @@ import org.apache.poi.hpsf.PropertySet;
 import org.apache.poi.hpsf.PropertySetFactory;
 import org.apache.poi.hpsf.SummaryInformation;
 import org.apache.poi.poifs.filesystem.DirectoryEntry;
+import org.apache.poi.poifs.filesystem.DirectoryNode;
 import org.apache.poi.poifs.filesystem.DocumentEntry;
 import org.apache.poi.poifs.filesystem.DocumentInputStream;
 import org.apache.poi.poifs.filesystem.Entry;
@@ -50,6 +51,8 @@ public abstract class POIDocument {
 	protected DocumentSummaryInformation dsInf;
 	/** The open POIFS FileSystem that contains our document */
 	protected POIFSFileSystem filesystem;
+	/**	The directory that our document lives in */
+	protected DirectoryNode directory;
 	
 	/** For our own logging use */
 	protected POILogger logger = POILogFactory.getLogger(this.getClass());
@@ -57,6 +60,15 @@ public abstract class POIDocument {
     /* Have the property streams been read yet? (Only done on-demand) */
     protected boolean initialized = false;
     
+
+    protected POIDocument(DirectoryNode dir, POIFSFileSystem fs) {
+    	this.filesystem = fs;
+    	this.directory = dir;
+    }
+    protected POIDocument(POIFSFileSystem fs) {
+    	this(fs.getRoot(), fs);
+    }
+
 	/**
 	 * Fetch the Document Summary Information of the document
 	 */
@@ -110,7 +122,7 @@ public abstract class POIDocument {
 		DocumentInputStream dis;
 		try {
 			// Find the entry, and get an input stream for it
-			dis = filesystem.createDocumentInputStream(setName);
+			dis = directory.createDocumentInputStream(setName);
 		} catch(IOException ie) {
 			// Oh well, doesn't exist
 			logger.log(POILogger.WARN, "Error getting property set with name " + setName + "\n" + ie);

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

@@ -139,6 +139,7 @@ public class HSSFWorkbook extends POIDocument
 
     protected HSSFWorkbook( Workbook book )
     {
+    	super(null, null);
         workbook = book;
         sheets = new ArrayList( INITIAL_CAPACITY );
         names = new ArrayList( INITIAL_CAPACITY );
@@ -164,8 +165,8 @@ public class HSSFWorkbook extends POIDocument
     public HSSFWorkbook(POIFSFileSystem fs, boolean preserveNodes)
             throws IOException
     {
+    	super(fs);
         this.preserveNodes = preserveNodes;
-        this.filesystem = fs;
         
         // If we're not preserving nodes, don't track the
         //  POIFS any more

src/java/org/apache/poi/poifs/filesystem/DirectoryNode.java

@@ -106,6 +106,31 @@ public class DirectoryNode
         return _path;
     }
     
+    /**
+     * open a document in the directory's entry's list of entries
+     *
+     * @param documentName the name of the document to be opened
+     *
+     * @return a newly opened DocumentInputStream
+     *
+     * @exception IOException if the document does not exist or the
+     *            name is that of a DirectoryEntry
+     */
+
+    public DocumentInputStream createDocumentInputStream(
+            final String documentName)
+        throws IOException
+    {
+        Entry document = getEntry(documentName);
+
+        if (!document.isDocumentEntry())
+        {
+            throw new IOException("Entry '" + documentName
+                                  + "' is not a DocumentEntry");
+        }
+        return new DocumentInputStream(( DocumentEntry ) document);
+    }
+
     /**
      * create a new DocumentEntry
      *

src/java/org/apache/poi/poifs/filesystem/POIFSFileSystem.java

@@ -422,7 +422,7 @@ public class POIFSFileSystem
      * @return the root entry
      */
 
-    public DirectoryEntry getRoot()
+    public DirectoryNode getRoot()
     {
         if (_root == null)
         {
@@ -446,14 +446,7 @@ public class POIFSFileSystem
             final String documentName)
         throws IOException
     {
-        Entry document = getRoot().getEntry(documentName);
+    	return getRoot().createDocumentInputStream(documentName);
-
-        if (!document.isDocumentEntry())
-        {
-            throw new IOException("Entry '" + documentName
-                                  + "' is not a DocumentEntry");
-        }
-        return new DocumentInputStream(( DocumentEntry ) document);
     }
 
     /**

src/scratchpad/src/org/apache/poi/hdgf/HDGFDiagram.java

@@ -53,7 +53,7 @@ public class HDGFDiagram extends POIDocument {
 	private PointerFactory ptrFactory;
 	
 	public HDGFDiagram(POIFSFileSystem fs) throws IOException {
-		filesystem = fs;
+		super(fs);
 		
 		DocumentEntry docProps =
 			(DocumentEntry)filesystem.getRoot().getEntry("VisioDocument");

src/scratchpad/src/org/apache/poi/hslf/HSLFSlideShow.java

@@ -124,7 +124,7 @@ public class HSLFSlideShow extends POIDocument
 	 */
 	public HSLFSlideShow(POIFSFileSystem filesystem) throws IOException
 	{
-		this.filesystem = filesystem;
+		super(filesystem);
 		
 		// First up, grab the "Current User" stream
 		// We need this before we can detect Encrypted Documents

src/scratchpad/src/org/apache/poi/hwpf/HWPFDocument.java

@@ -29,6 +29,7 @@ import java.io.ByteArrayInputStream;
 import java.util.Iterator;
 
 import org.apache.poi.POIDocument;
+import org.apache.poi.poifs.filesystem.DirectoryNode;
 import org.apache.poi.poifs.filesystem.POIFSFileSystem;
 import org.apache.poi.poifs.filesystem.DocumentEntry;
 import org.apache.poi.poifs.common.POIFSConstants;
@@ -95,7 +96,7 @@ public class HWPFDocument extends POIDocument
 
   protected HWPFDocument()
   {
-
+     super(null, null);
   }
 
   /**
@@ -141,16 +142,31 @@ public class HWPFDocument extends POIDocument
    *         in POIFSFileSystem.
    */
   public HWPFDocument(POIFSFileSystem pfilesystem) throws IOException
+  {
+	this(pfilesystem.getRoot(), pfilesystem);
+  }
+  
+  /**
+   * This constructor loads a Word document from a specific point
+   *  in a POIFSFileSystem, probably not the default.
+   * Used typically to open embeded documents.
+   *
+   * @param pfilesystem The POIFSFileSystem that contains the Word document.
+   * @throws IOException If there is an unexpected IOException from the passed
+   *         in POIFSFileSystem.
+   */
+  public HWPFDocument(DirectoryNode directory, POIFSFileSystem pfilesystem) throws IOException
   {
     // Sort out the hpsf properties
-    filesystem = pfilesystem;
+	super(directory, pfilesystem);
     readProperties();
     
     // read in the main stream.
-    DocumentEntry documentProps =
+    DocumentEntry documentProps = (DocumentEntry)
-       (DocumentEntry)filesystem.getRoot().getEntry("WordDocument");
+       directory.getEntry("WordDocument");
     _mainStream = new byte[documentProps.getSize()];
-    filesystem.createDocumentInputStream("WordDocument").read(_mainStream);
+    
+    directory.createDocumentInputStream("WordDocument").read(_mainStream);
 
     // use the fib to determine the name of the table stream.
     _fib = new FileInformationBlock(_mainStream);
@@ -165,14 +181,14 @@ public class HWPFDocument extends POIDocument
     DocumentEntry tableProps;
 	try {
 		tableProps =
-			(DocumentEntry)filesystem.getRoot().getEntry(name);
+			(DocumentEntry)directory.getEntry(name);
 	} catch(FileNotFoundException fnfe) {
 		throw new IllegalStateException("Table Stream '" + name + "' wasn't found - Either the document is corrupt, or is Word95 (or earlier)");
 	}
 
     // read in the table stream.
     _tableStream = new byte[tableProps.getSize()];
-    filesystem.createDocumentInputStream(name).read(_tableStream);
+    directory.createDocumentInputStream(name).read(_tableStream);
 
     _fib.fillVariableFields(_mainStream, _tableStream);
 
@@ -180,7 +196,7 @@ public class HWPFDocument extends POIDocument
     try
     {
       DocumentEntry dataProps =
-          (DocumentEntry) filesystem.getRoot().getEntry("Data");
+          (DocumentEntry)directory.getEntry("Data");
       _dataStream = new byte[dataProps.getSize()];
       filesystem.createDocumentInputStream("Data").read(_dataStream);
     }

src/scratchpad/testcases/org/apache/poi/hwpf/extractor/TestWordExtractor.java

@@ -23,6 +23,8 @@ import org.apache.poi.hwpf.HWPFDocument;
 import org.apache.poi.hwpf.model.TextPiece;
 import org.apache.poi.hwpf.usermodel.Paragraph;
 import org.apache.poi.hwpf.usermodel.Range;
+import org.apache.poi.poifs.filesystem.DirectoryNode;
+import org.apache.poi.poifs.filesystem.POIFSFileSystem;
 
 import junit.framework.TestCase;
 
@@ -54,12 +56,16 @@ public class TestWordExtractor extends TestCase {
 	private WordExtractor extractor;
 	// Corrupted document - can't do paragraph based stuff
 	private WordExtractor extractor2;
+	// A word doc embeded in an excel file
+	private String filename3;
 	
     protected void setUp() throws Exception {
 		String dirname = System.getProperty("HWPF.testdata.path");
+		String pdirname = System.getProperty("POIFS.testdata.path");
 		
 		String filename = dirname + "/test2.doc";
 		String filename2 = dirname + "/test.doc";
+		filename3 = pdirname + "/excel_with_embeded.xls";
 		extractor = new WordExtractor(new FileInputStream(filename));
 		extractor2 = new WordExtractor(new FileInputStream(filename2));
 		
@@ -101,4 +107,25 @@ public class TestWordExtractor extends TestCase {
     	String text = extractor.getTextFromPieces();
     	assertEquals(p_text1_block, text);
     }
+    
+    
+    /**
+     * Test that we can get data from an
+     *  embeded word document
+     * @throws Exception
+     */
+    public void testExtractFromEmbeded() throws Exception {
+    	POIFSFileSystem fs = new POIFSFileSystem(new FileInputStream(filename3));
+    	DirectoryNode dir = (DirectoryNode)
+    		fs.getRoot().getEntry("MBD03F25D8D");
+    	// Should have WordDocument and 1Table
+    	assertNotNull(dir.getEntry("1Table"));
+    	assertNotNull(dir.getEntry("WordDocument"));
+    	
+    	HWPFDocument doc = new HWPFDocument(dir, fs);
+    	WordExtractor extractor3 = new WordExtractor(doc);
+		
+    	assertNotNull(extractor3.getText());
+    	assertTrue(extractor3.getText().length() > 20);
+    }
 }