Author: tilman
Date: Thu Mar 13 17:51:32 2014
New Revision: 1577261
URL: http://svn.apache.org/r1577261
Log:
PDFBOX-1964: support non-sequential parser
Modified:
pdfbox/branches/1.8/pdfbox/src/main/java/org/apache/pdfbox/util/PDFMergerUtility.java
Modified:
pdfbox/branches/1.8/pdfbox/src/main/java/org/apache/pdfbox/util/PDFMergerUtility.java
URL:
http://svn.apache.org/viewvc/pdfbox/branches/1.8/pdfbox/src/main/java/org/apache/pdfbox/util/PDFMergerUtility.java?rev=1577261&r1=1577260&r2=1577261&view=diff
==============================================================================
---
pdfbox/branches/1.8/pdfbox/src/main/java/org/apache/pdfbox/util/PDFMergerUtility.java
(original)
+++
pdfbox/branches/1.8/pdfbox/src/main/java/org/apache/pdfbox/util/PDFMergerUtility.java
Thu Mar 13 17:51:32 2014
@@ -35,6 +35,7 @@ import org.apache.pdfbox.cos.COSNumber;
import org.apache.pdfbox.cos.COSStream;
import org.apache.pdfbox.cos.COSString;
import org.apache.pdfbox.exceptions.COSVisitorException;
+import org.apache.pdfbox.io.RandomAccess;
import org.apache.pdfbox.pdmodel.PDDocument;
import org.apache.pdfbox.pdmodel.PDDocumentCatalog;
import org.apache.pdfbox.pdmodel.PDDocumentInformation;
@@ -53,10 +54,11 @@ import org.apache.pdfbox.pdmodel.interac
import org.apache.pdfbox.pdmodel.interactive.form.PDFieldFactory;
/**
- * This class will take a list of pdf documents and merge them, saving the
result in a new document.
- *
+ * This class will take a list of pdf documents and merge them, saving the
+ * result in a new document.
+ *
* @author <a href="mailto:b...@benlitchfield.com";>Ben Litchfield</a>
- *
+ *
*/
public class PDFMergerUtility
{
@@ -77,7 +79,7 @@ public class PDFMergerUtility
/**
* Get the name of the destination file.
- *
+ *
* @return Returns the destination.
*/
public String getDestinationFileName()
@@ -87,7 +89,7 @@ public class PDFMergerUtility
/**
* Set the name of the destination file.
- *
+ *
* @param destination The destination to set.
*/
public void setDestinationFileName(String destination)
@@ -97,7 +99,7 @@ public class PDFMergerUtility
/**
* Get the destination OutputStream.
- *
+ *
* @return Returns the destination OutputStream.
*/
public OutputStream getDestinationStream()
@@ -107,7 +109,7 @@ public class PDFMergerUtility
/**
* Set the destination OutputStream.
- *
+ *
* @param destStream The destination to set.
*/
public void setDestinationStream(OutputStream destStream)
@@ -117,7 +119,7 @@ public class PDFMergerUtility
/**
* Add a source file to the list of files to merge.
- *
+ *
* @param source Full path and file name of source document.
*/
public void addSource(String source)
@@ -134,7 +136,7 @@ public class PDFMergerUtility
/**
* Add a source file to the list of files to merge.
- *
+ *
* @param source File representing source document
*/
public void addSource(File source)
@@ -151,7 +153,7 @@ public class PDFMergerUtility
/**
* Add a source to the list of documents to merge.
- *
+ *
* @param source InputStream representing source document
*/
public void addSource(InputStream source)
@@ -161,8 +163,9 @@ public class PDFMergerUtility
/**
* Add a list of sources to the list of documents to merge.
- *
- * @param sourcesList List of InputStream objects representing source
documents
+ *
+ * @param sourcesList List of InputStream objects representing source
+ * documents
*/
public void addSources(List<InputStream> sourcesList)
{
@@ -170,13 +173,37 @@ public class PDFMergerUtility
}
/**
- * Merge the list of source documents, saving the result in the
destination file.
- *
+ * Merge the list of source documents, saving the result in the destination
+ * file.
+ *
* @throws IOException If there is an error saving the document.
- * @throws COSVisitorException If an error occurs while saving the
destination file.
+ * @throws COSVisitorException If an error occurs while saving the
+ * destination file.
*/
public void mergeDocuments() throws IOException, COSVisitorException
{
+ mergeDocuments(false, null);
+ }
+
+ /**
+ * Merge the list of source documents with the non sequential parser,
saving
+ * the result in the destination file.
+ *
+ * @param scratchFile location to store temp PDFBox data for this output
+ * document
+ * @throws IOException If there is an error saving the document.
+ * @throws COSVisitorException If an error occurs while saving the
+ * destination file.
+ */
+ public void mergeDocumentsNonSeq(RandomAccess scratchFile)
+ throws IOException, COSVisitorException
+ {
+ mergeDocuments(true, scratchFile);
+ }
+
+ private void mergeDocuments(boolean isNonSeq, RandomAccess scratchFile)
+ throws IOException, COSVisitorException
+ {
PDDocument destination = null;
InputStream sourceFile;
PDDocument source;
@@ -188,12 +215,27 @@ public class PDFMergerUtility
{
Iterator<InputStream> sit = sources.iterator();
sourceFile = sit.next();
- destination = PDDocument.load(sourceFile);
+ if (isNonSeq)
+ {
+ destination = PDDocument.loadNonSeq(sourceFile,
scratchFile);
+ }
+ else
+ {
+ destination = PDDocument.load(sourceFile);
+ }
while (sit.hasNext())
{
sourceFile = sit.next();
- source = PDDocument.load(sourceFile);
+ if (isNonSeq)
+ {
+ source = PDDocument.loadNonSeq(sourceFile, null);
+ }
+ else
+ {
+ source = PDDocument.load(sourceFile);
+ }
+
tobeclosed.add(source);
appendDocument(destination, source);
}
@@ -222,13 +264,15 @@ public class PDFMergerUtility
/**
* append all pages from source to destination.
- *
+ *
* @param destination the document to receive the pages
* @param source the document originating the new pages
- *
- * @throws IOException If there is an error accessing data from either
document.
+ *
+ * @throws IOException If there is an error accessing data from either
+ * document.
*/
- public void appendDocument(PDDocument destination, PDDocument source)
throws IOException
+ public void appendDocument(PDDocument destination, PDDocument source)
+ throws IOException
{
if (destination.isEncrypted())
{
@@ -517,8 +561,9 @@ public class PDFMergerUtility
private int nextFieldNum = 1;
/**
- * Merge the contents of the source form into the destination form for the
destination file.
- *
+ * Merge the contents of the source form into the destination form for the
+ * destination file.
+ *
* @param cloner the object cloner for the destination document
* @param destAcroForm the destination form
* @param srcAcroForm the source form
@@ -555,7 +600,7 @@ public class PDFMergerUtility
/**
* Indicates if acroform errors are ignored or not.
- *
+ *
* @return true if acroform errors are ignored
*/
public boolean isIgnoreAcroFormErrors()
@@ -565,8 +610,9 @@ public class PDFMergerUtility
/**
* Set to true to ignore acroform errors.
- *
- * @param ignoreAcroFormErrorsValue true if acroform errors should be
ignored
+ *
+ * @param ignoreAcroFormErrorsValue true if acroform errors should be
+ * ignored
*/
public void setIgnoreAcroFormErrors(boolean ignoreAcroFormErrorsValue)
{
@@ -575,7 +621,7 @@ public class PDFMergerUtility
/**
* Update the Pg and Obj references to the new (merged) page.
- *
+ *
* @param parentTreeEntry
* @param objMapping mapping between old and new references
*/
@@ -626,7 +672,7 @@ public class PDFMergerUtility
/**
* Update the P reference to the new parent dictionary.
- *
+ *
* @param kArray the kids array
* @param newParent the new parent
*/
@@ -648,7 +694,7 @@ public class PDFMergerUtility
/**
* Update the StructParents and StructParent values in a PDPage.
- *
+ *
* @param page the new page
* @param structParentOffset the offset which should be applied
*/
