com.lowagie.text.pdf.PdfCopy.java Source code

Java tutorial

Introduction

Here is the source code for com.lowagie.text.pdf.PdfCopy.java

Source

/*
 * $Id: PdfCopy.java 3912 2009-04-26 08:38:15Z blowagie $
 *
 * Copyright (C) 2002 Mark Thompson
 *
 * The contents of this file are subject to the Mozilla Public License Version 1.1
 * (the "License"); you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at http://www.mozilla.org/MPL/
 *
 * Software distributed under the License is distributed on an "AS IS" basis,
 * WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
 * for the specific language governing rights and limitations under the License.
 *
 * The Original Code is 'iText, a free JAVA-PDF library'.
 *
 * The Initial Developer of the Original Code is Bruno Lowagie. Portions created by
 * the Initial Developer are Copyright (C) 1999, 2000, 2001, 2002 by Bruno Lowagie.
 * All Rights Reserved.
 * Co-Developer of the code is Paulo Soares. Portions created by the Co-Developer
 * are Copyright (C) 2000, 2001, 2002 by Paulo Soares. All Rights Reserved.
 *
 * Contributor(s): all the names of the contributors are added in the source code
 * where applicable.
 *
 * Alternatively, the contents of this file may be used under the terms of the
 * LGPL license (the "GNU LIBRARY GENERAL PUBLIC LICENSE"), in which case the
 * provisions of LGPL are applicable instead of those above.  If you wish to
 * allow use of your version of this file only under the terms of the LGPL
 * License and not to allow others to use your version of this file under
 * the MPL, indicate your decision by deleting the provisions above and
 * replace them with the notice and other provisions required by the LGPL.
 * If you do not delete the provisions above, a recipient may use your version
 * of this file under either the MPL or the GNU LIBRARY GENERAL PUBLIC LICENSE.
 *
 * This library is free software; you can redistribute it and/or modify it
 * under the terms of the MPL as stated above or under the terms of the GNU
 * Library General Public License as published by the Free Software Foundation;
 * either version 2 of the License, or any later version.
 *
 * This library is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
 * FOR A PARTICULAR PURPOSE. See the GNU Library general Public License for more
 * details.
 *
 * If you didn't download this code from the following link, you should check if
 * you aren't using an obsolete version:
 * http://www.lowagie.com/iText/
 */
package com.lowagie.text.pdf;

import java.io.IOException;
import java.io.OutputStream;
import java.util.HashMap;
import java.util.Iterator;

import com.lowagie.text.Document;
import com.lowagie.text.DocumentException;
import com.lowagie.text.ExceptionConverter;
import com.lowagie.text.Rectangle;
import java.util.ArrayList;

/**
 * Make copies of PDF documents. Documents can be edited after reading and
 * before writing them out.
 * @author Mark Thompson
 */

public class PdfCopy extends PdfWriter {
    /**
     * This class holds information about indirect references, since they are
     * renumbered by iText.
     */
    static class IndirectReferences {
        PdfIndirectReference theRef;
        boolean hasCopied;

        IndirectReferences(PdfIndirectReference ref) {
            theRef = ref;
            hasCopied = false;
        }

        void setCopied() {
            hasCopied = true;
        }

        boolean getCopied() {
            return hasCopied;
        }

        PdfIndirectReference getRef() {
            return theRef;
        }
    };

    protected HashMap indirects;
    protected HashMap indirectMap;
    protected int currentObjectNum = 1;
    protected PdfReader reader;
    protected PdfIndirectReference acroForm;
    protected int[] namePtr = { 0 };
    /** Holds value of property rotateContents. */
    private boolean rotateContents = true;
    protected PdfArray fieldArray;
    protected HashMap fieldTemplates;

    /**
     * A key to allow us to hash indirect references
     */
    protected static class RefKey {
        int num;
        int gen;

        RefKey(int num, int gen) {
            this.num = num;
            this.gen = gen;
        }

        RefKey(PdfIndirectReference ref) {
            num = ref.getNumber();
            gen = ref.getGeneration();
        }

        RefKey(PRIndirectReference ref) {
            num = ref.getNumber();
            gen = ref.getGeneration();
        }

        public int hashCode() {
            return (gen << 16) + num;
        }

        public boolean equals(Object o) {
            if (!(o instanceof RefKey))
                return false;
            RefKey other = (RefKey) o;
            return this.gen == other.gen && this.num == other.num;
        }

        public String toString() {
            return Integer.toString(num) + ' ' + gen;
        }
    }

    /**
     * Constructor
     * @param document
     * @param os outputstream
     */
    public PdfCopy(Document document, OutputStream os) throws DocumentException {
        super(new PdfDocument(), os);
        document.addDocListener(pdf);
        pdf.addWriter(this);
        indirectMap = new HashMap();
    }

    /** Getter for property rotateContents.
     * @return Value of property rotateContents.
     *
     */
    public boolean isRotateContents() {
        return this.rotateContents;
    }

    /** Setter for property rotateContents.
     * @param rotateContents New value of property rotateContents.
     *
     */
    public void setRotateContents(boolean rotateContents) {
        this.rotateContents = rotateContents;
    }

    /**
     * Grabs a page from the input document
     * @param reader the reader of the document
     * @param pageNumber which page to get
     * @return the page
     */
    public PdfImportedPage getImportedPage(PdfReader reader, int pageNumber) {
        if (currentPdfReaderInstance != null) {
            if (currentPdfReaderInstance.getReader() != reader) {
                try {
                    currentPdfReaderInstance.getReader().close();
                    currentPdfReaderInstance.getReaderFile().close();
                } catch (IOException ioe) {
                    // empty on purpose
                }
                currentPdfReaderInstance = reader.getPdfReaderInstance(this);
            }
        } else {
            currentPdfReaderInstance = reader.getPdfReaderInstance(this);
        }
        return currentPdfReaderInstance.getImportedPage(pageNumber);
    }

    /**
     * Translate a PRIndirectReference to a PdfIndirectReference
     * In addition, translates the object numbers, and copies the
     * referenced object to the output file.
     * NB: PRIndirectReferences (and PRIndirectObjects) really need to know what
     * file they came from, because each file has its own namespace. The translation
     * we do from their namespace to ours is *at best* heuristic, and guaranteed to
     * fail under some circumstances.
     */
    protected PdfIndirectReference copyIndirect(PRIndirectReference in) throws IOException, BadPdfFormatException {
        PdfIndirectReference theRef;
        RefKey key = new RefKey(in);
        IndirectReferences iRef = (IndirectReferences) indirects.get(key);
        if (iRef != null) {
            theRef = iRef.getRef();
            if (iRef.getCopied()) {
                return theRef;
            }
        } else {
            theRef = body.getPdfIndirectReference();
            iRef = new IndirectReferences(theRef);
            indirects.put(key, iRef);
        }
        PdfObject obj = PdfReader.getPdfObjectRelease(in);
        if (obj != null && obj.isDictionary()) {
            PdfObject type = PdfReader.getPdfObjectRelease(((PdfDictionary) obj).get(PdfName.TYPE));
            if (type != null && PdfName.PAGE.equals(type)) {
                return theRef;
            }
        }
        iRef.setCopied();
        obj = copyObject(obj);
        addToBody(obj, theRef);
        return theRef;
    }

    /**
     * Translate a PRDictionary to a PdfDictionary. Also translate all of the
     * objects contained in it.
     */
    protected PdfDictionary copyDictionary(PdfDictionary in) throws IOException, BadPdfFormatException {
        PdfDictionary out = new PdfDictionary();
        PdfObject type = PdfReader.getPdfObjectRelease(in.get(PdfName.TYPE));

        for (Iterator it = in.getKeys().iterator(); it.hasNext();) {
            PdfName key = (PdfName) it.next();
            PdfObject value = in.get(key);
            //       System.out.println("Copy " + key);
            if (type != null && PdfName.PAGE.equals(type)) {
                if (!key.equals(PdfName.B) && !key.equals(PdfName.PARENT))
                    out.put(key, copyObject(value));
            } else
                out.put(key, copyObject(value));
        }
        return out;
    }

    /**
     * Translate a PRStream to a PdfStream. The data part copies itself.
     */
    protected PdfStream copyStream(PRStream in) throws IOException, BadPdfFormatException {
        PRStream out = new PRStream(in, null);

        for (Iterator it = in.getKeys().iterator(); it.hasNext();) {
            PdfName key = (PdfName) it.next();
            PdfObject value = in.get(key);
            out.put(key, copyObject(value));
        }

        return out;
    }

    /**
     * Translate a PRArray to a PdfArray. Also translate all of the objects contained
     * in it
     */
    protected PdfArray copyArray(PdfArray in) throws IOException, BadPdfFormatException {
        PdfArray out = new PdfArray();

        for (Iterator i = in.listIterator(); i.hasNext();) {
            PdfObject value = (PdfObject) i.next();
            out.add(copyObject(value));
        }
        return out;
    }

    /**
     * Translate a PR-object to a Pdf-object
     */
    protected PdfObject copyObject(PdfObject in) throws IOException, BadPdfFormatException {
        if (in == null)
            return PdfNull.PDFNULL;
        switch (in.type) {
        case PdfObject.DICTIONARY:
            //           System.out.println("Dictionary: " + in.toString());
            return copyDictionary((PdfDictionary) in);
        case PdfObject.INDIRECT:
            return copyIndirect((PRIndirectReference) in);
        case PdfObject.ARRAY:
            return copyArray((PdfArray) in);
        case PdfObject.NUMBER:
        case PdfObject.NAME:
        case PdfObject.STRING:
        case PdfObject.NULL:
        case PdfObject.BOOLEAN:
        case 0:
            return in;
        case PdfObject.STREAM:
            return copyStream((PRStream) in);
        //                return in;
        default:
            if (in.type < 0) {
                String lit = ((PdfLiteral) in).toString();
                if (lit.equals("true") || lit.equals("false")) {
                    return new PdfBoolean(lit);
                }
                return new PdfLiteral(lit);
            }
            System.out.println("CANNOT COPY type " + in.type);
            return null;
        }
    }

    /**
     * convenience method. Given an imported page, set our "globals"
     */
    protected int setFromIPage(PdfImportedPage iPage) {
        int pageNum = iPage.getPageNumber();
        PdfReaderInstance inst = currentPdfReaderInstance = iPage.getPdfReaderInstance();
        reader = inst.getReader();
        setFromReader(reader);
        return pageNum;
    }

    /**
     * convenience method. Given a reader, set our "globals"
     */
    protected void setFromReader(PdfReader reader) {
        this.reader = reader;
        indirects = (HashMap) indirectMap.get(reader);
        if (indirects == null) {
            indirects = new HashMap();
            indirectMap.put(reader, indirects);
            PdfDictionary catalog = reader.getCatalog();
            PRIndirectReference ref = null;
            PdfObject o = catalog.get(PdfName.ACROFORM);
            if (o == null || o.type() != PdfObject.INDIRECT)
                return;
            ref = (PRIndirectReference) o;
            if (acroForm == null)
                acroForm = body.getPdfIndirectReference();
            indirects.put(new RefKey(ref), new IndirectReferences(acroForm));
        }
    }

    /**
     * Add an imported page to our output
     * @param iPage an imported page
     * @throws IOException, BadPdfFormatException
     */
    public void addPage(PdfImportedPage iPage) throws IOException, BadPdfFormatException {
        int pageNum = setFromIPage(iPage);

        PdfDictionary thePage = reader.getPageN(pageNum);
        PRIndirectReference origRef = reader.getPageOrigRef(pageNum);
        reader.releasePage(pageNum);
        RefKey key = new RefKey(origRef);
        PdfIndirectReference pageRef;
        IndirectReferences iRef = (IndirectReferences) indirects.get(key);
        if (iRef != null && !iRef.getCopied()) {
            pageReferences.add(iRef.getRef());
            iRef.setCopied();
        }
        pageRef = getCurrentPage();
        if (iRef == null) {
            iRef = new IndirectReferences(pageRef);
            indirects.put(key, iRef);
        }
        iRef.setCopied();
        PdfDictionary newPage = copyDictionary(thePage);
        root.addPage(newPage);
        ++currentPageNumber;
    }

    /**
     * Adds a blank page.
     * @param   rect The page dimension
     * @param   rotation The rotation angle in degrees
     * @since   2.1.5
     */
    public void addPage(Rectangle rect, int rotation) {
        PdfRectangle mediabox = new PdfRectangle(rect, rotation);
        PageResources resources = new PageResources();
        PdfPage page = new PdfPage(mediabox, new HashMap(), resources.getResources(), 0);
        page.put(PdfName.TABS, getTabs());
        root.addPage(page);
        ++currentPageNumber;
    }

    /**
     * Copy the acroform for an input document. Note that you can only have one,
     * we make no effort to merge them.
     * @param reader The reader of the input file that is being copied
     * @throws IOException, BadPdfFormatException
     */
    public void copyAcroForm(PdfReader reader) throws IOException, BadPdfFormatException {
        setFromReader(reader);

        PdfDictionary catalog = reader.getCatalog();
        PRIndirectReference hisRef = null;
        PdfObject o = catalog.get(PdfName.ACROFORM);
        if (o != null && o.type() == PdfObject.INDIRECT)
            hisRef = (PRIndirectReference) o;
        if (hisRef == null)
            return; // bugfix by John Englar
        RefKey key = new RefKey(hisRef);
        PdfIndirectReference myRef;
        IndirectReferences iRef = (IndirectReferences) indirects.get(key);
        if (iRef != null) {
            acroForm = myRef = iRef.getRef();
        } else {
            acroForm = myRef = body.getPdfIndirectReference();
            iRef = new IndirectReferences(myRef);
            indirects.put(key, iRef);
        }
        if (!iRef.getCopied()) {
            iRef.setCopied();
            PdfDictionary theForm = copyDictionary((PdfDictionary) PdfReader.getPdfObject(hisRef));
            addToBody(theForm, myRef);
        }
    }

    /*
     * the getCatalog method is part of PdfWriter.
     * we wrap this so that we can extend it
     */
    protected PdfDictionary getCatalog(PdfIndirectReference rootObj) {
        try {
            PdfDictionary theCat = pdf.getCatalog(rootObj);
            if (fieldArray == null) {
                if (acroForm != null)
                    theCat.put(PdfName.ACROFORM, acroForm);
            } else
                addFieldResources(theCat);
            return theCat;
        } catch (IOException e) {
            throw new ExceptionConverter(e);
        }
    }

    private void addFieldResources(PdfDictionary catalog) throws IOException {
        if (fieldArray == null)
            return;
        PdfDictionary acroForm = new PdfDictionary();
        catalog.put(PdfName.ACROFORM, acroForm);
        acroForm.put(PdfName.FIELDS, fieldArray);
        acroForm.put(PdfName.DA, new PdfString("/Helv 0 Tf 0 g "));
        if (fieldTemplates.isEmpty())
            return;
        PdfDictionary dr = new PdfDictionary();
        acroForm.put(PdfName.DR, dr);
        for (Iterator it = fieldTemplates.keySet().iterator(); it.hasNext();) {
            PdfTemplate template = (PdfTemplate) it.next();
            PdfFormField.mergeResources(dr, (PdfDictionary) template.getResources());
        }
        // if (dr.get(PdfName.ENCODING) == null) dr.put(PdfName.ENCODING, PdfName.WIN_ANSI_ENCODING);
        PdfDictionary fonts = dr.getAsDict(PdfName.FONT);
        if (fonts == null) {
            fonts = new PdfDictionary();
            dr.put(PdfName.FONT, fonts);
        }
        if (!fonts.contains(PdfName.HELV)) {
            PdfDictionary dic = new PdfDictionary(PdfName.FONT);
            dic.put(PdfName.BASEFONT, PdfName.HELVETICA);
            dic.put(PdfName.ENCODING, PdfName.WIN_ANSI_ENCODING);
            dic.put(PdfName.NAME, PdfName.HELV);
            dic.put(PdfName.SUBTYPE, PdfName.TYPE1);
            fonts.put(PdfName.HELV, addToBody(dic).getIndirectReference());
        }
        if (!fonts.contains(PdfName.ZADB)) {
            PdfDictionary dic = new PdfDictionary(PdfName.FONT);
            dic.put(PdfName.BASEFONT, PdfName.ZAPFDINGBATS);
            dic.put(PdfName.NAME, PdfName.ZADB);
            dic.put(PdfName.SUBTYPE, PdfName.TYPE1);
            fonts.put(PdfName.ZADB, addToBody(dic).getIndirectReference());
        }
    }

    /**
     * Signals that the <CODE>Document</CODE> was closed and that no other
     * <CODE>Elements</CODE> will be added.
     * <P>
     * The pages-tree is built and written to the outputstream.
     * A Catalog is constructed, as well as an Info-object,
     * the reference table is composed and everything is written
     * to the outputstream embedded in a Trailer.
     */

    public void close() {
        if (open) {
            PdfReaderInstance ri = currentPdfReaderInstance;
            pdf.close();
            super.close();
            if (ri != null) {
                try {
                    ri.getReader().close();
                    ri.getReaderFile().close();
                } catch (IOException ioe) {
                    // empty on purpose
                }
            }
        }
    }

    public PdfIndirectReference add(PdfOutline outline) {
        return null;
    }

    public void addAnnotation(PdfAnnotation annot) {
    }

    PdfIndirectReference add(PdfPage page, PdfContents contents) throws PdfException {
        return null;
    }

    public void freeReader(PdfReader reader) throws IOException {
        indirectMap.remove(reader);
        if (currentPdfReaderInstance != null) {
            if (currentPdfReaderInstance.getReader() == reader) {
                try {
                    currentPdfReaderInstance.getReader().close();
                    currentPdfReaderInstance.getReaderFile().close();
                } catch (IOException ioe) {
                    // empty on purpose
                }
                currentPdfReaderInstance = null;
            }
        }
    }

    /**
     * Create a page stamp. New content and annotations, including new fields, are allowed.
     * The fields added cannot have parents in another pages. This method modifies the PdfReader instance.<p>
     * The general usage to stamp something in a page is:
     * <p>
     * <pre>
     * PdfImportedPage page = copy.getImportedPage(reader, 1);
     * PdfCopy.PageStamp ps = copy.createPageStamp(page);
     * ps.addAnnotation(PdfAnnotation.createText(copy, new Rectangle(50, 180, 70, 200), "Hello", "No Thanks", true, "Comment"));
     * PdfContentByte under = ps.getUnderContent();
     * under.addImage(img);
     * PdfContentByte over = ps.getOverContent();
     * over.beginText();
     * over.setFontAndSize(bf, 18);
     * over.setTextMatrix(30, 30);
     * over.showText("total page " + totalPage);
     * over.endText();
     * ps.alterContents();
     * copy.addPage(page);
     * </pre>
     * @param iPage an imported page
     * @return the <CODE>PageStamp</CODE>
     */
    public PageStamp createPageStamp(PdfImportedPage iPage) {
        int pageNum = iPage.getPageNumber();
        PdfReader reader = iPage.getPdfReaderInstance().getReader();
        PdfDictionary pageN = reader.getPageN(pageNum);
        return new PageStamp(reader, pageN, this);
    }

    public static class PageStamp {

        PdfDictionary pageN;
        PdfCopy.StampContent under;
        PdfCopy.StampContent over;
        PageResources pageResources;
        PdfReader reader;
        PdfCopy cstp;

        PageStamp(PdfReader reader, PdfDictionary pageN, PdfCopy cstp) {
            this.pageN = pageN;
            this.reader = reader;
            this.cstp = cstp;
        }

        public PdfContentByte getUnderContent() {
            if (under == null) {
                if (pageResources == null) {
                    pageResources = new PageResources();
                    PdfDictionary resources = pageN.getAsDict(PdfName.RESOURCES);
                    pageResources.setOriginalResources(resources, cstp.namePtr);
                }
                under = new PdfCopy.StampContent(cstp, pageResources);
            }
            return under;
        }

        public PdfContentByte getOverContent() {
            if (over == null) {
                if (pageResources == null) {
                    pageResources = new PageResources();
                    PdfDictionary resources = pageN.getAsDict(PdfName.RESOURCES);
                    pageResources.setOriginalResources(resources, cstp.namePtr);
                }
                over = new PdfCopy.StampContent(cstp, pageResources);
            }
            return over;
        }

        public void alterContents() throws IOException {
            if (over == null && under == null)
                return;
            PdfArray ar = null;
            PdfObject content = PdfReader.getPdfObject(pageN.get(PdfName.CONTENTS), pageN);
            if (content == null) {
                ar = new PdfArray();
                pageN.put(PdfName.CONTENTS, ar);
            } else if (content.isArray()) {
                ar = (PdfArray) content;
            } else if (content.isStream()) {
                ar = new PdfArray();
                ar.add(pageN.get(PdfName.CONTENTS));
                pageN.put(PdfName.CONTENTS, ar);
            } else {
                ar = new PdfArray();
                pageN.put(PdfName.CONTENTS, ar);
            }
            ByteBuffer out = new ByteBuffer();
            if (under != null) {
                out.append(PdfContents.SAVESTATE);
                applyRotation(pageN, out);
                out.append(under.getInternalBuffer());
                out.append(PdfContents.RESTORESTATE);
            }
            if (over != null)
                out.append(PdfContents.SAVESTATE);
            PdfStream stream = new PdfStream(out.toByteArray());
            stream.flateCompress(cstp.getCompressionLevel());
            PdfIndirectReference ref1 = cstp.addToBody(stream).getIndirectReference();
            ar.addFirst(ref1);
            out.reset();
            if (over != null) {
                out.append(' ');
                out.append(PdfContents.RESTORESTATE);
                out.append(PdfContents.SAVESTATE);
                applyRotation(pageN, out);
                out.append(over.getInternalBuffer());
                out.append(PdfContents.RESTORESTATE);
                stream = new PdfStream(out.toByteArray());
                stream.flateCompress(cstp.getCompressionLevel());
                ar.add(cstp.addToBody(stream).getIndirectReference());
            }
            pageN.put(PdfName.RESOURCES, pageResources.getResources());
        }

        void applyRotation(PdfDictionary pageN, ByteBuffer out) {
            if (!cstp.rotateContents)
                return;
            Rectangle page = reader.getPageSizeWithRotation(pageN);
            int rotation = page.getRotation();
            switch (rotation) {
            case 90:
                out.append(PdfContents.ROTATE90);
                out.append(page.getTop());
                out.append(' ').append('0').append(PdfContents.ROTATEFINAL);
                break;
            case 180:
                out.append(PdfContents.ROTATE180);
                out.append(page.getRight());
                out.append(' ');
                out.append(page.getTop());
                out.append(PdfContents.ROTATEFINAL);
                break;
            case 270:
                out.append(PdfContents.ROTATE270);
                out.append('0').append(' ');
                out.append(page.getRight());
                out.append(PdfContents.ROTATEFINAL);
                break;
            }
        }

        private void addDocumentField(PdfIndirectReference ref) {
            if (cstp.fieldArray == null)
                cstp.fieldArray = new PdfArray();
            cstp.fieldArray.add(ref);
        }

        private void expandFields(PdfFormField field, ArrayList allAnnots) {
            allAnnots.add(field);
            ArrayList kids = field.getKids();
            if (kids != null) {
                for (int k = 0; k < kids.size(); ++k)
                    expandFields((PdfFormField) kids.get(k), allAnnots);
            }
        }

        public void addAnnotation(PdfAnnotation annot) {
            try {
                ArrayList allAnnots = new ArrayList();
                if (annot.isForm()) {
                    PdfFormField field = (PdfFormField) annot;
                    if (field.getParent() != null)
                        return;
                    expandFields(field, allAnnots);
                    if (cstp.fieldTemplates == null)
                        cstp.fieldTemplates = new HashMap();
                } else
                    allAnnots.add(annot);
                for (int k = 0; k < allAnnots.size(); ++k) {
                    annot = (PdfAnnotation) allAnnots.get(k);
                    if (annot.isForm()) {
                        if (!annot.isUsed()) {
                            HashMap templates = annot.getTemplates();
                            if (templates != null)
                                cstp.fieldTemplates.putAll(templates);
                        }
                        PdfFormField field = (PdfFormField) annot;
                        if (field.getParent() == null)
                            addDocumentField(field.getIndirectReference());
                    }
                    if (annot.isAnnotation()) {
                        PdfObject pdfobj = PdfReader.getPdfObject(pageN.get(PdfName.ANNOTS), pageN);
                        PdfArray annots = null;
                        if (pdfobj == null || !pdfobj.isArray()) {
                            annots = new PdfArray();
                            pageN.put(PdfName.ANNOTS, annots);
                        } else
                            annots = (PdfArray) pdfobj;
                        annots.add(annot.getIndirectReference());
                        if (!annot.isUsed()) {
                            PdfRectangle rect = (PdfRectangle) annot.get(PdfName.RECT);
                            if (rect != null && (rect.left() != 0 || rect.right() != 0 || rect.top() != 0
                                    || rect.bottom() != 0)) {
                                int rotation = reader.getPageRotation(pageN);
                                Rectangle pageSize = reader.getPageSizeWithRotation(pageN);
                                switch (rotation) {
                                case 90:
                                    annot.put(PdfName.RECT, new PdfRectangle(pageSize.getTop() - rect.bottom(),
                                            rect.left(), pageSize.getTop() - rect.top(), rect.right()));
                                    break;
                                case 180:
                                    annot.put(PdfName.RECT, new PdfRectangle(pageSize.getRight() - rect.left(),
                                            pageSize.getTop() - rect.bottom(), pageSize.getRight() - rect.right(),
                                            pageSize.getTop() - rect.top()));
                                    break;
                                case 270:
                                    annot.put(PdfName.RECT,
                                            new PdfRectangle(rect.bottom(), pageSize.getRight() - rect.left(),
                                                    rect.top(), pageSize.getRight() - rect.right()));
                                    break;
                                }
                            }
                        }
                    }
                    if (!annot.isUsed()) {
                        annot.setUsed();
                        cstp.addToBody(annot, annot.getIndirectReference());
                    }
                }
            } catch (IOException e) {
                throw new ExceptionConverter(e);
            }
        }
    }

    public static class StampContent extends PdfContentByte {
        PageResources pageResources;

        /** Creates a new instance of StampContent */
        StampContent(PdfWriter writer, PageResources pageResources) {
            super(writer);
            this.pageResources = pageResources;
        }

        /**
         * Gets a duplicate of this <CODE>PdfContentByte</CODE>. All
         * the members are copied by reference but the buffer stays different.
         *
         * @return a copy of this <CODE>PdfContentByte</CODE>
         */
        public PdfContentByte getDuplicate() {
            return new PdfCopy.StampContent(writer, pageResources);
        }

        PageResources getPageResources() {
            return pageResources;
        }
    }
}