
org.jpedal.objects.PdfResources Maven / Gradle / Ivy
/*
* ===========================================
* Java Pdf Extraction Decoding Access Library
* ===========================================
*
* Project Info: http://www.idrsolutions.com
* Help section for developers at http://www.idrsolutions.com/support/
*
* (C) Copyright 1997-2015 IDRsolutions and Contributors.
*
* This file is part of JPedal/JPDF2HTML5
*
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) 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
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*
* ---------------
* PdfResources.java
* ---------------
*/
package org.jpedal.objects;
import org.jpedal.exception.PdfException;
import org.jpedal.io.PdfObjectReader;
import org.jpedal.objects.layers.PdfLayerList;
import org.jpedal.objects.outlines.OutlineData;
import org.jpedal.objects.raw.PdfDictionary;
import org.jpedal.objects.raw.PdfObject;
import org.jpedal.parser.PdfStreamDecoder;
import org.jpedal.parser.ValueTypes;
import org.jpedal.utils.LogWriter;
import org.w3c.dom.Document;
public class PdfResources {
public static final int AcroFormObj = 1;
public static final int GlobalResources = 2;
public static final int StructTreeRootObj = 3;
public static final int MarkInfoObj = 4;
PdfLayerList layers;
/**objects read from root*/
private PdfObject metadataObj, acroFormObj, globalResources,PropertiesObj, structTreeRootObj, OCProperties,markInfoObj,OutlinesObj;
/**
* store outline data extracted from pdf
*/
private OutlineData outlineData;
/**
* initialise OC Content and other items before Page decoded but after Resources read
* @param current
* @param currentPdfFile
*/
public void setupResources(final PdfStreamDecoder current, final boolean alwaysCheck, final PdfObject Resources, final int pageNumber,
final PdfObjectReader currentPdfFile) throws PdfException {
if (globalResources != null){
current.readResources(globalResources,true);
final PdfObject propObj=globalResources.getDictionary(PdfDictionary.Properties);
if(propObj!=null) {
PropertiesObj = propObj;
}
}
/**read the resources for the page*/
if (Resources != null){
current.readResources(Resources,true);
final PdfObject propObj=Resources.getDictionary(PdfDictionary.Properties);
if(propObj!=null) {
PropertiesObj = propObj;
}
}
/**
* layers
*/
if(OCProperties!=null && (layers==null || pageNumber!=layers.getOCpageNumber() || alwaysCheck)){
currentPdfFile.checkResolved(OCProperties);
if(layers==null) {
layers = new PdfLayerList();
}
layers.init(OCProperties, PropertiesObj, currentPdfFile,pageNumber);
}
current.setObjectValue(ValueTypes.PdfLayerList,layers);
}
public PdfObject getPdfObject(final int key) {
PdfObject obj=null;
switch(key){
case AcroFormObj:
obj=acroFormObj;
break;
case GlobalResources:
obj=globalResources;
break;
case MarkInfoObj:
obj=markInfoObj;
break;
case StructTreeRootObj:
obj=structTreeRootObj;
break;
}
return obj;
}
public void setPdfObject(final int key, final PdfObject obj) {
switch(key){
case GlobalResources:
globalResources=obj;
break;
}
}
public void flush() {
globalResources=null;
}
public void flushObjects() {
//flush objects held
metadataObj=null;
acroFormObj=null;
markInfoObj=null;
PropertiesObj=null;
OCProperties=null;
structTreeRootObj=null;
OutlinesObj=null;
layers=null;
}
/**
* flag to show if PDF document contains an outline
*/
public final boolean hasOutline() {
return OutlinesObj != null;
}
public void setValues(final PdfObject pdfObject, final PdfObjectReader currentPdfFile) {
currentPdfFile.checkResolved(pdfObject);
metadataObj=pdfObject.getDictionary(PdfDictionary.Metadata);
acroFormObj=pdfObject.getDictionary(PdfDictionary.AcroForm);
currentPdfFile.checkResolved(acroFormObj);
markInfoObj=pdfObject.getDictionary(PdfDictionary.MarkInfo);
structTreeRootObj=pdfObject.getDictionary(PdfDictionary.StructTreeRoot);
OCProperties=pdfObject.getDictionary(PdfDictionary.OCProperties);
OutlinesObj=pdfObject.getDictionary(PdfDictionary.Outlines);
//set up outlines
outlineData = null;
}
/**
* provide direct access to outlineData object
* @return OutlineData
*/
public OutlineData getOutlineData() {
return outlineData;
}
public Document getOutlineAsXML(final PdfObjectReader currentPdfFile) {
if (outlineData == null && OutlinesObj != null) {
try {
currentPdfFile.checkResolved(OutlinesObj);
outlineData = new OutlineData();
outlineData.readOutlineFileMetadata(OutlinesObj, currentPdfFile);
} catch (final Exception e) {
if(LogWriter.isOutput()) {
LogWriter.writeLog("Exception " + e + " accessing outline ");
}
outlineData = null;
//
}
}
if (outlineData != null) {
return outlineData.getList();
} else {
return null;
}
}
public PdfFileInformation getMetaData(final PdfObjectReader currentPdfFile) {
if (currentPdfFile != null){
/**Information object holds information from file*/
return new PdfFileInformation().readPdfFileMetadata(metadataObj, currentPdfFile);
}else {
return null;
}
}
public boolean isForm() {
return acroFormObj!=null;
}
public PdfLayerList getPdfLayerList() {
return layers;
}
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy