
org.jpedal.examples.viewer.gui.BaseTransferHandler 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
*
* ---------------
* BaseTransferHandler.java
* ---------------
*/
package org.jpedal.examples.viewer.gui;
import java.awt.datatransfer.DataFlavor;
import java.awt.datatransfer.Transferable;
import java.io.BufferedReader;
import java.io.ByteArrayInputStream;
import java.io.IOException;
import java.io.Reader;
import javax.swing.JComponent;
import javax.swing.TransferHandler;
import javax.xml.parsers.DocumentBuilder;
import javax.xml.parsers.DocumentBuilderFactory;
import javax.xml.parsers.ParserConfigurationException;
import org.jpedal.examples.viewer.Commands;
import org.jpedal.examples.viewer.Values;
import org.jpedal.gui.GUIFactory;
import org.jpedal.utils.StringUtils;
import org.w3c.dom.Document;
import org.w3c.dom.Element;
import org.w3c.dom.NamedNodeMap;
import org.w3c.dom.Node;
import org.xml.sax.SAXException;
public class BaseTransferHandler extends TransferHandler {
protected final Commands currentCommands;
protected final GUIFactory currentGUI;
protected final Values commonValues;
public BaseTransferHandler(final Values commonValues, final GUIFactory currentGUI, final Commands currentCommands) {
this.commonValues = commonValues;
this.currentGUI = currentGUI;
this.currentCommands = currentCommands;
}
@Override
public boolean canImport(final JComponent dest, final DataFlavor[] flavors) {
return true;
}
protected Object getImport(final Transferable transferable) throws Exception{
final DataFlavor[] flavors = transferable.getTransferDataFlavors();
DataFlavor listFlavor = null;
final int lastFlavor = flavors.length - 1;
// Check the flavors and see if we find one we like.
// If we do, save it.
for (int f = 0; f <= lastFlavor; f++) {
if (flavors[f].isFlavorJavaFileListType()) {
listFlavor = flavors[f];
}
}
// Ok, now try to display the content of the drop.
try {
final DataFlavor bestTextFlavor = DataFlavor.selectBestTextFlavor(flavors);
if (bestTextFlavor != null) { // this could be a file from a web page being dragged in
final Reader r = bestTextFlavor.getReaderForText(transferable);
/** acquire the text data from the reader. */
String textData = readTextDate(r);
// System.out.println(textData);
/** need to remove all the 0 characters that will appear in the String when importing on Linux */
textData = removeChar(textData, (char) 0);
if(textData.contains("ftp:/")) {
currentGUI.showMessageDialog("Files cannot be opened via FTP");
return null;
}
/** get the URL from the text data */
textData = getURL(textData);
/** replace URL spaces */
textData = textData.replaceAll("%20", " ");
return textData;
} else if (listFlavor != null) { // this is most likely a file being dragged in
return transferable.getTransferData(listFlavor);
}
} catch (final Exception e) {
//
return null;
}
return null;
}
private static String removeChar(final String s, final char c) {
StringBuilder r = new StringBuilder();
for (int i = 0; i < s.length(); i++) {
if (s.charAt(i) != c) {
r.append(s.charAt(i));
}
}
return r.toString();
}
/**
* Returns the URL from the text data acquired from the transferable object.
* @param textData text data acquired from the transferable.
* @return the URL of the file to open
* @throws ParserConfigurationException
* @throws SAXException
* @throws IOException
*/
private static String getURL(String textData) throws ParserConfigurationException, SAXException, IOException {
if (!textData.startsWith("http://") && !textData.startsWith("file://")) { // its not a url so it must be a file
final DocumentBuilderFactory dbf = DocumentBuilderFactory.newInstance();
final DocumentBuilder db = dbf.newDocumentBuilder();
final Document doc = db.parse(new ByteArrayInputStream(StringUtils.toBytes(textData)));
final Element a = (Element) doc.getElementsByTagName("a").item(0);
textData = getHrefAttribute(a);
}
return textData;
}
/**
* Acquire text data from a reader.
* Firefox this will be some html containing an "a" element with the "href" attribute linking to the to the PDF.
* IE a simple one line String containing the URL will be returned
* @param r the reader to read from
* @return the text data from the reader
* @throws IOException
*/
private static String readTextDate(final Reader r) throws IOException {
final BufferedReader br = new BufferedReader(r);
StringBuilder textData = new StringBuilder();
String line = br.readLine();
while (line != null) {
textData.append(line);
line = br.readLine();
}
br.close();
return textData.toString();
}
/**
* Returns the URL held in the href attribute from an element
* @param element the element containing the href attribute
* @return the URL held in the href attribute
*/
private static String getHrefAttribute(final Element element) {
final NamedNodeMap attrs = element.getAttributes();
final Node nameNode = attrs.getNamedItem("href");
if (nameNode != null) {
return nameNode.getNodeValue();
}
return null;
}
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy