PDFPackager


1   /*
2    * PDFPackager.java
3    *
4    * Version: $Revision: 1.1 $
5    *
6    * Date: $Date: 2006/03/17 00:04:38 $
7    *
8    * Copyright (c) 2002-2005, Hewlett-Packard Company and Massachusetts
9    * Institute of Technology.  All rights reserved.
10   *
11   * Redistribution and use in source and binary forms, with or without
12   * modification, are permitted provided that the following conditions are
13   * met:
14   *
15   * - Redistributions of source code must retain the above copyright
16   * notice, this list of conditions and the following disclaimer.
17   *
18   * - Redistributions in binary form must reproduce the above copyright
19   * notice, this list of conditions and the following disclaimer in the
20   * documentation and/or other materials provided with the distribution.
21   *
22   * - Neither the name of the Hewlett-Packard Company nor the name of the
23   * Massachusetts Institute of Technology nor the names of their
24   * contributors may be used to endorse or promote products derived from
25   * this software without specific prior written permission.
26   *
27   * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
28   * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
29   * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
30   * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
31   * HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
32   * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
33   * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
34   * OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
35   * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR
36   * TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
37   * USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
38   * DAMAGE.
39   */
40  
41  package org.dspace.content.packager;
42  
43  import java.io.IOException  ;
44  import java.io.InputStream  ;
45  import java.io.OutputStream  ;
46  import java.sql.SQLException  ;
47  import java.util.Calendar  ;
48  
49  import org.apache.log4j.Logger;
50  import org.dspace.authorize.AuthorizeException;
51  import org.dspace.content.Bitstream;
52  import org.dspace.content.BitstreamFormat;
53  import org.dspace.content.Bundle;
54  import org.dspace.content.Collection;
55  import org.dspace.content.DCDate;
56  import org.dspace.content.DSpaceObject;
57  import org.dspace.content.Item;
58  import org.dspace.content.WorkspaceItem;
59  import org.dspace.content.crosswalk.CrosswalkException;
60  import org.dspace.content.crosswalk.MetadataValidationException;
61  import org.dspace.core.Constants;
62  import org.dspace.core.Context;
63  import org.dspace.core.LogManager;
64  import org.dspace.core.SelfNamedPlugin;
65  import org.dspace.core.Utils;
66  import org.pdfbox.cos.COSDocument;
67  import org.pdfbox.pdfparser.PDFParser;
68  import org.pdfbox.pdmodel.PDDocument;
69  import org.pdfbox.pdmodel.PDDocumentInformation;
70                                      
71  /**
72   * Accept a PDF file by itself as a SIP.
73   * <p>
74   * This is mainly a proof-of-concept to demonstrate the flexibility
75   * of the packager and crosswalk plugins.
76   * <p>
77   * To import, open up the PDF and try to extract sufficient metadata
78   * from its InfoDict.
79   * <p>
80   * Export is a crude hack: if the item has a bitstream containing PDF,
81   * send that, otherwise it fails. Do not attempt to insert metadata.
82   *
83   * @author Larry Stone
84   * @version $Revision: 1.1 $
85   */
86  public class PDFPackager
87         extends SelfNamedPlugin
88         implements PackageIngester, PackageDisseminator
89  {
90      /** log4j category */
91      private static Logger log = Logger.getLogger(PDFPackager.class);
92  
93      private final static String   BITSTREAM_FORMAT_NAME = "Adobe PDF";
94  
95      private static String   aliases[] = { "PDF", "Adobe PDF", "pdf", "application/pdf" };
96  
97      public static String  [] getPluginNames()
98      {
99          return aliases;
100     }
101 
102     // utility to grovel bitstream formats..
103     private static void setFormatToMIMEType(Context context, Bitstream bs, String   mimeType)
104         throws SQLException  
105     {
106         BitstreamFormat bf[] = BitstreamFormat.findNonInternal(context);
107         for (int i = 0; i < bf.length; ++i)
108         {
109             if (bf[i].getMIMEType().equalsIgnoreCase(mimeType))
110             {
111                 bs.setFormat(bf[i]);
112                 break;
113             }
114         }
115     }
116 
117     /**
118      * Create new Item out of the ingested package, in the indicated
119      * collection.  It creates a workspace item, which the application
120      * can then install if it chooses to bypass Workflow.
121      * <p>
122      * This is a VERY crude import of a single Adobe PDF (Portable
123      * Document Format) file, using the document's embedded metadata
124      * for package metadata.  If the PDF file hasn't got the minimal
125      * metadata available, it is rejected.
126      * <p>
127      * @param context  DSpace context.
128      * @param collection  collection under which to create new item.
129      * @param pkg  input stream containing package to ingest.
130      * @param params  package parameters (none recognized)
131      * @param license  may be null, which takes default license.
132      * @return workspace item created by ingest.
133      * @throws PackageException if package is unacceptable or there is
134      *  a fatal error turning it into an Item.
135      */
136     public WorkspaceItem ingest(Context context, Collection collection,
137                                 InputStream   pkg, PackageParameters params,
138                                 String   license)
139         throws PackageValidationException, CrosswalkException,
140                AuthorizeException, SQLException  , IOException  
141     {
142         InputStream   bis = null;
143         COSDocument cos = null;
144         boolean success = false;
145         Bundle original = null;
146         Bitstream bs = null;
147         WorkspaceItem wi = null;
148 
149         /** XXX comment out for now
150           // XXX for debugging of parameter handling
151           if (params != null)
152           {
153               Enumeration pe = params.propertyNames();
154               while (pe.hasMoreElements())
155               {
156                   String name = (String)pe.nextElement();
157                   String v[] = params.getProperties(name);
158                   StringBuffer msg = new StringBuffer("PackageParam: ");
159                   msg.append(name).append(" = ");
160                   for (int i = 0; i < v.length; ++i)
161                   {
162                       if (i > 0)
163                           msg.append(", ");
164                       msg.append(v[i]);
165                   }
166                   log.debug(msg);
167               }
168           }
169         **/
170            
171         try
172         {
173             // Save the PDF in a bitstream first, since the parser
174             // has to read it as well, and we cannot "rewind" it after that.
175             wi = WorkspaceItem.create(context, collection, false);
176             Item myitem = wi.getItem();
177             original = myitem.createBundle("ORIGINAL");
178             bs = original.createBitstream(pkg);
179             pkg.close();
180             bs.setName("package.pdf");
181             setFormatToMIMEType(context, bs, "application/pdf");
182             bs.update();
183             log.debug("Created bitstream ID="+String.valueOf(bs.getID())+", parsing...");
184 
185             crosswalkPDF(context, myitem, bs.retrieve());
186 
187             wi.update();
188             context.commit();
189             success = true;
190             log.info(LogManager.getHeader(context, "ingest",
191                 "Created new Item, db ID="+String.valueOf(myitem.getID())+
192                 ", WorkspaceItem ID="+String.valueOf(wi.getID())));
193             return wi;
194         }
195         finally
196         {
197             try
198             {
199                 // Close bitstream input stream and PDF file.
200                 if (bis != null)
201                     bis.close();
202                 if (cos != null)
203                     cos.close();
204             }
205             catch (IOException   ie)
206             { }
207 
208             // get rid of bitstream and item if ingest fails
209             if (!success)
210             {
211                 if (original != null && bs != null)
212                     original.removeBitstream(bs);
213                 if (wi != null)
214                     wi.deleteAll();
215             }
216             context.commit();
217         }
218     }
219 
220     /**
221      * Replace is not implemented.
222      */
223     public Item replace(Context ctx, Item item, InputStream   pckage, PackageParameters params)
224         throws PackageValidationException, CrosswalkException,
225                AuthorizeException, SQLException  , IOException  ,
226                UnsupportedOperationException  
227     {
228         throw new UnsupportedOperationException  ("The replace operation is not implemented.");
229     }
230 
231     /**
232      * VERY crude dissemination: just look for the first
233      * bitstream with the PDF package type, and toss it out.
234      * Works on packages importer with this packager, and maybe some others.
235      */
236     public void disseminate(Context context, DSpaceObject dso,
237                             PackageParameters params, OutputStream   out)
238         throws PackageValidationException, CrosswalkException,
239                AuthorizeException, SQLException  , IOException  
240     {
241         if (dso.getType() != Constants.ITEM)
242             throw new PackageValidationException("This disseminator can only handle objects of type ITEM.");
243 
244         Item item = (Item)dso;
245         try
246         {
247             BitstreamFormat pdff = BitstreamFormat.findByShortDescription(context,
248                                     BITSTREAM_FORMAT_NAME);
249             if (pdff == null)
250                 throw new PackageValidationException("Cannot find BitstreamFormat \""+BITSTREAM_FORMAT_NAME+"\"");
251             Bitstream pkgBs = PackageUtils.getBitstreamByFormat(item, pdff, Constants.DEFAULT_BUNDLE_NAME);
252             if (pkgBs == null)
253                 throw new PackageValidationException("Cannot find Bitstream with format \""+BITSTREAM_FORMAT_NAME+"\"");
254             Utils.copy(pkgBs.retrieve(), out);
255         }
256             finally {}
257     }
258 
259     /**
260      * Identifies the MIME-type of this package, i.e. "application/pdf".
261      *
262      * @return the MIME type (content-type header) of the package to be returned
263      */
264     public String   getMIMEType(PackageParameters params)
265     {
266         return "application/pdf";
267     }
268 
269     private void crosswalkPDF(Context context, Item item, InputStream   metadata)
270         throws CrosswalkException, IOException  , SQLException  , AuthorizeException
271     {
272         COSDocument cos = null;
273 
274         try
275         {
276             PDFParser parser = new PDFParser(metadata);
277             parser.parse();
278             cos = parser.getDocument();
279 
280             // sanity check: PDFBox breaks on encrypted documents, so give up.
281             if(cos.getEncryptionDictionary() != null)
282                 throw new MetadataValidationException("This packager cannot accept an encrypted PDF document.");
283 
284             /* PDF to DC "crosswalk":
285              *
286              * NOTE: This is not in a crosswalk plugin because (a) it isn't
287              * useful anywhere else, and more importantly, (b) the source
288              * data is not XML so it doesn't fit the plugin's interface.
289              *
290              * pattern of crosswalk -- PDF dict entries to DC:
291              *   Title -> title.null
292              *   Author -> contributor.author
293              *   CreationDate -> date.created
294              *   ModDate -> date.created
295              *   Creator -> description.provenance (application that created orig)
296              *   Producer -> description.provenance (convertor to pdf)
297              *   Subject -> description.abstract
298              *   Keywords -> subject.other
299              *    date is java.util.Calendar
300              */
301             PDDocument pd = new PDDocument(cos);
302             PDDocumentInformation docinfo = pd.getDocumentInformation();
303             String   title = docinfo.getTitle();
304 
305             // sanity check: item must have a title.
306             if (title == null)
307                 throw new MetadataValidationException("This PDF file is unacceptable, it does not have a value for \"Title\" in its Info dictionary.");
308             log.debug("PDF Info dict title=\""+title+"\"");
309             item.addDC("title", null, "en", title);
310             String   value;
311             Calendar   date;
312             if ((value = docinfo.getAuthor()) != null)
313             {
314                 item.addDC("contributor", "author", null, value);
315                 log.debug("PDF Info dict author=\""+value+"\"");
316             }
317             if ((value = docinfo.getCreator()) != null)
318                 item.addDC("description", "provenance", "en",
319                               "Application that created the original document: "+value);
320             if ((value = docinfo.getProducer()) != null)
321                 item.addDC("description", "provenance", "en",
322                               "Original document converted to PDF by: "+value);
323             if ((value = docinfo.getSubject()) != null)
324                 item.addDC("description", "abstract", null, value);
325             if ((value = docinfo.getKeywords()) != null)
326                 item.addDC("subject", "other", null, value);
327 
328             // Take either CreationDate or ModDate as "date.created",
329             // Too bad there's no place to put "last modified" in the DC.
330             Calendar   calValue;
331             if ((calValue = docinfo.getCreationDate()) == null)
332                 calValue = docinfo.getModificationDate();
333             if (calValue != null)
334                 item.addDC("date", "created", null,
335                              (new DCDate(calValue.getTime())).toString());
336             item.update();
337         }
338         finally
339         {
340             if (cos != null)
341                 cos.close();
342         }
343     }
344 }
345
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags