Examples of org.apache.pdfbox.pdmodel.PDDocumentInformation

org.apache.pdfbox.pdmodel.PDDocumentInformation
This is the document metadata. Each getXXX method will return the entry if it exists or null if it does not exist. If you pass in null for the setXXX method then it will clear the value. @author Ben Litchfield @author Gerardo Ortiz @version $Revision: 1.12 $

        }
    }


    private void extractMetadata(PDDocument document, Metadata metadata)
            throws TikaException {
        PDDocumentInformation info = document.getDocumentInformation();
        metadata.set(PagedText.N_PAGES, document.getNumberOfPages());
        addMetadata(metadata, TikaCoreProperties.TITLE, info.getTitle());
        addMetadata(metadata, TikaCoreProperties.CREATOR, info.getAuthor());
        addMetadata(metadata, TikaCoreProperties.CREATOR_TOOL, info.getCreator());
        addMetadata(metadata, TikaCoreProperties.KEYWORDS, info.getKeywords());
        addMetadata(metadata, "producer", info.getProducer());
        // TODO: Move to description in Tika 2.0
        addMetadata(metadata, TikaCoreProperties.TRANSITION_SUBJECT_TO_OO_SUBJECT, info.getSubject());
        addMetadata(metadata, "trapped", info.getTrapped());
        try {
            // TODO Remove these in Tika 2.0
            addMetadata(metadata, "created", info.getCreationDate());
            addMetadata(metadata, TikaCoreProperties.CREATED, info.getCreationDate());
        } catch (IOException e) {
            // Invalid date format, just ignore
        }
        try {
            Calendar modified = info.getModificationDate(); 
            addMetadata(metadata, Metadata.LAST_MODIFIED, modified);
            addMetadata(metadata, TikaCoreProperties.MODIFIED, modified);
        } catch (IOException e) {
            // Invalid date format, just ignore
        }
        
        // All remaining metadata is custom
        // Copy this over as-is
        List<String> handledMetadata = Arrays.asList(new String[] {
             "Author", "Creator", "CreationDate", "ModDate",
             "Keywords", "Producer", "Subject", "Title", "Trapped"
        });
        for(COSName key : info.getDictionary().keySet()) {
            String name = key.getName();
            if(! handledMetadata.contains(name)) {
          addMetadata(metadata, name, info.getDictionary().getDictionaryObject(key));
            }
        }
    }

View Full Code Here

        }
    }


    private void extractMetadata(PDDocument document, Metadata metadata)
            throws TikaException {
        PDDocumentInformation info = document.getDocumentInformation();
        metadata.set(PagedText.N_PAGES, document.getNumberOfPages());
        addMetadata(metadata, Metadata.TITLE, info.getTitle());
        addMetadata(metadata, Metadata.AUTHOR, info.getAuthor());
        addMetadata(metadata, Metadata.CREATOR, info.getCreator());
        addMetadata(metadata, Metadata.KEYWORDS, info.getKeywords());
        addMetadata(metadata, "producer", info.getProducer());
        addMetadata(metadata, Metadata.SUBJECT, info.getSubject());
        addMetadata(metadata, "trapped", info.getTrapped());
        try {
            addMetadata(metadata, "created", info.getCreationDate());
            addMetadata(metadata, Metadata.CREATION_DATE, info.getCreationDate());
        } catch (IOException e) {
            // Invalid date format, just ignore
        }
        try {
            Calendar modified = info.getModificationDate(); 
            addMetadata(metadata, Metadata.LAST_MODIFIED, modified);
        } catch (IOException e) {
            // Invalid date format, just ignore
        }
        
        // All remaining metadata is custom
        // Copy this over as-is
        List<String> handledMetadata = Arrays.asList(new String[] {
             "Author", "Creator", "CreationDate", "ModDate",
             "Keywords", "Producer", "Subject", "Title", "Trapped"
        });
        for(COSName key : info.getDictionary().keySet()) {
            String name = key.getName();
            if(! handledMetadata.contains(name)) {
          addMetadata(metadata, name, info.getDictionary().getDictionaryObject(key));
            }
        }
    }

View Full Code Here

        parse(stream, handler, metadata, new ParseContext());
    }


    private void extractMetadata(PDDocument document, Metadata metadata)
            throws TikaException {
        PDDocumentInformation info = document.getDocumentInformation();
        metadata.set(PagedText.N_PAGES, document.getNumberOfPages());
        addMetadata(metadata, Metadata.TITLE, info.getTitle());
        addMetadata(metadata, Metadata.AUTHOR, info.getAuthor());
        addMetadata(metadata, Metadata.CREATOR, info.getCreator());
        addMetadata(metadata, Metadata.KEYWORDS, info.getKeywords());
        addMetadata(metadata, "producer", info.getProducer());
        addMetadata(metadata, Metadata.SUBJECT, info.getSubject());
        addMetadata(metadata, "trapped", info.getTrapped());
        try {
            addMetadata(metadata, "created", info.getCreationDate());
            addMetadata(metadata, Metadata.CREATION_DATE, info.getCreationDate());
        } catch (IOException e) {
            // Invalid date format, just ignore
        }
        try {
            Calendar modified = info.getModificationDate(); 
            addMetadata(metadata, Metadata.LAST_MODIFIED, modified);
        } catch (IOException e) {
            // Invalid date format, just ignore
        }
        
        // All remaining metadata is custom
        // Copy this over as-is
        List<String> handledMetadata = Arrays.asList(new String[] {
             "Author", "Creator", "CreationDate", "ModDate",
             "Keywords", "Producer", "Subject", "Title", "Trapped"
        });
        for(COSName key : info.getDictionary().keySet()) {
            String name = key.getName();
            if(! handledMetadata.contains(name)) {
          addMetadata(metadata, name, info.getDictionary().getDictionaryObject(key));
            }
        }
    }

View Full Code Here

        }
        if( source.isEncrypted() )
        {
            throw new IOException( "Error: source PDF is encrypted, can't append encrypted PDF documents." );
        }
        PDDocumentInformation destInfo = destination.getDocumentInformation();
        PDDocumentInformation srcInfo = source.getDocumentInformation();
        destInfo.getDictionary().mergeInto( srcInfo.getDictionary() );


        PDDocumentCatalog destCatalog = destination.getDocumentCatalog();
        PDDocumentCatalog srcCatalog = source.getDocumentCatalog();


        if( destCatalog.getOpenAction() == null )

View Full Code Here

public class Info {
    public Info(Scrutinizer scrutinizer) {
        DocumentAdapter documentAdapter = scrutinizer.getDocumentAdapter();
        if (documentAdapter != null && documentAdapter.IsDocLoaded()) {
            PDDocument pd = documentAdapter.getDocument();
            PDDocumentInformation docinfo = pd.getDocumentInformation();
            if (docinfo != null) {
                Title = title = docinfo.getTitle();
                Author = author = docinfo.getAuthor();
                Subject = subject = docinfo.getSubject();
                Keywords = keywords = docinfo.getKeywords();
                Creator = creator = docinfo.getCreator();
                Producer = producer = docinfo.getProducer();
                try {
                    CreationDate = creationdate = docinfo.getCreationDate();
                    ModDate = moddate = docinfo.getModificationDate();
                } catch (IOException e) {
                    // should already be triggered at  de.pdf_scrutinizer.DocumentAdapter.getDocInfo
                }
                Trapped = trapped = docinfo.getTrapped();
            }
        }
    }

View Full Code Here

                ScriptableObject.putProperty(scope, "URL", Context.javaToJS(doc.URL, scope));
                ScriptableObject.putProperty(scope, "numPages", Context.javaToJS(doc.numPages, scope));


                // if document catalog contains custom metadata we have to put them into the context.
                if (scrutinizer.getDocumentAdapter() != null) {
                    PDDocumentInformation info = scrutinizer.getDocumentAdapter().getDocument().getDocumentInformation();
                    if (info != null) {
                        String[] docinfokeys = {"Author", "Creator", "CreationDate", "Subject", "Title", "Keywords", "ModDate", "Producer", "Trapped"};
                        Set<String> metadataKeys = info.getMetadataKeys();
                        metadataKeys.removeAll(Arrays.asList(docinfokeys)); // remove non-custom keys


                        for (String metadataKey : metadataKeys) {
                            log.debug(String.format("adding custom info metadata: %s value: %s", "info." + metadataKey, info.getCustomMetadataValue(metadataKey)));
                            ScriptableObject.putProperty((Scriptable) scope.get("info", scope), metadataKey, Context.javaToJS(info.getCustomMetadataValue(metadataKey), scope));
                        }
                    }
                }


                app.viewerVersion = Double.parseDouble(viewerVersion);

View Full Code Here

                dcSchema = xmp.getDublinCoreSchema();
            }
        } catch (IOException e) {
            //swallow
        }
        PDDocumentInformation info = document.getDocumentInformation();
        metadata.set(PagedText.N_PAGES, document.getNumberOfPages());
        extractMultilingualItems(metadata, TikaCoreProperties.TITLE, info.getTitle(), dcSchema);
        extractDublinCoreListItems(metadata, TikaCoreProperties.CREATOR, info.getAuthor(), dcSchema);
        extractDublinCoreListItems(metadata, TikaCoreProperties.CONTRIBUTOR, null, dcSchema);
        addMetadata(metadata, TikaCoreProperties.CREATOR_TOOL, info.getCreator());
        addMetadata(metadata, TikaCoreProperties.KEYWORDS, info.getKeywords());
        addMetadata(metadata, "producer", info.getProducer());
        extractMultilingualItems(metadata, TikaCoreProperties.DESCRIPTION, null, dcSchema);


        // TODO: Move to description in Tika 2.0
        addMetadata(metadata, TikaCoreProperties.TRANSITION_SUBJECT_TO_OO_SUBJECT, info.getSubject());
        addMetadata(metadata, "trapped", info.getTrapped());
        try {
            // TODO Remove these in Tika 2.0
            addMetadata(metadata, "created", info.getCreationDate());
            addMetadata(metadata, TikaCoreProperties.CREATED, info.getCreationDate());
        } catch (IOException e) {
            // Invalid date format, just ignore
        }
        try {
            Calendar modified = info.getModificationDate();
            addMetadata(metadata, Metadata.LAST_MODIFIED, modified);
            addMetadata(metadata, TikaCoreProperties.MODIFIED, modified);
        } catch (IOException e) {
            // Invalid date format, just ignore
        }
        
        // All remaining metadata is custom
        // Copy this over as-is
        List<String> handledMetadata = Arrays.asList("Author", "Creator", "CreationDate", "ModDate",
                "Keywords", "Producer", "Subject", "Title", "Trapped");
        for(COSName key : info.getDictionary().keySet()) {
            String name = key.getName();
            if(! handledMetadata.contains(name)) {
          addMetadata(metadata, name, info.getDictionary().getDictionaryObject(key));
            }
        }


        //try to get the various versions
        //Caveats:

View Full Code Here


                  if (props.isEmpty())
                  {
                     // The pdf doesn't contain any metadata, try to use the document
                     // information instead
                     PDDocumentInformation docInfo = pdDocument.getDocumentInformation();


                     if (docInfo != null)
                     {
                        try
                        {
                           if (docInfo.getAuthor() != null)
                              props.put(DCMetaData.CONTRIBUTOR, docInfo.getAuthor());
                        }
                        catch (Exception e)
                        {
                           LOG.warn("getAuthor failed: " + e.getMessage());
                        }
                        try
                        {
                           if (docInfo.getCreationDate() != null)
                              props.put(DCMetaData.DATE, docInfo.getCreationDate());
                        }
                        catch (Exception e)
                        {
                           LOG.warn("getCreationDate failed: " + e.getMessage());
                        }
                        try
                        {
                           if (docInfo.getCreator() != null)
                              props.put(DCMetaData.CREATOR, docInfo.getCreator());
                        }
                        catch (Exception e)
                        {
                           LOG.warn("getCreator failed: " + e.getMessage());
                        }
                        try
                        {


                           if (docInfo.getKeywords() != null)
                              props.put(DCMetaData.SUBJECT, docInfo.getKeywords());
                        }
                        catch (Exception e)
                        {
                           LOG.warn("getKeywords failed: " + e.getMessage());
                        }
                        try
                        {
                           if (docInfo.getModificationDate() != null)
                              props.put(DCMetaData.DATE, docInfo.getModificationDate());
                        }
                        catch (Exception e)
                        {
                           LOG.warn("getModificationDate failed: " + e.getMessage());
                        }
                        try
                        {
                           if (docInfo.getProducer() != null)
                              props.put(DCMetaData.PUBLISHER, docInfo.getProducer());
                        }
                        catch (Exception e)
                        {
                           LOG.warn("getProducer failed: " + e.getMessage());
                        }
                        try
                        {
                           if (docInfo.getSubject() != null)
                              props.put(DCMetaData.DESCRIPTION, docInfo.getSubject());
                        }
                        catch (Exception e)
                        {
                           LOG.warn("getSubject failed: " + e.getMessage());
                        }
                        try
                        {
                           if (docInfo.getTitle() != null)
                              props.put(DCMetaData.TITLE, docInfo.getTitle());
                        }
                        catch (Exception e)
                        {
                           LOG.warn("getTitle failed: " + e.getMessage());
                        }

View Full Code Here

0 1 2 3 4 5

TOP

Related Classes of org.apache.pdfbox.pdmodel.PDDocumentInformation

de.pdf_scrutinizer.API.app.doc.Info

de.pdf_scrutinizer.emulation.InterpreterEmulation

geopms.GeoPMSImportPDF

net.sf.regain.crawler.preparator.PdfBoxPreparator

net.yacy.document.parser.pdfParser

org.apache.camel.component.fop.FopHelper

org.apache.jackrabbit.core.query.pdf.PDFParser

org.apache.padaf.preflight.xmp.SynchronizedMetaDataValidation

org.apache.pdfbox.cos.COSDictionary

org.apache.pdfbox.examples.lucene.LucenePDFDocument

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.