metsmanifest.java

dspace 用j2ee架构的一个数字图书馆.开源程序

            /*
            XMLOutputter outputPretty = new XMLOutputter(Format.getPrettyFormat());
            log.debug("Got METS DOCUMENT:");
            log.debug(outputPretty.outputString(metsDocument));
              */
        }
        catch (JDOMException je)
        {
            throw new MetadataValidationException("Error validating METS in "
                    + is.toString(),  je);
        }

        return new METSManifest(builder, metsDocument.getRootElement());
    }

    /**
     * Gets name of the profile to which this METS document conforms.
     * @return value the PROFILE attribute of mets element, or null if none.
     */
    public String getProfile()
    {
        return mets.getAttributeValue("PROFILE");
    }

    /**
     * Gets all <code>file</code> elements which make up
     *   the item's content.
     * @return a List of <code>Element</code>s.
     */
    public List getContentFiles()
        throws MetadataValidationException
    {
        if (contentFiles != null)
            return contentFiles;

        Element fileSec = mets.getChild("fileSec", metsNS);
        if (fileSec == null)
            throw new MetadataValidationException("Invalid METS Manifest: DSpace requires a fileSec element, but it is missing.");

        contentFiles = new ArrayList();
        Iterator fgi = fileSec.getChildren("fileGrp", metsNS).iterator();
        while (fgi.hasNext())
        {
            Element fg = (Element)fgi.next();
            Iterator fi = fg.getChildren("file", metsNS).iterator();
            while (fi.hasNext())
            {
                Element f = (Element)fi.next();
                contentFiles.add(f);
            }
        }
        return contentFiles;
    }

    /**
     * Gets list of all <code>mdRef</code> elements in the METS
     *   document.  Used by ingester to e.g. check that all
     *   required files are present.
     * @return a List of <code>Element</code>s.
     */
    public List getMdFiles()
        throws MetadataValidationException
    {
        if (mdFiles == null)
        {
            try
            {
                // Use a special namespace with known prefix
                // so we get the right prefix.
                XPath xpath = XPath.newInstance("descendant::mets:mdRef");
                xpath.addNamespace(metsNS);
                mdFiles = xpath.selectNodes(mets);
            }
            catch (JDOMException je)
            {
                throw new MetadataValidationException("Failed while searching for mdRef elements in manifest: ", je);
            }
        }
        return mdFiles;
    }

    /**
     * Get the "original" file element for a derived file.
     * Finds the original from which this was derived by matching the GROUPID
     * attribute that binds it to its original.  For instance, the file for
     * a thumbnail image would have the same GROUPID as its full-size version.
     * <p>
     * NOTE: This pattern of relating derived files through the GROUPID
     * attribute is peculiar to the DSpace METS SIP profile, and may not be
     * generally useful with other sorts of METS documents.
     * @param file METS file element of derived file
     * @return file Element of original or null if none found.
     */
    public Element getOriginalFile(Element file)
    {
        String groupID = file.getAttributeValue("GROUPID");
        if (groupID == null || groupID.equals(""))
            return null;

        try
        {
            XPath xpath = XPath.newInstance(
"mets:fileSec/mets:fileGrp[@USE=\"CONTENT\"]/mets:file[@GROUPID=\""+groupID+"\"]");
            xpath.addNamespace(metsNS);
            List oFiles = xpath.selectNodes(mets);
            if (oFiles.size() > 0)
            {
                log.debug("Got ORIGINAL file for derived="+file.toString());
                return (Element)oFiles.get(0);
            }
            else
                return null;
        }
        catch (JDOMException je)
        {
            log.warn("Got exception on XPATH looking for Original file, "+je.toString());
            return null;
        }
    }

    // translate bundle name from METS to DSpace; METS may be "CONTENT"
    // or "ORIGINAL" for the DSPace "ORIGINAL", rest are left alone.
    private static String normalizeBundleName(String in)
    {
        if (in.equals("CONTENT"))
            return Constants.CONTENT_BUNDLE_NAME;
        else if (in.equals("MANIFESTMD"))
            return Constants.METADATA_BUNDLE_NAME;
        return in;
    }

    /**
     * Get the DSpace bundle name corresponding to the <code>USE</code> attribute of the file group enclosing this <code>file</code> element.
     * @return DSpace bundle name
     * @throws MetadataValidationException when there is no USE attribute on the enclosing fileGrp.
     */
    public static String getBundleName(Element file)
        throws MetadataValidationException
    {
        Element fg = file.getParentElement();
        String fgUse = fg.getAttributeValue("USE");
        if (fgUse == null)
            throw new MetadataValidationException("Invalid METS Manifest: every fileGrp element must have a USE attribute.");
        return normalizeBundleName(fgUse);
    }

    /**
     * Get the "local" file name of this <code>file</code> or <code>mdRef</code> element.
     * By "local" we mean the reference to the actual resource containing
     * the data for this file, e.g. a relative path within a Zip or tar archive
     * if the METS is serving as a manifest for that sort of package.
     * @return "local" file name (i.e.  relative to package or content
     *  directory) corresponding to this <code>file</code> or <code>mdRef</code> element.
     * @throws MetadataValidationException when there is not enough information to find a resource identifier.
     */
    public static String getFileName(Element file)
        throws MetadataValidationException
    {
        Element ref;
        if (file.getName().equals("file"))
        {
            ref = file.getChild("FLocat", metsNS);
            if (ref == null)
            {
                // check for forbidden FContent child first:
                if (file.getChild("FContent", metsNS) == null)
                    throw new MetadataValidationException("Invalid METS Manifest: Every file element must have FLocat child.");
                else
                    throw new MetadataValidationException("Invalid METS Manifest: file element has forbidden FContent child, only FLocat is allowed.");
            }
        }
        else if (file.getName().equals("mdRef"))
            ref = file;
        else
            throw new MetadataValidationException("getFileName() called with recognized element type: "+file.toString());
        String loctype = ref.getAttributeValue("LOCTYPE");
        if (loctype != null && loctype.equals("URL"))
        {
            String result = ref.getAttributeValue("href", xlinkNS);
            if (result == null)
                throw new MetadataValidationException("Invalid METS Manifest: FLocat/mdRef is missing the required xlink:href attribute.");
            return result;
        }
        throw new MetadataValidationException("Invalid METS Manifest: FLocat/mdRef does not have LOCTYPE=\"URL\" attribute.");
    }

    /**
     * Returns file element corresponding to primary bitstream.
     * There is <i>ONLY</i> a primary bitstream if the first <code>div</code> under
     * first </code>structMap</code> has an </code>fptr</code>.
     *
     * @return file element of Item's primary bitstream, or null if there is none.
     */
    public Element getPrimaryBitstream()
        throws MetadataValidationException
    {
        Element firstDiv = getFirstDiv();
        Element fptr = firstDiv.getChild("fptr", metsNS);
        if (fptr == null)
            return null;
        String id = fptr.getAttributeValue("FILEID");
        if (id == null)
            throw new MetadataValidationException("fptr for Primary Bitstream is missing the required FILEID attribute.");
        Element result = getElementByXPath("descendant::mets:file[@ID=\""+id+"\"]", false);
        if (result == null)
            throw new MetadataValidationException("Cannot find file element for Primary Bitstream: looking for ID="+id);
        return result;
    }

    /** Get the metadata type from within a *mdSec element.
     * @return metadata type name.
     */
    public String getMdType(Element mdSec)
        throws MetadataValidationException
    {
        Element md = mdSec.getChild("mdRef", metsNS);
        if (md == null)
            md = mdSec.getChild("mdWrap", metsNS);
        if (md == null)
            throw new MetadataValidationException("Invalid METS Manifest: ?mdSec element has neither mdRef nor mdWrap child.");
        String result = md.getAttributeValue("MDTYPE");
        if (result != null && result.equals("OTHER"))
            result = md.getAttributeValue("OTHERMDTYPE");
        if (result == null)
            throw new MetadataValidationException("Invalid METS Manifest: "+md.getName()+" has no MDTYPE or OTHERMDTYPE attribute.");
        return result;
    }

    /**
     *  Returns MIME type of metadata content, if available.
     *  @return MIMEtype word, or null if none is available.
     */
    public String getMdContentMimeType(Element mdSec)
        throws MetadataValidationException
    {
        Element mdWrap = mdSec.getChild("mdWrap", metsNS);
        if (mdWrap != null)
        {
            String mimeType = mdWrap.getAttributeValue("MIMETYPE");
            if (mimeType == null && mdWrap.getChild("xmlData", metsNS) != null)
            mimeType = "text/xml";
            return mimeType;
        }
        Element mdRef = mdSec.getChild("mdRef", metsNS);
        if (mdRef != null)
            return mdRef.getAttributeValue("MIMETYPE");
        return null;
    }

    /**
     * Return contents of *md element as List of XML Element objects.
     * Gets content, dereferecing mdRef if necessary, or decoding and parsing
     * a binData that contains XML.
     * @return contents of metadata section, or empty list if no XML content is available.
     * @throws MetadataValidationException if METS is invalid, or there is an error parsing the XML.
     */
    public List getMdContentAsXml(Element mdSec, Mdref callback)
        throws MetadataValidationException, IOException, SQLException, AuthorizeException
    {
        try
        {
            Element mdRef = null;
            Element mdWrap = mdSec.getChild("mdWrap", metsNS);
            if (mdWrap != null)
            {
                Element xmlData = mdWrap.getChild("xmlData", metsNS);
                if (xmlData == null)
                {
                    Element bin = mdWrap.getChild("binData", metsNS);
                    if (bin == null)
                        throw new MetadataValidationException("Invalid METS Manifest: mdWrap element with neither xmlData nor binData child.");

                    // if binData is actually XML, return it; otherwise ignore.
                    else
                    {
                        String mimeType = mdWrap.getAttributeValue("MIMETYPE");
                        if (mimeType != null && mimeType.equalsIgnoreCase("text/xml"))
                        {
                            byte value[] = Base64.decodeBase64(bin.getText().getBytes());
                            Document mdd = parser.build(new ByteArrayInputStream(value));
                            List result = new ArrayList(1);
                            result.add(mdd.getRootElement());
                            return result;
                        }
                        else
                        {
                            log.warn("Ignoring binData section because MIMETYPE is not XML, but: "+mimeType);
                            return new ArrayList(0);
                        }
                   }
                }
                else