metsmanifest.java

dspace 用j2ee架构的一个数字图书馆.开源程序

                {
                    return xmlData.getChildren();
                }
            }
            else if ((mdRef = mdSec.getChild("mdRef", metsNS)) != null)
            {
                String mimeType = mdRef.getAttributeValue("MIMETYPE");
                if (mimeType != null && mimeType.equalsIgnoreCase("text/xml"))
                {
                    Document mdd = parser.build(callback.getInputStream(mdRef));
                    List result = new ArrayList(1);
                    result.add(mdd.getRootElement());
                    return result;
                }
                else
                {
                    log.warn("Ignoring mdRef section because MIMETYPE is not XML, but: "+mimeType);
                    return new ArrayList(0);
                }
            }
            else
                throw new MetadataValidationException("Invalid METS Manifest: ?mdSec element with neither mdRef nor mdWrap child.");
        }
        catch (JDOMException je)
        {
            throw new MetadataValidationException("Error parsing or validating metadata section in mdRef or binData within "+mdSec.toString(), je);
        }

    }

    /**
     * Return contents of *md element as stream.
     * Gets content, dereferecing mdRef if necessary, or decoding
     * a binData element if necessary.
     * @return Stream containing contents of metadata section.  Never returns null.
     * @throws MetadataValidationException if METS format does not contain any metadata.
     */
    public InputStream getMdContentAsStream(Element mdSec, Mdref callback)
        throws MetadataValidationException, IOException, SQLException, AuthorizeException
    {
        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.");

                else
                {
                    byte value[] = Base64.decodeBase64(bin.getText().getBytes());
                    return new ByteArrayInputStream(value);
                }
            }
            else
            {
                XMLOutputter outputPretty = new XMLOutputter(Format.getPrettyFormat());
                return new ByteArrayInputStream(
                        outputPretty.outputString(xmlData.getChildren()).getBytes());
            }
        }
        else if ((mdRef = mdSec.getChild("mdRef", metsNS)) != null)
        {
            return callback.getInputStream(mdRef);
        }
        else
            throw new MetadataValidationException("Invalid METS Manifest: ?mdSec element with neither mdRef nor mdWrap child.");
    }


    // special call to crosswalk the guts of a metadata *Sec (dmdSec, amdSec)
    // because mdRef and mdWrap have to be handled differently.
    // It's a lot like getMdContentAsXml but cannot use that because xwalk
    // should be called with root element OR list depending on what was given.
    private void crosswalkMdContent(Element mdSec, Mdref callback,
                IngestionCrosswalk xwalk, Context context, DSpaceObject dso)
        throws CrosswalkException, IOException, SQLException, AuthorizeException
    {
        List xml = getMdContentAsXml(mdSec,callback);

        // if we get inappropriate metadata, e.g. PREMIS for Item, let it go.
        try
        {
            xwalk.ingest(context, dso, xml);
        }
        catch (CrosswalkObjectNotSupported e)
        {
            log.warn("Skipping metadata for inappropriate type of object: Object="+dso.toString()+", error="+e.toString());
        }
    }

    // return first <div> of first <structMap>;
    // in DSpace profile, this is where item-wide dmd and other metadata
    // lives as IDrefs.
    private Element getFirstDiv()
        throws MetadataValidationException
    {
        Element sm = mets.getChild("structMap", metsNS);
        if (sm == null)
            throw new MetadataValidationException("METS document is missing the required structMap element.");

        Element result = sm.getChild("div", metsNS);
        if (result == null)
            throw new MetadataValidationException("METS document is missing the required first div element in first structMap.");

        log.debug("Got firstDiv result="+result.toString());
        return (Element)result;
    }

    // return a single Element node found by one-off path.
    // use only when path varies each time you call it.
    private Element getElementByXPath(String path, boolean nullOk)
        throws MetadataValidationException
    {
        try
        {
            XPath xpath = XPath.newInstance(path);
            xpath.addNamespace(metsNS);
            xpath.addNamespace(xlinkNS);
            Object result = xpath.selectSingleNode(mets);
            if (result == null && nullOk)
                return null;
            else if (result instanceof Element)
                return (Element)result;
            else
                throw new MetadataValidationException("METSManifest: Failed to resolve XPath, path=\""+path+"\"");
        }
        catch (JDOMException je)
        {
            throw new MetadataValidationException("METSManifest: Failed to resolve XPath, path=\""+path+"\"", je);
        }
    }

    // Find crosswalk for the indicated metadata type (e.g. "DC", "MODS")
    // The crosswalk plugin name MAY be indirected in config file,
    // through an entry like
    //  mets.submission.crosswalk.{mdType} = {pluginName}
    //   e.g.
    //  mets.submission.crosswalk.DC = mysite-QDC
    private IngestionCrosswalk getCrosswalk(String type)
    {
        String xwalkName = ConfigurationManager.getProperty(CONFIG_METADATA_PREFIX + type);
        if (xwalkName == null)
            xwalkName = type;
        return (IngestionCrosswalk)
          PluginManager.getNamedPlugin(IngestionCrosswalk.class, xwalkName);
    }

    /**
     * Gets all dmdSec elements containing metadata for the DSpace Item.
     *
     * @return array of Elements, each a dmdSec.  May be empty but NOT null.
     * @throws MetadataValidationException if the METS is missing a reference to item-wide
     *          DMDs in the correct place.
     */
    public Element[] getItemDmds()
        throws MetadataValidationException
    {
        // div@DMDID is actually IDREFS, a space-separated list of IDs:
        Element firstDiv = getFirstDiv();
        String dmds = firstDiv.getAttributeValue("DMDID");
        if (dmds == null)
            throw new MetadataValidationException("Invalid METS: Missing reference to Item descriptive metadata, first div on first structmap must have a DMDID attribute.");
        String dmdID[] = dmds.split("\\s+");
        Element result[] = new Element[dmdID.length];

        for (int i = 0; i < dmdID.length; ++i)
            result[i] = getElementByXPath("mets:dmdSec[@ID=\""+dmdID[i]+"\"]", false);
        return result;
    }

    /**
     * Return rights metadata section(s) relevant to item as a whole.
     * @return array of rightsMd elements, possibly empty but never null.
     * @throws MetadataValidationException if METS is invalid, e.g. referenced amdSec is missing.
     */
    public Element[] getItemRightsMD()
        throws MetadataValidationException
    {
        // div@ADMID is actually IDREFS, a space-separated list of IDs:
        Element firstDiv = getFirstDiv();
        String amds = firstDiv.getAttributeValue("ADMID");
        if (amds == null)
        {
            log.debug("getItemRightsMD: No ADMID references found.");
            return new Element[0];
        }
        String amdID[] = amds.split("\\s+");
        List resultList = new ArrayList();
        for (int i = 0; i < amdID.length; ++i)
        {
            List rmds = getElementByXPath("mets:amdSec[@ID=\""+amdID[i]+"\"]", false).
                            getChildren("rightsMD", metsNS);
            if (rmds.size() > 0)
                resultList.addAll(rmds);
        }
        return (Element[])resultList.toArray(new Element[resultList.size()]);
    }

    /**
     * Invokes appropriate crosswalks on Item-wide descriptive metadata.
     */
    public void crosswalkItem(Context context, Item item, Element dmd, Mdref callback)
        throws MetadataValidationException,
               CrosswalkException, IOException, SQLException, AuthorizeException
    {
        String type = getMdType(dmd);
        IngestionCrosswalk xwalk = getCrosswalk(type);

        if (xwalk == null)
            throw new MetadataValidationException("Cannot process METS Manifest: "+
                "No crosswalk found for MDTYPE="+type);
        crosswalkMdContent(dmd, callback, xwalk, context, item);
    }

    /**
     * Crosswalk the metadata associated with a particular <code>file</code>
     * element into the bitstream it corresponds to.
     * @param context a dspace context.
     * @param bs bitstream target of the crosswalk
     * @param fileId value of ID attribute in the file element responsible
     *  for the contents of that bitstream.
     */
    public void crosswalkBitstream(Context context, Bitstream bitstream,
                                   String fileId, Mdref callback)
        throws MetadataValidationException,
               CrosswalkException, IOException, SQLException, AuthorizeException
    {
        Element file = getElementByXPath("descendant::mets:file[@ID=\""+fileId+"\"]", false);
        if (file == null)
            throw new MetadataValidationException("Failed in Bitstream crosswalk, Could not find file element with ID="+fileId);

        // In DSpace METS SIP spec, admin metadata is only "highly
        // recommended", not "required", so it is OK if there is no ADMID.
        String amds = file.getAttributeValue("ADMID");
        if (amds == null)
        {
            log.warn("Got no bitstream ADMID, file@ID="+fileId);
            return;
        }
        String amdID[] = amds.split("\\s+");
        for (int i = 0; i < amdID.length; ++i)
        {
            List techMDs = getElementByXPath("mets:amdSec[@ID=\""+amdID[i]+"\"]", false).
                                 getChildren("techMD", metsNS);
            Iterator ti = techMDs.iterator();
            while (ti.hasNext())
            {
                Element techMD = (Element)ti.next();
                if (techMD != null)
                {
                    String type = getMdType(techMD);
                    IngestionCrosswalk xwalk = getCrosswalk(type);
                    log.debug("Got bitstream techMD of type="+type+", for file ID="+fileId);
                     
                    if (xwalk == null)
                        throw new MetadataValidationException("Cannot process METS Manifest: "+
                            "No crosswalk found for techMD MDTYPE="+type);
                    crosswalkMdContent(techMD, callback, xwalk, context, bitstream);
                }
            }
        }
    }

    /**
     * Find Handle (if any) identifier labelling this manifest.
     * @return handle (never null)
     * @throws MetadataValidationException if no handle available.
     */
    public String getHandle()
        throws MetadataValidationException
    {
        // TODO: XXX Make configurable? Handle optionally passed in?
        // FIXME: Not sure if OBJID is really the right place

        String handle = mets.getAttributeValue("OBJID");

        if (handle != null && handle.startsWith("hdl:"))
        {
            return handle.substring(4);
        }
        else
        {
            throw new MetadataValidationException("Item has no valid Handle (OBJID)");
        }
    }
}