Examples with FileType eu.transkribus.core.model.beans.mets.FileType used on opensource projects

Example 6 with FileType

use of eu.transkribus.core.model.beans.mets.FileType in project TranskribusCore by Transkribus.

the class MetsUtil method getImagesToUpload.

public static List<PageUploadDescriptor> getImagesToUpload(Mets mets) {
    // check filesection. needs img group and xml group to distinguish them without going for mimetypes
    List<FileGrpType> typeGrps = getMasterFileGrp(mets);
    boolean hasXml = true;
    List<FileType> xmlGrp = null;
    List<FileType> imgGrp = null;
    for (FileGrpType type : typeGrps) {
        switch(type.getID()) {
            case TrpMetsBuilder.IMG_GROUP_ID:
                imgGrp = type.getFile();
                break;
            case TrpMetsBuilder.PAGE_GROUP_ID:
                xmlGrp = type.getFile();
                break;
            default:
                break;
        }
    }
    if (imgGrp == null) {
        throw new IllegalArgumentException("METS file has no image file list!");
    }
    if (xmlGrp == null) {
        logger.debug("METS file has no xml file list!");
    }
    List<DivType> pageDivs = getPageDivsFromStructMap(mets);
    if (pageDivs == null)
        throw new IllegalArgumentException("No valid StructMap was found!");
    List<PageUploadDescriptor> images = new ArrayList<PageUploadDescriptor>(pageDivs.size());
    for (DivType div : pageDivs) {
        PageUploadDescriptor image = buildUploadImage(div, imgGrp, xmlGrp);
        images.add(image);
    }
    return images;
}

Example 7 with FileType

use of eu.transkribus.core.model.beans.mets.FileType in project TranskribusCore by Transkribus.

the class MetsUtil method getTrpPages.

/**
 * Builds the set of TrpPage objects with
 * local file references from the mets master file group and structmap.
 * The method is strict regarding PAGE XML existence! Each image file must have a correspondent PAGE XML.
 * @param mets
 * @param parentDir
 * @return
 * @throws IOException
 */
public static List<TrpPage> getTrpPages(Mets mets, File parentDir) throws IOException {
    // check filesection. needs img group and xml group to distinguish them without going for mimetypes
    List<FileGrpType> typeGrps = getMasterFileGrp(mets);
    List<FileType> xmlGrp = null;
    List<FileType> imgGrp = null;
    for (FileGrpType type : typeGrps) {
        switch(type.getID()) {
            case TrpMetsBuilder.IMG_GROUP_ID:
                imgGrp = type.getFile();
                break;
            case TrpMetsBuilder.PAGE_GROUP_ID:
                xmlGrp = type.getFile();
                break;
            default:
                break;
        }
    }
    if (imgGrp == null)
        throw new IOException("METS file has no image file list!");
    if (xmlGrp == null)
        throw new IOException("METS file has no xml file list!");
    List<DivType> pageDivs = getPageDivsFromStructMap(mets);
    if (pageDivs == null) {
        throw new IOException("No valid StructMap was found!");
    }
    List<TrpPage> pages = new ArrayList<TrpPage>(pageDivs.size());
    for (DivType div : pageDivs) {
        TrpPage page = buildPage(div, imgGrp, xmlGrp, parentDir);
        pages.add(page);
    }
    return pages;
}

Example 8 with FileType

use of eu.transkribus.core.model.beans.mets.FileType in project TranskribusCore by Transkribus.

the class MetsUtil method buildPage.

private static TrpPage buildPage(DivType div, List<FileType> imgGrp, List<FileType> xmlGrp, File parentDir) throws IOException {
    TrpPage page = new TrpPage();
    int nr = div.getORDER().intValue();
    page.setPageNr(nr);
    File imgFile = null;
    File xmlFile = null;
    // FIXME this will only work for local files
    for (Fptr ptr : div.getFptr()) {
        FileType type = (FileType) ptr.getArea().getFILEID();
        if (imgGrp.contains(type)) {
            imgFile = MetsUtil.getFile(type, parentDir);
        } else if (xmlGrp.contains(type)) {
            xmlFile = MetsUtil.getFile(type, parentDir);
        }
    }
    if (imgFile == null) {
        logger.error("No master image mapped for page " + nr + " in the structmap!");
    } else {
        logger.info("Page " + page.getPageNr() + " image: " + imgFile.getAbsolutePath());
    }
    // FIXME NullpointerException if imgFile == null!
    page.setUrl(imgFile.toURI().toURL());
    page.setKey(null);
    page.setDocId(-1);
    page.setImgFileName(imgFile.getName());
    if (xmlFile == null) {
        logger.error("No master xml mapped for page " + nr + " in the structmap!");
    } else {
        logger.info("Page " + page.getPageNr() + " xml: " + xmlFile.getAbsolutePath());
    }
    // FIXME NullpointerException if xmlFile == null!
    TrpTranscriptMetadata tmd = new TrpTranscriptMetadata();
    tmd.setPageReferenceForLocalDocs(page);
    tmd.setPageId(page.getPageId());
    tmd.setUrl(xmlFile.toURI().toURL());
    tmd.setKey(null);
    tmd.setStatus(EditStatus.NEW);
    tmd.setTimestamp(new Date().getTime());
    tmd.setUserName("LocalDocReader");
    page.getTranscripts().add(tmd);
    return page;
}

Example 9 with FileType

use of eu.transkribus.core.model.beans.mets.FileType in project TranskribusCore by Transkribus.

the class TrpMetsBuilder method buildMets.

/**
 * Generate a METS containing
 * <ul>
 * <li>TrpDocMetadata embedded in sourceMd</li>
 * <li>all page images</li>
 * <li>the most recent PAGE XML files from the Doc</li>
 * </ul>
 *
 * If a local document is passed, all hrefs will contain the relative paths to files based on the localFolder!
 *
 * @param doc
 * @param exportImages
 * @param pageIndices
 * @return
 * @throws IOException if image/xml files can't be accessed for reading the mimetype etc.
 */
public static Mets buildMets(TrpDoc doc, boolean exportPage, boolean exportAlto, boolean exportImages, Set<Integer> pageIndices) throws IOException {
    Mets mets = new Mets();
    TrpDocMetadata md = doc.getMd();
    File localFolder = md.getLocalFolder();
    boolean isLocalDoc = localFolder != null;
    mets.setLABEL(md.getTitle());
    mets.setOBJID("" + md.getDocId());
    mets.setPROFILE(TRP_METS_PROFILE);
    // FIXME remove TYPE
    // mets.setTYPE(TRP_METS_PROFILE);
    // metsHdr
    MetsHdr hdr = buildMetsHdr(md);
    mets.setMetsHdr(hdr);
    // TODO dcmd_elec omitted meanwhile
    // md_orig
    AmdSecType amdSec = new AmdSecType();
    amdSec.setID(SOURCE_MD_ID_CONST);
    MdSecType sourceMdSec = buildSourceMdSec(md);
    amdSec.getSourceMD().add(sourceMdSec);
    mets.getAmdSec().add(amdSec);
    // structmap div, linking to the sourceMd section with dmd
    DivType div = new DivType();
    div.getADMID().add(sourceMdSec);
    div.setID(TRP_DOC_DIV_ID);
    FileSec fileSec = new FileSec();
    StructMapType structMap = new StructMapType();
    structMap.setID(TRP_STRUCTMAP_ID);
    structMap.setTYPE("MANUSCRIPT");
    structMap.setDiv(div);
    List<TrpPage> pages = doc.getPages();
    FimgStoreGetClient client = null;
    if (!isLocalDoc) {
        // TODO maybe we need this stuff in the docMetadata?
        URL url = pages.get(0).getUrl();
        client = new FimgStoreGetClient(url);
    }
    FileGrp masterGrp = new FileGrp();
    masterGrp.setID(MASTER_FILE_GRP_ID);
    FileGrpType imgGrp = new FileGrpType();
    imgGrp.setID(IMG_GROUP_ID);
    FileGrpType pageGrp = new FileGrpType();
    pageGrp.setID(PAGE_GROUP_ID);
    FileGrpType altoGrp = new FileGrpType();
    altoGrp.setID(ALTO_GROUP_ID);
    int i = -1;
    for (TrpPage p : pages) {
        i++;
        if (pageIndices != null && !pageIndices.contains(i)) {
            continue;
        }
        // build a page div for the structmap
        DivType pageDiv = new DivType();
        pageDiv.setID("PAGE_" + p.getPageNr());
        pageDiv.setTYPE("SINGLE_PAGE");
        pageDiv.setORDER(BigInteger.valueOf(p.getPageNr()));
        final String imgId = "IMG_" + p.getPageNr();
        final String xmlId = PAGE_GROUP_ID + "_" + p.getPageNr();
        final String altoId = ALTO_GROUP_ID + "_" + p.getPageNr();
        /* only the most recent transcript is added here for now
			 * 
			 * TODO how to deal with imagestore files? use orig image? right now, it's just the view file...
			 * TODO thumbnails not yet included
			*/
        if (exportImages) {
            FileType img = buildFileType(localFolder, imgId, p, p.getPageNr(), client);
            imgGrp.getFile().add(img);
            // linking images
            Fptr imgPtr = buildFptr(img);
            pageDiv.getFptr().add(imgPtr);
        }
        // TODO error handling.. if no transcript??
        if (exportPage) {
            // xmlfiletype: just add the most recent transcript
            TrpTranscriptMetadata tMd;
            // get the transcript chosen for export
            tMd = p.getCurrentTranscript();
            FileType xml = buildFileType(md.getLocalFolder(), xmlId, tMd, p.getPageNr(), client);
            pageGrp.getFile().add(xml);
            Fptr xmlPtr = buildFptr(xml);
            pageDiv.getFptr().add(xmlPtr);
        }
        // creat ALTO fileGrp
        if (exportAlto) {
            FileType altoFt = new FileType();
            altoFt.setCHECKSUMTYPE(ChecksumUtils.ChkSumAlg.MD5.toString());
            // TODO calculate checksum
            altoFt.setCHECKSUM("");
            FLocat fLocat = new FLocat();
            fLocat.setLOCTYPE("OTHER");
            fLocat.setOTHERLOCTYPE("FILE");
            altoFt.setID(altoId);
            altoFt.setSEQ(p.getPageNr());
            // String tmpImgName = img.getFLocat().get(0).getHref();
            String relAltoPath = "alto".concat(File.separator).concat(p.getImgFileName().substring(0, p.getImgFileName().lastIndexOf(".")).concat(".xml"));
            fLocat.setHref(relAltoPath);
            // String absAltoPath = tMd.getUrl().getPath().replace("page", "alto");
            final String path = FileUtils.toFile(p.getUrl()).getAbsolutePath();
            String absAltoPath = path.substring(0, path.lastIndexOf(File.separator));
            absAltoPath = absAltoPath.concat("/alto/").concat(p.getImgFileName().substring(0, p.getImgFileName().lastIndexOf(".")).concat(".xml"));
            // logger.info("alto path starts with: " + absAltoPath);
            if (absAltoPath.startsWith("\\")) /*|| absAltoPath.startsWith("/")*/
            {
                // logger.info("alto path starts with \\ or /");
                absAltoPath = absAltoPath.substring(1);
            }
            String mime = MimeTypes.getMimeType("xml");
            altoFt.setMIMETYPE(mime);
            File altoTmp = new File(absAltoPath);
            if (altoTmp.exists()) {
                // logger.info("alto file exist at " + absAltoPath);
                Date date = new Date(altoTmp.lastModified());
                XMLGregorianCalendar cal = JaxbUtils.getXmlCalendar(date);
                altoFt.setCREATED(cal);
            } else {
                logger.info("alto file does not exist at " + absAltoPath);
            }
            // System.out.println("relAltoPath " + relAltoPath);
            // System.out.println("absAltoPath " + absAltoPath);
            // System.in.read();
            altoFt.getFLocat().add(fLocat);
            altoGrp.getFile().add(altoFt);
            Fptr altoPtr = buildFptr(altoFt);
            pageDiv.getFptr().add(altoPtr);
        }
        div.getDiv().add(pageDiv);
    }
    fileSec.getFileGrp().add(masterGrp);
    mets.setFileSec(fileSec);
    if (exportImages) {
        masterGrp.getFileGrp().add(imgGrp);
    }
    if (exportPage) {
        masterGrp.getFileGrp().add(pageGrp);
    }
    if (exportAlto) {
        masterGrp.getFileGrp().add(altoGrp);
    }
    mets.getStructMap().add(structMap);
    return mets;
}

Example 10 with FileType

use of eu.transkribus.core.model.beans.mets.FileType in project TranskribusCore by Transkribus.

the class TrpMetsBuilder method buildFileType.

/**
 * @param localFolder null if isLocalDoc
 * @param id
 * @param o
 * @param client
 * @return
 * @throws IOException
 */
private static FileType buildFileType(File localFolder, String id, ITrpFile o, final int seq, FimgStoreGetClient client) throws IOException {
    FileType fType = new FileType();
    fType.setID(id);
    String mime = null;
    Date date = null;
    FLocat fLocat = new FLocat();
    String loc = null;
    if (localFolder != null) {
        URL url = o.getUrl();
        if (!url.getProtocol().contains("file")) {
            throw new IOException("Doc contains local folder reference but an URL refers to a non-local file! " + url.toString());
        }
        final String path = FileUtils.toFile(url).getAbsolutePath();
        File f = new File(path);
        mime = MimeTypes.getMimeType(FilenameUtils.getExtension(f.getName()));
        date = new Date(f.lastModified());
        fLocat.setLOCTYPE("OTHER");
        fLocat.setOTHERLOCTYPE("FILE");
        // remove protocol and localfolder, i.e. get relative path to this file
        // loc = path.substring(localFolder.getAbsolutePath().length() + 1); // BUG: localFolder != path!!
        loc = FilenameUtils.getName(path);
        if (id.startsWith(PAGE_GROUP_ID)) {
            // append relative folder for PAGE XML files
            loc = "page/" + loc;
        }
        logger.debug("loc = " + loc);
        if (o.getMd5Sum() != null) {
            fType.setCHECKSUMTYPE(ChecksumUtils.ChkSumAlg.MD5.toString());
            fType.setCHECKSUM(o.getMd5Sum());
        }
    } else {
        try {
            FimgStoreFileMd fMd = client.getFileMd(o.getKey());
            date = fMd.getUploadDate();
            mime = fMd.getMimetype();
            fLocat.setLOCTYPE("URL");
            // full URL in case of remote file
            loc = o.getUrl().toString();
        } catch (IOException e) {
            logger.error(e.getMessage(), e);
            throw new IOException("FileMetadata could not be retrieved from imagestore for key: " + o.getKey());
        }
    }
    fType.setMIMETYPE(mime);
    XMLGregorianCalendar cal = JaxbUtils.getXmlCalendar(date);
    fType.setCREATED(cal);
    fType.setSEQ(seq);
    fLocat.setHref(loc);
    fType.getFLocat().add(fLocat);
    return fType;
}