Examples with DigitalObjectsExtraction com.opentext.ia.sdk.sip.DigitalObjectsExtraction used on opensource projects

Example 1 with DigitalObjectsExtraction

use of com.opentext.ia.sdk.sip.DigitalObjectsExtraction in project infoarchive-sip-sdk by Enterprise-Content-Management.

the class FileArchiver method run.

private void run(String rootPath, String sip) throws IOException {
    File sipFile = new File(sip).getCanonicalFile();
    if (!sipFile.getParentFile().mkdirs() && !sipFile.getParentFile().isDirectory()) {
        throw new IllegalStateException("Could not create all required directories in path " + sipFile.getParentFile().getAbsolutePath());
    }
    System.out.printf("%nSample 2: Archiving files from %s into %s%n", rootPath, sipFile.getPath());
    // Tell InfoArchive where and how to archive the data
    URI entityUri = URI.create("urn:com.opentext.ia.sdk.sample.file:1.0");
    String entityName = "file";
    PackagingInformation prototype = PackagingInformation.builder().dss().application("fileApplication").holding("fileHolding").producer("SIP SDK").entity(entityName).schema(entityUri.toString()).end().build();
    // Define a mapping from our domain object to the PDI XML
    PdiAssembler<File> pdiAssembler = new XmlPdiAssembler<File>(entityUri, entityName) {

        @Override
        protected void doAdd(File file, Map<String, ContentInfo> contentInfo) {
            try {
                String path = relativePath(file, rootPath);
                getBuilder().element("path", path).element("size", Long.toString(file.length())).element("permissions", permissionsOf(file)).element("contentType", Files.probeContentType(file.toPath())).elements("hashes", "hash", contentInfo.get(path).getContentHashes(), (hash, builder) -> {
                    builder.attribute("algorithm", hash.getHashFunction()).attribute("encoding", hash.getEncoding()).attribute("value", hash.getValue());
                });
            } catch (IOException e) {
                throw new RuntimeIoException(e);
            }
        }
    };
    DigitalObjectsExtraction<File> contentAssembler = file -> Collections.singleton(DigitalObject.fromFile(relativePath(file, rootPath), file)).iterator();
    HashAssembler contentHashAssembler = new SingleHashAssembler(HashFunction.SHA256, Encoding.BASE64);
    // Assemble the SIP
    SipAssembler<File> assembler = SipAssembler.forPdiAndContentWithContentHashing(prototype, pdiAssembler, contentAssembler, contentHashAssembler);
    assembler.start(new FileBuffer(sipFile));
    try {
        addFilesIn(new File(rootPath), rootPath, relativePath(sipFile, rootPath), assembler);
    } finally {
        assembler.end();
    }
    // Show metrics about the assembly process
    SipMetrics metrics = assembler.getMetrics();
    System.out.printf("  Added %d files to SIP of %d bytes in %d ms%n", metrics.numDigitalObjects(), metrics.sipFileSize(), metrics.assemblyTime());
}

Example 2 with DigitalObjectsExtraction

use of com.opentext.ia.sdk.sip.DigitalObjectsExtraction in project infoarchive-sip-sdk by Enterprise-Content-Management.

the class YamlSipIngester method run.

private void run(String rootPath) throws IOException {
    System.out.printf("%nSample 3: Assemble SIP from %s and ingest into InfoArchive%n", rootPath);
    // Load the configuration
    YamlConfiguration configuration = new YamlConfiguration(new File(rootPath, "configuration.yml"));
    // Tell InfoArchive where and how to archive the data
    URI entityUri = URI.create(configuration.getPdiSchemaName());
    String entityName = "animal";
    PackagingInformation prototype = PackagingInformation.builder().dss().application(configuration.getApplicationName()).holding(configuration.getHoldingName()).producer("SIP-SDK").entity(entityName).schema(entityUri.toString()).end().build();
    // Define a mapping from our domain object to the PDI XML
    XmlPdiAssembler<File> pdiAssembler;
    try (InputStream schema = new StringStream(configuration.getPdiSchema())) {
        pdiAssembler = new XmlPdiAssembler<File>(entityUri, entityName, schema) {

            @Override
            protected void doAdd(File value, Map<String, ContentInfo> ignored) {
                String name = value.getName();
                getBuilder().element("animal_name", name.substring(0, name.lastIndexOf('.'))).element("file_path", relativePath(value, rootPath));
            }
        };
    }
    DigitalObjectsExtraction<File> contentAssembler = file -> Collections.singleton(DigitalObject.fromFile(relativePath(file, rootPath), file)).iterator();
    // Assemble the SIP
    SipAssembler<File> sipAssembler = SipAssembler.forPdiAndContent(prototype, pdiAssembler, contentAssembler);
    FileGenerator<File> generator = new FileGenerator<>(sipAssembler, FileSupplier.fromTemporaryDirectory());
    File f1 = new File(rootPath, "ape.dat");
    File f2 = new File(rootPath, "bear.dat");
    File f3 = new File(rootPath, "cobra.dat");
    FileGenerationMetrics metrics = generator.generate(Arrays.asList(f1, f2, f3));
    File assembledSip = metrics.getFile();
    // Get an ArchiveClient instance to interact with InfoArchive.
    // In this case, we start with a blank installation, and create a sample holding from scratch that will contain
    // the SIP we've just assembled.
    // Use ArchiveClients.usingAlreadyConfiguredServer() instead if you already configured the server with application,
    // holding, etc.
    ArchiveClient archiveClient = ArchiveClients.configuringApplicationUsing(new YamlBasedApplicationConfigurer(configuration), newArchiveConnection(rootPath));
    // Ingest the SIP into InfoArchive
    try (InputStream sip = Files.newInputStream(assembledSip.toPath(), StandardOpenOption.READ)) {
        String aipId = archiveClient.ingestDirect(sip);
        System.out.println("  SIP ingested as AIP " + aipId);
    }
}