Dependencies for CITS SIARD

keeps · Nov 28, 2024 · 27b515c · 27b515c
1 parent fee73f5
commit 27b515c
Show file tree

Hide file tree

Showing 13 changed files with 302 additions and 57 deletions.
diff --git a/src/main/java/org/roda_project/commons_ip/utils/IPEnums.java b/src/main/java/org/roda_project/commons_ip/utils/IPEnums.java
@@ -9,7 +9,7 @@
 
 public class IPEnums {
   public enum IPType {
-    SIP, AIP, DIP;
+    SIP, AIP, DIP, SIARD;
   }
 
   public enum IPStatus {
@@ -33,7 +33,7 @@ public static IPStatus parse(String value) {
   }
 
   public enum SipType {
-    EARK2S, EARK2
+    EARK2S, EARK2, SIARD
   }
 
 }
diff --git a/src/main/java/org/roda_project/commons_ip2/cli/Validate.java b/src/main/java/org/roda_project/commons_ip2/cli/Validate.java
@@ -11,7 +11,7 @@
 import javax.xml.parsers.ParserConfigurationException;
 
 import org.roda_project.commons_ip2.cli.model.ExitCodes;
-import org.roda_project.commons_ip2.cli.model.enums.ReportType;
+import org.roda_project.commons_ip2.cli.model.enums.ReportTypeEnums;
 import org.roda_project.commons_ip2.cli.model.exception.CLIException;
 import org.roda_project.commons_ip2.cli.model.exception.ValidationException;
 import org.roda_project.commons_ip2.cli.utils.CLI.ValidateCommandUtils;
@@ -27,6 +27,9 @@
 
 import picocli.CommandLine;
 
+import static org.roda_project.commons_ip2.cli.model.enums.ReportTypeEnums.ReportType.COMMONS_IP;
+import static org.roda_project.commons_ip2.cli.model.enums.ReportTypeEnums.ReportType.PYIP;
+
 /**
  * @author Miguel Guimarães <mguimaraes@keep.pt>
  */
@@ -48,7 +51,7 @@ public class Validate implements Callable<Integer> {
 
   @CommandLine.Option(names = {"-r",
     "--reporter-type"}, paramLabel = "<type>", description = "Report type (possible values: ${COMPLETION-CANDIDATES})")
-  ReportType reportType = ReportType.COMMONS_IP;
+  ReportTypeEnums.ReportType reportType = COMMONS_IP;
 
   @CommandLine.Option(names = {"-v",
     "--verbose"}, description = "Verbose command line output with all validation steps")
@@ -72,7 +75,7 @@ public Integer call() throws ValidationException, CLIException {
     return ExitCodes.EXIT_CODE_OK;
   }
 
-  private void handleSipValidation(final String sip, final String reportPathDir, final ReportType reportType,
+  private void handleSipValidation(final String sip, final String reportPathDir, final ReportTypeEnums.ReportType reportType,
     final boolean verbose)
     throws IOException, ParserConfigurationException, SAXException, CLIException, NoSuchAlgorithmException {
     final Path sipPath = Paths.get(sip);
@@ -83,27 +86,25 @@ private void handleSipValidation(final String sip, final String reportPathDir, f
 
     LogSystem.logOperatingSystemInfo();
     LOGGER.debug("command executed: {}", commandLineString);
-    switch (reportType) {
-      case COMMONS_IP -> {
-        final OutputStream outputStream = ValidateCommandUtils.createReportOutputStream(reportPath);
-        if (outputStream != null) {
-          final ValidationReportOutputJson jsonReporter = new ValidationReportOutputJson(sipPath, outputStream);
-          final EARKSIPValidator earksipValidator = new EARKSIPValidator(jsonReporter, version);
-          if (verbose) {
-            earksipValidator.addObserver(new ProgressValidationLoggerObserver());
-          }
-          earksipValidator.validate(version);
-        }
-      }
-      case PYIP -> {
-        final ValidationReportOutputJSONPyIP jsonReporter = new ValidationReportOutputJSONPyIP(reportPath, sipPath);
-        final EARKPyIPValidator earkPyIPValidator = new EARKPyIPValidator(jsonReporter, version);
+    if (reportType.equals(COMMONS_IP)) {
+      final OutputStream outputStream = ValidateCommandUtils.createReportOutputStream(reportPath);
+      if (outputStream != null) {
+        final ValidationReportOutputJson jsonReporter = new ValidationReportOutputJson(sipPath, outputStream);
+        final EARKSIPValidator earksipValidator = new EARKSIPValidator(jsonReporter, version);
         if (verbose) {
-          earkPyIPValidator.addObserver(new ProgressValidationLoggerObserver());
+          earksipValidator.addObserver(new ProgressValidationLoggerObserver());
         }
-        earkPyIPValidator.validate();
+        earksipValidator.validate(version);
+      }
+    } else if (reportType.equals(PYIP)) {
+      final ValidationReportOutputJSONPyIP jsonReporter = new ValidationReportOutputJSONPyIP(reportPath, sipPath);
+      final EARKPyIPValidator earkPyIPValidator = new EARKPyIPValidator(jsonReporter, version);
+      if (verbose) {
+        earkPyIPValidator.addObserver(new ProgressValidationLoggerObserver());
       }
-      default -> throw new CLIException("Unexpected value: " + reportType);
+      earkPyIPValidator.validate();
+    } else {
+      throw new CLIException("Unexpected value: " + reportType);
     }
     new CommandLine(this).getOut().printf("E-ARK SIP validation report at '%s'%n",
       reportPath.normalize().toAbsolutePath());

diff --git a/src/main/java/org/roda_project/commons_ip2/cli/model/enums/ReportType.java b/src/main/java/org/roda_project/commons_ip2/cli/model/enums/ReportType.java
diff --git a/src/main/java/org/roda_project/commons_ip2/cli/model/enums/ReportTypeEnums.java b/src/main/java/org/roda_project/commons_ip2/cli/model/enums/ReportTypeEnums.java
@@ -0,0 +1,21 @@
+package org.roda_project.commons_ip2.cli.model.enums;
+
+/**
+ * @author Miguel Guimarães <mguimaraes@keep.pt>
+ */
+public class ReportTypeEnums {
+  public enum ReportType {
+    COMMONS_IP("commons-ip"), PYIP("eark-validator"), SIARD("siard");
+
+    private final String type;
+
+    ReportType(String type) {
+      this.type = type;
+    }
+
+    @Override
+    public String toString() {
+      return type;
+    }
+  }
+}
diff --git a/src/main/java/org/roda_project/commons_ip2/cli/utils/SIPBuilderUtils.java b/src/main/java/org/roda_project/commons_ip2/cli/utils/SIPBuilderUtils.java
@@ -91,6 +91,15 @@ private static void addRepresentationToSIP(SIP sip, final RepresentationGroup re
     }
   }
 
+  public static void addRepresentationGroupsToErmsSIP(SIP sip, final List<RepresentationGroup> representationGroups,
+    boolean targetOnly) throws IPException {
+    if (representationGroups != null) {
+      for (RepresentationGroup representationGroup : representationGroups) {
+        addRepresentationToSIP(sip, representationGroup, targetOnly);
+      }
+    }
+  }
+
   private static void addFileToRepresentation(final IPRepresentation representation, final boolean targetOnly,
     final Path dataPath, final List<String> relativePath) {
     if (Files.isDirectory(dataPath)) {

diff --git a/src/main/java/org/roda_project/commons_ip2/model/IPContentInformationType.java b/src/main/java/org/roda_project/commons_ip2/model/IPContentInformationType.java
@@ -15,7 +15,7 @@ public class IPContentInformationType implements Serializable {
   private static final long serialVersionUID = 1191075605637022551L;
 
   public enum IPContentInformationTypeEnum {
-    ERMS, SIARD1, SIARD2, SIARDDK, GEODATA, MIXED, OTHER;
+    ERMS, citssiard_v1_0, SIARD1, SIARD2, SIARDDK, GEODATA, MIXED, OTHER, Dataset;
   }
 
   private IPContentInformationTypeEnum type;

diff --git a/src/main/java/org/roda_project/commons_ip2/model/IPContentType.java b/src/main/java/org/roda_project/commons_ip2/model/IPContentType.java
@@ -26,7 +26,7 @@ public enum IPContentTypeEnum {
     AUDIO_ON_TANGIBLE_MEDIUM("Audio – On Tangible Medium (digital or analog)"),
     AUDIO_MEDIA_INDEPENDENT("Audio – Media-independent (digital)"),
     MOTION_PICTURES("Motion Pictures – Digital and Physical Media"), VIDEO("Video – File-based and Physical Media"),
-    SOFTWARE("Software"), DATASETS("Datasets"), GEOSPATIAL_DATA("Geospatial Data"), DATABASES("Databases"),
+    SOFTWARE("Software"), DATASETS("Datasets"), DATASET("Dataset"), GEOSPATIAL_DATA("Geospatial Data"), DATABASES("Databases"),
     WEBSITES("Websites"), COLLECTION("Collection"), EVENT("Event"), INTERACTIVE_RESOURCE("Interactive resource"),
     PHYSICAL_OBJECT("Physical object"), SERVICE("Service"), MIXED("Mixed"), OTHER("Other");
 
@@ -131,4 +131,12 @@ public String asString() {
   public static IPContentType getMIXED() {
     return new IPContentType(IPContentTypeEnum.MIXED);
   }
+  public static IPContentType getDataset() {
+    return new IPContentType(IPContentTypeEnum.DATASET);
+  }
+
+  public static IPContentType getDatabase() {
+    return new IPContentType(IPContentTypeEnum.DATABASES);
+  }
+
 }
diff --git a/src/main/java/org/roda_project/commons_ip2/model/impl/eark/EARKMETSCreator.java b/src/main/java/org/roda_project/commons_ip2/model/impl/eark/EARKMETSCreator.java
@@ -11,6 +11,7 @@
 
 import javax.xml.datatype.DatatypeConfigurationException;
 import javax.xml.datatype.XMLGregorianCalendar;
+import javax.xml.namespace.QName;
 
 import org.apache.commons.lang3.StringUtils;
 import org.roda_project.commons_ip.utils.IPEnums;
@@ -57,6 +58,54 @@ public abstract class EARKMETSCreator {
   private static final Logger LOGGER = LoggerFactory.getLogger(EARKMETSCreator.class);
   private final Map<String, MetsType.FileSec.FileGrp> dataFileGrp = new HashMap<>();
 
+  public MetsWrapper generateMetsSiard(final String id, final String label, final String profile, final boolean mainMets,
+                                  final Optional<List<String>> ancestors, final Path metsPath, final IPHeader ipHeader, final String type,
+                                  final IPContentType contentType, final IPContentInformationType contentInformationType, final boolean isMetadata,
+                                  final boolean isMetadataOther, final boolean isSchemas, final boolean isDocumentation, final boolean isSubmission,
+                                  final boolean isRepresentations, final boolean isRepresentationsData) throws IPException {
+    final Mets mets = new Mets();
+    final MetsWrapper metsWrapper = new MetsWrapper(mets, metsPath);
+
+    // basic attributes
+    addBasicAttributesToMets(mets, id, label, profile, contentType, contentInformationType);
+    // header
+
+    addHeaderToMets(mets, ipHeader, type);
+
+    // administrative section
+    addAmdSecToMets(mets);
+
+    // file section
+
+    final MetsType.FileSec fileSec = createFileSec();
+
+    // Add data file grp
+    addDataFileGrpToMets(metsWrapper, fileSec, mainMets, isRepresentationsData);
+
+    // Add schemas, documentation, submission to main div
+    addCommonFileGrpToMets(metsWrapper, fileSec, isSchemas, isSubmission, isDocumentation, type);
+
+    if ((mainMets && isRepresentations) || !fileSec.getFileGrp().isEmpty()) {
+      mets.setFileSec(fileSec);
+    }
+
+    // E-ARK struct map
+    final StructMapType structMap = createStructMap();
+
+    final DivType mainDiv = addCommonDivsToMainDiv(metsWrapper, id, isMetadata, isMetadataOther, isSchemas,
+      isDocumentation, isSubmission, type);
+
+    // data div
+    addDataDivToMets(metsWrapper, mainDiv, mainMets, isRepresentationsData);
+
+    structMap.setDiv(mainDiv);
+    mets.getStructMap().add(structMap);
+
+    addAncestorsToMets(mets, ancestors);
+
+    return metsWrapper;
+  }
+
   public MetsWrapper generateMETS(final String id, final String label, final String profile, final boolean mainMets,
     final Optional<List<String>> ancestors, final Path metsPath, final IPHeader ipHeader, final String type,
     final IPContentType contentType, final IPContentInformationType contentInformationType, final boolean isMetadata,
@@ -251,6 +300,47 @@ public void addRepresentationMETSToZipAndToMainMETS(final Map<String, ZipEntryIn
     }
   }
 
+  public void addRepresentationSiardMETSToZipAndToMainMETS(final Map<String, ZipEntryInfo> zipEntries,
+    final MetsWrapper mainMETSWrapper, final String representationId, final MetsWrapper representationMETSWrapper,
+    final String representationMetsPath, final Path buildDir) throws IPException, InterruptedException {
+    try {
+      if (Thread.interrupted()) {
+        throw new InterruptedException();
+      }
+
+      // create mets pointer
+      final DivType.Mptr mptr = new DivType.Mptr();
+      mptr.setLOCTYPE(METSEnums.LocType.URL.toString());
+      mptr.setType(IPConstants.METS_TYPE_SIMPLE);
+      mptr.setHref(METSUtils.encodeHref(representationMetsPath));
+
+
+      // create file
+      final FileType fileType = new FileType();
+      fileType.setID(Utils.generateRandomAndPrefixedFileID());
+
+      addMETSToZip(zipEntries, representationMETSWrapper, representationMetsPath, buildDir, false, fileType);
+
+      // add to file group and then to file section
+      final MetsType.FileSec.FileGrp fileGrp = createFileGroup(
+        IPConstants.REPRESENTATIONS_WITH_FIRST_LETTER_CAPITAL + "/" + representationId);
+      final FileType.FLocat fileLocation = METSUtils.createFileLocation(representationMetsPath);
+      fileType.getFLocat().add(fileLocation);
+      fileGrp.getFile().add(fileType);
+      fileGrp.getOtherAttributes().put(QName.valueOf("csip:CONTENTINFORMATIONTYPE"), "citssiard_v1_0");
+      fileGrp.getOtherAttributes().put(QName.valueOf("csip:OTHERCONTENTINFORMATIONTYPE"), mainMETSWrapper.getMets().getOTHERCONTENTINFORMATIONTYPE());
+
+      mainMETSWrapper.getMets().getFileSec().getFileGrp().add(fileGrp);
+
+      // set mets pointer
+      final DivType representationDiv = createRepresentationDivForStructMap(representationId, mptr);
+      mptr.setTitle(fileGrp.getID());
+      mainMETSWrapper.getMainDiv().getDiv().add(representationDiv);
+    } catch (JAXBException | IOException e) {
+      throw new IPException("Error saving representation METS", e);
+    }
+  }
+
   protected void addMETSToZip(final Map<String, ZipEntryInfo> zipEntries, final MetsWrapper metsWrapper,
     final String metsPath, final Path buildDir, final boolean mainMets, final FileType fileType)
     throws JAXBException, IOException, IPException {

diff --git a/src/main/java/org/roda_project/commons_ip2/model/impl/eark/EARKSIP.java b/src/main/java/org/roda_project/commons_ip2/model/impl/eark/EARKSIP.java
@@ -147,6 +147,20 @@ public Path build(WriteStrategy writeStrategy) throws IPException, InterruptedEx
     return build(writeStrategy, false);
   }
 
+  /**
+   * Builds a SIP.
+   *
+   * @param destinationDirectory
+   *          the {@link Path} where the SIP should be build.
+   * @param onlyManifest
+   *          build only the manifest file? (<strong>this parameter is
+   *          ignored</strong>).
+   * @return the {@link Path}.
+   * @throws IPException
+   *           if some error occurs.
+   * @throws InterruptedException
+   *           if some error occurs.
+   */
   @Override
   public Path build(WriteStrategy writeStrategy, final boolean onlyManifest)
     throws IPException, InterruptedException {
@@ -190,12 +204,19 @@ public Path build(WriteStrategy writeStrategy, final String fileNameWithoutExten
       boolean isDocumentation = (this.getDocumentation() != null && !this.getDocumentation().isEmpty());
       boolean isSchemas = (this.getSchemas() != null && !this.getSchemas().isEmpty());
       boolean isRepresentations = (this.getRepresentations() != null && !this.getRepresentations().isEmpty());
-
-      MetsWrapper mainMETSWrapper = metsCreator.generateMETS(StringUtils.join(this.getIds(), " "),
-        this.getDescription(), this.getProfile(), true, Optional.ofNullable(this.getAncestors()), null,
-        this.getHeader(), this.getType(), this.getContentType(), this.getContentInformationType(), isMetadata,
-        isMetadataOther, isSchemas, isDocumentation, false, isRepresentations, false);
-
+      MetsWrapper mainMETSWrapper;
+
+      if (this.getType().equals("SIARD")) {
+        mainMETSWrapper = metsCreator.generateMetsSiard(StringUtils.join(this.getIds(), " "), this.getDescription(),
+          this.getProfile(), true, Optional.ofNullable(this.getAncestors()), null, this.getHeader(), this.getType(),
+          this.getContentType(), this.getContentInformationType(), isMetadata, isMetadataOther, isSchemas,
+          isDocumentation, false, isRepresentations, false);
+      } else {
+        mainMETSWrapper = metsCreator.generateMETS(StringUtils.join(this.getIds(), " "), this.getDescription(),
+          this.getProfile(), true, Optional.ofNullable(this.getAncestors()), null, this.getHeader(), this.getType(),
+          this.getContentType(), this.getContentInformationType(), isMetadata, isMetadataOther, isSchemas,
+          isDocumentation, false, isRepresentations, false);
+      }
       earkUtils.addDescriptiveMetadataToZipAndMETS(zipEntries, mainMETSWrapper, getDescriptiveMetadata(), null);
       earkUtils.addPreservationMetadataToZipAndMETS(zipEntries, mainMETSWrapper, getPreservationMetadata(), null);
       earkUtils.addOtherMetadataToZipAndMETS(zipEntries, mainMETSWrapper, getOtherMetadata(), null);