Dependencies for CITS ERMS

keeps · Nov 28, 2024 · 9a28fc6 · 9a28fc6
1 parent fee73f5
commit 9a28fc6
Show file tree

Hide file tree

Showing 12 changed files with 188 additions and 51 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
@@ -33,7 +33,7 @@ public static IPStatus parse(String value) {
   }
 
   public enum SipType {
-    EARK2S, EARK2
+    EARK2S, EARK2, ERMS
   }
 
 }
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 <[email protected]>
  */
@@ -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 <[email protected]>
+ */
+public class ReportTypeEnums {
+  public enum ReportType {
+    COMMONS_IP("commons-ip"), PYIP("eark-validator");
+
+    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, citserms_v2_1, 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 +131,8 @@ public String asString() {
   public static IPContentType getMIXED() {
     return new IPContentType(IPContentTypeEnum.MIXED);
   }
+  public static IPContentType getDataset() {
+    return new IPContentType(IPContentTypeEnum.DATASET);
+  }
+
 }
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 {
@@ -212,7 +226,6 @@ public Path build(WriteStrategy writeStrategy, final String fileNameWithoutExten
       throw e;
     } finally {
       ModelUtils.deleteBuildDir(buildDir);
-      notifySipBuildPackagingEnded();
     }
   }
 

diff --git a/src/main/java/org/roda_project/commons_ip2/model/impl/eark/EARKUtils.java b/src/main/java/org/roda_project/commons_ip2/model/impl/eark/EARKUtils.java
@@ -162,7 +162,14 @@ protected void addRepresentationsToZipAndMETS(IPInterface ip, List<IPRepresentat
         final boolean isRepresentationsData = (representation.getData() != null && !representation.getData().isEmpty());
         final IPHeader header = new IPHeader(IPEnums.IPStatus.NEW).setAgents(representation.getAgents());
         final MetsWrapper representationMETSWrapper;
-        if (!IPEnums.SipType.EARK2S.equals(sipType)) {
+
+        if (IPEnums.SipType.ERMS.equals(sipType)) {
+          representationMETSWrapper = metsGenerator.generateMETS(representationId, representation.getDescription(),
+            ip.getProfile(), false, Optional.empty(), null, header,
+            mainMETSWrapper.getMets().getMetsHdr().getOAISPACKAGETYPE(), representation.getContentType(),
+            representation.getContentInformationType(), isRepresentationMetadata, isRepresentationMetadataOther,
+            isRepresentationSchemas, isRepresentationDocumentation, false, false, isRepresentationsData);
+        } else if (!IPEnums.SipType.EARK2S.equals(sipType)) {
           representationMETSWrapper = metsGenerator.generateMETS(representationId, representation.getDescription(),
             ip.getProfile(), false, Optional.empty(), null, header,
             mainMETSWrapper.getMets().getMetsHdr().getOAISPACKAGETYPE(), representation.getContentType(),
@@ -178,8 +185,13 @@ protected void addRepresentationsToZipAndMETS(IPInterface ip, List<IPRepresentat
         representationMETSWrapper.getMainDiv().setTYPE(representation.getStatus().asString());
 
         // representation data
-        addRepresentationDataFilesToZipAndMETS(ip, zipEntries, representationMETSWrapper, representation,
-          representationId);
+        if (IPEnums.SipType.ERMS.equals(sipType)) {
+          addRepresentationDataFilesToZipErmsAndMETS(ip, zipEntries, representationMETSWrapper, representation,
+            representationId);
+        } else {
+          addRepresentationDataFilesToZipAndMETS(ip, zipEntries, representationMETSWrapper, representation,
+            representationId);
+        }
 
         // representation descriptive metadata
         addDescriptiveMetadataToZipAndMETS(zipEntries, representationMETSWrapper,
@@ -201,17 +213,62 @@ protected void addRepresentationsToZipAndMETS(IPInterface ip, List<IPRepresentat
           representationId);
 
         // add representation METS to Zip file and to main METS file
-        metsGenerator.addRepresentationMETSToZipAndToMainMETS(zipEntries, mainMETSWrapper, representationId,
-          representationMETSWrapper, IPConstants.REPRESENTATIONS_FOLDER + representationId
-            + IPConstants.ZIP_PATH_SEPARATOR + IPConstants.METS_FILE,
-          buildDir);
+        if (IPEnums.SipType.ERMS.equals(sipType)) {
+          metsGenerator.addRepresentationMETSToZipAndToMainMETS(zipEntries, mainMETSWrapper, representationId,
+            representationMETSWrapper,
+            IPConstants.REPRESENTATIONS_FOLDER + representationId + IPConstants.ZIP_PATH_SEPARATOR + IPConstants.DATA
+              + IPConstants.ZIP_PATH_SEPARATOR + IPConstants.METS_FILE,
+            buildDir);
+        } else {
+          metsGenerator.addRepresentationMETSToZipAndToMainMETS(zipEntries, mainMETSWrapper, representationId,
+            representationMETSWrapper, IPConstants.REPRESENTATIONS_FOLDER + representationId
+              + IPConstants.ZIP_PATH_SEPARATOR + IPConstants.METS_FILE,
+            buildDir);
+        }
 
         metsGenerator.cleanFileGrpStructure();
       }
       if (ip instanceof SIP) {
         ((SIP) ip).notifySipBuildRepresentationsProcessingEnded();
       }
     }
+
+  }
+
+  private void addRepresentationDataFilesToZipErmsAndMETS(IPInterface ip, Map<String, ZipEntryInfo> zipEntries,
+    MetsWrapper representationMETSWrapper, IPRepresentation representation, String representationId)
+    throws InterruptedException, IPException {
+    if (representation.getData() != null && !representation.getData().isEmpty()) {
+      if (ip instanceof SIP sip) {
+        sip.notifySipBuildRepresentationProcessingStarted(representation.getData().size());
+      }
+      int i = 0;
+      for (IPFileInterface file : representation.getData()) {
+        if (Thread.interrupted()) {
+          throw new InterruptedException();
+        }
+
+        if (file instanceof IPFile) {
+          String dataFilePath = ModelUtils.getFoldersFromList(file.getRelativeFolders()) + file.getFileName();
+          FileType fileType = metsGenerator.addDataFileToMETS(representationMETSWrapper, dataFilePath, file.getPath());
+
+          dataFilePath = IPConstants.DATA_FOLDER + dataFilePath;
+          dataFilePath = IPConstants.REPRESENTATIONS_FOLDER + representationId + IPConstants.ZIP_PATH_SEPARATOR
+            + dataFilePath;
+          ZIPUtils.addFileTypeFileToZip(zipEntries, file.getPath(), dataFilePath, fileType);
+        } else if (file instanceof IPFileShallow shallow && (shallow.getFileLocation() != null)) {
+          metsGenerator.addDataFileToMETS(representationMETSWrapper, shallow);
+        }
+
+        i++;
+        if (ip instanceof SIP sip) {
+          sip.notifySipBuildRepresentationProcessingCurrentStatus(i);
+        }
+      }
+      if (ip instanceof SIP sip) {
+        sip.notifySipBuildRepresentationProcessingEnded();
+      }
+    }
   }
 
   protected void addRepresentationDataFilesToZipAndMETS(IPInterface ip, Map<String, ZipEntryInfo> zipEntries,

diff --git a/src/main/java/org/roda_project/commons_ip2/validator/common/FolderManager.java b/src/main/java/org/roda_project/commons_ip2/validator/common/FolderManager.java
@@ -59,6 +59,21 @@ public InputStream getMetsRootInputStream(final Path path) throws FileNotFoundEx
     return new FileInputStream(metsPath);
   }
 
+  public InputStream getErmsInputStream(final Path path) throws FileNotFoundException {
+    folder = path.toFile();
+    String ermsPath = null;
+    for (File f : folder.listFiles()) {
+      if (f.getName().endsWith("erms.xml")) {
+        ermsPath = f.getPath();
+      }
+    }
+    if (ermsPath == null) {
+      LOGGER.debug("ERMS.xml not Found");
+      throw new FileNotFoundException("ERMS.xml not Found");
+    }
+    return new FileInputStream(ermsPath);
+  }
+
   public String getSipRootFolderName(final Path path) {
     final String[] tmp = path.toString().split("/");
     return tmp[tmp.length - 1];

diff --git a/src/main/java/org/roda_project/commons_ip2/validator/common/ZipManager.java b/src/main/java/org/roda_project/commons_ip2/validator/common/ZipManager.java
@@ -79,6 +79,38 @@ public InputStream getMetsRootInputStream(Path path) throws IOException {
     return zipFile.getInputStream(zipArchiveEntry);
   }
 
+  /**
+   * Get ERMS file in IP {@link InputStream}.
+   *
+   * @param path
+   *          {@link Path} to the IP
+   * @return {@link InputStream} to the file.
+   * @throws IOException
+   *           if some I/O error occurs
+   */
+  public InputStream getErmsInputStream(Path path) throws IOException {
+    if (zipFile == null) {
+      zipFile = new ZipFile(path.toFile());
+    }
+
+    Enumeration entries = zipFile.entries();
+    String entry = null;
+    while (entries.hasMoreElements()) {
+      ZipEntry entr = (ZipEntry) entries.nextElement();
+      if (entr.getName().endsWith("erms.xml")) {
+        if (entr.getName().split("/").length == 4) {
+          entry = entr.getName();
+        }
+      }
+    }
+    if (entry == null) {
+      LOGGER.debug("ERMS.xml not Found");
+      throw new IOException("ERMS.xml not Found");
+    }
+    ZipEntry zipArchiveEntry = zipFile.getEntry(entry);
+    return zipFile.getInputStream(zipArchiveEntry);
+  }
+
   public Enumeration getEntries() {
     return zipFile.entries();
   }

diff --git a/src/main/resources/controlledVocabularies/CSIPVocabularyContentCategory.xml b/src/main/resources/controlledVocabularies/CSIPVocabularyContentCategory.xml
@@ -66,6 +66,10 @@
             <Term lang="en">Datasets</Term>
             <Definition>Data encoded in a defined structure.</Definition>
         </Entry>
+        <Entry>
+            <Term lang="en">Dataset</Term>
+            <Definition>Data encoded in a defined structure.</Definition>
+        </Entry>
         <Entry>
             <Term lang="en">Geospatial Data</Term>
             <Definition></Definition>