prevent orig file size from being added to tabular files #6118

IQSS · Jul 16, 2020 · 115353c · 115353c
1 parent 1957776
commit 115353c
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 7 deletions.
diff --git a/src/main/java/edu/harvard/iq/dataverse/DatasetPage.java b/src/main/java/edu/harvard/iq/dataverse/DatasetPage.java
@@ -2903,7 +2903,8 @@ public void setSelectedNonDownloadallableFiles(List<FileMetadata> selectedNonDow
     public String getSizeOfDataset() {
         boolean countCachedFiles = false;
         boolean useOrigFileSize = false;
-        GetDatasetStorageSizeCommand cmd = new GetDatasetStorageSizeCommand(dvRequestService.getDataverseRequest(), dataset, countCachedFiles, useOrigFileSize, GetDatasetStorageSizeCommand.Mode.DOWNLOAD, workingVersion);
+        boolean dbOnly = true;
+        GetDatasetStorageSizeCommand cmd = new GetDatasetStorageSizeCommand(dvRequestService.getDataverseRequest(), dataset, countCachedFiles, dbOnly, useOrigFileSize, GetDatasetStorageSizeCommand.Mode.DOWNLOAD, workingVersion);
         try {
             long bytes = commandEngine.submit(cmd);
             return FileSizeChecker.bytesToHumanReadable(bytes);
@@ -2915,7 +2916,8 @@ public String getSizeOfDataset() {
     public String getSizeOfDatasetOrig() {
         boolean countCachedFiles = false;
         boolean useOrigFileSize = true;
-        GetDatasetStorageSizeCommand cmd = new GetDatasetStorageSizeCommand(dvRequestService.getDataverseRequest(), dataset, countCachedFiles, useOrigFileSize, GetDatasetStorageSizeCommand.Mode.DOWNLOAD, workingVersion);
+        boolean dbOnly = true;
+        GetDatasetStorageSizeCommand cmd = new GetDatasetStorageSizeCommand(dvRequestService.getDataverseRequest(), dataset, countCachedFiles, useOrigFileSize, dbOnly, GetDatasetStorageSizeCommand.Mode.DOWNLOAD, workingVersion);
         try {
             long bytes = commandEngine.submit(cmd);
             return FileSizeChecker.bytesToHumanReadable(bytes);

diff --git a/src/main/java/edu/harvard/iq/dataverse/DatasetServiceBean.java b/src/main/java/edu/harvard/iq/dataverse/DatasetServiceBean.java
@@ -882,7 +882,8 @@ public long findStorageSize(Dataset dataset, boolean countCachedExtras) throws I
 
     public long findStorageSize(Dataset dataset, boolean countCachedExtras, GetDatasetStorageSizeCommand.Mode mode, DatasetVersion version) throws IOException {
         boolean useOrigFileSize = false;
-        return findStorageSize(dataset, countCachedExtras, useOrigFileSize, mode, version);
+        boolean dbOnly = false;
+        return findStorageSize(dataset, countCachedExtras, useOrigFileSize, dbOnly, mode, version);
     }
     /**
      * Returns the total byte size of the files in this dataset 
@@ -892,13 +893,14 @@ public long findStorageSize(Dataset dataset, boolean countCachedExtras, GetDatas
      * @param useOrigFileSize allows original tabular file size to be used instead of derived archival file
      * @param mode String indicating whether we are getting the result for storage (entire dataset) or download version based
      * @param version optional param for dataset version
+     * @param dbOnly only get bytes from database, no countCachedExtras processing
      * @return total size 
      * @throws IOException if it can't access the objects via StorageIO 
      * (in practice, this can only happen when called with countCachedExtras=true; when run in the 
      * default mode, the method doesn't need to access the storage system, as the 
      * sizes of the main files are recorded in the database)
      */
-    public long findStorageSize(Dataset dataset, boolean countCachedExtras, boolean useOrigFileSize, GetDatasetStorageSizeCommand.Mode mode, DatasetVersion version) throws IOException {
+    public long findStorageSize(Dataset dataset, boolean countCachedExtras, boolean useOrigFileSize, boolean dbOnly, GetDatasetStorageSizeCommand.Mode mode, DatasetVersion version) throws IOException {
         long total = 0L; 
 
         if (dataset.isHarvested()) {
@@ -935,8 +937,13 @@ public long findStorageSize(Dataset dataset, boolean countCachedExtras, boolean
                 total += datafile.getFilesize();
             }
 
+            if (dbOnly) {
+                // Skip the rest. Don't add any more to the total. No cached extras.
+                continue;
+            }
+
             if (!countCachedExtras) {
-                if (!useOrigFileSize && datafile.isTabularData()) {
+                if (datafile.isTabularData()) {
                     // count the size of the stored original, in addition to the main tab-delimited file:
                     Long originalFileSize = datafile.getDataTable().getOriginalFileSize();
                     if (originalFileSize != null) {

diff --git a/src/main/java/edu/harvard/iq/dataverse/engine/command/impl/GetDatasetStorageSizeCommand.java b/src/main/java/edu/harvard/iq/dataverse/engine/command/impl/GetDatasetStorageSizeCommand.java
@@ -33,6 +33,7 @@ public class GetDatasetStorageSizeCommand extends AbstractCommand<Long> {
     private final Dataset dataset;
     private final Boolean countCachedFiles;
     private final boolean useOrigFileSize;
+    private final boolean dbOnly;
     private final Mode mode;
     private final DatasetVersion version;
 
@@ -46,6 +47,7 @@ public GetDatasetStorageSizeCommand(DataverseRequest aRequest, Dataset target) {
         dataset = target;
         countCachedFiles = false;
         this.useOrigFileSize = false;
+        this.dbOnly = false;
         mode = Mode.DOWNLOAD;
         version = null;
     }
@@ -55,15 +57,17 @@ public GetDatasetStorageSizeCommand(DataverseRequest aRequest, Dataset target, b
         dataset = target;
         this.countCachedFiles = countCachedFiles;
         this.useOrigFileSize = false;
+        this.dbOnly = false;
         this.mode = mode;
         this.version = version;
     }
 
-    public GetDatasetStorageSizeCommand(DataverseRequest aRequest, Dataset target, boolean countCachedFiles, boolean useOrigFileSize, Mode mode, DatasetVersion version) {
+    public GetDatasetStorageSizeCommand(DataverseRequest aRequest, Dataset target, boolean countCachedFiles, boolean useOrigFileSize, boolean dbOnly, Mode mode, DatasetVersion version) {
         super(aRequest, target);
         dataset = target;
         this.countCachedFiles = countCachedFiles;
         this.useOrigFileSize = useOrigFileSize;
+        this.dbOnly = dbOnly;
         this.mode = mode;
         this.version = version;
     }
@@ -78,7 +82,7 @@ public Long execute(CommandContext ctxt) throws CommandException {
         }
 
         try {
-            return ctxt.datasets().findStorageSize(dataset, countCachedFiles, useOrigFileSize, mode, version);
+            return ctxt.datasets().findStorageSize(dataset, countCachedFiles, useOrigFileSize, dbOnly, mode, version);
         } catch (IOException ex) {
             throw new CommandException(BundleUtil.getStringFromBundle("datasets.api.datasize.ioerror"), this);
         }