Merge pull request #11114 from QualitativeDataRepository/IQSS/11113-F…

…ix_file_perm_doc_creation

IQSS/11113 fix file perm doc creation

doc/release-notes/11113-avoid-orphan-perm-docs.md

@@ -0,0 +1,5 @@
+This release fixes a bug that caused Dataverse to generate unnecessary solr documents for files when a file is added/deleted from a draft dataset. These documents could accumulate and potentially impact performance.
+
+Assuming the upgrade to solr 9.7.0 also occurs in this release, there's nothing else needed for this PR. (Starting with a new solr insures the solr db is empty and that a reindex is already required.)
+
+

src/main/java/edu/harvard/iq/dataverse/DataFile.java

@@ -1142,4 +1142,12 @@ public boolean isDeaccessioned() {
         }
         return inDeaccessionedVersions; // since any published version would have already returned
     }
+    public boolean isInDatasetVersion(DatasetVersion version) {
+        for (FileMetadata fmd : getFileMetadatas()) {
+            if (fmd.getDatasetVersion().equals(version)) {
+                return true;
+            }
+        }
+        return false;
+    }
 } // end of class

src/main/java/edu/harvard/iq/dataverse/search/SolrIndexServiceBean.java

@@ -155,7 +155,15 @@ private List<DvObjectSolrDoc> constructDatafileSolrDocs(DataFile dataFile, Map<L
         Map<DatasetVersion.VersionState, Boolean> desiredCards = searchPermissionsService.getDesiredCards(dataFile.getOwner());
         for (DatasetVersion datasetVersionFileIsAttachedTo : datasetVersionsToBuildCardsFor(dataFile.getOwner())) {
             boolean cardShouldExist = desiredCards.get(datasetVersionFileIsAttachedTo.getVersionState());
-            if (cardShouldExist) {
+            /*
+             * Since datasetVersionFileIsAttachedTo should be a draft or the most recent
+             * released one, it could be more efficient to stop the search through
+             * FileMetadatas after those two (versus continuing through all prior versions
+             * as in isInDatasetVersion). Alternately, perhaps filesToReIndexPermissionsFor
+             * should not combine the list of files for the different datsetversions into a
+             * single list to start with.
+             */ 
+            if (cardShouldExist && dataFile.isInDatasetVersion(datasetVersionFileIsAttachedTo)) {
                 String solrIdStart = IndexServiceBean.solrDocIdentifierFile + dataFile.getId();
                 String solrIdEnd = getDatasetOrDataFileSolrEnding(datasetVersionFileIsAttachedTo.getVersionState());
                 String solrId = solrIdStart + solrIdEnd;