Merge pull request #10598 from Recherche-Data-Gouv/8941-adding-fileCo…

…unt-in-solr

8941 adding file count in solr (v2)

conf/solr/schema.xml

@@ -234,6 +234,7 @@
     <field name="datasetValid" type="boolean" stored="true" indexed="true" multiValued="false"/>
 
     <field name="license" type="string" stored="true" indexed="true" multiValued="false"/>
+    <field name="fileCount" type="plong" stored="true" indexed="true" multiValued="false"/>
 
     <!--
         METADATA SCHEMA FIELDS

doc/release-notes/8941-adding-fileCount-in-solr.md

@@ -0,0 +1,15 @@
+## Release Highlights
+
+### Adding fileCount as SOLR field
+
+A new search field called `fileCount` can be searched to discover the number of files per dataset. (#10598)
+
+## Upgrade Instructions
+
+1. Update your Solr `schema.xml` to include the new field.
+For details, please see https://guides.dataverse.org/en/latest/admin/metadatacustomization.html#updating-the-solr-schema
+
+2. Reindex Solr.
+Once the schema.xml is updated, Solr must be restarted and a reindex initiated.
+For details, see https://guides.dataverse.org/en/latest/admin/solr-search-index.html but here is the reindex command:
+`curl http://localhost:8080/api/admin/index`

src/main/java/edu/harvard/iq/dataverse/api/Search.java

@@ -175,7 +175,7 @@ public Response search(
             JsonArrayBuilder itemsArrayBuilder = Json.createArrayBuilder();
             List<SolrSearchResult> solrSearchResults = solrQueryResponse.getSolrSearchResults();
             for (SolrSearchResult solrSearchResult : solrSearchResults) {
-                itemsArrayBuilder.add(solrSearchResult.json(showRelevance, showEntityIds, showApiUrls, metadataFields, getDatasetFileCount(solrSearchResult)));
+                itemsArrayBuilder.add(solrSearchResult.json(showRelevance, showEntityIds, showApiUrls, metadataFields));
             }
 
             JsonObjectBuilder spelling_alternatives = Json.createObjectBuilder();
@@ -229,15 +229,6 @@ public Response search(
         }
     }
 
-    private Long getDatasetFileCount(SolrSearchResult solrSearchResult) {
-        DvObject dvObject = solrSearchResult.getEntity();
-        if (dvObject.isInstanceofDataset()) {
-            DatasetVersion datasetVersion = ((Dataset) dvObject).getVersionFromId(solrSearchResult.getDatasetVersionId());
-            return datasetVersionFilesServiceBean.getFileMetadataCount(datasetVersion);
-        }
-        return null;
-    }
-
     private User getUser(ContainerRequestContext crc) throws WrappedResponse {
         User userToExecuteSearchAs = GuestUser.get();
         try {

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

@@ -135,6 +135,9 @@ public class IndexServiceBean {
     @EJB
     DatasetFieldServiceBean datasetFieldService;
 
+    @Inject
+    DatasetVersionFilesServiceBean datasetVersionFilesServiceBean;
+
     public static final String solrDocIdentifierDataverse = "dataverse_";
     public static final String solrDocIdentifierFile = "datafile_";
     public static final String solrDocIdentifierDataset = "dataset_";
@@ -1018,6 +1021,8 @@ public SolrInputDocuments toSolrDocs(IndexableDataset indexableDataset, Set<Long
             solrInputDocument.addField(SearchFields.DATASET_CITATION, datasetVersion.getCitation(false));
             solrInputDocument.addField(SearchFields.DATASET_CITATION_HTML, datasetVersion.getCitation(true));
 
+            solrInputDocument.addField(SearchFields.FILE_COUNT, datasetVersionFilesServiceBean.getFileMetadataCount(datasetVersion));
+
             if (datasetVersion.isInReview()) {
                 solrInputDocument.addField(SearchFields.PUBLICATION_STATUS, IN_REVIEW_STRING);
             }

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

@@ -291,5 +291,6 @@ more targeted results for just datasets. The format is YYYY (i.e.
     public static final String DATASET_VALID = "datasetValid";
 
     public static final String DATASET_LICENSE = "license";
+    public static final String FILE_COUNT = "fileCount";
 
 }

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

@@ -497,7 +497,8 @@ public SolrQueryResponse search(
             Long retentionEndDate = (Long) solrDocument.getFieldValue(SearchFields.RETENTION_END_DATE);
             //
             Boolean datasetValid = (Boolean) solrDocument.getFieldValue(SearchFields.DATASET_VALID);
-
+            Long fileCount = (Long) solrDocument.getFieldValue(SearchFields.FILE_COUNT);
+
             List<String> matchedFields = new ArrayList<>();
 
             SolrSearchResult solrSearchResult = new SolrSearchResult(query, name);
@@ -570,6 +571,7 @@ public SolrQueryResponse search(
             solrSearchResult.setDeaccessionReason(deaccessionReason);
             solrSearchResult.setDvTree(dvTree);
             solrSearchResult.setDatasetValid(datasetValid);
+            solrSearchResult.setFileCount(fileCount);
 
             if (Boolean.TRUE.equals((Boolean) solrDocument.getFieldValue(SearchFields.IS_HARVESTED))) {
                 solrSearchResult.setHarvested(true);

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

@@ -78,6 +78,10 @@ public class SolrSearchResult {
     private String citation;
     private String citationHtml;
     private String datasetType;
+    /**
+    * Only Dataset can have a file count
+    */
+    private Long fileCount;
     /**
      * Files and datasets might have a UNF. Dataverses don't.
      */
@@ -456,10 +460,10 @@ public JsonObjectBuilder getJsonForMyData(boolean isValid) {
     } // getJsonForMydata
 
     public JsonObjectBuilder json(boolean showRelevance, boolean showEntityIds, boolean showApiUrls) {
-        return json(showRelevance, showEntityIds, showApiUrls, null, null);
+        return json(showRelevance, showEntityIds, showApiUrls, null);
     }
 
-    public JsonObjectBuilder json(boolean showRelevance, boolean showEntityIds, boolean showApiUrls, List<String> metadataFields, Long datasetFileCount) {
+    public JsonObjectBuilder json(boolean showRelevance, boolean showEntityIds, boolean showApiUrls, List<String> metadataFields) {
         if (this.type == null) {
             return jsonObjectBuilder();
         }
@@ -597,7 +601,7 @@ public JsonObjectBuilder json(boolean showRelevance, boolean showEntityIds, bool
                     subjects.add(subject);
                 }
                 nullSafeJsonBuilder.add("subjects", subjects);
-                nullSafeJsonBuilder.add("fileCount", datasetFileCount);
+                nullSafeJsonBuilder.add("fileCount", this.fileCount);
                 nullSafeJsonBuilder.add("versionId", dv.getId());
                 nullSafeJsonBuilder.add("versionState", dv.getVersionState().toString());
                 if (this.isPublishedState()) {
@@ -1348,4 +1352,12 @@ public boolean isValid(Predicate<SolrSearchResult> canUpdateDataset) {
         }
         return !canUpdateDataset.test(this);
     }
+
+    public Long getFileCount() {
+        return fileCount;
+    }
+
+    public void setFileCount(Long fileCount) {
+        this.fileCount = fileCount;
+    }
 }

src/test/java/edu/harvard/iq/dataverse/search/IndexServiceBeanTest.java

@@ -53,6 +53,7 @@ public void setUp() {
         indexService.dataverseService = Mockito.mock(DataverseServiceBean.class);
         indexService.datasetFieldService = Mockito.mock(DatasetFieldServiceBean.class);
         indexService.datasetVersionService = Mockito.mock(DatasetVersionServiceBean.class);
+        indexService.datasetVersionFilesServiceBean = Mockito.mock(DatasetVersionFilesServiceBean.class);
         BrandingUtil.injectServices(indexService.dataverseService, indexService.settingsService);
 
         Mockito.when(indexService.dataverseService.findRootDataverse()).thenReturn(dataverse);