Beta Candidate 1, May 8, 4:15pm

conf/solr/4.6.0/schema.xml

@@ -269,6 +269,10 @@
    <field name="description" type="text_en" multiValued="false" stored="true" indexed="true"/>
    <copyField source="description" dest="text" maxChars="3000"/>
 
+   <!-- Added for Dataverse 4.0 Beta: make variable names and labels searchable in basic search https://redmine.hmdc.harvard.edu/issues/3945 -->
+   <copyField source="variable_name_en" dest="text" maxChars="3000"/>
+   <copyField source="variable_label_en" dest="text" maxChars="3000"/>
+
    <!-- Added for Dataverse 4.0 alpha 1 from output of http://localhost:8080/api/config/solr/schema -->
    <field name="accessToSources" type="text_en" multiValued="false" stored="true" indexed="true"/>
    <field name="actionsToMinimizeLoss" type="text_en" multiValued="false" stored="true" indexed="true"/>
@@ -412,7 +416,8 @@
    <!-- <copyField source="cat" dest="text"/> -->
    <!-- Dataverse 4.0: we want the "name" field in the "catchall"  -->
    <copyField source="name" dest="text"/>
-   <!-- Dataverse 4.0: we want the "filename_without_extension_en" field in the "catchall" per https://redmine.hmdc.harvard.edu/issues/3848  -->
+   <!-- Dataverse 4.0: we want the "filetype_en" and "filename_without_extension_en" field in the "catchall" per https://redmine.hmdc.harvard.edu/issues/3848  -->
+   <copyField source="filetype_en" dest="text"/>
    <copyField source="filename_without_extension_en" dest="text"/>
    <!-- <copyField source="manu" dest="text"/> -->
    <!-- <copyField source="features" dest="text"/> -->

doc/Sphinx/source/User/appendix.rst

@@ -16,7 +16,7 @@ currently proposed metadata fields for 4.0 (per metadata block):
 - `Citation Metadata <https://docs.google.com/spreadsheet/ccc?key=0AjeLxEN77UZodDBaYTFPakhGaEpoa3hqZUJTOWZtclE&usp=sharing>`__ (compliant with `DDI 2.5 <http://www.ddialliance.org/>`__ and `DataCite 3.0 <http://schema.datacite.org/meta/kernel-3/index.html>`__) 
 - `Social Science & Humanities Metadata (DDI 2.5 compliant) <https://docs.google.com/spreadsheet/ccc?key=0AjeLxEN77UZodEppcTFHT1NnajNLV0tacE10NEdmUnc&usp=sharing>`__
 - `Astronomy and Astrophysics Metadata <https://docs.google.com/spreadsheet/ccc?key=0AjeLxEN77UZodEp4Qmp0QURkUWo1S0t4X3hia0FnZUE&usp=sharing>`__
-  (based on `Virtual Observatory (VO) Discovery and Provenance Metadata <http://www.wf4ever-project.org/wiki/download/attachments/1179927/DPmetadata.pdf?version=1&modificationDate=1337186963000>`__) 
+  : These metadata elements can be mapped/exported to the International Virtual Observatory Alliance’s (IVOA) Resource Metadata for the Virtual Observatory (VOResource Schema format) and based on `Virtual Observatory (VO) Discovery and Provenance Metadata <http://www.wf4ever-project.org/wiki/download/attachments/1179927/DPmetadata.pdf?version=1&modificationDate=1337186963000>`__) 
 - `Biomedical Metadata <https://docs.google.com/spreadsheet/ccc?key=0AjeLxEN77UZodExsRTB2SEpVWWd1Qmx6M09HSkExd3c&usp=sharing>`__ 
   (based on `ISA-Tab <http://isatab.sourceforge.net/format.html>`__ and `Stem Cell Commons <http://stemcellcommons.org/>`__)
 

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

@@ -35,6 +35,8 @@ public class AdvancedSearchPage {
     private String fileFieldName;
     private String fileFieldDescription;
     private String fileFieldFiletype;
+    private String fileFieldVariableName;
+    private String fileFieldVariableLabel;
 
     public void init() {
         /**
@@ -115,6 +117,14 @@ private String constructFileQuery() {
             queryStrings.add(constructQuery(SearchFields.FILE_TYPE_SEARCHABLE, fileFieldFiletype));
         }
 
+        if (!fileFieldVariableName.isEmpty()) {
+            queryStrings.add(constructQuery(SearchFields.VARIABLE_NAME, fileFieldVariableName));
+        }
+
+        if (!fileFieldVariableLabel.isEmpty()) {
+            queryStrings.add(constructQuery(SearchFields.VARIABLE_LABEL, fileFieldVariableLabel));
+        }
+
         return constructQuery(queryStrings, true);
     }
 
@@ -256,4 +266,20 @@ public void setFileFieldFiletype(String fileFieldFiletype) {
         this.fileFieldFiletype = fileFieldFiletype;
     }
 
+    public String getFileFieldVariableName() {
+        return fileFieldVariableName;
+    }
+
+    public void setFileFieldVariableName(String fileFieldVariableName) {
+        this.fileFieldVariableName = fileFieldVariableName;
+    }
+
+    public String getFileFieldVariableLabel() {
+        return fileFieldVariableLabel;
+    }
+
+    public void setFileFieldVariableLabel(String fileFieldVariableLabel) {
+        this.fileFieldVariableLabel = fileFieldVariableLabel;
+    }
+
 }

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

@@ -322,7 +322,7 @@ public void refresh(ActionEvent e) {
         // .getLatestVersion().getFileMetadatas() - because that's how the page is
         // accessing them. -- L.A.)
         //for (DataFile dataFile : dataset.getFiles()) {
-        for (FileMetadata fileMetadata : dataset.getLatestVersion().getFileMetadatas()) {
+        for (FileMetadata fileMetadata : getDisplayVersion().getFileMetadatas()) {
             DataFile dataFile = fileMetadata.getDataFile();
             // and see if any are marked as "ingest-in-progress":
             if (dataFile.isIngestInProgress()) {

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

@@ -714,6 +714,7 @@ private String addOrUpdateDataset(IndexableDataset indexableDataset) {
             datafileSolrInputDocument.addField(SearchFields.FILE_TYPE, FileUtil.getFacetFileType(fileMetadata.getDataFile()));
             datafileSolrInputDocument.addField(SearchFields.FILE_TYPE_SEARCHABLE, FileUtil.getFacetFileType(fileMetadata.getDataFile()));
             datafileSolrInputDocument.addField(SearchFields.DESCRIPTION, fileMetadata.getDescription());
+            datafileSolrInputDocument.addField(SearchFields.FILE_DESCRIPTION, fileMetadata.getDescription());
             datafileSolrInputDocument.addField(SearchFields.SUBTREE, dataversePaths);
 //            datafileSolrInputDocument.addField(SearchFields.HOST_DATAVERSE, dataFile.getOwner().getOwner().getName());
            // datafileSolrInputDocument.addField(SearchFields.PARENT_NAME, dataFile.getDataset().getTitle());
@@ -727,11 +728,8 @@ private String addOrUpdateDataset(IndexableDataset indexableDataset) {
 
             if (fileMetadata.getDataFile().isTabularData()) {
                 List<DataVariable> variables = fileMetadata.getDataFile().getDataTable().getDataVariables();
-                String variableNamesToIndex = null;
-                String variableLabelsToIndex = null; 
                 for (DataVariable var : variables) {
                     // Hard-coded search fields, for now: 
-                    // TODO: immediately: define these as constants in SearchFields;
                     // TODO: eventually: review, decide how datavariables should
                     // be handled for indexing purposes. (should it be a fixed
                     // setup, defined in the code? should it be flexible? unlikely
@@ -741,28 +739,12 @@ private String addOrUpdateDataset(IndexableDataset indexableDataset) {
                     // anyway -- needs to be reviewed. -- L.A. 4.0alpha1 
 
                     if (var.getName() != null && !var.getName().equals("")) {
-                        if (variableNamesToIndex == null) {
-                            variableNamesToIndex = var.getName();
-                        } else {
-                            variableNamesToIndex = variableNamesToIndex + " " + var.getName();
-                        }
+                        datafileSolrInputDocument.addField(SearchFields.VARIABLE_NAME, var.getName());
                     }
                     if (var.getLabel() != null && !var.getLabel().equals("")) {
-                        if (variableLabelsToIndex == null) {
-                            variableLabelsToIndex = var.getLabel();
-                        } else {
-                            variableLabelsToIndex = variableLabelsToIndex + " " + var.getLabel();
-                        }
+                        datafileSolrInputDocument.addField(SearchFields.VARIABLE_LABEL, var.getLabel());
                     }
                 }
-                if (variableNamesToIndex != null) {
-                    logger.info("indexing  " + variableNamesToIndex.length() + " bytes");
-                    datafileSolrInputDocument.addField("varname_s", variableNamesToIndex);
-                }
-                if (variableLabelsToIndex != null) {
-                    logger.info("indexing  " + variableLabelsToIndex.length() + " bytes");
-                    datafileSolrInputDocument.addField("varlabel_s", variableLabelsToIndex);
-                }
             }
 
             docs.add(datafileSolrInputDocument);
@@ -991,7 +973,8 @@ private List<String> findSolrDocIdsForDraftFilesToDelete(Dataset datasetWithDraf
         Long datasetId = datasetWithDraftFilesToDelete.getId();
         SolrServer solrServer = new HttpSolrServer("http://localhost:8983/solr");
         SolrQuery solrQuery = new SolrQuery();
-        solrQuery.setQuery("parentid:" + datasetId);
+        solrQuery.setRows(Integer.MAX_VALUE);
+        solrQuery.setQuery(SearchFields.PARENT_ID + ":" + datasetId);
         /**
          * @todo rather than hard coding "_draft" here, tie to
          * IndexableDataset(new DatasetVersion()).getDatasetState().getSuffix()
@@ -1000,6 +983,8 @@ private List<String> findSolrDocIdsForDraftFilesToDelete(Dataset datasetWithDraf
         solrQuery.addFilterQuery(SearchFields.ID + ":" + "*_draft");
         List<String> solrIdsOfFilesToDelete = new ArrayList<>();
         try {
+            // i.e. rows=2147483647&q=parentid%3A16&fq=id%3A*_draft
+            logger.info("passing this Solr query to find draft files to delete: " + solrQuery);
             QueryResponse queryResponse = solrServer.query(solrQuery);
             SolrDocumentList results = queryResponse.getResults();
             for (SolrDocument solrDocument : results) {