Add executor for multithreaded search

Author: Sebastian Benjamin

jbrowse/src/client/JBrowse/VariantSearch/components/VariantTableWidget.tsx

@@ -76,6 +76,13 @@ const VariantTableWidget = observer(props => {
         session.hideWidget(widget)
     }
 
+    function resetPaginationToFirstPage() {
+      setPageSizeModel(prev => ({
+        page: 0,
+        pageSize: prev.pageSize,
+      }));
+    }
+
     function handleQuery(passedFilters, pushToHistory, pageQueryModel = pageSizeModel, sortQueryModel = sortModel) {
         const { page = pageSizeModel.page, pageSize = pageSizeModel.pageSize } = pageQueryModel;
         const { field = "genomicPosition", sort = false } = sortQueryModel[0] ?? {};
@@ -461,7 +468,7 @@ const VariantTableWidget = observer(props => {
             columnVisibilityModel={columnVisibilityModel}
             pageSizeOptions={[10,25,50,100]}
             paginationModel={ pageSizeModel }
-            rowCount={ totalHits }
+            rowCount={ -1 }
             paginationMode="server"
             onPaginationModelChange = {(newModel) => {
                 setPageSizeModel(newModel)
@@ -485,6 +492,7 @@ const VariantTableWidget = observer(props => {
             onSortModelChange={(newModel) => {
                 setSortModel(newModel)
                 handleQuery(filters, true, { page: 0, pageSize: pageSizeModel.pageSize }, newModel);
+                resetPaginationToFirstPage()
             }}
             localeText={{
                 MuiTablePagination: {
@@ -515,7 +523,10 @@ const VariantTableWidget = observer(props => {
                 fieldTypeInfo: fieldTypeInfo,
                 allowedGroupNames: allowedGroupNames,
                 promotedFilters: promotedFilters,
-                handleQuery: (filters) => handleQuery(filters, true, { page: 0, pageSize: pageSizeModel.pageSize}, sortModel)
+                handleQuery: (filters) => {
+                    handleQuery(filters, true, { page: 0, pageSize: pageSizeModel.pageSize}, sortModel)
+                    resetPaginationToFirstPage()
+                }
             }}
         />
     );

jbrowse/src/client/JBrowse/utils.ts

@@ -366,7 +366,8 @@ export async function fetchLuceneQuery(filters, sessionId, trackGUID, offset, pa
             successCallback(jsonRes)
         },
         failure: function(res) {
-            failureCallback("There was an error: " + res.status + "\n Status Body: " + res.responseText + "\n Session ID:" + sessionId)
+            console.log("There was an error: " + res.status + "\n Status Body: " + res.responseText + "\n Session ID:" + sessionId)
+            failureCallback("There was an error: status " + res.status)
         },
         params: {
             "searchString": encoded,

jbrowse/src/org/labkey/jbrowse/JBrowseLuceneSearch.java

@@ -16,6 +16,7 @@
 import org.apache.lucene.queryparser.flexible.standard.config.PointsConfig;
 import org.apache.lucene.search.BooleanClause;
 import org.apache.lucene.search.BooleanQuery;
+import org.apache.lucene.search.ConstantScoreQuery;
 import org.apache.lucene.search.IndexSearcher;
 import org.apache.lucene.search.LRUQueryCache;
 import org.apache.lucene.search.MatchAllDocsQuery;
@@ -24,6 +25,7 @@
 import org.apache.lucene.search.ScoreDoc;
 import org.apache.lucene.search.Sort;
 import org.apache.lucene.search.SortField;
+import org.apache.lucene.search.TopDocs;
 import org.apache.lucene.search.TopFieldDocs;
 import org.apache.lucene.search.UsageTrackingQueryCachingPolicy;
 import org.apache.lucene.store.Directory;
@@ -65,6 +67,8 @@
 import java.util.Map;
 import java.util.Set;
 import java.util.StringTokenizer;
+import java.util.concurrent.ExecutorService;
+import java.util.concurrent.Executors;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
@@ -75,6 +79,7 @@
 public class JBrowseLuceneSearch
 {
     private static final Logger _log = LogHelper.getLogger(JBrowseLuceneSearch.class, "Logger related to JBrowse/Lucene indexing and queries");
+    private static final ExecutorService SEARCH_EXECUTOR = Executors.newFixedThreadPool(JBrowseServiceImpl.get().getCoresForLuceneSearches());
     private final JBrowseSession _session;
     private final JsonFile _jsonFile;
     private final User _user;
@@ -116,7 +121,7 @@ private static synchronized CacheEntry getCacheEntryForSession(String trackObjec
                 Directory indexDirectory = FSDirectory.open(indexPath.toPath());
                 LRUQueryCache queryCache = new LRUQueryCache(maxCachedQueries, maxRamBytesUsed);
                 IndexReader indexReader = DirectoryReader.open(indexDirectory);
-                IndexSearcher indexSearcher = new IndexSearcher(indexReader);
+                IndexSearcher indexSearcher = new IndexSearcher(indexReader, SEARCH_EXECUTOR);
                 indexSearcher.setQueryCache(queryCache);
                 indexSearcher.setQueryCachingPolicy(new ForceMatchAllDocsCachingPolicy());
                 cacheEntry = new CacheEntry(queryCache, indexSearcher, indexPath);
@@ -252,7 +257,7 @@ private SearchConfig createSearchConfig(User u, String searchString, final int p
 
         if (searchString.equals(ALL_DOCS))
         {
-            booleanQueryBuilder.add(new MatchAllDocsQuery(), BooleanClause.Occur.MUST);
+            booleanQueryBuilder.add(new ConstantScoreQuery(new MatchAllDocsQuery()), BooleanClause.Occur.MUST);
         }
 
         // Split input into tokens, 1 token per query separated by &
@@ -321,41 +326,46 @@ else if (numericQueryParserFields.containsKey(fieldName))
     }
 
     private JSONObject paginateJSON(SearchConfig c) throws IOException, ParseException {
-        // Get chunks of size {pageSize}. Default to 1 chunk -- add to the offset to get more.
-        // We then iterate over the range of documents we want based on the offset. This does grow in memory
-        // linearly with the number of documents, but my understanding is that these are just score,id pairs
-        // rather than full documents, so mem usage *should* still be pretty low.
-        // Perform the search with sorting
-        TopFieldDocs topDocs = c.cacheEntry.indexSearcher.search(c.query, c.pageSize * (c.offset + 1), c.sort);
+        IndexSearcher searcher = c.cacheEntry.indexSearcher;
+        TopDocs topDocs;
+
+        if (c.offset == 0) {
+            topDocs = searcher.search(c.query, c.pageSize, c.sort);
+        } else {
+            TopFieldDocs prev = searcher.search(c.query, c.pageSize * c.offset, c.sort);
+            long totalHits = prev.totalHits.value;
+            ScoreDoc[] prevHits = prev.scoreDocs;
+
+            if (prevHits.length < c.pageSize * c.offset)
+            {
+                JSONObject results = new JSONObject();
+                results.put("data", Collections.emptyList());
+                results.put("totalHits", totalHits);
+                return results;
+            }
+
+            ScoreDoc lastDoc = prevHits[c.pageSize * c.offset - 1];
+            topDocs = searcher.searchAfter(lastDoc, c.query, c.pageSize, c.sort);
+        }
+
         JSONObject results = new JSONObject();
+        List<JSONObject> data = new ArrayList<>(topDocs.scoreDocs.length);
 
-        // Iterate over the doc list, (either to the total end or until the page ends) grab the requested docs,
-        // and add to returned results
-        List<JSONObject> data = new ArrayList<>();
-        for (int i = c.pageSize * c.offset; i < Math.min(c.pageSize * (c.offset + 1), topDocs.scoreDocs.length); i++)
+        for (ScoreDoc sd : topDocs.scoreDocs)
         {
+            Document doc = searcher.storedFields().document(sd.doc);
             JSONObject elem = new JSONObject();
-            Document doc = c.cacheEntry.indexSearcher.storedFields().document(topDocs.scoreDocs[i].doc);
-
-            for (IndexableField field : doc.getFields())
+            for (IndexableField f : doc.getFields())
             {
-                String fieldName = field.name();
-                String[] fieldValues = doc.getValues(fieldName);
-                if (fieldValues.length > 1)
-                {
-                    elem.put(fieldName, fieldValues);
-                }
-                else
-                {
-                    elem.put(fieldName, fieldValues[0]);
-                }
+                String name = f.name();
+                String[] vals = doc.getValues(name);
+                elem.put(name, vals.length > 1 ? Arrays.asList(vals) : vals[0]);
             }
             data.add(elem);
         }
 
         results.put("data", data);
         results.put("totalHits", topDocs.totalHits.value);
-
         return results;
     }
 

jbrowse/src/org/labkey/jbrowse/JBrowseServiceImpl.java

@@ -435,4 +435,8 @@ public boolean isAvailable(Container c)
             return c.getActiveModules().contains(ModuleLoader.getInstance().getModule(JBrowseModule.class));
         }
     }
+
+    public int getCoresForLuceneSearches() {
+        return Runtime.getRuntime().availableProcessors();
+    }
 }