Resolves #175.

JULIELab · Jan 4, 2024 · c302155 · c302155
1 parent cb92ede
commit c302155
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 6 deletions.
diff --git a/jcore-elasticsearch-consumer/pom.xml b/jcore-elasticsearch-consumer/pom.xml
@@ -14,7 +14,7 @@
         perfectly valid ElasticSearch default JSON which does not require the plugin. Then, however, the preanalyzed
         data format cannot be used because this is the format that required the ElasticSearch plugin.
     </description>
-
+    <version>2.6.3-SNAPSHOT</version>
     <dependencies>
         <dependency>
             <groupId>de.julielab</groupId>

diff --git a/...ticsearch-consumer/src/main/java/de/julielab/jcore/consumer/es/ElasticSearchConsumer.java b/...ticsearch-consumer/src/main/java/de/julielab/jcore/consumer/es/ElasticSearchConsumer.java
@@ -69,7 +69,7 @@ public void initialize(UimaContext aContext) throws ResourceInitializationExcept
         indexName = (String) getContext().getConfigParameterValue(PARAM_INDEX_NAME);
         type = (String) getContext().getConfigParameterValue(PARAM_TYPE);
         batchSize = Optional.ofNullable((Integer) getContext().getConfigParameterValue(PARAM_BATCH_SIZE)).orElse(50);
-        bulkCommand = new ArrayList<>(4000);
+        bulkCommand = new ArrayList<>(batchSize*2);
         deleteDocsBeforeIndexing = (boolean) Optional.ofNullable(getContext().getConfigParameterValue(PARAM_DELETE_DOCS_BEFORE_INDEXING)).orElse(false);
         docIdField = (String) getContext().getConfigParameterValue(PARAM_DOC_ID_FIELD);
 
@@ -265,12 +265,13 @@ private void postBulkIndexAction() throws AnalysisEngineProcessException {
                 if (subList.isEmpty())
                     continue;
                 lastIndex += subList.size();
-                log.debug("Sending {} documents to index {}.", subList.size(), indexName);
+                log.debug("Sending {} documents to index {}.", subList.size()/2, indexName);
                 long time = System.currentTimeMillis();
                 // The bulk format requires us to have a newline also after the
                 // last
                 // line; it will be ignored otherwise!
                 String bulkCommandString = StringUtils.join(subList, "\n") + "\n";
+                log.trace(bulkCommandString);
                 StringEntity bulkIndexEntity = new StringEntity(bulkCommandString, "UTF-8");
                 indexPost.setEntity(bulkIndexEntity);
                 HttpResponse response = httpclient.execute(indexPost);

diff --git a/...ticsearch-consumer/src/main/java/de/julielab/jcore/consumer/es/filter/SnowballFilter.java b/...ticsearch-consumer/src/main/java/de/julielab/jcore/consumer/es/filter/SnowballFilter.java
@@ -2,19 +2,24 @@
 
 import org.tartarus.snowball.SnowballProgram;
 
+import java.lang.reflect.InvocationTargetException;
 import java.util.List;
 
 public class SnowballFilter extends AbstractFilter {
 
 	private SnowballProgram stemmer;
 
 	public SnowballFilter() {
+		this("org.tartarus.snowball.ext.EnglishStemmer");
+	}
+
+	public SnowballFilter(String snowballProgram) {
 		super();
 		Class<? extends SnowballProgram> stemClass;
 		try {
-			stemClass = Class.forName("org.tartarus.snowball.ext.EnglishStemmer").asSubclass(SnowballProgram.class);
-			stemmer = stemClass.newInstance();
-		} catch (ClassNotFoundException | InstantiationException | IllegalAccessException e) {
+			stemClass = Class.forName(snowballProgram).asSubclass(SnowballProgram.class);
+			stemmer = stemClass.getDeclaredConstructor().newInstance();
+		} catch (ClassNotFoundException | InstantiationException | IllegalAccessException | NoSuchMethodException | InvocationTargetException e) {
 			throw new RuntimeException(e);
 		}
 	}