Bug fix: clean Memory object to avoid memory leak.

Add support to JAVA_OPT into Docker instances.

src/main/docker/Dockerfile

@@ -20,5 +20,6 @@ ADD limsiSourceExtractor.jar limsiSourceExtractor.jar
 EXPOSE 8080
 
 RUN sh -c 'touch /limsiSourceExtractor.jar'
-ENV JAVA_OPTS="-Dspring.config.location=/configuration/"
-ENTRYPOINT [ "sh", "-c", "java $JAVA_OPTS -Djava.security.egd=file:/dev/./urandom -jar /limsiSourceExtractor.jar" ]
\ No newline at end of file
+#ENV JAVA_OPTS="-Dspring.config.location=/configuration/"
+#ENTRYPOINT [ "sh", "-c", "java $JAVA_OPTS -Dspring.config.location=/configuration/ -Djava.security.egd=file:/dev/./urandom -jar /limsiSourceExtractor.jar" ]
+ENTRYPOINT  exec java $JAVA_OPTS -Dspring.config.location=/configuration/ -Djava.security.egd=file:/dev/./urandom -jar /limsiSourceExtractor.jar
\ No newline at end of file

src/main/java/fr/limsi/sourceExtractor/Memory.java

@@ -15,5 +15,11 @@ public class Memory {
 		this.docTexts = docTexts;
 		this.sentenceOffsetsByFile = sentenceOffsetsByFile;
 	}
+	
+	public void cleanEntry(String fileId) {
+		this.parsedTexts.remove(fileId);
+		this.docTexts.remove(fileId);
+		this.sentenceOffsetsByFile.remove(fileId);
+	}
 
 }

src/main/java/fr/limsi/sourceExtractor/application/process/SimpleExtractSource.java

@@ -10,6 +10,7 @@ import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 import fr.limsi.sourceExtractor.DIRUtils;
+import fr.limsi.sourceExtractor.Memory;
 import fr.limsi.sourceExtractor.Paths;
 import fr.limsi.sourceExtractor.application.configuration.SourceExtractorConfig;
 import fr.limsi.sourceExtractor.application.configuration.SourceExtractorConstant;
@@ -66,6 +67,8 @@ public class SimpleExtractSource extends TrainingUnLabel {
 		String response = parseFileToAddAnnotationToBrat(fileId, annContent);
 		logger.debug("output: {}", response);
 		cleanDirContent(this.paths.DATA_UNLABELED_DIR, fileId);
+		//Remote Memory entries
+		extractorConfig.getMemory().cleanEntry(fileId);
 		return response;
 	}
 

src/main/java/fr/limsi/sourceExtractor/training/AProcess.java

@@ -29,6 +29,7 @@ import org.apache.commons.collections4.CollectionUtils;
 import com.google.common.base.Charsets;
 import com.google.common.io.Files;
 
+import edu.stanford.nlp.pipeline.Annotation;
 import fr.limsi.sourceExtractor.DIRUtils;
 import fr.limsi.sourceExtractor.Memory;
 import fr.limsi.sourceExtractor.Pair;
@@ -159,7 +160,9 @@ public abstract class AProcess extends AProcessSupport {
 			//inputTxtFile = new File(this.paths.DIR_INPUT_UNLABELED, fileId + ".txt");
 		}
 		//le fichier n'est jamais utilisé => remplacement par null (à tester possible effet de bord)
-		taggedTextOnlyOffset = parseText(fileId, SourceExtractorUtil.getDocText(fileId, null, memory), tools, resources, memory);
+		String text = SourceExtractorUtil.getDocText(fileId, null, memory);
+		Annotation annotation = new Annotation(text);
+		taggedTextOnlyOffset = parseText(annotation, fileId, text, tools, resources, memory);
 
 		HashMap<String, String> retrieveOffsetsOfSource = new HashMap<>();
 
@@ -446,8 +449,8 @@ public abstract class AProcess extends AProcessSupport {
 			text = SourceExtractorUtil.getDocText(fileId, new File(paths.DIR_INPUT_UNLABELED, fileId + ".txt"), memory);
 		}
 		
-
-		String tagWithOffset = parseText(fileId, text, tools, resources, memory);
+		Annotation annotation = new Annotation(text);	
+		String tagWithOffset = parseText(annotation, fileId, text, tools, resources, memory);
 
 		String resultWapiti = "";
 		String line;

src/main/java/fr/limsi/sourceExtractor/training/AProcessSupport.java

@@ -80,10 +80,11 @@ public abstract class AProcessSupport {
 	public String tagAndConvert(String fileId, String text, File outFileSEC, String ann, Tools tools, Resources resources, Memory memory)
 			throws IOException {
 		// we use malt parser to keep the subjects found
-		String maltParsertext = parseTextForMaltParser(text, tools, resources);
+		Annotation document = new Annotation(text);
+		String maltParsertext = parseTextForMaltParser(document, text, tools, resources);
 
 		// we parse the text to tag each word of this text.
-		String parsedText = parseText(fileId, text, tools, resources, memory);
+		String parsedText = parseText(document, fileId, text, tools, resources, memory);
 
 		// we join both parse text
 		String taggedText = tagFromMaltParserAndParseText(parsedText, maltParsertext, resources);
@@ -126,14 +127,14 @@ public abstract class AProcessSupport {
 	 *            Le text à tag.
 	 * @return Le StringBuilder créé avec toutes les informations nécessaires. (6 champs)
 	 */
-	protected String parseText(String fileId, String text, Tools tools, Resources resources, Memory memory) {
+	protected String parseText(Annotation document, String fileId, String text, Tools tools, Resources resources, Memory memory) {
 		String parsedText = memory.parsedTexts.get(fileId);
 
 		if (parsedText != null) {
 			return parsedText;
 		}
 		// create an empty Annotation just with the given text
-		Annotation document = new Annotation(text);
+		//Annotation document = new Annotation(text);
 
 		// run all Annotators on this text
 		tools.getCoreNLPParser().annotate(document);
@@ -516,9 +517,9 @@ public abstract class AProcessSupport {
 	 * @return
 	 * @throws IOException
 	 */
-	private String parseTextForMaltParser(String text, Tools tools, Resources resources) throws IOException {
+	private String parseTextForMaltParser(Annotation document, String text, Tools tools, Resources resources) throws IOException {
 		// create an empty Annotation just with the given text
-		Annotation document = new Annotation(text);
+		//Annotation document = new Annotation(text);
 
 		// run all Annotators on this text
 		tools.coreNLPParser.annotate(document);