diff --git a/src/main/java/org/olat/core/util/filter/impl/HtmlFilter.java b/src/main/java/org/olat/core/util/filter/impl/HtmlFilter.java
index 3efef4f1645a8fff835636bcddea189d4e4f32c1..1ce347a55ab940796ca23a497f013b842e5cad70 100644
--- a/src/main/java/org/olat/core/util/filter/impl/HtmlFilter.java
+++ b/src/main/java/org/olat/core/util/filter/impl/HtmlFilter.java
@@ -81,7 +81,7 @@ public class HtmlFilter implements Filter {
if (in == null) return null;
try {
HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
- HtmlHandler contentHandler = new HtmlHandler((int)(1000 * 0.66f), false);
+ HtmlHandler contentHandler = new HtmlHandler(8096, false);
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(in));
return contentHandler.getContent();
diff --git a/src/main/java/org/olat/search/service/indexer/FolderIndexerWorker.java b/src/main/java/org/olat/search/service/indexer/FolderIndexerWorker.java
index bac90bfba6ff48271fd66aa1fe393c796fb68d91..dcd5fea8c617e47122084cccd1ae6a302929bdef 100644
--- a/src/main/java/org/olat/search/service/indexer/FolderIndexerWorker.java
+++ b/src/main/java/org/olat/search/service/indexer/FolderIndexerWorker.java
@@ -29,10 +29,10 @@ package org.olat.search.service.indexer;
import java.io.IOException;
import java.util.concurrent.Callable;
+import org.apache.logging.log4j.Logger;
import org.apache.lucene.document.Document;
import org.olat.core.CoreSpringFactory;
import org.olat.core.commons.persistence.DBFactory;
-import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.vfs.VFSContainer;
import org.olat.core.util.vfs.VFSItem;
@@ -86,27 +86,28 @@ public class FolderIndexerWorker implements Callable<Boolean> {
throws IOException, InterruptedException {
// Items: List of VFSContainer & VFSLeaf
String myFilePath = fPath;
+ boolean debug = log.isDebugEnabled();
for (VFSItem item : cont.getItems(new VFSSystemItemFilter())) {
if (item instanceof VFSContainer) {
// ok it is a container go further
- if (log.isDebugEnabled()) log.debug(item.getName() + " is a VFSContainer => go further ");
+ if (debug) log.debug("{} is a VFSContainer => go further ", item.getName());
if(aRule.allowed(item)) {
doIndexVFSContainer(resourceContext, (VFSContainer)item, writer, myFilePath + "/" + ((VFSContainer)item).getName(), aRule);
}
} else if (item instanceof VFSLeaf) {
// ok it is a file => analyse it
- if (log.isDebugEnabled()) log.debug(item.getName() + " is a VFSLeaf => analyse file");
+ if (debug) log.debug("{} is a VFSLeaf => analyse file", item.getName() );
if(aRule.allowed(item)) {
doIndexVFSLeaf(resourceContext, (VFSLeaf)item, writer, myFilePath);
}
} else {
- log.warn("Unkown element in item-list class=" + item.getClass());
+ log.warn("Unkown element in item-list class={}", item.getClass());
}
}
}
protected void doIndexVFSLeaf(SearchResourceContext leafResourceContext, VFSLeaf leaf, OlatFullIndexer writer, String fPath) {
- if (log.isDebugEnabled()) log.debug("Analyse VFSLeaf=" + leaf.getName());
+ if (log.isDebugEnabled()) log.debug("Analyse VFSLeaf={}", leaf.getName());
try {
if (docFactory.isFileSupported(leaf)) {
String myFilePath = fPath + "/" + leaf.getName();
@@ -116,16 +117,16 @@ public class FolderIndexerWorker implements Callable<Boolean> {
writer.addDocument(document);
}
} else {
- if (log.isDebugEnabled()) log.debug("Documenttype not supported. file=" + leaf.getName());
+ if (log.isDebugEnabled()) log.debug("Documenttype not supported. file={}", leaf.getName());
}
} catch (DocumentAccessException e) {
- if (log.isDebugEnabled()) log.debug("Can not access document." + e.getMessage());
+ if (log.isDebugEnabled()) log.debug("Can not access document.", e);
} catch (InterruptedException e) {
- if (log.isDebugEnabled()) log.debug("InterruptedException: Can not index leaf=" + leaf.getName() + ";" + e.getMessage());
+ if (log.isDebugEnabled()) log.debug("InterruptedException: Can not index leaf={}", leaf.getName(), e);
} catch (IOException ioEx) {
- log.warn("IOException: Can not index leaf=" + leaf.getName(), ioEx);
+ log.warn("IOException: Can not index leaf={}", leaf.getName(), ioEx);
} catch (Exception ex) {
- log.warn("Exception: Can not index leaf=" + leaf.getName(), ex);
+ log.warn("Exception: Can not index leaf={}", leaf.getName(), ex);
}
}
diff --git a/src/main/java/org/olat/search/service/indexer/repository/ImsCPRepositoryIndexer.java b/src/main/java/org/olat/search/service/indexer/repository/ImsCPRepositoryIndexer.java
index a6f02ec2edc1499b14cab87862a6ed600928920c..201355be8e172787a8dd296c6556103bd982a36b 100644
--- a/src/main/java/org/olat/search/service/indexer/repository/ImsCPRepositoryIndexer.java
+++ b/src/main/java/org/olat/search/service/indexer/repository/ImsCPRepositoryIndexer.java
@@ -50,7 +50,7 @@ public class ImsCPRepositoryIndexer extends FolderIndexer {
private static final Logger log = Tracing.createLoggerFor(ImsCPRepositoryIndexer.class);
// Must correspond with LocalString_xx.properties
- // Do not use '_' because we want to seach for certain documenttype and lucene haev problems with '_'
+ // Do not use '_' because we want to search for certain document types and lucene has problems with '_'
public static final String TYPE = "type.repository.entry.imscp";
public static final String ORES_TYPE_CP = ImsCPFileResource.TYPE_NAME;
@@ -59,10 +59,7 @@ public class ImsCPRepositoryIndexer extends FolderIndexer {
public String getSupportedTypeName() {
return ORES_TYPE_CP;
}
-
- /**
- * @see org.olat.repository.handlers.RepositoryHandler#supportsDownload()
- */
+
@Override
public void doIndex(SearchResourceContext resourceContext, Object parentObject, OlatFullIndexer indexWriter) throws IOException,InterruptedException {
RepositoryEntry repositoryEntry = (RepositoryEntry) parentObject;
diff --git a/src/main/java/org/olat/search/service/indexer/repository/ScormRepositoryIndexer.java b/src/main/java/org/olat/search/service/indexer/repository/ScormRepositoryIndexer.java
index c5ec9503fd17066c309581ce9d728d84c7a733f2..aefd0ce3b09425a1698527de15984de7fb367596 100644
--- a/src/main/java/org/olat/search/service/indexer/repository/ScormRepositoryIndexer.java
+++ b/src/main/java/org/olat/search/service/indexer/repository/ScormRepositoryIndexer.java
@@ -21,7 +21,6 @@ package org.olat.search.service.indexer.repository;
import java.io.File;
import java.io.IOException;
-import java.util.ArrayList;
import java.util.Date;
import java.util.HashSet;
import java.util.List;
@@ -59,41 +58,26 @@ public class ScormRepositoryIndexer extends FolderIndexer {
private static final Logger log = Tracing.createLoggerFor(ScormRepositoryIndexer.class);
- public static Set<String> stopWords = new HashSet<>();
+ private static final Set<String> stopWords = new HashSet<>();
static {
stopWords.add("LOMv1.0");
stopWords.add("yes");
stopWords.add("NA");
}
- public static final List<String> forbiddenExtensions = new ArrayList<>();
- static {
- forbiddenExtensions.add("LOMv1.0");
- forbiddenExtensions.add(".xsd");
- forbiddenExtensions.add(".js");
- }
- public static final Set<String> forbiddenFiles = new HashSet<>();
- static {
- forbiddenFiles.add("imsmanifest.xml");
- }
-
-
+
public static final String TYPE = "type.repository.entry.scorm";
public static final String ORES_TYPE_SCORM = ScormCPFileResource.TYPE_NAME;
public ScormRepositoryIndexer() {
// Repository types
}
-
- /**
- *
- */
+
+ @Override
public String getSupportedTypeName() {
return ORES_TYPE_SCORM;
}
- /**
- * @see org.olat.repository.handlers.RepositoryHandler#supportsDownload()
- */
+ @Override
public void doIndex(SearchResourceContext resourceContext, Object parentObject, OlatFullIndexer indexWriter)
throws IOException,InterruptedException {
if (log.isDebugEnabled()) log.debug("Index Scorm package...");
@@ -123,7 +107,7 @@ public class ScormRepositoryIndexer extends FolderIndexer {
private Document createManifestDocument(VFSLeaf fManifest, Element rootElement, SearchResourceContext resourceContext) {
IMSMetadataDocument document = new IMSMetadataDocument();
document.setResourceUrl(resourceContext.getResourceUrl());
- if (log.isDebugEnabled()) log.debug("MM: URL=" + document.getResourceUrl());
+ if (log.isDebugEnabled()) log.debug("MM: URL={}", document.getResourceUrl());
document.setLastChange(new Date(fManifest.getLastModified()));
document.setDocumentType(resourceContext.getDocumentType());
if (StringHelper.containsNonWhitespace(resourceContext.getTitle())) {
@@ -134,8 +118,6 @@ public class ScormRepositoryIndexer extends FolderIndexer {
document.setParentContextType(resourceContext.getParentContextType());
document.setParentContextName(resourceContext.getParentContextName());
-
-
StringBuilder sb = new StringBuilder();
collectLangString(sb, rootElement);
document.setContent(sb.toString());
@@ -157,21 +139,14 @@ public class ScormRepositoryIndexer extends FolderIndexer {
}
}
- public class ScormFileAccess implements FolderIndexerAccess {
-
+ private static class ScormFileAccess implements FolderIndexerAccess {
@Override
public boolean allowed(VFSItem item) {
- String name = item.getName();
- if(forbiddenFiles.contains(name)) {
- return false;
- }
-
- for(String forbiddenExtension:forbiddenExtensions) {
- if(name.endsWith(forbiddenExtension)) {
- return false;
- }
+ String name = item.getName().toLowerCase();
+ if(item instanceof VFSContainer) {
+ return !name.equals("template");
}
- return true;
+ return name.endsWith(".htm") || name.endsWith(".html");
}
}
}
diff --git a/src/main/java/org/olat/search/service/indexer/repository/course/ScormCourseNodeIndexer.java b/src/main/java/org/olat/search/service/indexer/repository/course/ScormCourseNodeIndexer.java
index 9a8bbffd92ffd34b7fb61b2d6fb072ebb2fece5e..dc188f6f05af9d1de6951f27fd8db28680146573 100644
--- a/src/main/java/org/olat/search/service/indexer/repository/course/ScormCourseNodeIndexer.java
+++ b/src/main/java/org/olat/search/service/indexer/repository/course/ScormCourseNodeIndexer.java
@@ -43,8 +43,8 @@ import org.olat.search.service.indexer.repository.ScormRepositoryIndexer;
* @author srosse, stephane.rosse@frentix.com
*/
public class ScormCourseNodeIndexer extends ScormRepositoryIndexer implements CourseNodeIndexer {
- public final static String NODE_TYPE = "type.course.node.scorm";
- private final static String SUPPORTED_TYPE_NAME = "org.olat.course.nodes.ScormCourseNode";
+ public static final String NODE_TYPE = "type.course.node.scorm";
+ private static final String SUPPORTED_TYPE_NAME = "org.olat.course.nodes.ScormCourseNode";
@Override
public void doIndex(SearchResourceContext repositoryResourceContext, ICourse course, CourseNode courseNode, OlatFullIndexer indexWriter)