diff --git a/pom.xml b/pom.xml
index 13be36b7ce59bf2a49f219e2c5969d47f010b25a..6d19fd40d44970ae9691120386f3daa9d759b242 100644
--- a/pom.xml
+++ b/pom.xml
@@ -62,7 +62,7 @@
<properties>
<project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
<targetJdk>1.8</targetJdk>
- <org.springframework.version>5.1.6.RELEASE</org.springframework.version>
+ <org.springframework.version>5.1.7.RELEASE</org.springframework.version>
<org.hibernate.version>5.4.2.Final</org.hibernate.version>
<com.sun.jersey.version>1.19.4</com.sun.jersey.version>
<apache.cxf>3.3.2</apache.cxf>
@@ -2165,6 +2165,11 @@
<scope>test</scope>
</dependency>
<!-- core dependencies -->
+ <dependency>
+ <groupId>nu.validator</groupId>
+ <artifactId>htmlparser</artifactId>
+ <version>1.4.13</version>
+ </dependency>
<dependency>
<groupId>net.sourceforge.nekohtml</groupId>
<artifactId>nekohtml</artifactId>
diff --git a/src/main/java/org/olat/core/gui/components/form/flexible/impl/elements/richText/TextMode.java b/src/main/java/org/olat/core/gui/components/form/flexible/impl/elements/richText/TextMode.java
index 8cff3631d97755ee2d60b87b1be8aaf7f333cb2d..eabf6530c240c2cd51c0b66f8aa9ecda63b14cd5 100644
--- a/src/main/java/org/olat/core/gui/components/form/flexible/impl/elements/richText/TextMode.java
+++ b/src/main/java/org/olat/core/gui/components/form/flexible/impl/elements/richText/TextMode.java
@@ -23,7 +23,6 @@ import java.io.ByteArrayInputStream;
import java.util.ArrayList;
import java.util.List;
-import org.cyberneko.html.parsers.SAXParser;
import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.StringHelper;
@@ -32,6 +31,9 @@ import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
*
* Initial date: 19 juil. 2017<br>
@@ -118,12 +120,9 @@ public enum TextMode {
}
private static void parse(String text, DefaultHandler handler) throws Exception {
- SAXParser parser = new SAXParser();
- parser.setProperty("http://cyberneko.org/html/properties/names/elems", "lower");
- parser.setFeature("http://cyberneko.org/html/features/balance-tags/document-fragment", true);
- parser.setProperty("http://cyberneko.org/html/properties/default-encoding", "UTF-8");
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
parser.setContentHandler(handler);
- parser.parse(new InputSource(new ByteArrayInputStream(text.getBytes())));
+ parser.parseFragment(new InputSource(new ByteArrayInputStream(text.getBytes())), "");
}
private static final class TextAnalyser extends DefaultHandler {
diff --git a/src/main/java/org/olat/core/util/Formatter.java b/src/main/java/org/olat/core/util/Formatter.java
index c012fe1f28ba50bea2a68fec20d25c71632894cb..9f7ae71b3617538f4dfdf2984370542191fb6ff0 100644
--- a/src/main/java/org/olat/core/util/Formatter.java
+++ b/src/main/java/org/olat/core/util/Formatter.java
@@ -50,7 +50,7 @@ import org.apache.commons.lang.time.DurationFormatUtils;
import org.olat.core.dispatcher.impl.StaticMediaDispatcher;
import org.olat.core.gui.render.StringOutput;
import org.olat.core.helpers.Settings;
-import org.olat.core.util.filter.impl.NekoHTMLMathScanner;
+import org.olat.core.util.filter.impl.HtmlMathScanner;
/**
* enclosing_type Description: <br>
@@ -623,7 +623,7 @@ public class Formatter {
public static String formatLatexFormulas(String htmlFragment) {
if (htmlFragment == null) return "";
// optimize, reduce jsmath calls on client
- if (new NekoHTMLMathScanner().scan(htmlFragment)) {
+ if (new HtmlMathScanner().scan(htmlFragment)) {
// add math wrapper
String domid = "mw_" + CodeHelper.getRAMUniqueID();
String elem = htmlFragment.contains("<div") || htmlFragment.contains("<p") ? "div" : "span";
diff --git a/src/main/java/org/olat/core/util/StringHelper.java b/src/main/java/org/olat/core/util/StringHelper.java
index 349d0c0ed96c0cccea0fb8bb1d7994b6dd8667ea..0d184df22794c496d73917c0177f1b60f4c4bb42 100644
--- a/src/main/java/org/olat/core/util/StringHelper.java
+++ b/src/main/java/org/olat/core/util/StringHelper.java
@@ -51,7 +51,7 @@ import org.olat.core.id.Identity;
import org.olat.core.logging.AssertException;
import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
-import org.olat.core.util.filter.impl.NekoHTMLScanner;
+import org.olat.core.util.filter.impl.HtmlScanner;
import org.olat.core.util.filter.impl.OWASPAntiSamyXSSFilter;
import org.olat.user.UserManager;
@@ -337,9 +337,7 @@ public class StringHelper {
*/
public static boolean isHtml(String s) {
if (s == null) return false;
-
- boolean containsHtml = new NekoHTMLScanner().scan(s);
- return containsHtml;
+ return new HtmlScanner().scan(s);
}
/**
diff --git a/src/main/java/org/olat/core/util/filter/FilterFactory.java b/src/main/java/org/olat/core/util/filter/FilterFactory.java
index 8100f1cebb6fbae357a7bb71fd133728e31d62d2..681fac8ae174a756181a4d37d6ad9c3ad8ab4627 100644
--- a/src/main/java/org/olat/core/util/filter/FilterFactory.java
+++ b/src/main/java/org/olat/core/util/filter/FilterFactory.java
@@ -22,7 +22,7 @@ package org.olat.core.util.filter;
import org.olat.core.util.filter.impl.AddBaseURLToMediaRelativeURLFilter;
import org.olat.core.util.filter.impl.ConditionalHTMLCommentsFilter;
-import org.olat.core.util.filter.impl.NekoHTMLFilter;
+import org.olat.core.util.filter.impl.HtmlFilter;
import org.olat.core.util.filter.impl.OWASPAntiSamyXSSFilter;
import org.olat.core.util.filter.impl.OWASPAntiSamyXSSFilter.Variant;
import org.olat.core.util.filter.impl.SimpleHTMLTagsFilter;
@@ -44,7 +44,7 @@ import org.olat.core.util.filter.impl.XMLValidEntityFilter;
public class FilterFactory {
// the html tag filter is static, not stateful
private static final Filter htmlTagsFilter = new SimpleHTMLTagsFilter();
- private static final Filter htmlTagsAndDesescapingFilter = new NekoHTMLFilter();
+ private static final Filter htmlTagsAndDesescapingFilter = new HtmlFilter();
private static final Filter conditionalCommentsFilter = new ConditionalHTMLCommentsFilter();
private static final Filter xmlValidCharacterFilter = new XMLValidCharacterFilter();
private static final Filter smileysCssToDataUriFilter = new SmileysCssToDataUriFilter();
diff --git a/src/main/java/org/olat/core/util/filter/impl/NekoHTMLFilter.java b/src/main/java/org/olat/core/util/filter/impl/HtmlFilter.java
similarity index 83%
rename from src/main/java/org/olat/core/util/filter/impl/NekoHTMLFilter.java
rename to src/main/java/org/olat/core/util/filter/impl/HtmlFilter.java
index 7ada2bb70b7e910f26b26e1e93710221b686dc6a..3efef4f1645a8fff835636bcddea189d4e4f32c1 100644
--- a/src/main/java/org/olat/core/util/filter/impl/NekoHTMLFilter.java
+++ b/src/main/java/org/olat/core/util/filter/impl/HtmlFilter.java
@@ -19,7 +19,6 @@
*/
package org.olat.core.util.filter.impl;
-import java.io.IOException;
import java.io.InputStream;
import java.io.StringReader;
import java.util.Arrays;
@@ -27,27 +26,28 @@ import java.util.HashSet;
import java.util.Set;
import org.apache.logging.log4j.Logger;
-import org.cyberneko.html.parsers.SAXParser;
import org.olat.core.logging.Tracing;
import org.olat.core.util.filter.Filter;
import org.olat.core.util.io.LimitedContentWriter;
import org.olat.search.service.document.file.FileDocumentFactory;
import org.xml.sax.Attributes;
import org.xml.sax.InputSource;
-import org.xml.sax.SAXException;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
* Description:<br>
- * Filter the HTML code using Neko SAX parser and extract the content.
- * Neko parse the HTML entities too and deliver cleaned text.
+ * Filter the HTML code using HtmlParser SAX parser and extract the content.
+ * It parses the HTML entities too and deliver cleaned text.
*
* <P>
* Initial Date: 2 dec. 2009 <br>
* @author srosse
*/
-public class NekoHTMLFilter implements Filter {
- private static final Logger log = Tracing.createLoggerFor(NekoHTMLFilter.class);
+public class HtmlFilter implements Filter {
+ private static final Logger log = Tracing.createLoggerFor(HtmlFilter.class);
public static final Set<String> blockTags = new HashSet<>();
public static final Set<String> toBeSkippedTags = new HashSet<>();
@@ -66,48 +66,36 @@ public class NekoHTMLFilter implements Filter {
if(original.isEmpty()) return "";
try {
- SAXParser parser = new SAXParser();
- HTMLHandler contentHandler = new HTMLHandler((int)(original.length() * 0.66f), pretty);
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
+ HtmlHandler contentHandler = new HtmlHandler((int)(original.length() * 0.66f), pretty);
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(new StringReader(original)));
return contentHandler.toString();
- } catch (SAXException e) {
- log.error("", e);
- return null;
- } catch (IOException e) {
- log.error("", e);
- return null;
} catch (Exception e) {
log.error("", e);
return null;
}
}
- public NekoContent filter(InputStream in) {
+ public HtmlContent filter(InputStream in) {
if (in == null) return null;
try {
- SAXParser parser = new SAXParser();
- HTMLHandler contentHandler = new HTMLHandler((int)(1000 * 0.66f), false);
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
+ HtmlHandler contentHandler = new HtmlHandler((int)(1000 * 0.66f), false);
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(in));
return contentHandler.getContent();
- } catch (SAXException e) {
- log.error("", e);
- return null;
- } catch (IOException e) {
- log.error("", e);
- return null;
} catch (Exception e) {
log.error("", e);
return null;
}
}
- public static class NekoContent {
+ public static class HtmlContent {
private final String title;
private final LimitedContentWriter content;
- public NekoContent(String title, LimitedContentWriter content) {
+ public HtmlContent(String title, LimitedContentWriter content) {
this.title = title;
this.content = content;
}
@@ -121,7 +109,7 @@ public class NekoHTMLFilter implements Filter {
}
}
- private static class HTMLHandler extends DefaultHandler {
+ private static class HtmlHandler extends DefaultHandler {
private boolean collect = true;
private boolean consumeBlanck = false;
private boolean consumeTitle = true;
@@ -129,7 +117,7 @@ public class NekoHTMLFilter implements Filter {
private final LimitedContentWriter content;
private final StringBuilder title;
- public HTMLHandler(int size, boolean pretty) {
+ public HtmlHandler(int size, boolean pretty) {
this.pretty = pretty;
content = new LimitedContentWriter(size, FileDocumentFactory.getMaxFileSize());
title = new StringBuilder(32);
@@ -213,8 +201,8 @@ public class NekoHTMLFilter implements Filter {
}
}
- public NekoContent getContent() {
- return new NekoContent(title.toString(), content);
+ public HtmlContent getContent() {
+ return new HtmlContent(title.toString(), content);
}
@Override
diff --git a/src/main/java/org/olat/core/util/filter/impl/NekoHTMLMathScanner.java b/src/main/java/org/olat/core/util/filter/impl/HtmlMathScanner.java
similarity index 86%
rename from src/main/java/org/olat/core/util/filter/impl/NekoHTMLMathScanner.java
rename to src/main/java/org/olat/core/util/filter/impl/HtmlMathScanner.java
index 5ba6c8fa531c078b9a443816ff88eaa5e31e6749..71642b582aa26c0096aeb89f4b28f4466a0514b9 100644
--- a/src/main/java/org/olat/core/util/filter/impl/NekoHTMLMathScanner.java
+++ b/src/main/java/org/olat/core/util/filter/impl/HtmlMathScanner.java
@@ -22,7 +22,6 @@ package org.olat.core.util.filter.impl;
import java.io.StringReader;
import org.apache.logging.log4j.Logger;
-import org.cyberneko.html.parsers.SAXParser;
import org.olat.core.logging.Tracing;
import org.olat.core.util.WebappHelper;
import org.xml.sax.Attributes;
@@ -30,6 +29,9 @@ import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
* Detect the tag math and a CSS clas named math
*
@@ -38,17 +40,15 @@ import org.xml.sax.helpers.DefaultHandler;
* @author srosse, stephane.rosse@frentix.com, http://www.frentix.com
*
*/
-public class NekoHTMLMathScanner {
+public class HtmlMathScanner {
- private static final Logger log = Tracing.createLoggerFor(NekoHTMLMathScanner.class);
+ private static final Logger log = Tracing.createLoggerFor(HtmlMathScanner.class);
public boolean scan(String original) {
if (original == null) return false;
try {
- SAXParser parser = new SAXParser();
- parser.setFeature("http://cyberneko.org/html/features/balance-tags/document-fragment", true);
- parser.setFeature("http://cyberneko.org/html/features/balance-tags", false);
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
HTMLHandler contentHandler = new HTMLHandler();
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(new StringReader(original)));
@@ -68,7 +68,7 @@ public class NekoHTMLMathScanner {
@Override
public void startElement(String uri, String localName, String qName, Attributes attributes) {
- if("MATH".equals(localName)) {
+ if("MATH".equalsIgnoreCase(localName)) {
mathFound = true;
} else if(attributes != null) {
String css = attributes.getValue("class");
diff --git a/src/main/java/org/olat/core/util/filter/impl/NekoHTMLScanner.java b/src/main/java/org/olat/core/util/filter/impl/HtmlScanner.java
similarity index 74%
rename from src/main/java/org/olat/core/util/filter/impl/NekoHTMLScanner.java
rename to src/main/java/org/olat/core/util/filter/impl/HtmlScanner.java
index 5af3dca31628eb14561d4ad3d2b7f2eccf30204f..bc6d76cb10e249f9b80bd1cd697c8b5059f0d4aa 100644
--- a/src/main/java/org/olat/core/util/filter/impl/NekoHTMLScanner.java
+++ b/src/main/java/org/olat/core/util/filter/impl/HtmlScanner.java
@@ -19,18 +19,20 @@
*/
package org.olat.core.util.filter.impl;
-import java.io.IOException;
import java.io.StringReader;
import org.apache.logging.log4j.Logger;
-import org.cyberneko.html.parsers.SAXParser;
import org.olat.core.logging.Tracing;
import org.olat.core.util.StringHelper;
import org.xml.sax.Attributes;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
+import org.xml.sax.SAXParseException;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
* Detect tags.
*
@@ -38,27 +40,21 @@ import org.xml.sax.helpers.DefaultHandler;
* @author srosse, stephane.rosse@frentix.com, http://www.frentix.com
*
*/
-public class NekoHTMLScanner {
+public class HtmlScanner {
- private static final Logger log = Tracing.createLoggerFor(NekoHTMLScanner.class);
+ private static final Logger log = Tracing.createLoggerFor(HtmlScanner.class);
public boolean scan(String original) {
if (original == null) return false;
try {
- SAXParser parser = new SAXParser();
- parser.setFeature("http://cyberneko.org/html/features/balance-tags/document-fragment", true);
- parser.setFeature("http://cyberneko.org/html/features/balance-tags", false);
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALLOW);
+ parser.setCheckingNormalization(false);
HTMLHandler contentHandler = new HTMLHandler();
parser.setContentHandler(contentHandler);
- parser.parse(new InputSource(new StringReader(original)));
+ parser.setErrorHandler(contentHandler);
+ parser.parseFragment(new InputSource(new StringReader(original)), "");
return contentHandler.tagFound();
- } catch (SAXException e) {
- log.error("", e);
- return false;
- } catch (IOException e) {
- log.error("", e);
- return false;
} catch (Exception e) {
log.error("", e);
return false;
@@ -72,6 +68,14 @@ public class NekoHTMLScanner {
return tagFound;
}
+ @Override
+ public void error(SAXParseException exception) throws SAXException {
+ String msg = exception.getMessage();
+ if(msg != null && (msg.equals("Stray start tag \u201Chtml\\u201D.") || msg.equals("Stray start tag \u201Cbody\u201D."))) {
+ tagFound = true;
+ }
+ }
+
@Override
public void startElement(String uri, String localName, String qName, Attributes attributes) {
if(!tagFound && StringHelper.containsNonWhitespace(localName) && isTagAllowed(localName)) {
diff --git a/src/main/java/org/olat/core/util/filter/impl/OWASPAntiSamyXSSFilter.java b/src/main/java/org/olat/core/util/filter/impl/OWASPAntiSamyXSSFilter.java
index c6c9eebb7d3eb25432bf4442bc92a91cc31dbc97..067a9ebc933541a9dc9d992bc2b949ff192b39d2 100644
--- a/src/main/java/org/olat/core/util/filter/impl/OWASPAntiSamyXSSFilter.java
+++ b/src/main/java/org/olat/core/util/filter/impl/OWASPAntiSamyXSSFilter.java
@@ -27,7 +27,6 @@ import java.io.StringWriter;
import java.io.Writer;
import org.apache.logging.log4j.Logger;
-import org.cyberneko.html.parsers.SAXParser;
import org.olat.core.logging.OLATRuntimeException;
import org.olat.core.logging.Tracing;
import org.olat.core.util.filter.Filter;
@@ -42,6 +41,9 @@ import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
* Description:<br>
* OWASP AntiSamy XSSFilter
@@ -213,7 +215,7 @@ public class OWASPAntiSamyXSSFilter implements Filter {
private String cleanHtml(String original) {
try {
HTMLCleanerHandler handler = new HTMLCleanerHandler();
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
parser.setContentHandler(handler);
parser.parse(new InputSource(new StringReader(original)));
return handler.toString();
diff --git a/src/main/java/org/olat/core/util/filter/impl/SimpleHTMLTagsFilter.java b/src/main/java/org/olat/core/util/filter/impl/SimpleHTMLTagsFilter.java
index 2b36d64245965be6dbb7799feb60d74e58a6d252..b15b25a3c14a57688a036ebc130e4a4e92ca9478 100644
--- a/src/main/java/org/olat/core/util/filter/impl/SimpleHTMLTagsFilter.java
+++ b/src/main/java/org/olat/core/util/filter/impl/SimpleHTMLTagsFilter.java
@@ -19,11 +19,9 @@
*/
package org.olat.core.util.filter.impl;
-import java.io.IOException;
import java.io.StringReader;
import org.apache.logging.log4j.Logger;
-import org.cyberneko.html.parsers.SAXParser;
import org.olat.core.logging.Tracing;
import org.olat.core.util.StringHelper;
import org.olat.core.util.filter.Filter;
@@ -31,9 +29,11 @@ import org.olat.core.util.io.LimitedContentWriter;
import org.olat.search.service.document.file.FileDocumentFactory;
import org.xml.sax.Attributes;
import org.xml.sax.InputSource;
-import org.xml.sax.SAXException;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
* Description:<br>
* The html tags filter takes a string and filters all HTML tags. The filter
@@ -60,7 +60,7 @@ public class SimpleHTMLTagsFilter implements Filter {
if(original.isEmpty()) return "";
try {
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
HTMLHandler contentHandler = new HTMLHandler(original.length());
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(new StringReader(original)));
@@ -68,12 +68,6 @@ public class SimpleHTMLTagsFilter implements Filter {
text = text.replace('\u00a0', ' ');
text = StringHelper.escapeHtml(text);
return text;
- } catch (SAXException e) {
- log.error("", e);
- return null;
- } catch (IOException e) {
- log.error("", e);
- return null;
} catch (Exception e) {
log.error("", e);
return null;
@@ -99,7 +93,7 @@ public class SimpleHTMLTagsFilter implements Filter {
content.append(" ");
} else if("br".equals(elem)) {
content.append(" ");
- } else if(NekoHTMLFilter.blockTags.contains(elem) && content.length() > 0 && content.charAt(content.length() -1) != ' ' ) {
+ } else if(HtmlFilter.blockTags.contains(elem) && content.length() > 0 && content.charAt(content.length() -1) != ' ' ) {
consumeBlanck = true;
}
}
@@ -124,7 +118,7 @@ public class SimpleHTMLTagsFilter implements Filter {
collect = true;
} else if("li".equals(elem) || "p".equals(elem)) {
content.append(" ");
- } else if(NekoHTMLFilter.blockTags.contains(elem) && content.length() > 0 && content.charAt(content.length() -1) != ' ' ) {
+ } else if(HtmlFilter.blockTags.contains(elem) && content.length() > 0 && content.charAt(content.length() -1) != ' ' ) {
consumeBlanck = true;
}
}
diff --git a/src/main/java/org/olat/core/util/mail/manager/MailManagerImpl.java b/src/main/java/org/olat/core/util/mail/manager/MailManagerImpl.java
index a8cf3cc7bc91346eb6e0f17d9e67cc07fcac4216..7dce181ea3049890b6422f07b36b558a1949bfc1 100644
--- a/src/main/java/org/olat/core/util/mail/manager/MailManagerImpl.java
+++ b/src/main/java/org/olat/core/util/mail/manager/MailManagerImpl.java
@@ -93,7 +93,7 @@ import org.olat.core.logging.Tracing;
import org.olat.core.util.Encoder;
import org.olat.core.util.StringHelper;
import org.olat.core.util.WebappHelper;
-import org.olat.core.util.filter.impl.NekoHTMLFilter;
+import org.olat.core.util.filter.impl.HtmlFilter;
import org.olat.core.util.mail.ContactList;
import org.olat.core.util.mail.MailAttachment;
import org.olat.core.util.mail.MailBundle;
@@ -1772,7 +1772,7 @@ public class MailManagerImpl implements MailManager, InitializingBean {
private Multipart createMultipartAlternative(String text)
throws MessagingException {
- String pureText = new NekoHTMLFilter().filter(text, true);
+ String pureText = new HtmlFilter().filter(text, true);
MimeBodyPart textPart = new MimeBodyPart();
textPart.setText(pureText, "utf-8");
diff --git a/src/main/java/org/olat/core/util/openxml/OpenXMLDocument.java b/src/main/java/org/olat/core/util/openxml/OpenXMLDocument.java
index 894527a88ba25e0442bcf89b19c50d069ba90668..4ea05b61e1f2668473ad11ccb910cd707c042110 100644
--- a/src/main/java/org/olat/core/util/openxml/OpenXMLDocument.java
+++ b/src/main/java/org/olat/core/util/openxml/OpenXMLDocument.java
@@ -44,10 +44,9 @@ import javax.xml.parsers.DocumentBuilder;
import javax.xml.parsers.DocumentBuilderFactory;
import org.apache.commons.io.IOUtils;
-import org.cyberneko.html.parsers.SAXParser;
+import org.apache.logging.log4j.Logger;
import org.olat.core.commons.services.image.ImageUtils;
import org.olat.core.commons.services.image.Size;
-import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.StringHelper;
import org.olat.core.util.io.ShieldInputStream;
@@ -64,6 +63,7 @@ import org.xml.sax.SAXException;
import fmath.conversion.ConvertFromLatexToMathML;
import fmath.conversion.ConvertFromMathMLToWord;
+import nu.validator.htmlparser.sax.HtmlParser;
/**
* The page are A4 format, with 2.54cm margins on top, bottom, left and right.
@@ -419,12 +419,10 @@ public class OpenXMLDocument {
if(!StringHelper.containsNonWhitespace(html)) return;
try {
html = cleanUpHTML(html);
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser();
parser.setContentHandler(new HTMLToOpenXMLHandler(this, spacing));
parser.parse(new InputSource(new StringReader(html)));
- } catch (SAXException e) {
- log.error("", e);
- } catch (IOException e) {
+ } catch (SAXException | IOException e) {
log.error("", e);
}
}
@@ -433,13 +431,11 @@ public class OpenXMLDocument {
if(!StringHelper.containsNonWhitespace(html)) return;
try {
html = cleanUpHTML(html);
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser();
Element paragraphEl = getParagraphToAppendTo(newParagraph);
parser.setContentHandler(new HTMLToOpenXMLHandler(this, null, paragraphEl, true));
parser.parse(new InputSource(new StringReader(html)));
- } catch (SAXException e) {
- log.error("", e);
- } catch (IOException e) {
+ } catch (SAXException | IOException e) {
log.error("", e);
}
}
@@ -448,20 +444,18 @@ public class OpenXMLDocument {
if(!StringHelper.containsNonWhitespace(html)) return;
try {
html = cleanUpHTML(html);
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser();
Element paragraphEl = getParagraphToAppendTo(newParagraph);
handler.setInitialParagraph(paragraphEl);
parser.setContentHandler(handler);
parser.parse(new InputSource(new StringReader(html)));
- } catch (SAXException e) {
- log.error("", e);
- } catch (IOException e) {
+ } catch (SAXException | IOException e) {
log.error("", e);
}
}
/**
- * The Neko HTMl parser has some issues with <p/>.
+ * The HTMl parser has / had some issues with <p/>.
*
* @param html The HTML to clean up
* @return HTML code which Neko understands
diff --git a/src/main/java/org/olat/course/nodes/gta/ui/HTMLZippedMediaResource.java b/src/main/java/org/olat/course/nodes/gta/ui/HTMLZippedMediaResource.java
index 27806d5974879aee6be7e055637b526f690d17a5..73b4b2e2a6c8dd4c40a41fa3a9358065e497bb73 100644
--- a/src/main/java/org/olat/course/nodes/gta/ui/HTMLZippedMediaResource.java
+++ b/src/main/java/org/olat/course/nodes/gta/ui/HTMLZippedMediaResource.java
@@ -31,10 +31,9 @@ import java.util.zip.ZipOutputStream;
import javax.servlet.http.HttpServletResponse;
-import org.cyberneko.html.parsers.SAXParser;
+import org.apache.logging.log4j.Logger;
import org.olat.core.gui.media.MediaResource;
import org.olat.core.gui.media.ServletUtil;
-import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.FileUtils;
import org.olat.core.util.StringHelper;
@@ -43,6 +42,9 @@ import org.xml.sax.Attributes;
import org.xml.sax.InputSource;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
*
* Initial date: 19.11.2015<br>
@@ -128,7 +130,7 @@ public class HTMLZippedMediaResource implements MediaResource {
public HTMLHandler filter(File file) {
try(InputStream in = new FileInputStream(file);
BufferedInputStream bis = new BufferedInputStream(in, FileUtils.BSIZE)) {
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
HTMLHandler contentHandler = new HTMLHandler();
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(bis));
diff --git a/src/main/java/org/olat/ims/qti/QTI12MetadataController.java b/src/main/java/org/olat/ims/qti/QTI12MetadataController.java
index 307ed8bee6d493b30e29174701ce08cd21664cfd..5e836fbf6935800a1529e3dff01b4c6118c83798 100644
--- a/src/main/java/org/olat/ims/qti/QTI12MetadataController.java
+++ b/src/main/java/org/olat/ims/qti/QTI12MetadataController.java
@@ -31,7 +31,7 @@ import org.olat.core.gui.control.WindowControl;
import org.olat.core.util.Formatter;
import org.olat.core.util.StringHelper;
import org.olat.core.util.Util;
-import org.olat.core.util.filter.impl.NekoHTMLFilter;
+import org.olat.core.util.filter.impl.HtmlFilter;
import org.olat.ims.qti.editor.ItemNodeTabbedFormController;
import org.olat.ims.qti.editor.QTIEditHelper;
import org.olat.ims.qti.editor.beecom.objects.ChoiceQuestion;
@@ -192,7 +192,7 @@ public class QTI12MetadataController extends FormBasicController {
} else {
responseSummary = response.getContent().renderAsText();
if(responseSummary.length() > 128) {
- responseSummary = new NekoHTMLFilter().filter(responseSummary);
+ responseSummary = new HtmlFilter().filter(responseSummary);
if(responseSummary.length() > 128) {
responseSummary = responseSummary.substring(0, 125) + "...";
}
diff --git a/src/main/java/org/olat/ims/qti/qpool/QTIExportProcessor.java b/src/main/java/org/olat/ims/qti/qpool/QTIExportProcessor.java
index 7198d19135537d3dada150ac90e44e5a3c14a5f6..2cae2d429ec20000077b599de04c2ae65267fce2 100644
--- a/src/main/java/org/olat/ims/qti/qpool/QTIExportProcessor.java
+++ b/src/main/java/org/olat/ims/qti/qpool/QTIExportProcessor.java
@@ -31,7 +31,7 @@ import java.util.zip.ZipEntry;
import java.util.zip.ZipOutputStream;
import org.apache.commons.io.IOUtils;
-import org.cyberneko.html.parsers.SAXParser;
+import org.apache.logging.log4j.Logger;
import org.dom4j.Attribute;
import org.dom4j.CDATA;
import org.dom4j.Document;
@@ -41,7 +41,6 @@ import org.dom4j.Element;
import org.dom4j.Node;
import org.dom4j.io.OutputFormat;
import org.dom4j.io.XMLWriter;
-import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.CodeHelper;
import org.olat.core.util.FileUtils;
@@ -58,6 +57,9 @@ import org.olat.modules.qpool.QuestionItemFull;
import org.olat.modules.qpool.manager.QPoolFileStorage;
import org.xml.sax.InputSource;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
*
* Initial date: 11.03.2013<br>
@@ -279,7 +281,7 @@ public class QTIExportProcessor {
*/
private List<String> findMaterialInMatText(String content) {
try {
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
QTI12HtmlHandler contentHandler = new QTI12HtmlHandler();
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(new StringReader(content)));
diff --git a/src/main/java/org/olat/ims/qti/qpool/QTIImportProcessor.java b/src/main/java/org/olat/ims/qti/qpool/QTIImportProcessor.java
index 5aa1729f7122ec436d985a9210d3e3bf990e9353..c45f7c354a0b668694d369acdfb8e27a8a2fa9dc 100644
--- a/src/main/java/org/olat/ims/qti/qpool/QTIImportProcessor.java
+++ b/src/main/java/org/olat/ims/qti/qpool/QTIImportProcessor.java
@@ -43,7 +43,7 @@ import java.util.zip.ZipEntry;
import java.util.zip.ZipInputStream;
import org.apache.commons.io.IOUtils;
-import org.cyberneko.html.parsers.SAXParser;
+import org.apache.logging.log4j.Logger;
import org.dom4j.Attribute;
import org.dom4j.Document;
import org.dom4j.DocumentFactory;
@@ -55,7 +55,6 @@ import org.dom4j.io.XMLWriter;
import org.olat.core.CoreSpringFactory;
import org.olat.core.commons.persistence.DB;
import org.olat.core.id.Identity;
-import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.FileUtils;
import org.olat.core.util.PathUtils;
@@ -88,6 +87,9 @@ import org.springframework.beans.factory.annotation.Autowired;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
* This class is NOT thread-safe
*
@@ -524,7 +526,7 @@ class QTIImportProcessor {
*/
protected void findMaterialInMatText(String content, List<String> materialPath) {
try {
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
QTI12HtmlHandler contentHandler = new QTI12HtmlHandler(materialPath);
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(new StringReader(content)));
diff --git a/src/main/java/org/olat/ims/qti21/manager/openxml/QTI21WordExport.java b/src/main/java/org/olat/ims/qti21/manager/openxml/QTI21WordExport.java
index d06a8478200cb03141f2214d367b4acc1b8c6534..3e296e0754b9c7d8932415d65fb7c0768c72fdeb 100644
--- a/src/main/java/org/olat/ims/qti21/manager/openxml/QTI21WordExport.java
+++ b/src/main/java/org/olat/ims/qti21/manager/openxml/QTI21WordExport.java
@@ -39,10 +39,9 @@ import java.util.zip.ZipOutputStream;
import javax.servlet.http.HttpServletResponse;
-import org.cyberneko.html.parsers.SAXParser;
+import org.apache.logging.log4j.Logger;
import org.olat.core.gui.media.MediaResource;
import org.olat.core.gui.translator.Translator;
-import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.StringHelper;
import org.olat.core.util.Util;
@@ -79,6 +78,7 @@ import org.xml.sax.Attributes;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
+import nu.validator.htmlparser.sax.HtmlParser;
import uk.ac.ed.ph.jqtiplus.attribute.Attribute;
import uk.ac.ed.ph.jqtiplus.node.QtiNode;
import uk.ac.ed.ph.jqtiplus.node.content.basic.Block;
@@ -420,6 +420,9 @@ public class QTI21WordExport implements MediaResource {
break;
case "textentryinteraction":
startTextEntryInteraction(tag, attributes);
+ flushText();
+ Style[] currentStyles = popStyle(tag);
+ unsetTextPreferences(currentStyles);
break;
case "extendedtextinteraction":
startExtendedTextInteraction(attributes);
@@ -507,9 +510,6 @@ public class QTI21WordExport implements MediaResource {
endSimpleChoice();
break;
case "textentryinteraction":
- flushText();
- Style[] currentStyles = popStyle(tag);
- unsetTextPreferences(currentStyles);
break;
case "extendedtextinteraction":
//auto closing tag
@@ -1108,7 +1108,7 @@ public class QTI21WordExport implements MediaResource {
return Collections.emptyList();
}
try {
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser();
HTMLToOpenXMLHandler handler = new HTMLToOpenXMLHandler(factory, relPath, wrapEl, false);
handler.setMaxWidthCm(widthCm);
parser.setContentHandler(handler);
diff --git a/src/main/java/org/olat/ims/qti21/model/xml/AssessmentHtmlBuilder.java b/src/main/java/org/olat/ims/qti21/model/xml/AssessmentHtmlBuilder.java
index 0daf21a3e4acbfc8111d9520f4823823fad0aff3..f719a6a0446829bcb07e7e937df7cf67902f9700 100644
--- a/src/main/java/org/olat/ims/qti21/model/xml/AssessmentHtmlBuilder.java
+++ b/src/main/java/org/olat/ims/qti21/model/xml/AssessmentHtmlBuilder.java
@@ -19,7 +19,6 @@
*/
package org.olat.ims.qti21.model.xml;
-import java.io.ByteArrayInputStream;
import java.io.StringReader;
import java.util.ArrayList;
import java.util.List;
@@ -29,11 +28,10 @@ import javax.xml.parsers.DocumentBuilderFactory;
import javax.xml.transform.stream.StreamResult;
import org.apache.logging.log4j.Logger;
-import org.cyberneko.html.parsers.SAXParser;
import org.olat.core.gui.render.StringOutput;
import org.olat.core.logging.Tracing;
import org.olat.core.util.StringHelper;
-import org.olat.core.util.filter.impl.NekoHTMLFilter;
+import org.olat.core.util.filter.impl.HtmlFilter;
import org.w3c.dom.Attr;
import org.w3c.dom.Document;
import org.w3c.dom.Element;
@@ -42,6 +40,8 @@ import org.xml.sax.Attributes;
import org.xml.sax.InputSource;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
import uk.ac.ed.ph.jqtiplus.JqtiExtensionManager;
import uk.ac.ed.ph.jqtiplus.exception.QtiModelException;
import uk.ac.ed.ph.jqtiplus.node.AbstractNode;
@@ -82,7 +82,8 @@ public class AssessmentHtmlBuilder {
if(!StringHelper.containsNonWhitespace(html)) return false;
try {
- SAXParser parser = new SAXParser();
+ // return always true? Neko send always an html tag -> content true
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
ContentDetectionHandler contentHandler = new ContentDetectionHandler();
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(new StringReader(html)));
@@ -154,11 +155,12 @@ public class AssessmentHtmlBuilder {
} else {
htmlFragment = "<p>" + htmlFragment + "</p>";
}
- //wrap around <html> to have a root element for neko
- Document document = filter("<html>" + htmlFragment + "</html>");
+ //root element is an <html> element
+ Document document = filter(htmlFragment);
if(document != null) {
Element docElement = document.getDocumentElement();
cleanUpNamespaces(docElement);
+ // load the elements under the <html> root element
parent.getNodeGroups().load(docElement, new HTMLLoadingContext());
}
}
@@ -174,6 +176,16 @@ public class AssessmentHtmlBuilder {
element.removeAttribute("xmlns");
}
+ // escaped by the HTML parser -> always remove them
+ Attr schemaLocationAttr = element.getAttributeNode("xsiU00003Aschemalocation");
+ if(schemaLocationAttr != null) {
+ element.removeAttribute("xsiU00003Aschemalocation");
+ }
+ Attr xsiattrEscaped = element.getAttributeNode("xmlnsU00003Axsi");
+ if(xsiattrEscaped != null) {
+ element.removeAttribute("xmlnsU00003Axsi");
+ }
+
for(Node child=element.getFirstChild(); child != null; child = child.getNextSibling()) {
if(child instanceof Element) {
cleanUpNamespaces((Element)child);
@@ -181,19 +193,25 @@ public class AssessmentHtmlBuilder {
}
}
+ /**
+ * The method filters the content of textEntryInteraction (rename it correctly
+ * and remove TinyMCE attributes) and repackage the video from TinyMCE format
+ * to an object suited for QTI 2.1.
+ *
+ * @param content The content to filter
+ * @return A document object with an HTML root element with the actual
+ * blocks under this root element.
+ */
private Document filter(String content) {
try {
- SAXParser parser = new SAXParser();
- parser.setProperty("http://cyberneko.org/html/properties/names/elems", "lower");
- parser.setFeature("http://cyberneko.org/html/features/balance-tags/document-fragment", true);
- parser.setProperty("http://cyberneko.org/html/properties/default-encoding", "UTF-8");
-
+ // Alter infoset because of special namespace on tag p
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
DocumentBuilder builder = factory.newDocumentBuilder();
Document document = builder.newDocument();
HtmlToDomBuilderHandler contentHandler = new HtmlToDomBuilderHandler(document);
parser.setContentHandler(contentHandler);
- parser.parse(new InputSource(new ByteArrayInputStream(content.getBytes())));
+ parser.parse(new InputSource(new StringReader(content)));
return document;
} catch (Exception e) {
log.error("", e);
@@ -222,9 +240,11 @@ public class AssessmentHtmlBuilder {
@Override
public void startElement(String uri, String localName, String qName, Attributes attributes) {
- if(video) return;
+ if(video || "head".equalsIgnoreCase(localName) || "body".equalsIgnoreCase(localName)) {
+ return;
+ }
- if("textentryinteraction".equals(localName)) {
+ if("textentryinteraction".equalsIgnoreCase(localName)) {
localName = qName = "textEntryInteraction";
AttributesImpl attributesCleaned = new AttributesImpl("");
@@ -239,13 +259,16 @@ public class AssessmentHtmlBuilder {
}
attributes = new AttributesDelegate(attributesCleaned);
- } else if("span".equals(localName)) {
+ super.startElement(uri, localName, qName, attributes);
+ super.endElement(uri, localName, qName);
+ return;
+ } else if("span".equalsIgnoreCase(localName)) {
String cssClass = attributes.getValue("class");
if(cssClass != null && "olatFlashMovieViewer".equals(cssClass)) {
video = true;
return;
}
- } else if("u".equals(localName)) {
+ } else if("u".equalsIgnoreCase(localName)) {
qName = "span";
AttributesImpl underlineAttributes = new AttributesImpl("");
underlineAttributes.addAttributes(attributes);
@@ -266,13 +289,16 @@ public class AssessmentHtmlBuilder {
@Override
public void endElement(String uri, String localName, String qName) {
+ if("head".equalsIgnoreCase(localName) || "body".equalsIgnoreCase(localName)) {
+ return;
+ }
if(video) {
- if("span".equals(localName)) {
+ if("span".equalsIgnoreCase(localName)) {
String content = scriptBuffer.toString();
String startScript = "BPlayer.insertPlayer(";
int start = content.indexOf(startScript);
if(start >= 0) {
- int end = content.indexOf(")", start);
+ int end = content.indexOf(')', start);
String parameters = content.substring(start + startScript.length(), end);
translateToObject(uri, parameters);
}
@@ -280,10 +306,9 @@ public class AssessmentHtmlBuilder {
}
return;
}
-
- if("textentryinteraction".equals(localName)) {
- localName = qName = "textEntryInteraction";
- }
+ if("textentryinteraction".equalsIgnoreCase(localName)) {
+ return;
+ }
super.endElement(uri, localName, qName);
}
@@ -297,7 +322,7 @@ public class AssessmentHtmlBuilder {
String type = array[6].replace("\"", "");
String ooData = parameters.replace("\"", "'");
- AttributesImpl attributes = new AttributesImpl(uri);
+ AttributesImpl attributes = new AttributesImpl("");
attributes.addAttribute("data", data);
attributes.addAttribute("id", id);
attributes.addAttribute("class", "olatFlashMovieViewer");
@@ -306,8 +331,8 @@ public class AssessmentHtmlBuilder {
attributes.addAttribute("type", type);
attributes.addAttribute("data-oo-movie", ooData);
- super.startElement(uri, "object", "object", attributes);
- super.endElement(uri, "object", "object");
+ super.startElement("", "object", "object", attributes);
+ super.endElement("", "object", "object");
}
}
@@ -542,7 +567,7 @@ public class AssessmentHtmlBuilder {
String elem = localName.toLowerCase();
if("script".equals(elem)) {
collect = false;
- } else if(!NekoHTMLFilter.blockTags.contains(localName)) {
+ } else if(!HtmlFilter.blockTags.contains(localName)) {
content = true;
}
}
@@ -562,7 +587,7 @@ public class AssessmentHtmlBuilder {
String elem = localName.toLowerCase();
if("script".equals(elem)) {
collect = true;
- } else if(!NekoHTMLFilter.blockTags.contains(localName)) {
+ } else if(!HtmlFilter.blockTags.contains(localName)) {
content = true;
}
}
diff --git a/src/main/java/org/olat/ims/qti21/model/xml/TestFeedbackBuilder.java b/src/main/java/org/olat/ims/qti21/model/xml/TestFeedbackBuilder.java
index 4c24a728d7df619abee792648024caea3841762e..e54c3541f27a361cde3cf66430f7711947ee4316 100644
--- a/src/main/java/org/olat/ims/qti21/model/xml/TestFeedbackBuilder.java
+++ b/src/main/java/org/olat/ims/qti21/model/xml/TestFeedbackBuilder.java
@@ -22,7 +22,6 @@ package org.olat.ims.qti21.model.xml;
import java.io.StringReader;
import java.util.List;
-import org.cyberneko.html.parsers.SAXParser;
import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.StringHelper;
@@ -32,6 +31,8 @@ import org.xml.sax.Attributes;
import org.xml.sax.InputSource;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
import uk.ac.ed.ph.jqtiplus.node.expression.Expression;
import uk.ac.ed.ph.jqtiplus.node.expression.general.BaseValue;
import uk.ac.ed.ph.jqtiplus.node.expression.general.Variable;
@@ -235,7 +236,7 @@ public class TestFeedbackBuilder {
}
try {
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
HTMLHandler contentHandler = new HTMLHandler();
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(new StringReader(value)));
diff --git a/src/main/java/org/olat/ims/qti21/pool/QTI12To21Converter.java b/src/main/java/org/olat/ims/qti21/pool/QTI12To21Converter.java
index cdc84c511d2cabad2b92de005527f6bddd42bda3..9f125c3d47daec53eb457f5c53fa0737d6890b82 100644
--- a/src/main/java/org/olat/ims/qti21/pool/QTI12To21Converter.java
+++ b/src/main/java/org/olat/ims/qti21/pool/QTI12To21Converter.java
@@ -43,11 +43,10 @@ import javax.xml.stream.XMLStreamException;
import javax.xml.stream.XMLStreamWriter;
import org.apache.commons.lang.StringEscapeUtils;
-import org.cyberneko.html.parsers.SAXParser;
+import org.apache.logging.log4j.Logger;
import org.olat.core.CoreSpringFactory;
import org.olat.core.gui.translator.Translator;
import org.olat.core.helpers.Settings;
-import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.FileUtils;
import org.olat.core.util.StringHelper;
@@ -109,6 +108,8 @@ import org.olat.resource.OLATResource;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
import uk.ac.ed.ph.jqtiplus.node.AssessmentObject;
import uk.ac.ed.ph.jqtiplus.node.content.variable.RubricBlock;
import uk.ac.ed.ph.jqtiplus.node.content.xhtml.text.P;
@@ -777,7 +778,7 @@ public class QTI12To21Converter {
XMLOutputFactory xof = XMLOutputFactory.newInstance();
XMLStreamWriter xtw = xof.createXMLStreamWriter(out);
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser();
QTI12To21HtmlHandler handler = new QTI12To21HtmlHandler(xtw);
parser.setContentHandler(handler);
parser.parse(new InputSource(new StringReader(trimmedText)));
@@ -797,7 +798,7 @@ public class QTI12To21Converter {
private void collectMaterial(String content) {
try {
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
QTI12HtmlHandler contentHandler = new QTI12HtmlHandler(materialPath);
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(new StringReader(content)));
diff --git a/src/main/java/org/olat/ims/qti21/ui/editor/interactions/FIBEditorController.java b/src/main/java/org/olat/ims/qti21/ui/editor/interactions/FIBEditorController.java
index 0cf20c1d59a50afb16ea37432d201f73c52d1b95..167e76b14b60eb0ed9d59738adff5bc1714872d5 100644
--- a/src/main/java/org/olat/ims/qti21/ui/editor/interactions/FIBEditorController.java
+++ b/src/main/java/org/olat/ims/qti21/ui/editor/interactions/FIBEditorController.java
@@ -22,7 +22,6 @@ package org.olat.ims.qti21.ui.editor.interactions;
import java.io.ByteArrayInputStream;
import java.io.File;
-import org.cyberneko.html.parsers.SAXParser;
import org.json.JSONException;
import org.json.JSONObject;
import org.olat.core.gui.UserRequest;
@@ -54,6 +53,9 @@ import org.xml.sax.Attributes;
import org.xml.sax.InputSource;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
*
* Initial date: 24.02.2016<br>
@@ -331,10 +333,7 @@ public class FIBEditorController extends FormBasicController {
private void extractSolution(String content) {
try {
- SAXParser parser = new SAXParser();
- parser.setProperty("http://cyberneko.org/html/properties/names/elems", "lower");
- parser.setFeature("http://cyberneko.org/html/features/balance-tags/document-fragment", true);
- parser.setProperty("http://cyberneko.org/html/properties/default-encoding", "UTF-8");
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
parser.setContentHandler(new SolutionExtractorHandler());
parser.parse(new InputSource(new ByteArrayInputStream(content.getBytes())));
} catch (Exception e) {
diff --git a/src/main/java/org/olat/modules/cp/CPPrintMapper.java b/src/main/java/org/olat/modules/cp/CPPrintMapper.java
index c5e71d97aedbbc8a2be734c1608aacb2b2a860b0..246bdfc0c4433c280c0a8ed2124d86e01627d763 100644
--- a/src/main/java/org/olat/modules/cp/CPPrintMapper.java
+++ b/src/main/java/org/olat/modules/cp/CPPrintMapper.java
@@ -34,7 +34,6 @@ import javax.servlet.http.HttpServletRequest;
import org.apache.batik.css.parser.ParseException;
import org.apache.batik.css.parser.Parser;
import org.apache.logging.log4j.Logger;
-import org.cyberneko.html.parsers.SAXParser;
import org.olat.core.dispatcher.impl.StaticMediaDispatcher;
import org.olat.core.dispatcher.mapper.Mapper;
import org.olat.core.gui.components.tree.TreeNode;
@@ -51,12 +50,15 @@ import org.olat.core.util.vfs.VFSLeaf;
import org.olat.core.util.vfs.VFSMediaResource;
import org.xml.sax.InputSource;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
*
* Description:<br>
* Deliver the CP as a single page. All the HTML Pages are parser and the
* attribute href/src are rewritten to absolute /olat/m/xxxx urls. The HTML parser
- * used is NekoHTML. For CSS there is the same process. We use the Batik CSS
+ * used is validator.HtmlParser. For CSS there is the same process. We use the Batik CSS
* parser for SAC.
*
* <P>
@@ -125,18 +127,18 @@ public class CPPrintMapper implements Mapper {
}
private MediaResource deliverCompositePage(HttpServletRequest request) {
- List<NekoHtmlPageHandler> parsedPages = composePrintPage(selectedNodeIds);
+ List<HtmlPageHandler> parsedPages = composePrintPage(selectedNodeIds);
StringBuilder sb = new StringBuilder();
sb.append("<html><head>");
- for(NekoHtmlPageHandler page:parsedPages) {
+ for(HtmlPageHandler page:parsedPages) {
sb.append("<!-- Header of -->");
sb.append(page.getHeader()).append("\n\n");
}
injectJavascriptAndCss(sb);
sb.append("</head><body onload='window.focus();window.print()'>");
- for(NekoHtmlPageHandler page:parsedPages) {
+ for(HtmlPageHandler page:parsedPages) {
if(page.isEmpty()) {
String title = page.getTitle();
if(StringHelper.containsNonWhitespace(title)) {
@@ -165,18 +167,18 @@ public class CPPrintMapper implements Mapper {
}
}
- protected void bodyDecorator(NekoHtmlPageHandler page, StringBuilder sb) {
+ protected void bodyDecorator(HtmlPageHandler page, StringBuilder sb) {
sb.append("<!-- Body of ").append(page.getDocument().getName()).append("-->");
sb.append("<div class=\"o_cp_print_page\" style='clear:both; position:relative;page-break-after:always;'>\n");
sb.append(page.getBody());
sb.append("\n</div>");
}
- private List<NekoHtmlPageHandler> composePrintPage(List<String> nodeIds) {
- List<NekoHtmlPageHandler> pages = new ArrayList<>();
+ private List<HtmlPageHandler> composePrintPage(List<String> nodeIds) {
+ List<HtmlPageHandler> pages = new ArrayList<>();
for(String nodeId:nodeIds) {
- NekoHtmlPageHandler parsedPage = null;
+ HtmlPageHandler parsedPage = null;
TreeNode treeNode = ctm.getNodeById(nodeId);
String identifierRes = (String)treeNode.getUserObject();
if(StringHelper.containsNonWhitespace(identifierRes)) {
@@ -190,15 +192,15 @@ public class CPPrintMapper implements Mapper {
}
}
if(parsedPage == null) {
- parsedPage = new NekoHtmlPageHandler(treeNode, null, rootDir, baseUri);
+ parsedPage = new HtmlPageHandler(treeNode, null, rootDir, baseUri);
}
pages.add(parsedPage);
}
return pages;
}
- private NekoHtmlPageHandler parsePage(String identifierRes, VFSLeaf document, TreeNode node) {
- NekoHtmlPageHandler page = new NekoHtmlPageHandler(node, document, rootDir, baseUri);
+ private HtmlPageHandler parsePage(String identifierRes, VFSLeaf document, TreeNode node) {
+ HtmlPageHandler page = new HtmlPageHandler(node, document, rootDir, baseUri);
int index = identifierRes.lastIndexOf('/');
if(index > 0) {
String relativePath = identifierRes.substring(0, index+1);
@@ -219,7 +221,7 @@ public class CPPrintMapper implements Mapper {
rawContent = FileUtils.load(document.getInputStream(), content.getEncoding());
}
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
parser.setContentHandler(page);
parser.parse(new InputSource(new StringReader(rawContent)));
return page;
diff --git a/src/main/java/org/olat/modules/cp/NekoHtmlPageHandler.java b/src/main/java/org/olat/modules/cp/HtmlPageHandler.java
similarity index 89%
rename from src/main/java/org/olat/modules/cp/NekoHtmlPageHandler.java
rename to src/main/java/org/olat/modules/cp/HtmlPageHandler.java
index c213cda15920e9c0577e7993812dcd1cfdce4d6b..edc4e8170444a47fe5905695a077ebcb6f7d1ab4 100644
--- a/src/main/java/org/olat/modules/cp/NekoHtmlPageHandler.java
+++ b/src/main/java/org/olat/modules/cp/HtmlPageHandler.java
@@ -32,20 +32,19 @@ import org.xml.sax.helpers.DefaultHandler;
/**
*
* Description:<br>
- * SAX handler for the NekoParser
+ * SAX handler for the HtmlParser
*
* <P>
* Initial Date: 18 mars 2011 <br>
* @author srosse, stephane.rosse@frentix.com, http://www.frentix.com
*/
-//fxdiff VCRP-14: print cp
-public class NekoHtmlPageHandler extends DefaultHandler {
- private final StringBuilder header = new StringBuilder();
- private final StringBuilder body = new StringBuilder();
+public class HtmlPageHandler extends DefaultHandler {
+ private final StringBuilder header = new StringBuilder(4096);
+ private final StringBuilder body = new StringBuilder(8192);
- private final String HEAD = "HEAD";
- private final String BODY = "BODY";
- private final String SCRIPT = "SCRIPT";
+ private static final String HEAD = "head";
+ private static final String BODY = "body";
+ private static final String SCRIPT = "script";
private boolean pauseOutput;
private StringBuilder output;
@@ -56,7 +55,7 @@ public class NekoHtmlPageHandler extends DefaultHandler {
private final VFSLeaf document;
private final VFSContainer rootContainer;
- public NekoHtmlPageHandler(TreeNode node, VFSLeaf document, VFSContainer rootDir, String baseUri) {
+ public HtmlPageHandler(TreeNode node, VFSLeaf document, VFSContainer rootDir, String baseUri) {
this.document = document;
this.rootContainer = rootDir;
this.baseUri = baseUri;
@@ -108,11 +107,11 @@ public class NekoHtmlPageHandler extends DefaultHandler {
@Override
public final void startElement(String uri, String localName, String qName, Attributes attributes) {
- if (HEAD.equals(localName)) {
+ if (HEAD.equalsIgnoreCase(localName)) {
output = header;
- } else if (BODY.equals(localName)) {
+ } else if (BODY.equalsIgnoreCase(localName)) {
output = body;
- } else if (SCRIPT.equals(localName)) {
+ } else if (SCRIPT.equalsIgnoreCase(localName)) {
pauseOutput = true;
} else if (output != null) {
pauseOutput = false;
diff --git a/src/main/java/org/olat/modules/fo/QuoteAndTagFilter.java b/src/main/java/org/olat/modules/fo/QuoteAndTagFilter.java
index 079f67e7a593838ab1bb55fd75dff9f9b1c4499a..7fcc78057d33d12058056f58984cbb30d594d77b 100644
--- a/src/main/java/org/olat/modules/fo/QuoteAndTagFilter.java
+++ b/src/main/java/org/olat/modules/fo/QuoteAndTagFilter.java
@@ -22,36 +22,30 @@ package org.olat.modules.fo;
import java.io.IOException;
import java.io.StringReader;
-import org.cyberneko.html.parsers.DOMParser;
import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.StringHelper;
import org.olat.core.util.filter.Filter;
-import org.w3c.dom.Document;
+import org.w3c.dom.DocumentFragment;
import org.w3c.dom.Node;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
+import nu.validator.htmlparser.dom.HtmlDocumentBuilder;
+
public class QuoteAndTagFilter implements Filter {
private static final Logger log = Tracing.createLoggerFor(QuoteAndTagFilter.class);
- /**
- * @see org.olat.core.util.filter.Filter#filter(java.lang.String)
- */
@Override
public String filter(String original) {
try {
- DOMParser parser = new DOMParser();
- parser.parse(new InputSource(new StringReader(original)));
- Document document = parser.getDocument();
+ HtmlDocumentBuilder parser = new HtmlDocumentBuilder();
+ DocumentFragment document = parser.parseFragment(new InputSource(new StringReader(original)), "");
StringBuilder sb = new StringBuilder();
scanNode(document, sb);
return sb.toString();
- } catch (SAXException e) {
- log.error("", e);
- return null;
- } catch (IOException e) {
+ } catch (SAXException | IOException e) {
log.error("", e);
return null;
}
diff --git a/src/main/java/org/olat/modules/portfolio/ui/export/ExportBinderAsCPResource.java b/src/main/java/org/olat/modules/portfolio/ui/export/ExportBinderAsCPResource.java
index d46183a461ccc1df1a5b79290e47beaa7184c721..233e290962eb704de9e834a7e001112767c74c52 100644
--- a/src/main/java/org/olat/modules/portfolio/ui/export/ExportBinderAsCPResource.java
+++ b/src/main/java/org/olat/modules/portfolio/ui/export/ExportBinderAsCPResource.java
@@ -40,7 +40,7 @@ import javax.xml.bind.JAXBContext;
import javax.xml.bind.JAXBException;
import javax.xml.bind.Marshaller;
-import org.cyberneko.html.parsers.SAXParser;
+import org.apache.logging.log4j.Logger;
import org.olat.core.CoreSpringFactory;
import org.olat.core.commons.services.commentAndRating.CommentAndRatingSecurityCallback;
import org.olat.core.commons.services.commentAndRating.ui.UserCommentsController;
@@ -63,7 +63,6 @@ import org.olat.core.gui.util.SyntheticUserRequest;
import org.olat.core.gui.util.WindowControlMocker;
import org.olat.core.helpers.Settings;
import org.olat.core.id.OLATResourceable;
-import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.FileUtils;
import org.olat.core.util.StringHelper;
@@ -110,6 +109,9 @@ import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
*
* Initial date: 17 août 2017<br>
@@ -421,7 +423,7 @@ public class ExportBinderAsCPResource implements MediaResource {
public String exportMedia(String html, ZipOutputStream zout) {
try {
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
ExportMedia contentHandler = new ExportMedia(zout);
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(new StringReader(html)));
diff --git a/src/main/java/org/olat/modules/portfolio/ui/export/ExportBinderAsPDFResource.java b/src/main/java/org/olat/modules/portfolio/ui/export/ExportBinderAsPDFResource.java
index 85f27b8f5f6ea7bd268b698ba480156a3d30a7cf..46f117bb68b9e6353e1d32eae73e869f1c5ec802 100644
--- a/src/main/java/org/olat/modules/portfolio/ui/export/ExportBinderAsPDFResource.java
+++ b/src/main/java/org/olat/modules/portfolio/ui/export/ExportBinderAsPDFResource.java
@@ -35,7 +35,7 @@ import java.util.UUID;
import javax.servlet.http.HttpServletResponse;
import org.apache.commons.io.IOUtils;
-import org.cyberneko.html.parsers.SAXParser;
+import org.apache.logging.log4j.Logger;
import org.olat.core.CoreSpringFactory;
import org.olat.core.commons.services.pdf.PdfService;
import org.olat.core.dispatcher.DispatcherModule;
@@ -56,7 +56,6 @@ import org.olat.core.gui.translator.Translator;
import org.olat.core.gui.util.SyntheticUserRequest;
import org.olat.core.gui.util.WindowControlMocker;
import org.olat.core.helpers.Settings;
-import org.apache.logging.log4j.Logger;
import org.olat.core.logging.Tracing;
import org.olat.core.util.FileUtils;
import org.olat.core.util.StringHelper;
@@ -74,6 +73,9 @@ import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
import org.xml.sax.helpers.DefaultHandler;
+import nu.validator.htmlparser.common.XmlViolationPolicy;
+import nu.validator.htmlparser.sax.HtmlParser;
+
/**
* It makes a flat html file, copy all medias around it
* and render it as PDF via phantomjs.
@@ -238,7 +240,7 @@ public class ExportBinderAsPDFResource implements MediaResource {
public String exportMedia(String html, File outputDir) {
try {
- SAXParser parser = new SAXParser();
+ HtmlParser parser = new HtmlParser(XmlViolationPolicy.ALTER_INFOSET);
ExportMedia contentHandler = new ExportMedia(outputDir);
parser.setContentHandler(contentHandler);
parser.parse(new InputSource(new StringReader(html)));
diff --git a/src/main/java/org/olat/registration/RegistrationManager.java b/src/main/java/org/olat/registration/RegistrationManager.java
index 13e7a9a53b4f3d211b19cfd4ce528a32bcad8775..d8e84fb1bf1b832db4b6b4847bfb8e72188a6150 100644
--- a/src/main/java/org/olat/registration/RegistrationManager.java
+++ b/src/main/java/org/olat/registration/RegistrationManager.java
@@ -63,7 +63,7 @@ import org.olat.core.util.StringHelper;
import org.olat.core.util.Util;
import org.olat.core.util.WebappHelper;
import org.olat.core.util.filter.FilterFactory;
-import org.olat.core.util.filter.impl.NekoHTMLFilter;
+import org.olat.core.util.filter.impl.HtmlFilter;
import org.olat.core.util.i18n.I18nModule;
import org.olat.core.util.mail.MailManager;
import org.olat.core.util.mail.MailerResult;
@@ -603,7 +603,7 @@ public class RegistrationManager implements UserDataDeletable, UserDataExportabl
sb.append(translator.translate("disclaimer.paragraph1"))
.append("\n")
.append(translator.translate("disclaimer.paragraph2"));
- String disclaimer = new NekoHTMLFilter().filter(sb.toString(), true);
+ String disclaimer = new HtmlFilter().filter(sb.toString(), true);
out.write(disclaimer);
}
} catch (IOException e) {
diff --git a/src/main/java/org/olat/search/service/document/file/HtmlDocument.java b/src/main/java/org/olat/search/service/document/file/HtmlDocument.java
index a1a5573c50384fe4a86e4ab79f50e036cb451f76..ac56baefc8686833b68f1d28d01f934c2195e809 100644
--- a/src/main/java/org/olat/search/service/document/file/HtmlDocument.java
+++ b/src/main/java/org/olat/search/service/document/file/HtmlDocument.java
@@ -32,8 +32,8 @@ import org.apache.logging.log4j.Logger;
import org.apache.lucene.document.Document;
import org.olat.core.gui.util.CSSHelper;
import org.olat.core.logging.Tracing;
-import org.olat.core.util.filter.impl.NekoHTMLFilter;
-import org.olat.core.util.filter.impl.NekoHTMLFilter.NekoContent;
+import org.olat.core.util.filter.impl.HtmlFilter;
+import org.olat.core.util.filter.impl.HtmlFilter.HtmlContent;
import org.olat.core.util.vfs.VFSLeaf;
import org.olat.search.service.SearchResourceContext;
@@ -64,7 +64,7 @@ public class HtmlDocument extends FileDocument {
protected FileContent readContent(VFSLeaf leaf) throws DocumentException {
try(InputStream is = leaf.getInputStream()) {
// Remove all HTML and Tags
- NekoContent output = new NekoHTMLFilter().filter(is);
+ HtmlContent output = new HtmlFilter().filter(is);
if (log.isDebugEnabled())
log.debug("HTML content without tags :" + output);
diff --git a/src/main/java/org/olat/search/service/document/file/XmlDocument.java b/src/main/java/org/olat/search/service/document/file/XmlDocument.java
index a9b15e4a7482917a3918d5893cbe39193b39ca07..5b6806777a74227bbc4b67f40b52f161997631b8 100644
--- a/src/main/java/org/olat/search/service/document/file/XmlDocument.java
+++ b/src/main/java/org/olat/search/service/document/file/XmlDocument.java
@@ -33,8 +33,8 @@ import org.apache.lucene.document.Document;
import org.olat.core.gui.util.CSSHelper;
import org.olat.core.logging.Tracing;
import org.olat.core.util.FileUtils;
-import org.olat.core.util.filter.impl.NekoHTMLFilter;
-import org.olat.core.util.filter.impl.NekoHTMLFilter.NekoContent;
+import org.olat.core.util.filter.impl.HtmlFilter;
+import org.olat.core.util.filter.impl.HtmlFilter.HtmlContent;
import org.olat.core.util.vfs.VFSLeaf;
import org.olat.search.service.SearchResourceContext;
@@ -61,13 +61,12 @@ public class XmlDocument extends FileDocument {
return htmlDocument.getLuceneDocument();
}
- //fxdiff FXOLAT-97: index run in infinite loop
protected FileContent readContent(VFSLeaf leaf) throws IOException {
InputStream is = leaf.getInputStream();
// Remove all HTML and Tags
- NekoContent output;
+ HtmlContent output;
try {
- output = new NekoHTMLFilter().filter(is);
+ output = new HtmlFilter().filter(is);
if (log.isDebugEnabled() ) log.debug("HTML content without tags :" + output);
} catch (Exception e) {
throw new IOException(e);
diff --git a/src/main/java/org/olat/search/ui/SearchLRUCache.java b/src/main/java/org/olat/search/ui/SearchLRUCache.java
deleted file mode 100644
index a63d0f1b26bd87e5828f64f42278fb2ee44f8fd3..0000000000000000000000000000000000000000
--- a/src/main/java/org/olat/search/ui/SearchLRUCache.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/**
- * <a href="http://www.openolat.org">
- * OpenOLAT - Online Learning and Training</a><br>
- * <p>
- * Licensed under the Apache License, Version 2.0 (the "License"); <br>
- * you may not use this file except in compliance with the License.<br>
- * You may obtain a copy of the License at the
- * <a href="http://www.apache.org/licenses/LICENSE-2.0">Apache homepage</a>
- * <p>
- * Unless required by applicable law or agreed to in writing,<br>
- * software distributed under the License is distributed on an "AS IS" BASIS, <br>
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. <br>
- * See the License for the specific language governing permissions and <br>
- * limitations under the License.
- * <p>
- * Initial code contributed and copyrighted by<br>
- * frentix GmbH, http://www.frentix.com
- * <p>
- */
-package org.olat.search.ui;
-
-import org.apache.commons.collections.map.LRUMap;
-import org.olat.search.SearchResults;
-
-/**
- * A LRU map
- *
- * @author srosse, stephane.rosse@frentix.com, http://www.frentix.com
- */
-class SearchLRUCache extends LRUMap {
-
- private static final long serialVersionUID = -6748874217812035135L;
-
- public SearchLRUCache() {
- super(5);
- }
-
- @Override
- public SearchResults get(Object key) {
- SearchEntry searchEntry = (SearchEntry)super.get(key);
- if(searchEntry != null) {
- if(searchEntry.isUpToDate()) {
- return searchEntry.getSearchResults();
- }
- remove(key);
- }
- return null;
- }
-
- @Override
- public Object put(Object key, Object value) {
- return super.put(key, new SearchEntry((SearchResults)value));
- }
-
- public class SearchEntry {
- private final SearchResults results;
- private long timestamp = System.currentTimeMillis();
-
- public SearchEntry(SearchResults results) {
- this.results = results;
- }
-
- public SearchResults getSearchResults() {
- return results;
- }
-
- public boolean isUpToDate() {
- return System.currentTimeMillis() - timestamp < 300000;
- }
- }
-}
diff --git a/src/test/java/org/olat/core/util/FormatLatexFormulasTest.java b/src/test/java/org/olat/core/util/FormatLatexFormulasTest.java
index 2dbc4a3076fc77278611b6822a1fd04cdd3e4577..4143c54beda0f62407cf8cd218ba9e653a514794 100644
--- a/src/test/java/org/olat/core/util/FormatLatexFormulasTest.java
+++ b/src/test/java/org/olat/core/util/FormatLatexFormulasTest.java
@@ -60,7 +60,7 @@ public class FormatLatexFormulasTest {
{ "<span class='math\"'\"></span>", Boolean.FALSE },
{ "<span class='math></\"span>", Boolean.FALSE },
{ "<span class='math></'span>", Boolean.FALSE },
- { "<span class='math", Boolean.TRUE },
+ { "<span class='math", Boolean.FALSE },
{ "<span class=math\"", Boolean.FALSE }
});
}
diff --git a/src/test/java/org/olat/core/util/StringHelperTest.java b/src/test/java/org/olat/core/util/StringHelperTest.java
index 61185ec8372bdedf1e3d4cbc16b61c2d76d3b566..b90a0cf7ba5c448a30dc3bb4aa508387f8fa5e36 100644
--- a/src/test/java/org/olat/core/util/StringHelperTest.java
+++ b/src/test/java/org/olat/core/util/StringHelperTest.java
@@ -174,13 +174,13 @@ public class StringHelperTest {
Assert.assertFalse(StringHelper.isHtml("http://some.domain:8080/olat/dmz/registration/index.html?key=b67a28bd5e5820155b3ba496ef16d1d9&lang=de"));
//good and bad html code
- Assert.assertTrue(StringHelper.isHtml("<html><head></head><body>Hello world</body></html>"));
Assert.assertTrue(StringHelper.isHtml("Hello <p>world</p>"));
Assert.assertTrue(StringHelper.isHtml("<ul><li>Hello<li>world</ul>"));
Assert.assertTrue(StringHelper.isHtml("Hello<br>world"));
+ Assert.assertTrue(StringHelper.isHtml("<html><head></head><body>Hello world</body></html>"));
+ Assert.assertTrue(StringHelper.isHtml("<html><head></head><body><p>Hello world</p></body></html>"));
}
-
@Test
public void formatAsCSVString() {
List<String> entries = new ArrayList<>();
diff --git a/src/test/java/org/olat/core/util/filter/impl/NekoHTMLFilterTest.java b/src/test/java/org/olat/core/util/filter/impl/HtmlFilterTest.java
similarity index 90%
rename from src/test/java/org/olat/core/util/filter/impl/NekoHTMLFilterTest.java
rename to src/test/java/org/olat/core/util/filter/impl/HtmlFilterTest.java
index a909db60567d1fa48681f0f346ccd79b2ec403f9..580cb01ff8cbcfb7f65a377e5c84b63b150c17d0 100644
--- a/src/test/java/org/olat/core/util/filter/impl/NekoHTMLFilterTest.java
+++ b/src/test/java/org/olat/core/util/filter/impl/HtmlFilterTest.java
@@ -21,34 +21,21 @@ package org.olat.core.util.filter.impl;
import java.io.ByteArrayInputStream;
-import org.junit.After;
import org.junit.Assert;
-import org.junit.Before;
import org.junit.Test;
-import org.junit.runner.RunWith;
-import org.junit.runners.JUnit4;
-import org.olat.core.util.filter.impl.NekoHTMLFilter.NekoContent;
+import org.olat.core.util.filter.impl.HtmlFilter.HtmlContent;
/**
* Description:<br>
- * This test case tests the NEko HTML tags filter
+ * This test case tests the HTML tags filter.
*
* <P>
* Initial Date: 14.07.2009 <br>
* @author gnaegi
*/
-@RunWith(JUnit4.class)
-public class NekoHTMLFilterTest{
+public class HtmlFilterTest {
- protected NekoHTMLFilter filter;
-
- @Before public void setup() {
- filter = new NekoHTMLFilter();
- }
-
- @After public void tearDown() {
- filter = null;
- }
+ private final HtmlFilter filter = new HtmlFilter();
private void t(String input, String result) {
Assert.assertEquals(result, filter.filter(input));
@@ -61,6 +48,7 @@ public class NekoHTMLFilterTest{
public void escaping() {
String output = filter.filter("Test ä test");
System.out.println(output);
+ Assert.assertEquals("Test \u00E4 test", output);
}
@Test public void testPlainText() {
@@ -158,8 +146,8 @@ public class NekoHTMLFilterTest{
p("<a href=\"index.html\">GO HERE</a>", "GO HERE");
t("<a href=\"/go/here/index.html\">GO HERE</a>", "GO HERE");
p("<a href=\"/go/here/index.html\">GO HERE</a>", "GO HERE");
- t("<a href=\"http://www.openolat.org/go/here/index.html\" title=\\\"super site\\\">GO HERE</a>", "GO HERE");
- p("<a href=\"http://www.openolat.org/go/here/index.html\" title=\\\"super site\\\">GO HERE</a>", "http://www.openolat.org/go/here/index.html GO HERE");
+ t("<a href=\"http://www.openolat.org/go/here/index.html\" title=\"super site\">GO HERE</a>", "GO HERE");
+ p("<a href=\"http://www.openolat.org/go/here/index.html\" title=\"super site\">GO HERE</a>", "http://www.openolat.org/go/here/index.html GO HERE");
t("<a href=\"https://www.openolat.org/go/here/index.html\" title=\"super site\">GO HERE</a>", "GO HERE");
p("<a href=\"https://www.openolat.org/go/here/index.html\" title=\"super site\">GO HERE</a>", "https://www.openolat.org/go/here/index.html GO HERE");
@@ -174,7 +162,7 @@ public class NekoHTMLFilterTest{
+ "</body></html>"; // Text = 'Dies ist der Test Text'
String text = "Hello Neko Test HTML Seite fuer JUnit Test Dies ist der Test\u00A0Text"; // must include '\u00A0' !!! 19.5.2010/cg
- NekoContent content = filter.filter(new ByteArrayInputStream(htmlText.getBytes()));
+ HtmlContent content = filter.filter(new ByteArrayInputStream(htmlText.getBytes()));
Assert.assertNotNull(content);
Assert.assertEquals("Hello Neko", content.getTitle());
Assert.assertEquals(text, content.getContent());
diff --git a/src/test/java/org/olat/core/util/filter/impl/HtmlMathScannerTest.java b/src/test/java/org/olat/core/util/filter/impl/HtmlMathScannerTest.java
new file mode 100644
index 0000000000000000000000000000000000000000..979ef768d005dfff36a5e2489198a239b7564caf
--- /dev/null
+++ b/src/test/java/org/olat/core/util/filter/impl/HtmlMathScannerTest.java
@@ -0,0 +1,84 @@
+/**
+ * <a href="http://www.openolat.org">
+ * OpenOLAT - Online Learning and Training</a><br>
+ * <p>
+ * Licensed under the Apache License, Version 2.0 (the "License"); <br>
+ * you may not use this file except in compliance with the License.<br>
+ * You may obtain a copy of the License at the
+ * <a href="http://www.apache.org/licenses/LICENSE-2.0">Apache homepage</a>
+ * <p>
+ * Unless required by applicable law or agreed to in writing,<br>
+ * software distributed under the License is distributed on an "AS IS" BASIS, <br>
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. <br>
+ * See the License for the specific language governing permissions and <br>
+ * limitations under the License.
+ * <p>
+ * Initial code contributed and copyrighted by<br>
+ * frentix GmbH, http://www.frentix.com
+ * <p>
+ */
+package org.olat.core.util.filter.impl;
+
+import java.util.Arrays;
+import java.util.Collection;
+
+import org.junit.After;
+import org.junit.Assert;
+import org.junit.Before;
+import org.junit.Test;
+import org.junit.runner.RunWith;
+import org.junit.runners.Parameterized;
+import org.junit.runners.Parameterized.Parameters;
+import org.olat.core.util.WebappHelper;
+
+/**
+ *
+ * Initial date: 18 mai 2019<br>
+ * @author srosse, stephane.rosse@frentix.com, http://www.frentix.com
+ *
+ */
+@RunWith(Parameterized.class)
+public class HtmlMathScannerTest {
+
+ @Parameters
+ public static Collection<Object[]> data() {
+ return Arrays.asList(new Object[][] {
+ { true, "<p><span class=\"math\" title=\"%5Coverbrace%7Bx+%5Ccdots+x%5C%2C%7D%5E%7Bk%5Crm%5C%3Btimes%7D%20%5Cqquad%0A%5Cunderbrace%7Bx+y+z%5C%2C%7D_%7B%3E%5C%2C0%7D\">\\overbrace{x+\\cdots+x\\,}^{k\\rm\\;times} \\qquad \\underbrace{x+y+z\\,}_{>\\,0}</span></p>" },
+ { true, "\\(math\\)" },
+ { true, "Ceci <p>\\[math\\]</p>" },
+ { true, "Was ist das <p>$$math$$</p>" },
+ { true, "<span><math xmlns=\"http://www.w3.org/1998/Math/MathML\"><semantics><mrow><mi>Ï€</mi><mo>=</mo><mfrac><mi>C</mi><mi>d</mi></mfrac></mrow><annotation encoding=\"SnuggleTeX\">$\\pi = \\frac{C}{d}</annotation></semantics></math>" },
+ { false, "math" },
+ { false, "<p>math</p>" }
+ });
+ }
+
+
+ private final HtmlMathScanner filter = new HtmlMathScanner();
+
+ private boolean hasMath;
+ private String input;
+ private boolean markersSettings;
+
+ public HtmlMathScannerTest(boolean hasMath, String input) {
+ this.hasMath = hasMath;
+ this.input = input;
+ }
+
+ @Before
+ public void setMathJaxMarkers() {
+ markersSettings = WebappHelper.isMathJaxMarkers();
+ new WebappHelper().setMathJaxMarkers(true);
+ }
+
+ @After
+ public void resetMathJaxMarkers() {
+ new WebappHelper().setMathJaxMarkers(markersSettings);
+ }
+
+ @Test
+ public void hasMath() {
+ boolean output = filter.scan(input);
+ Assert.assertEquals(hasMath, output);
+ }
+}
diff --git a/src/test/java/org/olat/core/util/filter/impl/SimpleHTMLTagsFilterTest.java b/src/test/java/org/olat/core/util/filter/impl/SimpleHTMLTagsFilterTest.java
index 99703b8ffc133c05e2801de388ebcf61b4f5509d..317e057b79463bd5bc706fb8365c5d063703532a 100644
--- a/src/test/java/org/olat/core/util/filter/impl/SimpleHTMLTagsFilterTest.java
+++ b/src/test/java/org/olat/core/util/filter/impl/SimpleHTMLTagsFilterTest.java
@@ -92,7 +92,7 @@ public class SimpleHTMLTagsFilterTest {
}
// Boundary test: this filter does NOT decode HTML entities. Use the
- // NekoHTMLFilter if you need this feature!
+ // HtmlFilter if you need this feature!
@Test
public void testTagsWithEntities() {
t("Gnägi", "Gnägi");
diff --git a/src/test/java/org/olat/ims/qti21/model/xml/AssessmentHtmlBuilderTest.java b/src/test/java/org/olat/ims/qti21/model/xml/AssessmentHtmlBuilderTest.java
index a684a8dd086aafd8190dc1deb5aed922addbc97e..83b7a9b8e1370f9acd1dac82b2ecef19fc487163 100644
--- a/src/test/java/org/olat/ims/qti21/model/xml/AssessmentHtmlBuilderTest.java
+++ b/src/test/java/org/olat/ims/qti21/model/xml/AssessmentHtmlBuilderTest.java
@@ -32,12 +32,15 @@ import uk.ac.ed.ph.jqtiplus.JqtiExtensionManager;
import uk.ac.ed.ph.jqtiplus.node.content.ItemBody;
import uk.ac.ed.ph.jqtiplus.node.content.basic.Block;
import uk.ac.ed.ph.jqtiplus.node.content.basic.TextRun;
+import uk.ac.ed.ph.jqtiplus.node.content.xhtml.table.Table;
import uk.ac.ed.ph.jqtiplus.node.content.xhtml.text.P;
import uk.ac.ed.ph.jqtiplus.node.item.AssessmentItem;
import uk.ac.ed.ph.jqtiplus.node.item.interaction.Interaction;
import uk.ac.ed.ph.jqtiplus.node.item.interaction.TextEntryInteraction;
import uk.ac.ed.ph.jqtiplus.node.item.interaction.choice.SimpleChoice;
+import uk.ac.ed.ph.jqtiplus.node.item.interaction.content.Hottext;
import uk.ac.ed.ph.jqtiplus.serialization.QtiSerializer;
+import uk.ac.ed.ph.jqtiplus.utils.QueryUtils;
/**
* Test the conversion from TinyMCE HTML code to the QTI Works object
@@ -62,9 +65,28 @@ public class AssessmentHtmlBuilderTest {
String content = new AssessmentHtmlBuilder().flowStaticString(helper.getFlowStatics());
Assert.assertTrue(content.contains(">Hello world<"));
}
-
+
+ @Test
+ public void isEmpty() {
+ String html = "Hello world";
+ boolean contains = new AssessmentHtmlBuilder().containsSomething(html);
+ Assert.assertTrue(contains);
+
+ html = "<p> </p>";
+ contains = new AssessmentHtmlBuilder().containsSomething(html);
+ Assert.assertTrue(contains);
+
+ html = "<p></p>";
+ contains = new AssessmentHtmlBuilder().containsSomething(html);
+ Assert.assertTrue(contains);
+
+ html = " ";
+ contains = new AssessmentHtmlBuilder().containsSomething(html);
+ Assert.assertFalse(contains);
+ }
+
@Test
- public void filter_alt() {
+ public void appendHtml() throws IOException {
String content = "<p>Test <textEntryInteraction responseIdentifier=\"RESPONSE_1\"/> </p>";
AssessmentItem item = new AssessmentItem();
@@ -78,10 +100,112 @@ public class AssessmentHtmlBuilderTest {
Assert.assertTrue(interaction instanceof TextEntryInteraction);
Assert.assertNotNull(interaction.getResponseIdentifier());
Assert.assertEquals("RESPONSE_1", interaction.getResponseIdentifier().toString());
+
+ // The serializer can throw some exceptions if it doesn't like the model
+ // we want to serialize.
+ StringOutput sb = new StringOutput();
+ QtiSerializer qtiSerializer = new QtiSerializer(new JqtiExtensionManager());
+ qtiSerializer.serializeJqtiObject(helper, new StreamResult(sb));
+ String serializedQti = sb.toString();
+ Assert.assertTrue(serializedQti.contains("textEntryInteraction"));
+ sb.close();
+ }
+
+ @Test
+ public void appendHtml_namespace() throws IOException {
+ String content = "<p xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xmlns=\"http://www.imsglobal.org/xsd/imsqti_v2p1\" xsi:schemaLocation=\"http://www.imsglobal.org/xsd/imsqti_v2p1 http://www.imsglobal.org/xsd/imsqti_v2p1.xsd\">European rocket <textEntryInteraction responseIdentifier=\"RESPONSE_2\" data-qti-solution=\"Ariane\" openolatType=\"string\" placeholderText=\"ari\"/>New text<textEntryInteraction responseIdentifier=\"RESPONSE_1\" data-qti-solution=\"Falcon9\" openolatType=\"string\" placeholderText=\"falc\"/></p>";
+
+ AssessmentItem item = new AssessmentItem();
+ ItemBody helper = new ItemBody(item);
+ new AssessmentHtmlBuilder().appendHtml(helper, content);
+
+ List<Interaction> interactions = helper.findInteractions();
+ Assert.assertNotNull(interactions);
+ Assert.assertEquals(2, interactions.size());
+ TextEntryInteraction entry1 = (TextEntryInteraction)interactions.get(0);
+ Assert.assertEquals("RESPONSE_2", entry1.getResponseIdentifier().toString());
+ TextEntryInteraction entry2 = (TextEntryInteraction)interactions.get(1);
+ Assert.assertEquals("RESPONSE_1", entry2.getResponseIdentifier().toString());
+
+ List<Block> blocks = helper.getBlocks();
+ Assert.assertNotNull(blocks);
+ Assert.assertEquals(1, blocks.size());
+ Assert.assertTrue(blocks.get(0) instanceof P);
+
+ // The serializer can throw some exceptions if it doesn't like the model
+ // we want to serialize.
+ StringOutput sb = new StringOutput();
+ QtiSerializer qtiSerializer = new QtiSerializer(new JqtiExtensionManager());
+ qtiSerializer.serializeJqtiObject(helper, new StreamResult(sb));
+ String serializedQti = sb.toString();
+ Assert.assertTrue(serializedQti.contains("textEntryInteraction"));
+ sb.close();
+ }
+
+ @Test
+ public void appendHtml_textEntryInteractions() throws IOException {
+ String content = "<p>Usefull for circles <textentryinteraction responseidentifier=\"RESPONSE_2\" data-qti-solution=\"Pi\" data-qti-solution-empty=\"false\"></textentryinteraction>New <br>text<textentryinteraction responseidentifier=\"RESPONSE_1\" data-qti-solution=\"Ln\" data-qti-solution-empty=\"false\"></textentryinteraction></p>";
+
+ AssessmentItem item = new AssessmentItem();
+ ItemBody helper = new ItemBody(item);
+ new AssessmentHtmlBuilder().appendHtml(helper, content);
+
+ List<Interaction> interactions = helper.findInteractions();
+ Assert.assertNotNull(interactions);
+ Assert.assertEquals(2, interactions.size());
+ TextEntryInteraction entry1 = (TextEntryInteraction)interactions.get(0);
+ Assert.assertEquals("RESPONSE_2", entry1.getResponseIdentifier().toString());
+ TextEntryInteraction entry2 = (TextEntryInteraction)interactions.get(1);
+ Assert.assertEquals("RESPONSE_1", entry2.getResponseIdentifier().toString());
+
+ List<Block> blocks = helper.getBlocks();
+ Assert.assertNotNull(blocks);
+ Assert.assertEquals(1, blocks.size());
+ Assert.assertTrue(blocks.get(0) instanceof P);
+
+ // The serializer can throw some exceptions if it doesn't like the model
+ // we want to serialize.
+ StringOutput sb = new StringOutput();
+ QtiSerializer qtiSerializer = new QtiSerializer(new JqtiExtensionManager());
+ qtiSerializer.serializeJqtiObject(helper, new StreamResult(sb));
+ String serializedQti = sb.toString();
+ Assert.assertTrue(serializedQti.contains("textEntryInteraction"));
+ sb.close();
+ }
+
+ @Test
+ public void appendHtml_textEntryInteractionsAutoClosed() throws IOException {
+ String content = "<p xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xmlns=\"http://www.imsglobal.org/xsd/imsqti_v2p1\" xsi:schemaLocation=\"http://www.imsglobal.org/xsd/imsqti_v2p1 http://www.imsglobal.org/xsd/imsqti_v2p1.xsd\">Usefull for circles <textEntryInteraction responseIdentifier=\"RESPONSE_2\" data-qti-solution=\"Pi\" openolatType=\"string\" placeholderText=\"314\" />New text<textEntryInteraction responseIdentifier=\"RESPONSE_1\" data-qti-solution=\"Ln\" openolatType=\"string\" placeholderText=\"lognat\" /></p>";
+
+ AssessmentItem item = new AssessmentItem();
+ ItemBody helper = new ItemBody(item);
+ new AssessmentHtmlBuilder().appendHtml(helper, content);
+
+ List<Interaction> interactions = helper.findInteractions();
+ Assert.assertNotNull(interactions);
+ Assert.assertEquals(2, interactions.size());
+ TextEntryInteraction entry1 = (TextEntryInteraction)interactions.get(0);
+ Assert.assertEquals("RESPONSE_2", entry1.getResponseIdentifier().toString());
+ TextEntryInteraction entry2 = (TextEntryInteraction)interactions.get(1);
+ Assert.assertEquals("RESPONSE_1", entry2.getResponseIdentifier().toString());
+
+ List<Block> blocks = helper.getBlocks();
+ Assert.assertNotNull(blocks);
+ Assert.assertEquals(1, blocks.size());
+ Assert.assertTrue(blocks.get(0) instanceof P);
+
+ // The serializer can throw some exceptions if it doesn't like the model
+ // we want to serialize.
+ StringOutput sb = new StringOutput();
+ QtiSerializer qtiSerializer = new QtiSerializer(new JqtiExtensionManager());
+ qtiSerializer.serializeJqtiObject(helper, new StreamResult(sb));
+ String serializedQti = sb.toString();
+ Assert.assertTrue(serializedQti.contains("textEntryInteraction"));
+ sb.close();
}
@Test
- public void filter() throws IOException {
+ public void appendHtml_accidentalHtmlRootElement() throws IOException {
String content = "<html><p>Test \u00EA<strong><span><img src='img.jpg'></span></strong></p><p>Test 2</p></html>";
AssessmentItem item = new AssessmentItem();
@@ -91,19 +215,73 @@ public class AssessmentHtmlBuilderTest {
List<Block> paragraphs = helper.getBlocks();
Assert.assertNotNull(paragraphs);
Assert.assertEquals(2, paragraphs.size());
+ Assert.assertTrue(paragraphs.get(0) instanceof P);
+ Assert.assertTrue(paragraphs.get(1) instanceof P);
- // The serializer can throw some exceptions if it doens't like the model
+ // The serializer can throw some exceptions if it doesn't like the model
// we want to serialize.
StringOutput sb = new StringOutput();
QtiSerializer qtiSerializer = new QtiSerializer(new JqtiExtensionManager());
qtiSerializer.serializeJqtiObject(helper, new StreamResult(sb));
String serializedQti = sb.toString();
Assert.assertTrue(serializedQti.contains("img.jpg"));
+ Assert.assertTrue(serializedQti.contains("Test \u00EA"));// check encoding
+ Assert.assertTrue(serializedQti.contains("Test 2"));
+ Assert.assertFalse(serializedQti.contains("<html"));
+ sb.close();
+ }
+
+ @Test
+ public void appendHtml_table() throws IOException {
+ String content = "<p>This is a hot<hottext identifier=\"ht2efcb885fc4149bdf3fcf567b44a78\">text</hottext> et un dans une table:</p>\n" +
+ "<table class=\"b_default\" style=\"height: 109px;\" width=\"440\">\n" +
+ "<tbody>\n" +
+ "<tr>\n" +
+ "<td style=\"width: 146.34375px;\">Ceci</td>\n" +
+ "<td style=\"width: 146.34375px;\">est</td>\n" +
+ "<td style=\"width: 146.34375px;\">une</td>\n" +
+ "</tr>\n" +
+ "<tr>\n" +
+ "<td style=\"width: 146.34375px;\">table</td>\n" +
+ "<td style=\"width: 146.34375px;\">donc</td>\n" +
+ "<td style=\"width: 146.34375px; border-color: #ed9595; background-color: #edc5c5;\"><hottext identifier=\"ht8096a52c17f663ab9d73b7071609\">table</hottext></td>\n" +
+ "</tr>\n" +
+ "</tbody>\n" +
+ "</table>";
+
+ AssessmentItem item = new AssessmentItem();
+ ItemBody helper = new ItemBody(item);
+ new AssessmentHtmlBuilder().appendHtml(helper, content);
+
+ List<Block> paragraphs = helper.getBlocks();
+ Assert.assertNotNull(paragraphs);
+ Assert.assertEquals(2, paragraphs.size());
+ Assert.assertTrue(paragraphs.get(0) instanceof P);
+ Assert.assertTrue(paragraphs.get(1) instanceof Table);
+
+ List<Hottext> hottexts = QueryUtils.search(Hottext.class, helper.getBlocks());
+ Assert.assertNotNull(hottexts);
+ Assert.assertEquals(2, hottexts.size());
+ Assert.assertEquals("ht2efcb885fc4149bdf3fcf567b44a78", hottexts.get(0).getIdentifier().toString());
+ Assert.assertEquals("ht8096a52c17f663ab9d73b7071609", hottexts.get(1).getIdentifier().toString());
+
+ // The serializer can throw some exceptions if it doesn't like the model
+ // we want to serialize.
+ StringOutput sb = new StringOutput();
+ QtiSerializer qtiSerializer = new QtiSerializer(new JqtiExtensionManager());
+ qtiSerializer.serializeJqtiObject(helper, new StreamResult(sb));
+ String serializedQti = sb.toString();
+ Assert.assertTrue(serializedQti.contains("ht2efcb885fc4149bdf3fcf567b44a78"));
+ Assert.assertTrue(serializedQti.contains("ht8096a52c17f663ab9d73b7071609"));
+ Assert.assertTrue(serializedQti.contains("background-color: #edc5c5;"));
+ Assert.assertTrue(serializedQti.contains("width: 146.34375px;"));
+ Assert.assertFalse(serializedQti.contains("ns:identifier"));// check namespaced attributes
+ Assert.assertFalse(serializedQti.contains("ns:style"));
sb.close();
}
@Test
- public void serializeVideo() throws IOException {
+ public void appendHtml_serializeVideo() throws IOException {
String content = "<p><span id=\"olatFlashMovieViewer213060\" class=\"olatFlashMovieViewer\" style=\"display:block;border:solid 1px #000; width:320px; height:240px;\">\n"
+ "<script src=\"/raw/fx-111111x11/movie/player.js\" type=\"text/javascript\"></script>\n"
+ "<script type=\"text/javascript\" defer=\"defer\">// <![CDATA[\n"
@@ -118,7 +296,7 @@ public class AssessmentHtmlBuilderTest {
List<Block> paragraphs = helper.getBlocks();
Assert.assertNotNull(paragraphs);
Assert.assertEquals(1, paragraphs.size());
-
+
StringOutput sb = new StringOutput();
QtiSerializer qtiSerializer = new QtiSerializer(new JqtiExtensionManager());
qtiSerializer.serializeJqtiObject(helper, new StreamResult(sb));
@@ -127,6 +305,7 @@ public class AssessmentHtmlBuilderTest {
Assert.assertTrue(serializedQti.contains("object"));
Assert.assertFalse(serializedQti.contains("span"));
Assert.assertFalse(serializedQti.contains("script"));
+ Assert.assertFalse(serializedQti.contains("ns:data"));
sb.close();
}
}
diff --git a/src/test/java/org/olat/ims/qti21/model/xml/TestFeedbackBuilderTest.java b/src/test/java/org/olat/ims/qti21/model/xml/TestFeedbackBuilderTest.java
new file mode 100644
index 0000000000000000000000000000000000000000..ad0e8ffbe7ca384ad7ffb17eddc62d11e06209a4
--- /dev/null
+++ b/src/test/java/org/olat/ims/qti21/model/xml/TestFeedbackBuilderTest.java
@@ -0,0 +1,68 @@
+/**
+ * <a href="http://www.openolat.org">
+ * OpenOLAT - Online Learning and Training</a><br>
+ * <p>
+ * Licensed under the Apache License, Version 2.0 (the "License"); <br>
+ * you may not use this file except in compliance with the License.<br>
+ * You may obtain a copy of the License at the
+ * <a href="http://www.apache.org/licenses/LICENSE-2.0">Apache homepage</a>
+ * <p>
+ * Unless required by applicable law or agreed to in writing,<br>
+ * software distributed under the License is distributed on an "AS IS" BASIS, <br>
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. <br>
+ * See the License for the specific language governing permissions and <br>
+ * limitations under the License.
+ * <p>
+ * Initial code contributed and copyrighted by<br>
+ * frentix GmbH, http://www.frentix.com
+ * <p>
+ */
+package org.olat.ims.qti21.model.xml;
+
+import java.util.Arrays;
+import java.util.Collection;
+
+import org.junit.Assert;
+import org.junit.Test;
+import org.junit.runner.RunWith;
+import org.junit.runners.Parameterized;
+import org.junit.runners.Parameterized.Parameters;
+
+
+/**
+ *
+ * Initial date: 18 mai 2019<br>
+ * @author srosse, stephane.rosse@frentix.com, http://www.frentix.com
+ *
+ */
+@RunWith(Parameterized.class)
+public class TestFeedbackBuilderTest {
+
+ @Parameters
+ public static Collection<Object[]> data() {
+ return Arrays.asList(new Object[][] {
+ { true, "<p>\n </p>" },
+ { true, "\n" },
+ { true, "<span></span>" },
+ { false, "Hello" },
+ { false, "<span class='olatFlashMovieViewer'></span>" },
+ { false, "<img src='img.jpg' />" },
+ { false, "<img src='img.jpg'></img>" },
+ { false, "<p>Hello</p>" }
+ });
+ }
+
+ private boolean empty;
+ private String input;
+
+ public TestFeedbackBuilderTest(boolean empty, String input) {
+ this.empty = empty;
+ this.input = input;
+ }
+
+ @Test
+ public void isEmpty() {
+ boolean output = TestFeedbackBuilder.isEmpty(input);
+ Assert.assertEquals(empty, output);
+ }
+}
diff --git a/src/test/java/org/olat/search/service/document/file/HtmlDocumentTest.java b/src/test/java/org/olat/search/service/document/file/HtmlDocumentTest.java
new file mode 100644
index 0000000000000000000000000000000000000000..728b0d895be4913895368e63a7cad1e3ed47b735
--- /dev/null
+++ b/src/test/java/org/olat/search/service/document/file/HtmlDocumentTest.java
@@ -0,0 +1,53 @@
+/**
+ * <a href="http://www.openolat.org">
+ * OpenOLAT - Online Learning and Training</a><br>
+ * <p>
+ * Licensed under the Apache License, Version 2.0 (the "License"); <br>
+ * you may not use this file except in compliance with the License.<br>
+ * You may obtain a copy of the License at the
+ * <a href="http://www.apache.org/licenses/LICENSE-2.0">Apache homepage</a>
+ * <p>
+ * Unless required by applicable law or agreed to in writing,<br>
+ * software distributed under the License is distributed on an "AS IS" BASIS, <br>
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. <br>
+ * See the License for the specific language governing permissions and <br>
+ * limitations under the License.
+ * <p>
+ * Initial code contributed and copyrighted by<br>
+ * frentix GmbH, http://www.frentix.com
+ * <p>
+ */
+package org.olat.search.service.document.file;
+
+import java.io.File;
+import java.io.IOException;
+import java.net.URISyntaxException;
+import java.net.URL;
+
+import org.junit.Assert;
+import org.junit.Test;
+import org.olat.core.util.vfs.VFSLeaf;
+import org.olat.test.VFSJavaIOFile;
+
+/**
+ *
+ * @author srosse, stephane.rosse@frentix.com, http://www.frentix.com
+ */
+public class HtmlDocumentTest {
+
+ @Test
+ public void testHtmlDocument()
+ throws DocumentException, DocumentAccessException, URISyntaxException, IOException {
+ URL url = HtmlDocumentTest.class.getResource("test.html");
+ Assert.assertNotNull(url);
+
+ VFSLeaf doc = new VFSJavaIOFile("test.html", new File(url.toURI()));
+ HtmlDocument document = new HtmlDocument();
+ FileContent content = document.readContent(doc);
+ Assert.assertNotNull(content);
+ Assert.assertEquals("Hello", content.getTitle().trim());
+ String body = content.getContent();
+ Assert.assertTrue(body.trim().contains("Hello, this is HTML"));
+ }
+
+}
diff --git a/src/test/java/org/olat/search/service/document/file/XmlDocumentTest.java b/src/test/java/org/olat/search/service/document/file/XmlDocumentTest.java
new file mode 100644
index 0000000000000000000000000000000000000000..75a3e122e8c483646875cbeeaf9f5a495c997ba8
--- /dev/null
+++ b/src/test/java/org/olat/search/service/document/file/XmlDocumentTest.java
@@ -0,0 +1,52 @@
+/**
+ * <a href="http://www.openolat.org">
+ * OpenOLAT - Online Learning and Training</a><br>
+ * <p>
+ * Licensed under the Apache License, Version 2.0 (the "License"); <br>
+ * you may not use this file except in compliance with the License.<br>
+ * You may obtain a copy of the License at the
+ * <a href="http://www.apache.org/licenses/LICENSE-2.0">Apache homepage</a>
+ * <p>
+ * Unless required by applicable law or agreed to in writing,<br>
+ * software distributed under the License is distributed on an "AS IS" BASIS, <br>
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. <br>
+ * See the License for the specific language governing permissions and <br>
+ * limitations under the License.
+ * <p>
+ * Initial code contributed and copyrighted by<br>
+ * frentix GmbH, http://www.frentix.com
+ * <p>
+ */
+package org.olat.search.service.document.file;
+
+import java.io.File;
+import java.io.IOException;
+import java.net.URISyntaxException;
+import java.net.URL;
+
+import org.junit.Assert;
+import org.junit.Test;
+import org.olat.core.util.vfs.VFSLeaf;
+import org.olat.test.VFSJavaIOFile;
+
+/**
+ *
+ * @author srosse, stephane.rosse@frentix.com, http://www.frentix.com
+ */
+public class XmlDocumentTest {
+
+ @Test
+ public void testXmlDocument()
+ throws DocumentException, DocumentAccessException, URISyntaxException, IOException {
+ URL url = XmlDocumentTest.class.getResource("test.xml");
+ Assert.assertNotNull(url);
+
+ VFSLeaf doc = new VFSJavaIOFile("test.xml", new File(url.toURI()));
+ XmlDocument document = new XmlDocument();
+ FileContent content = document.readContent(doc);
+ Assert.assertNotNull(content);
+ String body = content.getContent();
+ Assert.assertTrue(body.trim().contains("This is XML"));
+ }
+
+}
diff --git a/src/test/java/org/olat/search/service/document/file/test.html b/src/test/java/org/olat/search/service/document/file/test.html
index e1153f6bba59fe3fab57b9f5dbe23aabf955292b..b8e9d517dca196d511cedc791640a6148da94699 100644
--- a/src/test/java/org/olat/search/service/document/file/test.html
+++ b/src/test/java/org/olat/search/service/document/file/test.html
@@ -3,6 +3,6 @@
<title>Hello</title>
</head>
<body>
- <h4>Hello</h4>
+ <h4>Hello, this is HTML</h4>
</body>
</html>
\ No newline at end of file
diff --git a/src/test/java/org/olat/search/service/document/file/test.xml b/src/test/java/org/olat/search/service/document/file/test.xml
new file mode 100644
index 0000000000000000000000000000000000000000..a477f6cbf5143a8f4881c73daa955d2514a2f1ff
--- /dev/null
+++ b/src/test/java/org/olat/search/service/document/file/test.xml
@@ -0,0 +1,4 @@
+<some>
+ <tag>Hello</tag>
+ <frentix>This is XML</frentix>
+</some>
\ No newline at end of file
diff --git a/src/test/java/org/olat/search/service/document/file/test2.html b/src/test/java/org/olat/search/service/document/file/test2.html
new file mode 100644
index 0000000000000000000000000000000000000000..e1153f6bba59fe3fab57b9f5dbe23aabf955292b
--- /dev/null
+++ b/src/test/java/org/olat/search/service/document/file/test2.html
@@ -0,0 +1,8 @@
+<html>
+ <head>
+ <title>Hello</title>
+ </head>
+ <body>
+ <h4>Hello</h4>
+ </body>
+</html>
\ No newline at end of file
diff --git a/src/test/java/org/olat/test/AllTestsJunit4.java b/src/test/java/org/olat/test/AllTestsJunit4.java
index fdc4532321cc62795cd38c6ddc5eaf31f8a49154..c7b09caa2b7ecb441586fd621b344767e0f7b90b 100644
--- a/src/test/java/org/olat/test/AllTestsJunit4.java
+++ b/src/test/java/org/olat/test/AllTestsJunit4.java
@@ -56,7 +56,8 @@ import org.junit.runners.Suite;
org.olat.core.util.filter.impl.XSSFilterParamTest.class,
org.olat.core.util.filter.impl.AddBaseURLToMediaRelativeURLFilterTest.class,
org.olat.core.util.filter.impl.SimpleHTMLTagsFilterTest.class,
- org.olat.core.util.filter.impl.NekoHTMLFilterTest.class,
+ org.olat.core.util.filter.impl.HtmlFilterTest.class,
+ org.olat.core.util.filter.impl.HtmlMathScannerTest.class,
org.olat.core.util.filter.impl.ConditionalHtmlCommentsFilterTest.class,
org.olat.core.util.filter.impl.XMLValidCharacterFilterTest.class,
org.olat.core.util.filter.impl.XMLValidEntityFilterTest.class,
@@ -295,6 +296,7 @@ import org.junit.runners.Suite;
org.olat.properties.PropertyTest.class,
org.olat.search.service.document.file.FileDocumentFactoryTest.class,
org.olat.search.service.indexer.repository.course.SPCourseNodeIndexerTest.class,
+ org.olat.search.service.document.file.HtmlDocumentTest.class,
org.olat.search.service.document.file.PDFDocumentTest.class,
org.olat.search.service.document.file.OfficeDocumentTest.class,
org.olat.core.commons.services.notifications.manager.NotificationsManagerTest.class,
@@ -318,6 +320,7 @@ import org.junit.runners.Suite;
org.olat.ims.qti21.model.xml.AssessmentItemBuilderTest.class,
org.olat.ims.qti21.model.xml.MultipleChoiceAssessmentItemBuilderTest.class,
org.olat.ims.qti21.model.xml.SingleChoiceAssessmentItemBuilderTest.class,
+ org.olat.ims.qti21.model.xml.TestFeedbackBuilderTest.class,
org.olat.ims.qti21.model.xml.HottextAssessmentItemBuilderTest.class,
org.olat.ims.qti21.model.xml.AssessmentHtmlBuilderTest.class,
org.olat.ims.qti21.model.xml.AssessmentItemPackageTest.class,