X4O: Fixed html void element auto folding to allowed tags only

nx01-x4o-sax3/src/main/java/org/x4o/sax3/SAX3WriterHtml.java

@@ -24,7 +24,10 @@ package org.x4o.sax3;
 
 import java.io.IOException;
 import java.io.Writer;
+import java.util.Arrays;
 import java.util.Calendar;
+import java.util.Collections;
+import java.util.List;
 
 import org.x4o.sax3.io.ContentCloseable;
 import org.x4o.sax3.io.SAX3PropertyConfig;
@@ -40,8 +43,11 @@ import org.xml.sax.helpers.AttributesImpl;
  */
 public class SAX3WriterHtml extends SAX3WriterEnum<SAX3WriterHtml.Tag, SAX3WriterXml> {
 
+	static public final List<String> HTML_VOID_TAGS = Collections.unmodifiableList(Tag.valuesVoidElement());
+	
 	public SAX3WriterHtml(Writer out, String encoding) {
 		super(new SAX3WriterXml(out, encoding), "", SAX3XMLConstants.NULL_NS_URI);
+		getPropertyConfig().setProperty(SAX3WriterXml.OUTPUT_FOLD_EMPTY_TAGS, HTML_VOID_TAGS);
 	}
 
 	public SAX3PropertyConfig getPropertyConfig() {
@@ -247,7 +253,7 @@ public class SAX3WriterHtml extends SAX3WriterEnum<SAX3WriterHtml.Tag, SAX3Write
 
 		/* Deprecated TAGS */
 		frameset,
-		frame,
+		frame(true),
 		noframes,
 		tt,
 		font,
@@ -259,14 +265,17 @@ public class SAX3WriterHtml extends SAX3WriterEnum<SAX3WriterHtml.Tag, SAX3Write
 		acronym,
 		applet,
 		iframe,
+		menuitem(true),
+		keygen(true),
+		command(true),
 
 		/* HTML 4 TAGS */
 		html,
 		head,
 		title,
-		meta,
+		meta(true),
-		link,
+		link(true),
-		base,
+		base(true),
 		body,
 		script,
 		style,
@@ -282,9 +291,9 @@ public class SAX3WriterHtml extends SAX3WriterEnum<SAX3WriterHtml.Tag, SAX3Write
 		span,
 		p,
 		pre,
-		img,
+		img(true),
-		hr,
+		hr(true),
-		br,
+		br(true),
 		b,
 		em,
 		strong,
@@ -308,12 +317,12 @@ public class SAX3WriterHtml extends SAX3WriterEnum<SAX3WriterHtml.Tag, SAX3Write
 
 		abbr,
 		address,
-		area,
+		area(true),
 		bdo,
 		blockquote,
 		cite,
 		code,
-		col,
+		col(true),
 		colgroup,
 		del,
 		dfn,
@@ -324,7 +333,7 @@ public class SAX3WriterHtml extends SAX3WriterEnum<SAX3WriterHtml.Tag, SAX3Write
 		map,
 		menu,
 		object,
-		param,
+		param(true),
 		optgroup,
 		q,
 		s,
@@ -335,7 +344,7 @@ public class SAX3WriterHtml extends SAX3WriterEnum<SAX3WriterHtml.Tag, SAX3Write
 
 		form,
 		fieldset,
-		input,
+		input(true),
 		option,
 		label,
 		button,
@@ -346,16 +355,14 @@ public class SAX3WriterHtml extends SAX3WriterEnum<SAX3WriterHtml.Tag, SAX3Write
 		canvas,
 		audio,
 		video,
-		source,
+		source(true),
-		embed,
+		embed(true),
-		track,
+		track(true),
 		datalist,
-		keygen,
 		output,
 		article,
 		aside,
 		bdi,
-		command,
 		details,
 		dialog,
 		summary,
@@ -373,7 +380,26 @@ public class SAX3WriterHtml extends SAX3WriterEnum<SAX3WriterHtml.Tag, SAX3Write
 		rp,
 		section,
 		time,
-		wbr,;
+		wbr(true),
+		;
+		
+		private final boolean voidElement;
+		
+		private Tag() {
+			this(false);
+		}
+		
+		private Tag(boolean voidElement) {
+			this.voidElement = voidElement;
+		}
+		
+		public boolean voidElement() {
+			return voidElement;
+		}
+		
+		static public List<String> valuesVoidElement() {
+			return Arrays.stream(values()).filter(v -> v.voidElement()).map(v -> v.name()).toList();
+		}
 	}
 
 	private final static String DOCTYPE_NAME = "HTML PUBLIC";

nx01-x4o-sax3/src/main/java/org/x4o/sax3/io/AbstractContentWriterHandler.java

@@ -34,6 +34,7 @@ import java.io.Writer;
 import java.nio.charset.Charset;
 import java.util.ArrayList;
 import java.util.HashMap;
+import java.util.HashSet;
 import java.util.List;
 import java.util.Map;
 import java.util.Objects;
@@ -65,6 +66,7 @@ public class AbstractContentWriterHandler implements ContentHandler, Closeable {
 	private boolean printReturn = false;
 	private String lastElement = null;
 	private Stack<String> elements = null;
+	private Set<String> foldEmptyTags = null;
 
 	//@formatter:off
 	private final static String PROPERTY_CONTEXT_PREFIX = SAX3PropertyConfig.X4O_PROPERTIES_PREFIX + "content/"; // TODO: change to "writer/xml"
@@ -77,6 +79,7 @@ public class AbstractContentWriterHandler implements ContentHandler, Closeable {
 	public final static String OUTPUT_COMMENT_AUTO_SPACE  = PROPERTY_CONTEXT_PREFIX + "output/comment-auto-space";
 	public final static String OUTPUT_LINE_BREAK_WIDTH    = PROPERTY_CONTEXT_PREFIX + "output/line-break-width";
 	public final static String OUTPUT_LINE_PER_ATTRIBUTE  = PROPERTY_CONTEXT_PREFIX + "output/line-per-attribute";
+	public final static String OUTPUT_FOLD_EMPTY_TAGS     = PROPERTY_CONTEXT_PREFIX + "output/fold-empty-tags";
 	public final static String PROLOG_LICENCE_FILE        = PROPERTY_CONTEXT_PREFIX + "prolog/licence-file";
 	public final static String PROLOG_LICENCE_RESOURCE    = PROPERTY_CONTEXT_PREFIX + "prolog/licence-resource";
 	public final static String PROLOG_LICENCE_ENCODING    = PROPERTY_CONTEXT_PREFIX + "prolog/licence-encoding";
@@ -96,6 +99,7 @@ public class AbstractContentWriterHandler implements ContentHandler, Closeable {
 				new PropertyConfigItem(OUTPUT_COMMENT_AUTO_SPACE,  Boolean.class,   true),
 				new PropertyConfigItem(OUTPUT_LINE_BREAK_WIDTH,    Integer.class,   -1),
 				new PropertyConfigItem(OUTPUT_LINE_PER_ATTRIBUTE,  Boolean.class,   false),
+				new PropertyConfigItem(OUTPUT_FOLD_EMPTY_TAGS,     List.class       ), // if null|empty than all empty tags are folded
 				new PropertyConfigItem(PROLOG_LICENCE_ENCODING,    String.class,    SAX3XMLConstants.XML_DEFAULT_ENCODING),
 				new PropertyConfigItem(PROLOG_LICENCE_FILE,        File.class       ),
 				new PropertyConfigItem(PROLOG_LICENCE_RESOURCE,    String.class     ),
@@ -429,27 +433,29 @@ public class AbstractContentWriterHandler implements ContentHandler, Closeable {
 			throw new SAXException("Unexpected end tag: " + localName + " should be: " + elements.peek());
 		}
 		elements.pop();
-
-		if (startElement != null) {
-			String tag = startElement.toString();
-			write(tag.substring(0, tag.length() - 1));// rm normal close
-			write(SAX3XMLConstants.TAG_CLOSE_EMPTY);
-			startElement = null;
-			indent--;
-			return;
-		}
-
 		indent--;
-		if (printReturn || !localName.equals(lastElement)) {
+		if (startElement != null) {
-			write(getPropertyConfig().getPropertyString(OUTPUT_CHAR_NEWLINE));
+			// no child element of other content, thus use the empty body closing tag 
-			writeIndent();
+			if (allowEndElementFolding(localName)) {
+				String tag = startElement.toString();
+				write(tag.substring(0, tag.length() - 1));// rm normal close
+				write(SAX3XMLConstants.TAG_CLOSE_EMPTY);
+				startElement = null;
+				return;
+			}
+			// void element not allowed, so print start + empty body on same line
+			autoCloseStartElement();
 		} else {
-			printReturn = true;
+			if (printReturn || !localName.equals(lastElement)) {
+				write(getPropertyConfig().getPropertyString(OUTPUT_CHAR_NEWLINE));
+				writeIndent();
+			} else {
+				printReturn = true;
+			}
 		}
 		if (localName == null) {
 			localName = "null";
 		}
-
 		write(SAX3XMLConstants.TAG_OPEN_END);
 		if (SAX3XMLConstants.NULL_NS_URI.equals(uri) || uri == null) {
 			write(localName);
@@ -470,6 +476,19 @@ public class AbstractContentWriterHandler implements ContentHandler, Closeable {
 		}
 	}
 
+	private boolean allowEndElementFolding(String tag) {
+		if (foldEmptyTags == null) {
+			foldEmptyTags = new HashSet<>();
+			if (getPropertyConfig().getProperty(OUTPUT_FOLD_EMPTY_TAGS) != null) {
+				foldEmptyTags.addAll(getPropertyConfig().getPropertyList(OUTPUT_FOLD_EMPTY_TAGS));
+			}
+		}
+		if (foldEmptyTags.isEmpty()) {
+			return true;
+		}
+		return foldEmptyTags.contains(tag);
+	}
+	
 	/**
 	 * Starts the prefix mapping of an xml namespace uri.
 	 *