/** * This method gets called when an end tag is encountered. * * @param uri the Uniform Resource Identifier * @param lname the local name (without prefix), or the empty string if Namespace processing is * not being performed. * @param name the name of the tag that ends */ public void endElement(String uri, String lname, String name) { // System.err.println("End: " + name); name = name.toLowerCase(); if (ElementTags.PARAGRAPH.equals(name)) { try { document.add((Element) stack.pop()); return; } catch (DocumentException e) { throw new ExceptionConverter(e); } } if (HtmlTagMap.isHead(name)) { // we do nothing return; } if (HtmlTagMap.isTitle(name)) { if (currentChunk != null) { bodyAttributes.put(ElementTags.TITLE, currentChunk.getContent()); } return; } if (HtmlTagMap.isMeta(name)) { // we do nothing return; } if (HtmlTagMap.isLink(name)) { // we do nothing return; } if (HtmlTagMap.isBody(name)) { // we do nothing return; } if (myTags.containsKey(name)) { XmlPeer peer = (XmlPeer) myTags.get(name); if (ElementTags.TABLE.equals(peer.getTag())) { tableBorder = false; } super.handleEndingTags(peer.getTag()); return; } // super.handleEndingTags is replaced with handleEndingTags // suggestion by Ken Auer handleEndingTags(name); }
/** * This method gets called when a start tag is encountered. * * @param uri the Uniform Resource Identifier * @param lname the local name (without prefix), or the empty string if Namespace processing is * not being performed. * @param name the name of the tag that is encountered * @param attrs the list of attributes */ public void startElement(String uri, String lname, String name, Attributes attrs) { // System.err.println("Start: " + name); // super.handleStartingTags is replaced with handleStartingTags // suggestion by Vu Ngoc Tan/Hop name = name.toLowerCase(); if (HtmlTagMap.isHtml(name)) { // we do nothing return; } if (HtmlTagMap.isHead(name)) { // we do nothing return; } if (HtmlTagMap.isTitle(name)) { // we do nothing return; } if (HtmlTagMap.isMeta(name)) { // we look if we can change the body attributes String meta = null; String content = null; if (attrs != null) { for (int i = 0; i < attrs.getLength(); i++) { String attribute = attrs.getQName(i); if (attribute.equalsIgnoreCase(HtmlTags.CONTENT)) content = attrs.getValue(i); else if (attribute.equalsIgnoreCase(HtmlTags.NAME)) meta = attrs.getValue(i); } } if (meta != null && content != null) { bodyAttributes.put(meta, content); } return; } if (HtmlTagMap.isLink(name)) { // we do nothing for the moment, in a later version we could extract // the style sheet return; } if (HtmlTagMap.isBody(name)) { // maybe we could extract some info about the document: color, // margins,... // but that's for a later version... XmlPeer peer = new XmlPeer(ElementTags.ITEXT, name); peer.addAlias(ElementTags.TOP, HtmlTags.TOPMARGIN); peer.addAlias(ElementTags.BOTTOM, HtmlTags.BOTTOMMARGIN); peer.addAlias(ElementTags.RIGHT, HtmlTags.RIGHTMARGIN); peer.addAlias(ElementTags.LEFT, HtmlTags.LEFTMARGIN); bodyAttributes.putAll(peer.getAttributes(attrs)); handleStartingTags(peer.getTag(), bodyAttributes); return; } if (myTags.containsKey(name)) { XmlPeer peer = (XmlPeer) myTags.get(name); if (ElementTags.TABLE.equals(peer.getTag()) || ElementTags.CELL.equals(peer.getTag())) { Properties p = peer.getAttributes(attrs); String value; if (ElementTags.TABLE.equals(peer.getTag()) && (value = p.getProperty(ElementTags.BORDERWIDTH)) != null) { if (Float.parseFloat(value + "f") > 0) { tableBorder = true; } } if (tableBorder) { p.put(ElementTags.LEFT, String.valueOf(true)); p.put(ElementTags.RIGHT, String.valueOf(true)); p.put(ElementTags.TOP, String.valueOf(true)); p.put(ElementTags.BOTTOM, String.valueOf(true)); } handleStartingTags(peer.getTag(), p); return; } handleStartingTags(peer.getTag(), peer.getAttributes(attrs)); return; } Properties attributes = new Properties(); if (attrs != null) { for (int i = 0; i < attrs.getLength(); i++) { String attribute = attrs.getQName(i).toLowerCase(); attributes.setProperty(attribute, attrs.getValue(i).toLowerCase()); } } handleStartingTags(name, attributes); }