public List<Newsitem> parseContent(String content) throws Exception {
List<Newsitem> newsitems = new ArrayList<Newsitem>();
Tag newsDiv = this.extractTagByClassName(this.stripHtmlComments(content), "box_news");
NodeList nodes = this.extractTagsByClassName(newsDiv.toHtml(), "subItem");
for (int i = 0; i < nodes.size(); i++) {
NewsitemImpl newsitem = new NewsitemImpl();
Tag itemTable = (Tag) nodes.elementAt(i);
Tag titleTag = this.extractTagByClassName(itemTable.toHtml(), "subItemtitle");
newsitem.setTitle(titleTag.toPlainTextString());
Node descriptionSpan =
titleTag.getNextSibling().getNextSibling().getNextSibling().getNextSibling();
newsitem.setDescription(
descriptionSpan
.toPlainTextString()
.replaceAll("[^\\u0000-\\u00FF]", " ")
.replace(" Read More...", "")
.trim());
Tag linkTag = (Tag) extractLinks(itemTable.toHtml(), "/index.php.*").elementAt(0);
newsitem.setUrl(URL_PREFIX + linkTag.getAttribute("href"));
newsitems.add(newsitem);
}
return newsitems;
}
public void testParentConnections() throws ParserException {
String tag1 = "<custom>";
String tag2 = "<custom>something</custom>";
String tag3 = "</custom>";
createParser(tag1 + tag2 + tag3);
parser.setNodeFactory(
new PrototypicalNodeFactory(
new Tag[] {
new CustomTag(false), new AnotherTag(false),
}));
parseAndAssertNodeCount(3);
CustomTag customTag = (CustomTag) node[0];
assertStringEquals("first custom tag html", tag1 + "</custom>", customTag.toHtml());
assertNull("first custom tag should have no parent", customTag.getParent());
customTag = (CustomTag) node[1];
assertStringEquals("second custom tag html", tag2, customTag.toHtml());
assertNull("second custom tag should have no parent", customTag.getParent());
Node firstChild = customTag.childAt(0);
assertType("firstChild", Text.class, firstChild);
Node parent = firstChild.getParent();
assertNotNull("first child parent should not be null", parent);
assertSame("parent and custom tag should be the same", customTag, parent);
Tag endTag = (Tag) node[2];
assertStringEquals("third custom tag html", tag3, endTag.toHtml());
assertNull("end tag should have no parent", endTag.getParent());
}
/**
* A bug in the freshmeat page - really bad html tag - <A>Revision<\a> Reported by
* Mazlan Mat Note: Actually, this is completely legal HTML - Derrick
*/
public void testFreshMeatBug() throws ParserException {
String html = "<a>Revision</a>";
createParser(html, "http://www.yahoo.com");
parseAndAssertNodeCount(1);
assertTrue("Node 0 should be a tag", node[0] instanceof Tag);
Tag tag = (Tag) node[0];
assertEquals("Tag Contents", html, tag.toHtml());
assertEquals("Node 0 should have one child", 1, tag.getChildren().size());
assertTrue("The child should be a string node", tag.getChildren().elementAt(0) instanceof Text);
Text stringNode = (Text) tag.getChildren().elementAt(0);
assertEquals("Text Contents", "Revision", stringNode.getText());
}
public void testCompositeTagWithSelfChildren() throws ParserException {
String tag1 = "<custom>";
String tag2 = "<custom>something</custom>";
String tag3 = "</custom>";
createParser(tag1 + tag2 + tag3);
parser.setNodeFactory(
new PrototypicalNodeFactory(
new Tag[] {
new CustomTag(false), new AnotherTag(false),
}));
parseAndAssertNodeCount(3);
CustomTag customTag = (CustomTag) node[0];
assertEquals("child count", 0, customTag.getChildCount());
assertFalse("custom tag should not be xml end tag", customTag.isEmptyXmlTag());
assertStringEquals("first custom tag html", tag1 + "</custom>", customTag.toHtml());
customTag = (CustomTag) node[1];
assertStringEquals("second custom tag html", tag2, customTag.toHtml());
Tag endTag = (Tag) node[2];
assertStringEquals("third custom tag html", tag3, endTag.toHtml());
}
/**
* Given an input, analyze each HTML tag and remove unsecured attributes from them.
*
* @param contents The content to verify
* @return the content, secure.
*/
public String ensureAllAttributesAreSafe(String contents) {
StringBuffer sb = new StringBuffer(contents.length());
try {
Lexer lexer = new Lexer(contents);
Node node;
while ((node = lexer.nextNode()) != null) {
if (node instanceof Tag) {
Tag tag = (Tag) node;
this.checkAndValidateAttributes(tag, false);
sb.append(tag.toHtml());
} else {
sb.append(node.toHtml());
}
}
} catch (Exception e) {
throw new ForumException("Problems while parsing HTML: " + e, e);
}
return sb.toString();
}
