/** Test scan with data which is of diff nodes type */
public void testScan() throws ParserException {
createParser(
"<A HREF=\"mytest.html\"><IMG SRC=\"abcd.jpg\">Hello World</A>", "http://www.yahoo.com");
parser.setNodeFactory(
new PrototypicalNodeFactory(
new Tag[] {
new LinkTag(), new ImageTag(),
}));
parseAndAssertNodeCount(1);
assertTrue("Node should be a link node", node[0] instanceof LinkTag);
LinkTag linkTag = (LinkTag) node[0];
// Get the link data and cross-check
Node[] dataNode = new Node[10];
int i = 0;
for (SimpleNodeIterator e = linkTag.children(); e.hasMoreNodes(); ) {
dataNode[i++] = e.nextNode();
}
assertEquals("Number of data nodes", new Integer(2), new Integer(i));
assertTrue("First data node should be an Image Node", dataNode[0] instanceof ImageTag);
assertTrue("Second data node shouls be a String Node", dataNode[1] instanceof Text);
// Check the contents of each data node
ImageTag imageTag = (ImageTag) dataNode[0];
assertEquals("Image URL", "http://www.yahoo.com/abcd.jpg", imageTag.getImageURL());
Text stringNode = (Text) dataNode[1];
assertEquals("String Contents", "Hello World", stringNode.getText());
}
public void testCompositeTagWithDeadlock() throws ParserException {
createParser(
"<custom>"
+ "<another>something"
+ "</custom>"
+ "<custom>"
+ "<another>else</another>"
+ "</custom>");
parser.setNodeFactory(
new PrototypicalNodeFactory(
new Tag[] {
new CustomTag(), new AnotherTag(true),
}));
parseAndAssertNodeCount(2);
assertType("node", CustomTag.class, node[0]);
CustomTag customTag = (CustomTag) node[0];
assertEquals("child count", 1, customTag.getChildCount());
assertFalse("custom tag should not be xml end tag", customTag.isEmptyXmlTag());
assertEquals("starting loc", 0, customTag.getStartPosition());
assertEquals("ending loc", 8, customTag.getEndPosition());
assertEquals("starting line position", 0, customTag.getStartingLineNumber());
assertEquals("ending line position", 0, customTag.getEndingLineNumber());
AnotherTag anotherTag = (AnotherTag) customTag.childAt(0);
assertEquals("anotherTag child count", 1, anotherTag.getChildCount());
Text stringNode = (Text) anotherTag.childAt(0);
assertStringEquals("anotherTag child text", "something", stringNode.toPlainTextString());
assertStringEquals(
"first custom tag html",
"<custom><another>something</another></custom>",
customTag.toHtml());
customTag = (CustomTag) node[1];
assertStringEquals(
"second custom tag html", "<custom><another>else</another></custom>", customTag.toHtml());
}
public void testCompositeTagWithTwoNestedTags() throws ParserException {
createParser(
"<Custom>"
+ "<Another>"
+ "Hello"
+ "</Another>"
+ "<unknown>"
+ "World"
+ "</unknown>"
+ "<Custom/>"
+ "</Custom>"
+ "<Custom/>");
parser.setNodeFactory(
new PrototypicalNodeFactory(
new Tag[] {
new CustomTag(), new AnotherTag(false),
}));
parseAndAssertNodeCount(2);
assertType("first node", CustomTag.class, node[0]);
assertType("second node", CustomTag.class, node[1]);
CustomTag customTag = (CustomTag) node[0];
assertEquals("first custom tag children count", 5, customTag.getChildCount());
Node node = customTag.childAt(0);
assertType("first child", AnotherTag.class, node);
AnotherTag anotherTag = (AnotherTag) node;
assertEquals("another tag children count", 1, anotherTag.getChildCount());
node = anotherTag.childAt(0);
assertType("nested child", Text.class, node);
Text text = (Text) node;
assertEquals("text", "Hello", text.toPlainTextString());
}
/**
* 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 testErroneousLinkBug() throws ParserException {
createParser(
"Site Comments?<br>"
+ "<a href=\"mailto:[email protected]?subject=Site Comments\">"
+ "Mail Us"
+ "<a>");
parseAndAssertNodeCount(4);
// The first node should be a Text
assertTrue("First node should be a Text", node[0] instanceof Text);
Text stringNode = (Text) node[0];
assertEquals("Text of the Text", "Site Comments?", stringNode.getText());
assertTrue("Second node should be a tag", node[1] instanceof Tag);
assertTrue("Third node should be a link", node[2] instanceof LinkTag);
// LinkScanner.evaluate() says no HREF means it isn't a link:
assertTrue("Fourth node should be a tag", node[3] instanceof Tag);
}
/**
* Bug reported by Raj Sharma,5-Apr-2002, upon parsing http://www.samachar.com, the entire page
* could not be picked up. The problem was occurring after parsing a particular link after which
* the parsing would not proceed. This link was spread over three lines. The bug has been
* reproduced and fixed.
*/
public void testMultipleLineBug() throws ParserException {
createParser(
"<LI><font color=\"FF0000\" size=-1><b>Tech Samachar:</b></font><a \n"
+ "href=\"http://ads.samachar.com/bin/redirect/tech.txt?http://www.samachar.com/tech\n"
+ "nical.html\"> Journalism 3.0</a> by Rajesh Jain");
parser.setNodeFactory(new PrototypicalNodeFactory(new LinkTag()));
parseAndAssertNodeCount(8);
assertTrue("Seventh node should be a link tag", node[6] instanceof LinkTag);
LinkTag linkTag = (LinkTag) node[6];
String exp =
new String(
"http://ads.samachar.com/bin/redirect/tech.txt?http://www.samachar.com/technical.html");
// assertEquals("Length of link tag",exp.length(), linkTag.getLink().length());
assertStringEquals("Link URL of link tag", exp, linkTag.getLink());
assertEquals("Link Text of link tag", " Journalism 3.0", linkTag.getLinkText());
assertTrue("Eight node should be a string node", node[7] instanceof Text);
Text stringNode = (Text) node[7];
assertEquals("String node contents", " by Rajesh Jain", stringNode.getText());
}
public String toString() {
return delegate.toString();
}
public String toHtml() {
return delegate.toHtml();
}
public Node getNextSibling() {
return delegate.getNextSibling();
}
public Node getLastChild() {
return delegate.getLastChild();
}
/**
* Get the children of this node.
*
* @return The list of children contained by this node, if it's been set, <code>null</code>
* otherwise.
*/
public NodeList getChildren() {
return (delegate.getChildren());
}
public String getText() {
return delegate.getText();
}
public void accept(NodeVisitor visitor) {
delegate.accept(visitor);
}
@Override
public void visitStringNode(Text string) {
Element parent = getParent(string.getParent());
parent.setInnerText(string.getText());
}
public boolean equals(Object arg0) {
return delegate.equals(arg0);
}
public Node getParent() {
return delegate.getParent();
}
public void collectInto(NodeList list, NodeFilter filter) {
delegate.collectInto(list, filter);
}
public void setParent(Node node) {
delegate.setParent(node);
}
/**
* Gets the starting position of the node.
*
* @return The start position.
*/
public int getStartPosition() {
return (delegate.getStartPosition());
}
/**
* Set the children of this node.
*
* @param children The new list of children this node contains.
*/
public void setChildren(NodeList children) {
delegate.setChildren(children);
}
/**
* Sets the starting position of the node.
*
* @param position The new start position.
*/
public void setStartPosition(int position) {
delegate.setStartPosition(position);
}
public Node getPreviousSibling() {
return delegate.getPreviousSibling();
}
/**
* Gets the ending position of the node.
*
* @return The end position.
*/
public int getEndPosition() {
return (delegate.getEndPosition());
}
public void setText(String text) {
delegate.setText(text);
}
/**
* Sets the ending position of the node.
*
* @param position The new end position.
*/
public void setEndPosition(int position) {
delegate.setEndPosition(position);
}
public String toPlainTextString() {
return delegate.toPlainTextString();
}
/**
* Get the page this node came from.
*
* @return The page that supplied this node.
*/
public Page getPage() {
return (delegate.getPage());
}
public void doSemanticAction() throws ParserException {
delegate.doSemanticAction();
}
/**
* Set the page this node came from.
*
* @param page The page that supplied this node.
*/
public void setPage(Page page) {
delegate.setPage(page);
}
