public void testLogicalRangesOfUnclosedOpenTags() throws ParseException {
HtmlParseResult result =
parse(
"<!DOCTYPE html>"
+ "<html>"
+ "<head>"
+ "<title>hello</title>"
+ "</head>"
+ "<body>"
+ "<table>"
+ "</html>");
Node root = result.root();
// NodeUtils.dumpTree(root);
assertNotNull(root);
OpenTag htmlOpen = ElementUtils.query(root, "html");
assertNotNull(htmlOpen);
CloseTag htmlEnd = htmlOpen.matchingCloseTag();
assertNotNull(htmlEnd);
assertNotNull(ElementUtils.query(root, "html/head"));
assertNotNull(ElementUtils.query(root, "html/head/title"));
OpenTag body = ElementUtils.query(root, "html/body");
assertNotNull(body);
OpenTag table = ElementUtils.query(root, "html/body/table");
assertNotNull(table);
// both body and table should be logically closed at the beginning of the html end tag
assertEquals(htmlEnd.from(), body.semanticEnd());
assertEquals(htmlEnd.from(), table.semanticEnd());
}
public void testDivLogicalEndAtTheEOF() throws ParseException {
String code = "<!doctype html><div><div></div>";
// 0123456789012345678901234567890123456789
// ^
// NodeTreeBuilder.DEBUG = true;
HtmlParseResult result = parse(code);
Node root = result.root();
// NodeUtils.dumpTree(root);
// the 't' node is foster parented, so it goes to the table's parent, not table itself
OpenTag div = ElementUtils.query(root, "html/body/div");
assertNotNull(div);
assertEquals(30, div.semanticEnd());
code = "<!doctype html><div><div></</div>";
// 0123456789012345678901234567890123456789
// ^
// NodeTreeBuilder.DEBUG = true;
result = parse(code);
root = result.root();
// NodeUtils.dumpTree(root);
// the 't' node is foster parented, so it goes to the table's parent, not table itself
div = ElementUtils.query(root, "html/body/div");
assertNotNull(div);
assertEquals(32, div.semanticEnd());
code = "<!doctype html><div></";
// 0123456789012345678901234567890123456789
// ^
// NodeTreeBuilder.DEBUG = true;
result = parse(code);
root = result.root();
// NodeUtils.dumpTree(root);
// the 't' node is foster parented, so it goes to the table's parent, not table itself
div = ElementUtils.query(root, "html/body/div");
assertNotNull(div);
assertEquals(21, div.semanticEnd());
}
public void testSimpleDocument() throws ParseException {
String code =
"<!doctype html><html><head><title>x</title></head><body><div onclick=\"alert();\"/></body></html>";
// 012345678901234567890123456789012345678901234567890123456789012345678 901234567
// 8901234567890123456789
// 0 1 2 3 4 5 6 7
// 8 9
// NodeTreeBuilder.DEBUG_STATES = true;
HtmlParseResult result = parse(code);
Node root = result.root();
assertNotNull(root);
// NodeUtils.dumpTree(result.root());
OpenTag html = ElementUtils.query(root, "html");
assertEquals("html", html.name());
assertEquals(15, html.from());
assertEquals(21, html.to());
assertEquals(15, html.from());
assertEquals(95, html.semanticEnd());
OpenTag body = ElementUtils.query(root, "html/body");
assertEquals("body", body.name());
assertEquals(50, body.from());
assertEquals(56, body.to());
assertEquals(50, body.from());
assertEquals(88, body.semanticEnd());
CloseTag bodyEndTag = body.matchingCloseTag();
assertNotNull(bodyEndTag);
assertSame(body, bodyEndTag.matchingOpenTag());
assertSame(bodyEndTag, body.matchingCloseTag());
OpenTag title = ElementUtils.query(root, "html/head/title");
assertEquals("title", title.name());
assertEquals(27, title.from());
assertEquals(34, title.to());
assertEquals(27, title.from());
assertEquals(43, title.semanticEnd());
CloseTag titleEndTag = title.matchingCloseTag();
assertNotNull(titleEndTag);
assertSame(title, titleEndTag.matchingOpenTag());
assertSame(titleEndTag, title.matchingCloseTag());
}