#1046 Tag balancing differences in HtmlUnit and real browsers

[Joerg Werner]

In case of incomplete/unbalanced HTML source code, browsers like Firefox and Chrome might create different DOM trees depending on the document type. For example,

"<p><table></table>"

is balanced to

"<p><table></table></p>"

in Quirks mode and to

"<p></p><table></table>"

in Standard mode. HtmlUnit seems to apply the Quirks balancing strategy in Standard mode as well. This is not a big deal except that XPaths copied from one of these browsers might or might not work in HtmlUnit without modification.

[Joerg Werner]

test case

[Joerg Werner]

updated test case

[Joerg Werner]

There are more examples, especially if the HTML code is not valid (with respect to the specs). For example, the following snippet (btw. generated by a GUI framework):

<a><table>...</table></a>

was rebalanced to:

<a></a><table>...</table>

FF leaves this as is. So when clicking somewhere inside the table:

• the XPath needed is different across browsers (causes problems when Selenium is used with different drivers) and
• the anchor will not be triggered when using HtmlUnit.

I extended the test case to cover this issue as well.

[RBRi]

• Description has changed:

Diff:

--- old
+++ new
@@ -1 +1,19 @@
-In case of incomplete/unbalanced HTML source code, browsers like Firefox and Chrome might create different DOM trees depending on the document type. For example, "<p><table></table>" is balanced to "<p><table></table></p>" in Quirks mode and to "<p></p><table></table>" in Standard mode. HtmlUnit seems to apply the Quirks balancing strategy in Standard mode as well. This is not a big deal except that XPaths copied from one of these browsers might or might not work in HtmlUnit without modification.
+In case of incomplete/unbalanced HTML source code, browsers like Firefox and Chrome might create different DOM trees depending on the document type. For example,
+
+~~~~~~
+"<p><table></table>"
+~~~~~~
+
+is balanced to
+
+~~~~~~
+"<p><table></table></p>"
+~~~~~~
+
+in Quirks mode and to
+
+~~~~~~
+"<p></p><table></table>"
+~~~~~~
+
+in Standard mode. HtmlUnit seems to apply the Quirks balancing strategy in Standard mode as well. This is not a big deal except that XPaths copied from one of these browsers might or might not work in HtmlUnit without modification.

[Mark H. Butler]

I am seeing this bug as well. It may be a duplicate of https://sourceforge.net/p/htmlunit/bugs/1423/

[RBRi]

Ok, this seems to be fixed at least with our own neko fork.
Tescase added.