CDATA parsing is not correct when it is not in xmlMode

[yukinying]

It looks like the parsing code is considering CDATA block as comment block when xmlMode is not set with

this.oncomment("[CDATA[" + value + "]]");

(as in 1123da8).

This logic appears to be incorrect as "HTML bogus comment" blocks would end whenever the first > is seen. Thus for value like > <div>xxx</div>, the parsing logic will be incorrect.

[fb55]

In the context of a bogus comment, this is the right behavior. You can separately enable CDATA parsing, have a look at the options page in the wiki.

[yukinying]

I have verified with Gumbo and Parse5 such that the correct behavior for parsing input like <![CDATA[ > <div> xxx </div>]]> would be <![CDATA[ > (bogus comment) <div> (tag), xxx (text), </div> (tag) and ]]> (text).

In htmlparser2 it is parsed as <![CDATA[ > <div> xxx </div>]]> (bogus comment).

[fb55]

Ahh, right. I forgot it behaved this way. This will be fixed with #114.

[yukinying]

Thanks!