Inconsistent handling of U+0000 NULLs in input stream #2395

A solitary U+0000 NULL within a <p> tag is removed when parsing. However, that only happens if there is no other text; with other text, the U+0000 is preserved.

This code illustrates the issue, using v1.21.2:

public static void main(String[] args) {
    Document doc = Parser.parse("<p>\u0000</p><p>\u0000\u0000</p><p>Hi\u0000</p><p>\u0000<b>\u0000</b></p>", "");
    System.out.println(doc.child(0).child(1));
}

which has output:

<body>
 <p></p>
 <p>&#x0;&#x0;</p>
 <p>Hi&#x0;</p>
 <p><b></b></p>
</body>

Removing the tags is (I believe) correct, section 13.2.6.4.7 The "in body" insertion mode of the spec says that U+0000 NULL should be ignored.

TokeniserState.Date has a switch statement with a case for U+0000 which adds an error, but the text is appended to the token, and HtmlTreeBuilderState.InBody (line 289), checks whether the token equals a singleton U+0000.

See also
https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inbody