Please review a relatively simple upgrade to the DocCommentParser for
handling standalone HTML files.
In a standalone file, HTML content is treated as being in 3 parts ...
the preamble, the body, and the postamble, where the body is the
content of an equivalent doc comment. Traditionally, the preamble ends
at the end of the opening tag for the `body` element, and the body ends
at the beginning of the start of the closing tag for the `body` element.
In other words, the body has traditionally been the inner HTML of the
enclosing `body` element.
Since then, a style has evolved where authors are wrapping the content
in a `main` element as well, presumably to satisfy accessibility
checkers (which is good). But this conflicts with the traditional
determination of the content of the file, because (amongst other
reasons) there can only be one `main` element in a generated file.
The change is for the preamble to also include the opening tag of a
`main` element if it immediately follows the opening tag of the `body`
element (allowing for inter-element whitespace. The change is also for
the body to stop at the closing tag of a `main` element if one is
encountered.
Two test cases are added to an existing test. The test cases consist of
HTML files containing `main` elements as well as `body` elements. The
corresponding .out files are dumps of the doc comment tree, showing the
content of the preamble, body and postamble.
The change was also testing by building JDK API docs, running doccheck,
and reviewing the results, to confirm that affected files which
previously contained errors no longer do so.
-- Jon
JBS: https://bugs.openjdk.java.net/browse/JDK-8223805
Webrev:
file:///w/jjg/work/jdk.closed.dev/8223805/webrev.00/webrev/index.html
