RFR: JDK-8276964: Better indicate a snippet that could not be processed [v2]

[Jonathan Gibbons]

On Mon, 29 Nov 2021 13:55:39 GMT, Hannes Wallnöfer <hannesw at openjdk.org> wrote:

>> The primary purpose of this change is to make it easier to spot invalid snippets in generated documentation. 
>> 
>> This adds a new `Content invalidTagOutput(String summary, Optional<String> detail)` method to the `TagletWriter` class that returns HTML to display the summary and optionally a detail message. The method is only used for snippet tags for now, but is generic and could be used for other invalid tags in the future. 
>> 
>> If the `detail` argument is empty or contains a blank string, a `<span>` element is returned containing the `summary` argument. If a detail argument is provided, a HTML5 `<details>` element is returned containing a `<summary>` element with the `summary` argument and a `<pre>` element containing the `details` argument. In both cases the returned element is styled with a thin border and a light red background.
>> 
>> In its current use the `detail` argument is always provided by the message of the `ParseException` or `BadSnippetException` that was thrown and caught. 
>> 
>> Example output is available here: http://cr.openjdk.java.net/~hannesw/8276964/api.01/snippet_errormessages/A.html
>> 
>> I added output checks to some but not all of the negative tests. In addition I slightly reformatted `TestSnippetTag.java` to add indentation to some previously unindeted text blocks.
>
> Hannes Wallnöfer has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Normalize newlines on detail message

Generally great.

Two questions, bordering on suggestions ....

1. Consider allowing the `detail` to be `Optional<Content>` so there is more flexibility in how the detail is presented.  You can provide and use convenience methods to wrap plain text or preformatted text into `Content`.

2. While I generally prefer simplifying and removing stuff from `HtmlDocletWriter`, I wonder if this is a case where the method would be better there than `TagletWriter[Impl]`. The specific use case I have in mind is the presentation of `ErroneousTree`.  This is currently handled in `HtmlDocletWriter.commentTagsToContent`, line 1621, and the erroneous text is simply presented as regular `Text`. (line 1631).  At a push `HtmlDocletWriter` could create and use a `TagletWriter` but then the overtones of "invalid tag" are a bit "smelly".

src/jdk.javadoc/share/classes/jdk/javadoc/internal/doclets/toolkit/taglets/TagletWriter.java line 247:

> 245:      *
> 246:      * @param summary the single-line summary message
> 247:      * @param detail the optional detail message which may contain preformatted text

This would be more flexible if it used `Content` instead of `Text`

-------------

PR: https://git.openjdk.java.net/jdk/pull/6579