WRAP_AUTO_LINKS defaults to false, Markdown loses a potential useful link

[gsora]

Describe the bug

Rendering HTML to Markdown using FlexmarkHtmlParser.parse() ignores anchors in which the href value equals the wrapped value, thus ignoring a potential useful link being generated.

• Parser
• HtmlRenderer
• Formatter
• FlexmarkHtmlParser
• DocxRenderer
• PdfConverterExtension
• extension(s)

To Reproduce

This sample codes converts HTML to Markdown using the example approach detailed here:

String originalText = "<p>This doesn't get rendered: <a href=\"https://google.com\" rel=\"nofollow\">https://google.com</a> <a href=\"https://google.com\" rel=\"nofollow\">This does.</a></p> \n";
String output = FlexmarkHtmlParser.parse(originalText);

Expected behavior: the first anchor should be rendered to a Markdown link.

Resulting Output: it just gets transformed to text.

Additional context: to fix this issue, the following approach has been employed:

String originalText = "<p>This doesn't get rendered: <a href=\"https://google.com\" rel=\"nofollow\">https://google.com</a> <a href=\"https://google.com\" rel=\"nofollow\">This does.</a></p> \n";
String output = FlexmarkHtmlParser.parse(
                originalText,
                1,
                new MutableDataSet()
                        .set(FlexmarkHtmlParser.WRAP_AUTO_LINKS, true)
        );

Setting FlexmarkHtmlParser.WRAP_AUTO_LINKS to true renders the anchor successfully.

[vsch]

@gsora, WRAP_AUTO_LINKS controls whether links which look like auto-links, ie. the URL and link text match, get wrapped in <> or are left bare.

The bare links will be correctly interpreted as an auto-link when using the flexmark-ext-autolink extension.

I think it is a good idea to change the default for WRAP_AUTO_LINKS option to true but I will do this for version 0.50 which has many backward compatibility breaking changes. I do not want to introduce these breaking changes into the master branch.

[vsch]

@gsora, you can also remove auto-link conversion so all links convert to plain links by setting EXTRACT_AUTO_LINKS to false.

[gsora]

Thanks for your reply.

The bare links will be correctly interpreted as an auto-link when using the flexmark-ext-autolink extension.

What's the recommended way to use flexmark-ext-autolink with FlexmarkHtmlParser?

[vsch]

@gsora, the auto-link extension is a parser extension.

It has a post-processor which scans all text elements in the AST after parsing looking for anything that looks like a valid URI.

It is actually slow for large files (2x to 3x slower parsing) so if you do not need unwrapped auto-links then I would recommend not using it but instead set WRAP_AUTO_LINKS to true for HTML to markdown conversion.

If you do want plain auto-links recognized in markdown files then you should include the AutolinkExtension.create() in your parser options EXTENSIONS list option.