better filtering of discovered links

[willnorris]

• links to itself (typically the permalink of the post) can be excluded
• category and tag links should probably be excluded
• add option to exclude all links for the same domain
• others?

[nekr0z]

In static-webmentions (that uses this library under the hood) I am filtering out the links to the page itself, and also give a user two filter options:

• to specify a list of URLs they don't want to be webmention targets (relative, absolute, and schemes are accepted — you wouldn't want to send webmentions to URLs starting with mailto://, would you?);
• to specify a list of CSS selectors on their pages they want to exclude from target discovery (together with all the children of the said selectors), which is useful for people that display received webmentions inside their h-entries and want to avoid webmention loops.

These suit my own needs just fine, but I heard from at least one user that he extensively uses the option to save the targets and the sources for the webmentions that are planned to be sent to a JSON file (and later send them with a separate command) and uses jq with complex regexps to filter out the webmentions he doesn't want to send. I guess there just isn't a way to satisfy everybody. ;)