[feature] Allow anchor href to work for footnotes, use ID prefix to avoid clashes (#4298)

Updates markdown parser + sanitizer to allow footnote anchors to work properly, with appropriate roles. Footnote anchor IDs and backrefs use the status ID as a prefix to avoid clashes, so that footnotes don't break when multiple footnoted statuses are rendered on the same page (eg., in a thread or on the account's home page).

closes https://codeberg.org/superseriousbusiness/gotosocial/issues/4296

Reviewed-on: https://codeberg.org/superseriousbusiness/gotosocial/pulls/4298
Co-authored-by: tobi <tobi.smethurst@protonmail.com>
Co-committed-by: tobi <tobi.smethurst@protonmail.com>

internal/text/markdown.go

@@ -24,6 +24,7 @@ import (
 	"strings"
 
 	"code.superseriousbusiness.org/gotosocial/internal/gtsmodel"
+	"code.superseriousbusiness.org/gotosocial/internal/id"
 	"code.superseriousbusiness.org/gotosocial/internal/log"
 	"code.superseriousbusiness.org/gotosocial/internal/regexes"
 	"codeberg.org/gruf/go-byteutil"
@@ -118,6 +119,18 @@ func (f *Formatter) fromMarkdown(
 		}
 	}
 
+	// Inject a footnote ID prefix to avoid
+	// footnote ID clashes. StatusID isn't
+	// always set (eg., when parsing instance
+	// description markdown), so take a random
+	// ULID if it's not.
+	var footnoteIDPrefix string
+	if statusID != "" {
+		footnoteIDPrefix = statusID + "-"
+	} else {
+		footnoteIDPrefix = id.NewULID() + "-"
+	}
+
 	// Instantiate goldmark parser for
 	// markdown, using custom renderer
 	// to add hashtag/mention links.
@@ -141,7 +154,9 @@ func (f *Formatter) fromMarkdown(
 			extension.NewLinkify(
 				extension.WithLinkifyURLRegexp(regexes.URLLike),
 			),
-			extension.Footnote,
+			extension.NewFootnote(
+				extension.WithFootnoteIDPrefix(footnoteIDPrefix),
+			),
 			extension.Strikethrough,
 		),
 	)

internal/text/markdown_test.go

@@ -61,7 +61,9 @@ const (
 	mdCodeBlockWithNewlines            = "some code coming up\n\n```\n\n\n\n```\nthat was some code"
 	mdCodeBlockWithNewlinesExpected    = "<p>some code coming up</p><pre><code>\n\n\n</code></pre><p>that was some code</p>"
 	mdWithFootnote                     = "fox mulder,fbi.[^1]\n\n[^1]: federated bureau of investigation"
-	mdWithFootnoteExpected             = "<p>fox mulder,fbi.<sup id=\"fnref:1\"><a class=\"footnote-ref\">1</a></sup></p><div><hr><ol><li id=\"fn:1\"><p>federated bureau of investigation <a class=\"footnote-backref\">↩︎</a></p></li></ol></div>"
+	mdWithFootnoteExpected             = "<p>fox mulder,fbi.<sup id=\"dummy_status_ID-fnref:1\"><a href=\"#dummy_status_ID-fn:1\" class=\"footnote-ref\" role=\"doc-noteref\">1</a></sup></p><div><hr><ol><li id=\"dummy_status_ID-fn:1\"><p>federated bureau of investigation <a href=\"#dummy_status_ID-fnref:1\" class=\"footnote-backref\" role=\"doc-backlink\">↩︎</a></p></li></ol></div>"
+	mdWithAttemptedRelative            = "hello this is a cheeky relative link: <a href=\"../sneaky.html\">click it!</a>"
+	mdWithAttemptedRelativeExpected    = "<p>hello this is a cheeky relative link: click it!</p>"
 	mdWithBlockQuote                   = "get ready, there's a block quote coming:\n\n>line1\n>line2\n>\n>line3\n\n"
 	mdWithBlockQuoteExpected           = "<p>get ready, there's a block quote coming:</p><blockquote><p>line1<br>line2</p><p>line3</p></blockquote>"
 	mdHashtagAndCodeBlock              = "#Hashtag\n\n```\n#Hashtag\n```"
@@ -156,6 +158,11 @@ func (suite *MarkdownTestSuite) TestParseWithFootnote() {
 	suite.Equal(mdWithFootnoteExpected, formatted.HTML)
 }
 
+func (suite *MarkdownTestSuite) TestParseWithAttemptedRelative() {
+	formatted := suite.FromMarkdown(mdWithAttemptedRelative)
+	suite.Equal(mdWithAttemptedRelativeExpected, formatted.HTML)
+}
+
 func (suite *MarkdownTestSuite) TestParseWithBlockquote() {
 	formatted := suite.FromMarkdown(mdWithBlockQuote)
 	suite.Equal(mdWithBlockQuoteExpected, formatted.HTML)

internal/text/sanitize.go

@@ -124,17 +124,28 @@ var regular *bluemonday.Policy = func() *bluemonday.Policy {
 	*/
 
 	// Permit hyperlinks.
-	p.AllowAttrs("class", "href", "rel").OnElements("a")
+	p.AllowAttrs("class", "rel").OnElements("a")
+
+	// Permit footnote roles on anchor elements.
+	p.AllowAttrs("role").Matching(regexp.MustCompile("^doc-noteref$")).OnElements("a")
+	p.AllowAttrs("role").Matching(regexp.MustCompile("^doc-backlink$")).OnElements("a")
 
 	// URLs must be parseable by net/url.Parse().
 	p.RequireParseableURLs(true)
 
-	// Most common URL schemes only.
+	// Relative URLs are OK as we
+	// need fragments for footnotes.
+	p.AllowRelativeURLs(true)
+
+	// However *only* allow common schemes, and also
+	// relative URLs beginning with "#", ie., fragments.
+	// We don't want URL's like "../../peepee.html".
 	p.AllowURLSchemes("mailto", "http", "https")
+	p.AllowAttrs("href").Matching(regexp.MustCompile("^(?:#|mailto|https://|http://).+$")).OnElements("a")
 
 	// Force rel="noreferrer".
 	// See: https://developer.mozilla.org/en-US/docs/Web/HTML/Attributes/rel/noreferrer
-	p.RequireNoReferrerOnLinks(true)
+	p.RequireNoReferrerOnFullyQualifiedLinks(true)
 
 	// Add rel="nofollow" on all fully qualified (not relative) links.
 	// See: https://developer.mozilla.org/en-US/docs/Web/HTML/Attributes/rel#nofollow