markdown: add footnote support

Author: rsc

footnote.go

@@ -0,0 +1,188 @@
+// Copyright 2024 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package markdown
+
+import (
+	"strconv"
+	"strings"
+)
+
+type Footnote struct {
+	Position
+	Label  string
+	Blocks []Block
+}
+
+type FootnoteLink struct {
+	Label    string
+	Footnote *Footnote
+}
+
+type printedNote struct {
+	num  string
+	note *Footnote
+	refs []string
+}
+
+func (*FootnoteLink) Inline() {}
+
+func (x *Footnote) printed(p *printer) *printedNote {
+	if p.footnotes == nil {
+		p.footnotes = make(map[*Footnote]*printedNote)
+	}
+	pr, ok := p.footnotes[x]
+	if !ok {
+		pr = &printedNote{
+			num:  strconv.Itoa(len(p.footnotes) + 1),
+			note: x,
+		}
+		p.footnotes[x] = pr
+		p.footnotelist = append(p.footnotelist, pr)
+	}
+	ref := pr.num
+	if len(pr.refs) > 0 {
+		ref += "-" + strconv.Itoa(len(pr.refs)+1)
+	}
+	pr.refs = append(pr.refs, ref)
+	return pr
+}
+
+func (x *FootnoteLink) printHTML(p *printer) {
+	note := x.Footnote
+	if note == nil {
+		return
+	}
+	pr := note.printed(p)
+	ref := pr.refs[len(pr.refs)-1]
+	p.html(`<sup class="fn"><a id="fnref-`, ref, `" href="#fn-`, pr.num, `">`, pr.num, `</a></sup>`)
+}
+
+func (x *FootnoteLink) printMarkdown(p *printer) {
+	note := x.Footnote
+	if note == nil {
+		return
+	}
+	note.printed(p) // add to list for printFootnoteMarkdown
+	p.text(`[^`, x.Label, `]`)
+}
+
+func (x *FootnoteLink) printText(p *printer) {
+	p.text(`[^`, x.Label, `]`)
+}
+
+func printFootnoteHTML(p *printer) {
+	if len(p.footnotelist) == 0 {
+		return
+	}
+
+	p.html(`<div class="footnotes">Footnotes</div>`, "\n")
+	p.html("<ol>\n")
+	for num, note := range p.footnotelist {
+		num++
+		str := strconv.Itoa(num)
+		p.html(`<li id="fn-`, str, `">`, "\n")
+		for _, b := range note.note.Blocks {
+			b.printHTML(p)
+		}
+		if !p.eraseCloseP() {
+			p.html("<p>\n")
+		}
+		for _, ref := range note.refs {
+			p.html("\n", `<a class="fnref" href="#fnref-`, ref, `">↩</a>`)
+		}
+		p.html("</p>\n")
+		p.html("</li>\n")
+	}
+	p.html("</ol>\n")
+}
+
+func (x *Footnote) printMarkdown(p *printer) {
+	p.md(`[^`, x.Label, `]: `)
+	defer p.pop(p.push("  "))
+	printMarkdownBlocks(x.Blocks, p)
+}
+
+func printFootnoteMarkdown(p *printer) {
+	if len(p.footnotelist) == 0 {
+		return
+	}
+
+	p.maybeNL()
+	for _, note := range p.footnotelist {
+		p.nl()
+		note.note.printMarkdown(p)
+	}
+}
+
+func parseFootnoteRef(p *parser, s string, start int) (x Inline, end int, ok bool) {
+	if !p.Footnote || start+1 >= len(s) || s[start+1] != '^' {
+		return
+	}
+	end = strings.Index(s[start:], "]")
+	if end < 0 {
+		return
+	}
+	end += start + 1
+	label := s[start+2 : end-1]
+	note, ok := p.footnotes[normalizeLabel(label)]
+	if !ok {
+		return
+	}
+	return &FootnoteLink{label, note}, end, true
+}
+
+func startFootnote(p *parser, s line) (line, bool) {
+	t := s
+	t.trimSpace(0, 3, false)
+	if !t.trim('[') || !t.trim('^') {
+		return s, false
+	}
+	label := t.string()
+	i := strings.Index(label, "]")
+	if i < 0 || i+1 >= len(label) && label[i+1] != ':' {
+		return s, false
+	}
+	label = label[:i]
+	for j := 0; j < i; j++ {
+		c := label[j]
+		if c == ' ' || c == '\r' || c == '\n' || c == 0x00 || c == '\t' {
+			return s, false
+		}
+	}
+	t.skip(i + 2)
+
+	if _, ok := p.footnotes[normalizeLabel(label)]; ok {
+		// Already have a footnote with this label.
+		// cmark-gfm ignores all future references,
+		// dropping them from the document,
+		// but it seems more helpful to not treat it
+		// as a footnote.
+		p.corner = true
+		return s, false
+	}
+
+	fb := &footnoteBuilder{label}
+	p.addBlock(fb)
+	return t, true
+}
+
+type footnoteBuilder struct {
+	label string
+}
+
+func (b *footnoteBuilder) extend(p *parser, s line) (line, bool) {
+	if !s.trimSpace(4, 4, true) {
+		return s, false
+	}
+	return s, true
+}
+
+func (b *footnoteBuilder) build(p *parser) Block {
+	if p.footnotes == nil {
+		p.footnotes = make(map[string]*Footnote)
+	}
+	p.footnotes[normalizeLabel(b.label)] = &Footnote{p.pos(), b.label, p.blocks()}
+	return &Empty{}
+}

line.go

@@ -100,6 +100,13 @@ func (s *line) trim(c byte) bool {
 	return false
 }
 
+func (s *line) skip(n int) {
+	s.i += n
+	if s.nonblank < s.i {
+		s.setNonblank()
+	}
+}
+
 func (s *line) string() string {
 	switch s.spaces {
 	case 0:

link.go

@@ -136,6 +136,11 @@ func (x *Image) printText(p *printer) {
 // parseLinkOpen is an [inlineParser] for a link open [.
 // The caller has checked that s[start] == '['.
 func parseLinkOpen(p *parser, s string, start int) (x Inline, end int, ok bool) {
+	if p.Footnote {
+		if x, end, ok := parseFootnoteRef(p, s, start); ok {
+			return x, end, ok
+		}
+	}
 	return &openPlain{Plain{s[start : start+1]}, start + 1}, start + 1, true
 }
 

parse.go

@@ -93,6 +93,8 @@ type parser struct {
 	// texts to apply inline processing to
 	texts []textRaw
 
+	footnotes map[string]*Footnote
+
 	// inline parsing
 	s       string
 	emitted int // s[:emitted] has been emitted into list
@@ -343,4 +345,5 @@ var starters = []starter{
 	startThematicBreak,
 	startListItem,
 	startHTMLBlock,
+	startFootnote,
 }

print.go

@@ -20,6 +20,8 @@ type printer struct {
 	prefixOlder []byte
 	trimLimit   int
 	listOut
+	footnotes    map[*Footnote]*printedNote
+	footnotelist []*printedNote
 }
 
 type listOut struct {
@@ -83,12 +85,14 @@ func ToHTML(b Block) string {
 	var p printer
 	p.writeMode = writeHTML
 	b.printHTML(&p)
+	printFootnoteHTML(&p)
 	return p.buf.String()
 }
 
 func Format(b Block) string {
 	var p printer
 	b.printMarkdown(&p)
+	printFootnoteMarkdown(&p)
 	// TODO footnotes?
 	return p.buf.String()
 }

testdata/footnote.txt

@@ -0,0 +1,52 @@
+-- parser.json --
+{"Footnote": true}
+-- 1.md --
+Here is a simple footnote[^1][^4].
+
+A footnote can also[^3] have multiple lines[^4].
+
+[^1]: My reference.
+[^4]: To add line breaks within a footnote, prefix new lines with 2 spaces.
+  This is a second line.
+-- 1.html --
+<p>Here is a simple footnote<sup class="fn"><a id="fnref-1" href="#fn-1">1</a></sup><sup class="fn"><a id="fnref-2" href="#fn-2">2</a></sup>.</p>
+<p>A footnote can also[^3] have multiple lines<sup class="fn"><a id="fnref-2-2" href="#fn-2">2</a></sup>.</p>
+<div class="footnotes">Footnotes</div>
+<ol>
+<li id="fn-1">
+<p>My reference.
+<a class="fnref" href="#fnref-1">↩</a></p>
+</li>
+<li id="fn-2">
+<p>To add line breaks within a footnote, prefix new lines with 2 spaces.
+This is a second line.
+<a class="fnref" href="#fnref-2">↩</a>
+<a class="fnref" href="#fnref-2-2">↩</a></p>
+</li>
+</ol>
+-- 2.md --
+Footnote[^abc].
+
+[^aBc]: Hi.
+-- 2.html --
+<p>Footnote<sup class="fn"><a id="fnref-1" href="#fn-1">1</a></sup>.</p>
+<div class="footnotes">Footnotes</div>
+<ol>
+<li id="fn-1">
+<p>Hi.
+<a class="fnref" href="#fnref-1">↩</a></p>
+</li>
+</ol>
+-- 3.md --
+Footnote[^aBc].
+
+[^abC]: Hi.
+-- 3.html --
+<p>Footnote<sup class="fn"><a id="fnref-1" href="#fn-1">1</a></sup>.</p>
+<div class="footnotes">Footnotes</div>
+<ol>
+<li id="fn-1">
+<p>Hi.
+<a class="fnref" href="#fnref-1">↩</a></p>
+</li>
+</ol>