Merge pull request #487 from n-peugnet/patch-1

Fix GitHub actions badge URL

README.md

@@ -2,7 +2,7 @@ goldmark
 ==========================================
 
 [![https://pkg.go.dev/github.com/yuin/goldmark](https://pkg.go.dev/badge/github.com/yuin/goldmark.svg)](https://pkg.go.dev/github.com/yuin/goldmark)
-[![https://github.com/yuin/goldmark/actions?query=workflow:test](https://github.com/yuin/goldmark/workflows/test/badge.svg?branch=master&event=push)](https://github.com/yuin/goldmark/actions?query=workflow:test)
+[![https://github.com/yuin/goldmark/actions?query=workflow:test](https://github.com/yuin/goldmark/actions/workflows/test.yaml/badge.svg?branch=master&event=push)](https://github.com/yuin/goldmark/actions?query=workflow:test)
 [![https://coveralls.io/github/yuin/goldmark](https://coveralls.io/repos/github/yuin/goldmark/badge.svg?branch=master)](https://coveralls.io/github/yuin/goldmark)
 [![https://goreportcard.com/report/github.com/yuin/goldmark](https://goreportcard.com/badge/github.com/yuin/goldmark)](https://goreportcard.com/report/github.com/yuin/goldmark)
 
@@ -493,7 +493,7 @@ Extensions
 - [goldmark-d2](https://github.com/FurqanSoftware/goldmark-d2): Adds support for [D2](https://d2lang.com/) diagrams.
 - [goldmark-katex](https://github.com/FurqanSoftware/goldmark-katex): Adds support for [KaTeX](https://katex.org/) math and equations.
 - [goldmark-img64](https://github.com/tenkoh/goldmark-img64): Adds support for embedding images into the document as DataURL (base64 encoded).
-- [goldmark-enclave](https://github.com/quail-ink/goldmark-enclave): Adds support for embedding youtube/bilibili video, X's [oembed tweet](https://publish.twitter.com/), [tradingview](https://www.tradingview.com/widget/)'s chart, [quail](https://quail.ink)'s widget into the document.
+- [goldmark-enclave](https://github.com/quailyquaily/goldmark-enclave): Adds support for embedding youtube/bilibili video, X's [oembed X](https://publish.x.com/), [tradingview chart](https://www.tradingview.com/widget/)'s chart, [quaily widget](https://quaily.com), [spotify embeds](https://developer.spotify.com/documentation/embeds), [dify embed](https://dify.ai/) and html audio into the document.
 - [goldmark-wiki-table](https://github.com/movsb/goldmark-wiki-table): Adds support for embedding Wiki Tables.
 - [goldmark-tgmd](https://github.com/Mad-Pixels/goldmark-tgmd): A Telegram markdown renderer that can be passed to `goldmark.WithRenderer()`.
 

_benchmark/go/go.mod

@@ -6,7 +6,7 @@ require (
 	github.com/88250/lute v1.7.5
 	github.com/gomarkdown/markdown v0.0.0-20230322041520-c84983bdbf2a
 	github.com/russross/blackfriday/v2 v2.1.0
-	github.com/yuin/goldmark v1.2.1
+	github.com/yuin/goldmark v0.0.0
 	gitlab.com/golang-commonmark/markdown v0.0.0-20211110145824-bf3e522c626a
 )
 
@@ -22,3 +22,4 @@ require (
 )
 
 replace gopkg.in/russross/blackfriday.v2 v2.0.1 => github.com/russross/blackfriday/v2 v2.0.1
+replace github.com/yuin/goldmark v0.0.0 => ../../

_test/extra.txt

@@ -385,7 +385,8 @@ a* b c d *e*
 //- - - - - - - - -//
 		x
 //- - - - - - - - -//
-<pre><code>	x</code></pre>
+<pre><code>	x
+</code></pre>
 //= = = = = = = = = = = = = = = = = = = = = = = =//
 
 26: NUL bytes must be replaced with U+FFFD
@@ -787,3 +788,58 @@ text" /></p>
 //- - - - - - - - -//
 <p><img src="https://example.com/img.png" alt="`alt" /></p>
 //= = = = = = = = = = = = = = = = = = = = = = = =//
+
+63: Emphasis in link label
+//- - - - - - - - -//
+[*[a]*](b)
+//- - - - - - - - -//
+<p><a href="b"><em>[a]</em></a></p>
+//= = = = = = = = = = = = = = = = = = = = = = = =//
+
+64: Nested list under an empty list item
+//- - - - - - - - -//
+- 
+  - foo
+//- - - - - - - - -//
+<ul>
+<li>
+<ul>
+<li>foo</li>
+</ul>
+</li>
+</ul>
+//= = = = = = = = = = = = = = = = = = = = = = = =//
+
+65: Nested fenced code block with tab
+//- - - - - - - - -//
+>  ```
+>	0
+>  ```
+//- - - - - - - - -//
+<blockquote>
+<pre><code> 0
+</code></pre>
+</blockquote>
+//= = = = = = = = = = = = = = = = = = = = = = = =//
+
+66: EOF should be rendered as a newline with an unclosed block(w/ TAB)
+//- - - - - - - - -//
+>  ```
+>	0
+//- - - - - - - - -//
+<blockquote>
+<pre><code> 0
+</code></pre>
+</blockquote>
+//= = = = = = = = = = = = = = = = = = = = = = = =//
+
+67: EOF should be rendered as a newline with an unclosed block
+//- - - - - - - - -//
+>  ```
+>   0
+//- - - - - - - - -//
+<blockquote>
+<pre><code> 0
+</code></pre>
+</blockquote>
+//= = = = = = = = = = = = = = = = = = = = = = = =//

ast/ast.go

@@ -123,6 +123,12 @@ type Node interface {
 	Dump(source []byte, level int)
 
 	// Text returns text values of this node.
+	// This method is valid only for some inline nodes.
+	// If this node is a block node, Text returns a text value as reasonable as possible.
+	// Notice that there are no 'correct' text values for the block nodes.
+	// Result for the block nodes may be different from your expectation.
+	//
+	// Deprecated: Use other properties of the node to get the text value(i.e. Pragraph.Lines, Text.Value).
 	Text(source []byte) []byte
 
 	// HasBlankPreviousLines returns true if the row before this node is blank,
@@ -374,11 +380,18 @@ func (n *BaseNode) OwnerDocument() *Document {
 	return nil
 }
 
-// Text implements Node.Text  .
+// Text implements Node.Text .
+//
+// Deprecated: Use other properties of the node to get the text value(i.e. Pragraph.Lines, Text.Value).
 func (n *BaseNode) Text(source []byte) []byte {
 	var buf bytes.Buffer
 	for c := n.firstChild; c != nil; c = c.NextSibling() {
 		buf.Write(c.Text(source))
+		if sb, ok := c.(interface {
+			SoftLineBreak() bool
+		}); ok && sb.SoftLineBreak() {
+			buf.WriteByte('\n')
+		}
 	}
 	return buf.Bytes()
 }

ast/ast_test.go

@@ -5,21 +5,6 @@ import (
 	"testing"
 )
 
-func TestRemoveChildren(t *testing.T) {
-	root := NewDocument()
-
-	node1 := NewDocument()
-
-	node2 := NewDocument()
-
-	root.AppendChild(root, node1)
-	root.AppendChild(root, node2)
-
-	root.RemoveChildren(root)
-
-	t.Logf("%+v", node2.PreviousSibling())
-}
-
 func TestWalk(t *testing.T) {
 	tests := []struct {
 		name   string

ast/block.go

@@ -130,6 +130,13 @@ func (n *TextBlock) Kind() NodeKind {
 	return KindTextBlock
 }
 
+// Text implements Node.Text.
+//
+// Deprecated: Use other properties of the node to get the text value(i.e. TextBlock.Lines).
+func (n *TextBlock) Text(source []byte) []byte {
+	return n.Lines().Value(source)
+}
+
 // NewTextBlock returns a new TextBlock node.
 func NewTextBlock() *TextBlock {
 	return &TextBlock{
@@ -155,6 +162,13 @@ func (n *Paragraph) Kind() NodeKind {
 	return KindParagraph
 }
 
+// Text implements Node.Text.
+//
+// Deprecated: Use other properties of the node to get the text value(i.e. Paragraph.Lines).
+func (n *Paragraph) Text(source []byte) []byte {
+	return n.Lines().Value(source)
+}
+
 // NewParagraph returns a new Paragraph node.
 func NewParagraph() *Paragraph {
 	return &Paragraph{
@@ -249,6 +263,13 @@ func (n *CodeBlock) Kind() NodeKind {
 	return KindCodeBlock
 }
 
+// Text implements Node.Text.
+//
+// Deprecated: Use other properties of the node to get the text value(i.e. CodeBlock.Lines).
+func (n *CodeBlock) Text(source []byte) []byte {
+	return n.Lines().Value(source)
+}
+
 // NewCodeBlock returns a new CodeBlock node.
 func NewCodeBlock() *CodeBlock {
 	return &CodeBlock{
@@ -304,6 +325,13 @@ func (n *FencedCodeBlock) Kind() NodeKind {
 	return KindFencedCodeBlock
 }
 
+// Text implements Node.Text.
+//
+// Deprecated: Use other properties of the node to get the text value(i.e. FencedCodeBlock.Lines).
+func (n *FencedCodeBlock) Text(source []byte) []byte {
+	return n.Lines().Value(source)
+}
+
 // NewFencedCodeBlock return a new FencedCodeBlock node.
 func NewFencedCodeBlock(info *Text) *FencedCodeBlock {
 	return &FencedCodeBlock{
@@ -498,6 +526,17 @@ func (n *HTMLBlock) Kind() NodeKind {
 	return KindHTMLBlock
 }
 
+// Text implements Node.Text.
+//
+// Deprecated: Use other properties of the node to get the text value(i.e. HTMLBlock.Lines).
+func (n *HTMLBlock) Text(source []byte) []byte {
+	ret := n.Lines().Value(source)
+	if n.HasClosure() {
+		ret = append(ret, n.ClosureLine.Value(source)...)
+	}
+	return ret
+}
+
 // NewHTMLBlock returns a new HTMLBlock node.
 func NewHTMLBlock(typ HTMLBlockType) *HTMLBlock {
 	return &HTMLBlock{

ast/inline.go

@@ -143,17 +143,25 @@ func (n *Text) Merge(node Node, source []byte) bool {
 }
 
 // Text implements Node.Text.
+//
+// Deprecated: Use other properties of the node to get the text value(i.e. Text.Value).
 func (n *Text) Text(source []byte) []byte {
 	return n.Segment.Value(source)
 }
 
+// Value returns a value of this node.
+// SoftLineBreaks are not included in the returned value.
+func (n *Text) Value(source []byte) []byte {
+	return n.Segment.Value(source)
+}
+
 // Dump implements Node.Dump.
 func (n *Text) Dump(source []byte, level int) {
 	fs := textFlagsString(n.flags)
 	if len(fs) != 0 {
 		fs = "(" + fs + ")"
 	}
-	fmt.Printf("%sText%s: \"%s\"\n", strings.Repeat("    ", level), fs, strings.TrimRight(string(n.Text(source)), "\n"))
+	fmt.Printf("%sText%s: \"%s\"\n", strings.Repeat("    ", level), fs, strings.TrimRight(string(n.Value(source)), "\n"))
 }
 
 // KindText is a NodeKind of the Text node.
@@ -258,6 +266,8 @@ func (n *String) SetCode(v bool) {
 }
 
 // Text implements Node.Text.
+//
+// Deprecated: Use other properties of the node to get the text value(i.e. String.Value).
 func (n *String) Text(source []byte) []byte {
 	return n.Value
 }
@@ -492,15 +502,22 @@ func (n *AutoLink) URL(source []byte) []byte {
 		ret := make([]byte, 0, len(n.Protocol)+s.Len()+3)
 		ret = append(ret, n.Protocol...)
 		ret = append(ret, ':', '/', '/')
-		ret = append(ret, n.value.Text(source)...)
+		ret = append(ret, n.value.Value(source)...)
 		return ret
 	}
-	return n.value.Text(source)
+	return n.value.Value(source)
 }
 
 // Label returns a label of this node.
 func (n *AutoLink) Label(source []byte) []byte {
-	return n.value.Text(source)
+	return n.value.Value(source)
+}
+
+// Text implements Node.Text.
+//
+// Deprecated: Use other properties of the node to get the text value(i.e. AutoLink.Label).
+func (n *AutoLink) Text(source []byte) []byte {
+	return n.value.Value(source)
 }
 
 // NewAutoLink returns a new AutoLink node.
@@ -541,6 +558,13 @@ func (n *RawHTML) Kind() NodeKind {
 	return KindRawHTML
 }
 
+// Text implements Node.Text.
+//
+// Deprecated: Use other properties of the node to get the text value(i.e. RawHTML.Segments).
+func (n *RawHTML) Text(source []byte) []byte {
+	return n.Segments.Value(source)
+}
+
 // NewRawHTML returns a new RawHTML node.
 func NewRawHTML() *RawHTML {
 	return &RawHTML{

ast_test.go

@@ -0,0 +1,204 @@
+package goldmark_test
+
+import (
+	"bytes"
+	"testing"
+
+	. "github.com/yuin/goldmark"
+	"github.com/yuin/goldmark/testutil"
+	"github.com/yuin/goldmark/text"
+)
+
+func TestASTBlockNodeText(t *testing.T) {
+	var cases = []struct {
+		Name   string
+		Source string
+		T1     string
+		T2     string
+		C      bool
+	}{
+		{
+			Name: "AtxHeading",
+			Source: `# l1
+
+a
+
+# l2`,
+			T1: `l1`,
+			T2: `l2`,
+		},
+		{
+			Name: "SetextHeading",
+			Source: `l1
+l2
+===============
+
+a
+
+l3
+l4
+==============`,
+			T1: `l1
+l2`,
+			T2: `l3
+l4`,
+		},
+		{
+			Name: "CodeBlock",
+			Source: `    l1
+    l2
+
+a
+
+    l3
+	l4`,
+			T1: `l1
+l2
+`,
+			T2: `l3
+l4
+`,
+		},
+		{
+			Name: "FencedCodeBlock",
+			Source: "```" + `
+l1
+l2
+` + "```" + `
+
+a
+
+` + "```" + `
+l3
+l4`,
+			T1: `l1
+l2
+`,
+			T2: `l3
+l4
+`,
+		},
+		{
+			Name: "Blockquote",
+			Source: `> l1
+> l2
+
+a
+
+> l3
+> l4`,
+			T1: `l1
+l2`,
+			T2: `l3
+l4`,
+		},
+		{
+			Name: "List",
+			Source: `- l1
+  l2
+
+a
+
+- l3
+  l4`,
+			T1: `l1
+l2`,
+			T2: `l3
+l4`,
+			C: true,
+		},
+		{
+			Name: "HTMLBlock",
+			Source: `<div>
+l1
+l2
+</div>
+
+a
+
+<div>
+l3
+l4`,
+			T1: `<div>
+l1
+l2
+</div>
+`,
+			T2: `<div>
+l3
+l4`,
+		},
+	}
+
+	for _, cs := range cases {
+		t.Run(cs.Name, func(t *testing.T) {
+			s := []byte(cs.Source)
+			md := New()
+			n := md.Parser().Parse(text.NewReader(s))
+			c1 := n.FirstChild()
+			c2 := c1.NextSibling().NextSibling()
+			if cs.C {
+				c1 = c1.FirstChild()
+				c2 = c2.FirstChild()
+			}
+			if !bytes.Equal(c1.Text(s), []byte(cs.T1)) { // nolint: staticcheck
+
+				t.Errorf("%s unmatch: %s", cs.Name, testutil.DiffPretty(c1.Text(s), []byte(cs.T1))) // nolint: staticcheck
+
+			}
+			if !bytes.Equal(c2.Text(s), []byte(cs.T2)) { // nolint: staticcheck
+
+				t.Errorf("%s(EOF) unmatch: %s", cs.Name, testutil.DiffPretty(c2.Text(s), []byte(cs.T2))) // nolint: staticcheck
+
+			}
+		})
+	}
+
+}
+
+func TestASTInlineNodeText(t *testing.T) {
+	var cases = []struct {
+		Name   string
+		Source string
+		T1     string
+	}{
+		{
+			Name:   "CodeSpan",
+			Source: "`c1`",
+			T1:     `c1`,
+		},
+		{
+			Name:   "Emphasis",
+			Source: `*c1 **c2***`,
+			T1:     `c1 c2`,
+		},
+		{
+			Name:   "Link",
+			Source: `[label](url)`,
+			T1:     `label`,
+		},
+		{
+			Name:   "AutoLink",
+			Source: `<http://url>`,
+			T1:     `http://url`,
+		},
+		{
+			Name:   "RawHTML",
+			Source: `<span>c1</span>`,
+			T1:     `<span>`,
+		},
+	}
+
+	for _, cs := range cases {
+		t.Run(cs.Name, func(t *testing.T) {
+			s := []byte(cs.Source)
+			md := New()
+			n := md.Parser().Parse(text.NewReader(s))
+			c1 := n.FirstChild().FirstChild()
+			if !bytes.Equal(c1.Text(s), []byte(cs.T1)) { // nolint: staticcheck
+				t.Errorf("%s unmatch:\n%s", cs.Name, testutil.DiffPretty(c1.Text(s), []byte(cs.T1))) // nolint: staticcheck
+			}
+		})
+	}
+
+}

extension/_test/definition_list.txt

@@ -150,7 +150,8 @@ on two lines.</p>
 //- - - - - - - - -//
 <dl>
 <dt>0</dt>
-<dd><pre><code>	0</code></pre>
+<dd><pre><code>	0
+</code></pre>
 </dd>
 </dl>
 //= = = = = = = = = = = = = = = = = = = = = = = =//

extension/ast_test.go

@@ -0,0 +1,123 @@
+package extension
+
+import (
+	"bytes"
+	"testing"
+
+	"github.com/yuin/goldmark"
+	"github.com/yuin/goldmark/renderer/html"
+	"github.com/yuin/goldmark/testutil"
+	"github.com/yuin/goldmark/text"
+)
+
+func TestASTBlockNodeText(t *testing.T) {
+	var cases = []struct {
+		Name   string
+		Source string
+		T1     string
+		T2     string
+		C      bool
+	}{
+		{
+			Name: "DefinitionList",
+			Source: `c1
+:   c2
+    c3
+
+a
+
+c4
+:   c5
+    c6`,
+			T1: `c1c2
+c3`,
+			T2: `c4c5
+c6`,
+		},
+		{
+			Name: "Table",
+			Source: `| h1 | h2 |
+| -- | -- |
+| c1 | c2 |
+
+a
+
+
+| h3 | h4 |
+| -- | -- |
+| c3 | c4 |`,
+
+			T1: `h1h2c1c2`,
+			T2: `h3h4c3c4`,
+		},
+	}
+
+	for _, cs := range cases {
+		t.Run(cs.Name, func(t *testing.T) {
+			s := []byte(cs.Source)
+			md := goldmark.New(
+				goldmark.WithRendererOptions(
+					html.WithUnsafe(),
+				),
+				goldmark.WithExtensions(
+					DefinitionList,
+					Table,
+				),
+			)
+			n := md.Parser().Parse(text.NewReader(s))
+			c1 := n.FirstChild()
+			c2 := c1.NextSibling().NextSibling()
+			if cs.C {
+				c1 = c1.FirstChild()
+				c2 = c2.FirstChild()
+			}
+			if !bytes.Equal(c1.Text(s), []byte(cs.T1)) { // nolint: staticcheck
+
+				t.Errorf("%s unmatch:\n%s", cs.Name, testutil.DiffPretty(c1.Text(s), []byte(cs.T1))) // nolint: staticcheck
+
+			}
+			if !bytes.Equal(c2.Text(s), []byte(cs.T2)) { // nolint: staticcheck
+
+				t.Errorf("%s(EOF) unmatch: %s", cs.Name, testutil.DiffPretty(c2.Text(s), []byte(cs.T2))) // nolint: staticcheck
+
+			}
+		})
+	}
+
+}
+
+func TestASTInlineNodeText(t *testing.T) {
+	var cases = []struct {
+		Name   string
+		Source string
+		T1     string
+	}{
+		{
+			Name:   "Strikethrough",
+			Source: `~c1 *c2*~`,
+			T1:     `c1 c2`,
+		},
+	}
+
+	for _, cs := range cases {
+		t.Run(cs.Name, func(t *testing.T) {
+			s := []byte(cs.Source)
+			md := goldmark.New(
+				goldmark.WithRendererOptions(
+					html.WithUnsafe(),
+				),
+				goldmark.WithExtensions(
+					Strikethrough,
+				),
+			)
+			n := md.Parser().Parse(text.NewReader(s))
+			c1 := n.FirstChild().FirstChild()
+			if !bytes.Equal(c1.Text(s), []byte(cs.T1)) { // nolint: staticcheck
+
+				t.Errorf("%s unmatch:\n%s", cs.Name, testutil.DiffPretty(c1.Text(s), []byte(cs.T1))) // nolint: staticcheck
+
+			}
+		})
+	}
+
+}

extension/table.go

@@ -184,11 +184,11 @@ func (b *tableParagraphTransformer) Transform(node *gast.Paragraph, reader text.
 func (b *tableParagraphTransformer) parseRow(segment text.Segment,
 	alignments []ast.Alignment, isHeader bool, reader text.Reader, pc parser.Context) *ast.TableRow {
 	source := reader.Source()
+	segment = segment.TrimLeftSpace(source)
+	segment = segment.TrimRightSpace(source)
 	line := segment.Value(source)
 	pos := 0
-	pos += util.TrimLeftSpaceLength(line)
 	limit := len(line)
-	limit -= util.TrimRightSpaceLength(line)
 	row := ast.NewTableRow(alignments)
 	if len(line) > 0 && line[pos] == '|' {
 		pos++

extension/table_test.go

@@ -355,3 +355,40 @@ bar | baz
 		t,
 	)
 }
+
+func TestTableFuzzedPanics(t *testing.T) {
+	markdown := goldmark.New(
+		goldmark.WithRendererOptions(
+			html.WithXHTML(),
+			html.WithUnsafe(),
+		),
+		goldmark.WithExtensions(
+			NewTable(),
+		),
+	)
+	testutil.DoTestCase(
+		markdown,
+		testutil.MarkdownTestCase{
+			No:          1,
+			Description: "This should not panic",
+			Markdown:    "* 0\n-|\n\t0",
+			Expected: `<ul>
+<li>
+<table>
+<thead>
+<tr>
+<th>0</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td>0</td>
+</tr>
+</tbody>
+</table>
+</li>
+</ul>`,
+		},
+		t,
+	)
+}

markdown.go

@@ -2,12 +2,13 @@
 package goldmark
 
 import (
+	"io"
+
 	"github.com/yuin/goldmark/parser"
 	"github.com/yuin/goldmark/renderer"
 	"github.com/yuin/goldmark/renderer/html"
 	"github.com/yuin/goldmark/text"
 	"github.com/yuin/goldmark/util"
-	"io"
 )
 
 // DefaultParser returns a new Parser that is configured by default values.

parser/blockquote.go

@@ -28,12 +28,13 @@ func (b *blockquoteParser) process(reader text.Reader) bool {
 		reader.Advance(pos)
 		return true
 	}
-	if line[pos] == ' ' || line[pos] == '\t' {
-		pos++
-	}
 	reader.Advance(pos)
-	if line[pos-1] == '\t' {
-		reader.SetPadding(2)
+	if line[pos] == ' ' || line[pos] == '\t' {
+		padding := 0
+		if line[pos] == '\t' {
+			padding = util.TabWidth(reader.LineOffset()) - 1
+		}
+		reader.AdvanceAndSetPadding(1, padding)
 	}
 	return true
 }

parser/code_block.go

@@ -35,6 +35,7 @@ func (b *codeBlockParser) Open(parent ast.Node, reader text.Reader, pc Context)
 	if segment.Padding != 0 {
 		preserveLeadingTabInCodeBlock(&segment, reader, 0)
 	}
+	segment.ForceNewline = true
 	node.Lines().Append(segment)
 	reader.Advance(segment.Len() - 1)
 	return node, NoChildren
@@ -59,6 +60,7 @@ func (b *codeBlockParser) Continue(node ast.Node, reader text.Reader, pc Context
 		preserveLeadingTabInCodeBlock(&segment, reader, 0)
 	}
 
+	segment.ForceNewline = true
 	node.Lines().Append(segment)
 	reader.Advance(segment.Len() - 1)
 	return Continue | NoChildren

parser/fcode_block.go

@@ -100,6 +100,7 @@ func (b *fencedCodeBlockParser) Continue(node ast.Node, reader text.Reader, pc C
 	if padding != 0 {
 		preserveLeadingTabInCodeBlock(&seg, reader, fdata.indent)
 	}
+	seg.ForceNewline = true // EOF as newline
 	node.Lines().Append(seg)
 	reader.AdvanceAndSetPadding(segment.Stop-segment.Start-pos-1, padding)
 	return Continue | NoChildren

parser/link.go

@@ -126,13 +126,13 @@ func (s *linkParser) Parse(parent ast.Node, block text.Reader, pc Context) ast.N
 	if line[0] == '!' {
 		if len(line) > 1 && line[1] == '[' {
 			block.Advance(1)
-			pc.Set(linkBottom, pc.LastDelimiter())
+			pushLinkBottom(pc)
 			return processLinkLabelOpen(block, segment.Start+1, true, pc)
 		}
 		return nil
 	}
 	if line[0] == '[' {
-		pc.Set(linkBottom, pc.LastDelimiter())
+		pushLinkBottom(pc)
 		return processLinkLabelOpen(block, segment.Start, false, pc)
 	}
 
@@ -143,6 +143,7 @@ func (s *linkParser) Parse(parent ast.Node, block text.Reader, pc Context) ast.N
 	}
 	last := tlist.(*linkLabelState).Last
 	if last == nil {
+		_ = popLinkBottom(pc)
 		return nil
 	}
 	block.Advance(1)
@@ -151,11 +152,13 @@ func (s *linkParser) Parse(parent ast.Node, block text.Reader, pc Context) ast.N
 	//  > A link label can have at most 999 characters inside the square brackets.
 	if linkLabelStateLength(tlist.(*linkLabelState)) > 998 {
 		ast.MergeOrReplaceTextSegment(last.Parent(), last, last.Segment)
+		_ = popLinkBottom(pc)
 		return nil
 	}
 
 	if !last.IsImage && s.containsLink(last) { // a link in a link text is not allowed
 		ast.MergeOrReplaceTextSegment(last.Parent(), last, last.Segment)
+		_ = popLinkBottom(pc)
 		return nil
 	}
 
@@ -169,6 +172,7 @@ func (s *linkParser) Parse(parent ast.Node, block text.Reader, pc Context) ast.N
 		link, hasValue = s.parseReferenceLink(parent, last, block, pc)
 		if link == nil && hasValue {
 			ast.MergeOrReplaceTextSegment(last.Parent(), last, last.Segment)
+			_ = popLinkBottom(pc)
 			return nil
 		}
 	}
@@ -182,12 +186,14 @@ func (s *linkParser) Parse(parent ast.Node, block text.Reader, pc Context) ast.N
 		//  > A link label can have at most 999 characters inside the square brackets.
 		if len(maybeReference) > 999 {
 			ast.MergeOrReplaceTextSegment(last.Parent(), last, last.Segment)
+			_ = popLinkBottom(pc)
 			return nil
 		}
 
 		ref, ok := pc.Reference(util.ToLinkReference(maybeReference))
 		if !ok {
 			ast.MergeOrReplaceTextSegment(last.Parent(), last, last.Segment)
+			_ = popLinkBottom(pc)
 			return nil
 		}
 		link = ast.NewLink()
@@ -230,11 +236,7 @@ func processLinkLabelOpen(block text.Reader, pos int, isImage bool, pc Context)
 }
 
 func (s *linkParser) processLinkLabel(parent ast.Node, link *ast.Link, last *linkLabelState, pc Context) {
-	var bottom ast.Node
-	if v := pc.Get(linkBottom); v != nil {
-		bottom = v.(ast.Node)
-	}
-	pc.Set(linkBottom, nil)
+	bottom := popLinkBottom(pc)
 	ProcessDelimiters(bottom, pc)
 	for c := last.NextSibling(); c != nil; {
 		next := c.NextSibling()
@@ -395,6 +397,43 @@ func parseLinkTitle(block text.Reader) ([]byte, bool) {
 	return nil, false
 }
 
+func pushLinkBottom(pc Context) {
+	bottoms := pc.Get(linkBottom)
+	b := pc.LastDelimiter()
+	if bottoms == nil {
+		pc.Set(linkBottom, b)
+		return
+	}
+	if s, ok := bottoms.([]ast.Node); ok {
+		pc.Set(linkBottom, append(s, b))
+		return
+	}
+	pc.Set(linkBottom, []ast.Node{bottoms.(ast.Node), b})
+}
+
+func popLinkBottom(pc Context) ast.Node {
+	bottoms := pc.Get(linkBottom)
+	if bottoms == nil {
+		return nil
+	}
+	if v, ok := bottoms.(ast.Node); ok {
+		pc.Set(linkBottom, nil)
+		return v
+	}
+	s := bottoms.([]ast.Node)
+	v := s[len(s)-1]
+	n := s[0 : len(s)-1]
+	switch len(n) {
+	case 0:
+		pc.Set(linkBottom, nil)
+	case 1:
+		pc.Set(linkBottom, n[0])
+	default:
+		pc.Set(linkBottom, s[0:len(s)-1])
+	}
+	return v
+}
+
 func (s *linkParser) CloseBlock(parent ast.Node, block text.Reader, pc Context) {
 	pc.Set(linkBottom, nil)
 	tlist := pc.Get(linkLabelStateKey)

parser/list_item.go

@@ -58,7 +58,7 @@ func (b *listItemParser) Continue(node ast.Node, reader text.Reader, pc Context)
 	}
 
 	offset := lastOffset(node.Parent())
-	isEmpty := node.ChildCount() == 0
+	isEmpty := node.ChildCount() == 0 && pc.Get(emptyListItemWithBlankLines) != nil
 	indent, _ := util.IndentWidth(line, reader.LineOffset())
 	if (isEmpty || indent < offset) && indent < 4 {
 		_, typ := matchesListItem(line, true)

parser/parser.go

@@ -882,6 +882,7 @@ func (p *parser) Parse(reader text.Reader, opts ...ParseOption) ast.Node {
 	for _, at := range p.astTransformers {
 		at.Transform(root, reader, pc)
 	}
+
 	// root.Dump(reader.Source(), 0)
 	return root
 }
@@ -1256,4 +1257,5 @@ func (p *parser) parseBlock(block text.BlockReader, parent ast.Node, pc Context)
 	for _, ip := range p.closeBlockers {
 		ip.CloseBlock(parent, block, pc)
 	}
+
 }

renderer/html/html.go

@@ -680,7 +680,7 @@ func (r *Renderer) renderImage(w util.BufWriter, source []byte, node ast.Node, e
 		_, _ = w.Write(util.EscapeHTML(util.URLEscape(n.Destination, true)))
 	}
 	_, _ = w.WriteString(`" alt="`)
-	r.renderAttribute(w, source, n)
+	r.renderTexts(w, source, n)
 	_ = w.WriteByte('"')
 	if n.Title != nil {
 		_, _ = w.WriteString(` title="`)
@@ -737,7 +737,7 @@ func (r *Renderer) renderText(w util.BufWriter, source []byte, node ast.Node, en
 			if r.EastAsianLineBreaks != EastAsianLineBreaksNone && len(value) != 0 {
 				sibling := node.NextSibling()
 				if sibling != nil && sibling.Kind() == ast.KindText {
-					if siblingText := sibling.(*ast.Text).Text(source); len(siblingText) != 0 {
+					if siblingText := sibling.(*ast.Text).Value(source); len(siblingText) != 0 {
 						thisLastRune := util.ToRune(value, len(value)-1)
 						siblingFirstRune, _ := utf8.DecodeRune(siblingText)
 						if r.EastAsianLineBreaks.softLineBreak(thisLastRune, siblingFirstRune) {
@@ -770,19 +770,14 @@ func (r *Renderer) renderString(w util.BufWriter, source []byte, node ast.Node,
 	return ast.WalkContinue, nil
 }
 
-func (r *Renderer) renderAttribute(w util.BufWriter, source []byte, n ast.Node) {
+func (r *Renderer) renderTexts(w util.BufWriter, source []byte, n ast.Node) {
 	for c := n.FirstChild(); c != nil; c = c.NextSibling() {
 		if s, ok := c.(*ast.String); ok {
 			_, _ = r.renderString(w, source, s, true)
-		} else if t, ok := c.(*ast.String); ok {
+		} else if t, ok := c.(*ast.Text); ok {
 			_, _ = r.renderText(w, source, t, true)
-		} else if !c.HasChildren() {
-			r.Writer.Write(w, c.Text(source))
-			if t, ok := c.(*ast.Text); ok && t.SoftLineBreak() {
-				w.WriteByte('\n')
-			}
 		} else {
-			r.renderAttribute(w, source, c)
+			r.renderTexts(w, source, c)
 		}
 	}
 }

text/segment.go

@@ -2,6 +2,7 @@ package text
 
 import (
 	"bytes"
+
 	"github.com/yuin/goldmark/util"
 )
 
@@ -18,6 +19,20 @@ type Segment struct {
 
 	// Padding is a padding length of the segment.
 	Padding int
+
+	// ForceNewline is true if the segment should be ended with a newline.
+	// Some elements(i.e. CodeBlock, FencedCodeBlock) does not trim trailing
+	// newlines. Spec defines that EOF is treated as a newline, so we need to
+	// add a newline to the end of the segment if it is not empty.
+	//
+	// i.e.:
+	//
+	//     ```go
+	//     const test = "test"
+	//
+	// This code does not close the code block and ends with EOF. In this case,
+	// we need to add a newline to the end of the last line like `const test = "test"\n`.
+	ForceNewline bool
 }
 
 // NewSegment return a new Segment.
@@ -40,12 +55,18 @@ func NewSegmentPadding(start, stop, n int) Segment {
 
 // Value returns a value of the segment.
 func (t *Segment) Value(buffer []byte) []byte {
+	var result []byte
 	if t.Padding == 0 {
-		return buffer[t.Start:t.Stop]
+		result = buffer[t.Start:t.Stop]
+	} else {
+		result = make([]byte, 0, t.Padding+t.Stop-t.Start+1)
+		result = append(result, bytes.Repeat(space, t.Padding)...)
+		result = append(result, buffer[t.Start:t.Stop]...)
 	}
-	result := make([]byte, 0, t.Padding+t.Stop-t.Start+1)
-	result = append(result, bytes.Repeat(space, t.Padding)...)
-	return append(result, buffer[t.Start:t.Stop]...)
+	if t.ForceNewline && len(result) > 0 && result[len(result)-1] != '\n' {
+		result = append(result, '\n')
+	}
+	return result
 }
 
 // Len returns a length of the segment.
@@ -207,3 +228,12 @@ func (s *Segments) Unshift(v Segment) {
 	s.values = append(s.values[0:1], s.values[0:]...)
 	s.values[0] = v
 }
+
+// Value returns a string value of the collection.
+func (s *Segments) Value(buffer []byte) []byte {
+	var result []byte
+	for _, v := range s.values {
+		result = append(result, v.Value(buffer)...)
+	}
+	return result
+}

util/util.go

@@ -166,7 +166,13 @@ func IndentPositionPadding(bs []byte, currentPos, paddingv, width int) (pos, pad
 	w := 0
 	i := 0
 	l := len(bs)
+	p := paddingv
 	for ; i < l; i++ {
+		if p > 0 {
+			p--
+			w++
+			continue
+		}
 		if bs[i] == '\t' && w < width {
 			w += TabWidth(currentPos + w)
 		} else if bs[i] == ' ' && w < width {

util/util_unsafe_go120.go

@@ -1,5 +1,5 @@
-//go:build !appengine && !js
-// +build !appengine,!js
+//go:build !appengine && !js && !go1.21
+// +build !appengine,!js,!go1.21
 
 package util
 

util/util_unsafe_go121.go

@@ -0,0 +1,18 @@
+//go:build !appengine && !js && go1.21
+// +build !appengine,!js,go1.21
+
+package util
+
+import (
+	"unsafe"
+)
+
+// BytesToReadOnlyString returns a string converted from given bytes.
+func BytesToReadOnlyString(b []byte) string {
+	return unsafe.String(unsafe.SliceData(b), len(b))
+}
+
+// StringToReadOnlyBytes returns bytes converted from given string.
+func StringToReadOnlyBytes(s string) []byte {
+	return unsafe.Slice(unsafe.StringData(s), len(s))
+}