Fix bugs found in fuzzing

.gitignore

@@ -11,3 +11,9 @@
 
 # Output of the go coverage tool, specifically when used with LiteIDE
 *.out
+
+.DS_Store
+fuzz/corpus
+fuzz/crashers
+fuzz/suppressions
+fuzz/fuzz-fuzz.zip

Makefile

@@ -1,7 +1,16 @@
-.PHONY: test
+.PHONY: test fuzz
 
 test:
 	go test -coverprofile=profile.out -coverpkg=github.com/yuin/goldmark,github.com/yuin/goldmark/ast,github.com/yuin/goldmark/extension,github.com/yuin/goldmark/extension/ast,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 ./...
 
 cov: test
 	go tool cover -html=profile.out
+
+fuzz:
+	which go-fuzz 2>&1 > /dev/null || (GO111MODULE=off go get -u github.com/dvyukov/go-fuzz/go-fuzz github.com/dvyukov/go-fuzz/go-fuzz-build; GO111MODULE=off go get -d github.com/dvyukov/go-fuzz-corpus; true)
+	cd ./fuzz && go-fuzz-build
+	rm -rf ./fuzz/corpus
+	rm -rf ./fuzz/crashers
+	rm -rf ./fuzz/suppressions
+	rm -f ./fuzz/fuzz-fuzz.zip
+	cd ./fuzz && go-fuzz

ast/ast.go

@@ -348,7 +348,6 @@ func (n *BaseNode) SetAttribute(name, value []byte) {
 		}
 	}
 	n.attributes = append(n.attributes, Attribute{name, value})
-	return
 }
 
 // Attribute implements Node.Attribute.
@@ -396,10 +395,8 @@ func DumpHelper(v Node, source []byte, level int, kv map[string]string, cb func(
 		fmt.Printf("\"\n")
 		fmt.Printf("%sHasBlankPreviousLines: %v\n", indent2, v.HasBlankPreviousLines())
 	}
-	if kv != nil {
+	for name, value := range kv {
-		for name, value := range kv {
+		fmt.Printf("%s%s: %s\n", indent2, name, value)
-			fmt.Printf("%s%s: %s\n", indent2, name, value)
-		}
 	}
 	if cb != nil {
 		cb(level + 1)

extension/definition_list.go

@@ -28,7 +28,7 @@ func (b *definitionListParser) Open(parent gast.Node, reader text.Reader, pc par
 	}
 	line, _ := reader.PeekLine()
 	pos := pc.BlockOffset()
-	if line[pos] != ':' {
+	if pos < 0 || line[pos] != ':' {
 		return nil, parser.NoChildren
 	}
 
@@ -105,10 +105,13 @@ func NewDefinitionDescriptionParser() parser.BlockParser {
 func (b *definitionDescriptionParser) Open(parent gast.Node, reader text.Reader, pc parser.Context) (gast.Node, parser.State) {
 	line, _ := reader.PeekLine()
 	pos := pc.BlockOffset()
-	if line[pos] != ':' {
+	if pos < 0 || line[pos] != ':' {
 		return nil, parser.NoChildren
 	}
 	list, _ := parent.(*ast.DefinitionList)
+	if list == nil {
+		return nil, parser.NoChildren
+	}
 	para := list.TemporaryParagraph
 	list.TemporaryParagraph = nil
 	if para != nil {
@@ -183,18 +186,18 @@ func (r *DefinitionListHTMLRenderer) RegisterFuncs(reg renderer.NodeRendererFunc
 
 func (r *DefinitionListHTMLRenderer) renderDefinitionList(w util.BufWriter, source []byte, n gast.Node, entering bool) (gast.WalkStatus, error) {
 	if entering {
-		w.WriteString("<dl>\n")
+		_, _ = w.WriteString("<dl>\n")
 	} else {
-		w.WriteString("</dl>\n")
+		_, _ = w.WriteString("</dl>\n")
 	}
 	return gast.WalkContinue, nil
 }
 
 func (r *DefinitionListHTMLRenderer) renderDefinitionTerm(w util.BufWriter, source []byte, n gast.Node, entering bool) (gast.WalkStatus, error) {
 	if entering {
-		w.WriteString("<dt>")
+		_, _ = w.WriteString("<dt>")
 	} else {
-		w.WriteString("</dt>\n")
+		_, _ = w.WriteString("</dt>\n")
 	}
 	return gast.WalkContinue, nil
 }
@@ -203,12 +206,12 @@ func (r *DefinitionListHTMLRenderer) renderDefinitionDescription(w util.BufWrite
 	if entering {
 		n := node.(*ast.DefinitionDescription)
 		if n.IsTight {
-			w.WriteString("<dd>")
+			_, _ = w.WriteString("<dd>")
 		} else {
-			w.WriteString("<dd>\n")
+			_, _ = w.WriteString("<dd>\n")
 		}
 	} else {
-		w.WriteString("</dd>\n")
+		_, _ = w.WriteString("</dd>\n")
 	}
 	return gast.WalkContinue, nil
 }

extension/footnote.go

@@ -29,7 +29,7 @@ func NewFootnoteBlockParser() parser.BlockParser {
 func (b *footnoteBlockParser) Open(parent gast.Node, reader text.Reader, pc parser.Context) (gast.Node, parser.State) {
 	line, segment := reader.PeekLine()
 	pos := pc.BlockOffset()
-	if line[pos] != '[' {
+	if pos < 0 || line[pos] != '[' {
 		return nil, parser.NoChildren
 	}
 	pos++
@@ -37,7 +37,7 @@ func (b *footnoteBlockParser) Open(parent gast.Node, reader text.Reader, pc pars
 		return nil, parser.NoChildren
 	}
 	open := pos + 1
-	closes := -1
+	closes := 0
 	closure := util.FindClosure(line[pos+1:], '[', ']', false, false)
 	if closure > -1 {
 		closes = pos + 1 + closure
@@ -52,10 +52,15 @@ func (b *footnoteBlockParser) Open(parent gast.Node, reader text.Reader, pc pars
 	if util.IsBlank(label) {
 		return nil, parser.NoChildren
 	}
+	item := ast.NewFootnote(label)
+
 	pos = pos + 2 + closes - open + 2
+	if pos >= len(line) {
+		reader.Advance(pos)
+		return item, parser.NoChildren
+	}
 	childpos, padding := util.IndentPosition(line[pos:], pos, 1)
 	reader.AdvanceAndSetPadding(pos+childpos, padding)
-	item := ast.NewFootnote(label)
 	return item, parser.HasChildren
 }
 
@@ -207,13 +212,13 @@ func (r *FootnoteHTMLRenderer) renderFootnoteLink(w util.BufWriter, source []byt
 	if entering {
 		n := node.(*ast.FootnoteLink)
 		is := strconv.Itoa(n.Index)
-		w.WriteString(`<sup id="fnref:`)
+		_, _ = w.WriteString(`<sup id="fnref:`)
-		w.WriteString(is)
+		_, _ = w.WriteString(is)
-		w.WriteString(`"><a href="#fn:`)
+		_, _ = w.WriteString(`"><a href="#fn:`)
-		w.WriteString(is)
+		_, _ = w.WriteString(is)
-		w.WriteString(`" class="footnote-ref" role="doc-noteref">`)
+		_, _ = w.WriteString(`" class="footnote-ref" role="doc-noteref">`)
-		w.WriteString(is)
+		_, _ = w.WriteString(is)
-		w.WriteString(`</a></sup>`)
+		_, _ = w.WriteString(`</a></sup>`)
 	}
 	return gast.WalkContinue, nil
 }
@@ -222,12 +227,12 @@ func (r *FootnoteHTMLRenderer) renderFootnote(w util.BufWriter, source []byte, n
 	n := node.(*ast.Footnote)
 	is := strconv.Itoa(n.Index)
 	if entering {
-		w.WriteString(`<li id="fn:`)
+		_, _ = w.WriteString(`<li id="fn:`)
-		w.WriteString(is)
+		_, _ = w.WriteString(is)
-		w.WriteString(`" role="doc-endnote">`)
+		_, _ = w.WriteString(`" role="doc-endnote">`)
-		w.WriteString("\n")
+		_, _ = w.WriteString("\n")
 	} else {
-		w.WriteString("</li>\n")
+		_, _ = w.WriteString("</li>\n")
 	}
 	return gast.WalkContinue, nil
 }
@@ -238,20 +243,20 @@ func (r *FootnoteHTMLRenderer) renderFootnoteList(w util.BufWriter, source []byt
 		tag = "div"
 	}
 	if entering {
-		w.WriteString("<")
+		_, _ = w.WriteString("<")
-		w.WriteString(tag)
+		_, _ = w.WriteString(tag)
-		w.WriteString(` class="footnotes" role="doc-endnotes">`)
+		_, _ = w.WriteString(` class="footnotes" role="doc-endnotes">`)
 		if r.Config.XHTML {
-			w.WriteString("\n<hr />\n")
+			_, _ = w.WriteString("\n<hr />\n")
 		} else {
-			w.WriteString("\n<hr>\n")
+			_, _ = w.WriteString("\n<hr>\n")
 		}
-		w.WriteString("<ol>\n")
+		_, _ = w.WriteString("<ol>\n")
 	} else {
-		w.WriteString("</ol>\n")
+		_, _ = w.WriteString("</ol>\n")
-		w.WriteString("<")
+		_, _ = w.WriteString("<")
-		w.WriteString(tag)
+		_, _ = w.WriteString(tag)
-		w.WriteString(">\n")
+		_, _ = w.WriteString(">\n")
 	}
 	return gast.WalkContinue, nil
 }

extension/table.go

@@ -169,22 +169,22 @@ func (r *TableHTMLRenderer) RegisterFuncs(reg renderer.NodeRendererFuncRegistere
 
 func (r *TableHTMLRenderer) renderTable(w util.BufWriter, source []byte, n gast.Node, entering bool) (gast.WalkStatus, error) {
 	if entering {
-		w.WriteString("<table>\n")
+		_, _ = w.WriteString("<table>\n")
 	} else {
-		w.WriteString("</table>\n")
+		_, _ = w.WriteString("</table>\n")
 	}
 	return gast.WalkContinue, nil
 }
 
 func (r *TableHTMLRenderer) renderTableHeader(w util.BufWriter, source []byte, n gast.Node, entering bool) (gast.WalkStatus, error) {
 	if entering {
-		w.WriteString("<thead>\n")
+		_, _ = w.WriteString("<thead>\n")
-		w.WriteString("<tr>\n")
+		_, _ = w.WriteString("<tr>\n")
 	} else {
-		w.WriteString("</tr>\n")
+		_, _ = w.WriteString("</tr>\n")
-		w.WriteString("</thead>\n")
+		_, _ = w.WriteString("</thead>\n")
 		if n.NextSibling() != nil {
-			w.WriteString("<tbody>\n")
+			_, _ = w.WriteString("<tbody>\n")
 		}
 	}
 	return gast.WalkContinue, nil
@@ -192,11 +192,11 @@ func (r *TableHTMLRenderer) renderTableHeader(w util.BufWriter, source []byte, n
 
 func (r *TableHTMLRenderer) renderTableRow(w util.BufWriter, source []byte, n gast.Node, entering bool) (gast.WalkStatus, error) {
 	if entering {
-		w.WriteString("<tr>\n")
+		_, _ = w.WriteString("<tr>\n")
 	} else {
-		w.WriteString("</tr>\n")
+		_, _ = w.WriteString("</tr>\n")
 		if n.Parent().LastChild() == n {
-			w.WriteString("</tbody>\n")
+			_, _ = w.WriteString("</tbody>\n")
 		}
 	}
 	return gast.WalkContinue, nil

fuzz/fuzz.go

@@ -0,0 +1,28 @@
+package fuzz
+
+import (
+	"bytes"
+	"github.com/yuin/goldmark"
+	"github.com/yuin/goldmark/extension"
+	"github.com/yuin/goldmark/renderer/html"
+)
+
+func Fuzz(data []byte) int {
+	markdown := goldmark.New(
+		goldmark.WithRendererOptions(
+			html.WithUnsafe(),
+		),
+		goldmark.WithExtensions(
+			extension.DefinitionList,
+			extension.Footnote,
+			extension.GFM,
+			extension.Typographer,
+		),
+	)
+	var b bytes.Buffer
+	if err := markdown.Convert(data, &b); err != nil {
+		return 0
+	}
+
+	return 1
+}

fuzz/fuzz_test.go

@@ -0,0 +1,41 @@
+package fuzz
+
+import (
+	"bytes"
+	"fmt"
+	"io/ioutil"
+	"testing"
+
+	"github.com/yuin/goldmark"
+	"github.com/yuin/goldmark/extension"
+	"github.com/yuin/goldmark/renderer/html"
+	"github.com/yuin/goldmark/util"
+)
+
+var _ = fmt.Printf
+
+func TestFuzz(t *testing.T) {
+	crasher := "6dff3d03167cb144d4e2891edac76ee740a77bc7"
+	data, err := ioutil.ReadFile("crashers/" + crasher)
+	if err != nil {
+		return
+	}
+	fmt.Printf("%s\n", util.VisualizeSpaces(data))
+	fmt.Println("||||||||||||||||||||||")
+	markdown := goldmark.New(
+		goldmark.WithRendererOptions(
+			html.WithUnsafe(),
+		),
+		goldmark.WithExtensions(
+			extension.DefinitionList,
+			extension.Footnote,
+			extension.GFM,
+			extension.Typographer,
+		),
+	)
+	var b bytes.Buffer
+	if err := markdown.Convert(data, &b); err != nil {
+		panic(err)
+	}
+	fmt.Println(b.String())
+}

parser/atx_heading.go

@@ -77,6 +77,9 @@ func NewATXHeadingParser(opts ...HeadingOption) BlockParser {
 func (b *atxHeadingParser) Open(parent ast.Node, reader text.Reader, pc Context) (ast.Node, State) {
 	line, segment := reader.PeekLine()
 	pos := pc.BlockOffset()
+	if pos < 0 {
+		return nil, NoChildren
+	}
 	i := pos
 	for ; i < len(line) && line[i] == '#'; i++ {
 	}

parser/code_block.go

@@ -55,7 +55,7 @@ func (b *codeBlockParser) Close(node ast.Node, reader text.Reader, pc Context) {
 	lines := node.Lines()
 	length := lines.Len() - 1
 	source := reader.Source()
-	for {
+	for length >= 0 {
 		line := lines.At(length)
 		if util.IsBlank(line.Value(source)) {
 			length--

parser/fcode_block.go

@@ -30,7 +30,7 @@ var fencedCodeBlockInfoKey = NewContextKey()
 func (b *fencedCodeBlockParser) Open(parent ast.Node, reader text.Reader, pc Context) (ast.Node, State) {
 	line, segment := reader.PeekLine()
 	pos := pc.BlockOffset()
-	if line[pos] != '`' && line[pos] != '~' {
+	if pos < 0 || (line[pos] != '`' && line[pos] != '~') {
 		return nil, NoChildren
 	}
 	findent := pos

parser/html_block.go

@@ -109,7 +109,7 @@ func (b *htmlBlockParser) Open(parent ast.Node, reader text.Reader, pc Context)
 	var node *ast.HTMLBlock
 	line, segment := reader.PeekLine()
 	last := pc.LastOpenedBlock().Node
-	if pos := pc.BlockOffset(); line[pos] != '<' {
+	if pos := pc.BlockOffset(); pos < 0 || line[pos] != '<' {
 		return nil, NoChildren
 	}
 

parser/list.go

@@ -35,7 +35,7 @@ func parseListItem(line []byte) ([6]int, listItemType) {
 	ret[1] = i
 	ret[2] = i
 	var typ listItemType
-	if i < l && line[i] == '-' || line[i] == '*' || line[i] == '+' {
+	if i < l && (line[i] == '-' || line[i] == '*' || line[i] == '+') {
 		i++
 		ret[3] = i
 		typ = bulletList
@@ -46,7 +46,7 @@ func parseListItem(line []byte) ([6]int, listItemType) {
 		if ret[3] == ret[2] || ret[3]-ret[2] > 9 {
 			return ret, notList
 		}
-		if i < l && line[i] == '.' || line[i] == ')' {
+		if i < l && (line[i] == '.' || line[i] == ')') {
 			i++
 			ret[3] = i
 		} else {
@@ -56,12 +56,17 @@ func parseListItem(line []byte) ([6]int, listItemType) {
 	} else {
 		return ret, notList
 	}
-	if line[i] != '\n' {
+	if i < l && line[i] != '\n' {
 		w, _ := util.IndentWidth(line[i:], 0)
 		if w == 0 {
 			return ret, notList
 		}
 	}
+	if i >= l {
+		ret[4] = -1
+		ret[5] = -1
+		return ret, typ
+	}
 	ret[4] = i
 	ret[5] = len(line)
 	if line[ret[5]-1] == '\n' && line[i] != '\n' {
@@ -80,7 +85,7 @@ func matchesListItem(source []byte, strict bool) ([6]int, listItemType) {
 
 func calcListOffset(source []byte, match [6]int) int {
 	offset := 0
-	if util.IsBlank(source[match[4]:]) { // list item starts with a blank line
+	if match[4] < 0 || util.IsBlank(source[match[4]:]) { // list item starts with a blank line
 		offset = 1
 	} else {
 		offset, _ = util.IndentWidth(source[match[4]:], match[4])

parser/list_item.go

@@ -36,7 +36,7 @@ func (b *listItemParser) Open(parent ast.Node, reader text.Reader, pc Context) (
 	}
 	itemOffset := calcListOffset(line, match)
 	node := ast.NewListItem(match[3] + itemOffset)
-	if match[5]-match[4] == 1 {
+	if match[4] < 0 || match[5]-match[4] == 1 {
 		return node, NoChildren
 	}
 

parser/parser.go

@@ -156,6 +156,7 @@ type Context interface {
 
 	// BlockOffset returns a first non-space character position on current line.
 	// This value is valid only for BlockParser.Open.
+	// BlockOffset returns -1 if current line is blank.
 	BlockOffset() int
 
 	// BlockOffset sets a first non-space character position on current line.
@@ -833,7 +834,11 @@ retry:
 			//currentLineNum, _ = reader.Position()
 			line, _ = reader.PeekLine()
 			w, pos = util.IndentWidth(line, 0)
-			pc.SetBlockOffset(pos)
+			if w >= len(line) {
+				pc.SetBlockOffset(-1)
+			} else {
+				pc.SetBlockOffset(pos)
+			}
 			shouldPeek = false
 			if line == nil || line[0] == '\n' {
 				break

parser/setext_headings.go

@@ -65,7 +65,7 @@ func (b *setextHeadingParser) Open(parent ast.Node, reader text.Reader, pc Conte
 	}
 	node := ast.NewHeading(level)
 	node.Lines().Append(segment)
-	pc.Set(temporaryParagraphKey, paragraph)
+	pc.Set(temporaryParagraphKey, last)
 	return node, NoChildren
 }
 

renderer/html/html.go

@@ -577,28 +577,30 @@ func (d *defaultWriter) Write(writer util.BufWriter, source []byte) {
 			next := i + 1
 			if next < limit && source[next] == '#' {
 				nnext := next + 1
-				nc := source[nnext]
+				if nnext < limit {
-				// code point like #x22;
+					nc := source[nnext]
-				if nnext < limit && nc == 'x' || nc == 'X' {
+					// code point like #x22;
-					start := nnext + 1
+					if nnext < limit && nc == 'x' || nc == 'X' {
-					i, ok = util.ReadWhile(source, [2]int{start, limit}, util.IsHexDecimal)
+						start := nnext + 1
-					if ok && i < limit && source[i] == ';' {
+						i, ok = util.ReadWhile(source, [2]int{start, limit}, util.IsHexDecimal)
-						v, _ := strconv.ParseUint(util.BytesToReadOnlyString(source[start:i]), 16, 32)
+						if ok && i < limit && source[i] == ';' {
-						d.RawWrite(writer, source[n:pos])
+							v, _ := strconv.ParseUint(util.BytesToReadOnlyString(source[start:i]), 16, 32)
-						n = i + 1
+							d.RawWrite(writer, source[n:pos])
-						escapeRune(writer, rune(v))
+							n = i + 1
-						continue
+							escapeRune(writer, rune(v))
-					}
+							continue
-					// code point like #1234;
+						}
-				} else if nc >= '0' && nc <= '9' {
+						// code point like #1234;
-					start := nnext
+					} else if nc >= '0' && nc <= '9' {
-					i, ok = util.ReadWhile(source, [2]int{start, limit}, util.IsNumeric)
+						start := nnext
-					if ok && i < limit && i-start < 8 && source[i] == ';' {
+						i, ok = util.ReadWhile(source, [2]int{start, limit}, util.IsNumeric)
-						v, _ := strconv.ParseUint(util.BytesToReadOnlyString(source[start:i]), 0, 32)
+						if ok && i < limit && i-start < 8 && source[i] == ';' {
-						d.RawWrite(writer, source[n:pos])
+							v, _ := strconv.ParseUint(util.BytesToReadOnlyString(source[start:i]), 0, 32)
-						n = i + 1
+							d.RawWrite(writer, source[n:pos])
-						escapeRune(writer, rune(v))
+							n = i + 1
-						continue
+							escapeRune(writer, rune(v))
+							continue
+						}
 					}
 				}
 			} else {

text/reader.go

@@ -326,6 +326,9 @@ func (r *blockReader) PrecendingCharacter() rune {
 			break
 		}
 	}
+	if i < 0 {
+		return rune('\n')
+	}
 	rn, _ := utf8.DecodeRune(r.source[i:])
 	return rn
 }

util/util.go

@@ -567,7 +567,7 @@ func URLEscape(v []byte, resolveReference bool) []byte {
 		i += int(u8len)
 		n = i
 	}
-	if cob.IsCopied() {
+	if cob.IsCopied() && n < limit {
 		cob.Write(v[n:])
 	}
 	return cob.Bytes()