mirror/hugo
1
0
Fork 0
mirror of https://github.com/gohugoio/hugo.git synced 2025-08-30 22:39:58 +02:00
Code Issues Packages Projects Releases Wiki Activity

markup/goldmark: Add an optional Blackfriday auto ID strategy

Browse Source 
Fixes #6707
This commit is contained in:
Bjørn Erik Pedersen
2020-01-05 11:52:00 +01:00
parent 8f071fc159
commit 16e7c11203
7 changed files with 141 additions and 34 deletions
Download Patch File Download Diff File Expand all files Collapse all files

57
markup/goldmark/autoid.go
View File

@@ -19,6 +19,8 @@ import (
"unicode"
"unicode/utf8"
"github.com/gohugoio/hugo/markup/blackfriday"
"github.com/gohugoio/hugo/markup/goldmark/goldmark_config"
"github.com/gohugoio/hugo/common/text"
@@ -30,34 +32,41 @@ import (
bp "github.com/gohugoio/hugo/bufferpool"
)
func sanitizeAnchorNameString(s string, asciiOnly bool) string {
return string(sanitizeAnchorName([]byte(s), asciiOnly))
func sanitizeAnchorNameString(s string, idType string) string {
return string(sanitizeAnchorName([]byte(s), idType))
}
func sanitizeAnchorName(b []byte, asciiOnly bool) []byte {
return sanitizeAnchorNameWithHook(b, asciiOnly, nil)
func sanitizeAnchorName(b []byte, idType string) []byte {
return sanitizeAnchorNameWithHook(b, idType, nil)
}
func sanitizeAnchorNameWithHook(b []byte, asciiOnly bool, hook func(buf *bytes.Buffer)) []byte {
func sanitizeAnchorNameWithHook(b []byte, idType string, hook func(buf *bytes.Buffer)) []byte {
buf := bp.GetBuffer()
if asciiOnly {
// Normalize it to preserve accents if possible.
b = text.RemoveAccents(b)
}
if idType == goldmark_config.AutoHeadingIDTypeBlackfriday {
// TODO(bep) make it more efficient.
buf.WriteString(blackfriday.SanitizedAnchorName(string(b)))
} else {
asciiOnly := idType == goldmark_config.AutoHeadingIDTypeGitHubAscii
for len(b) > 0 {
r, size := utf8.DecodeRune(b)
switch {
case asciiOnly && size != 1:
case r == '-' || isSpace(r):
buf.WriteRune('-')
case isAlphaNumeric(r):
buf.WriteRune(unicode.ToLower(r))
default:
if asciiOnly {
// Normalize it to preserve accents if possible.
b = text.RemoveAccents(b)
}
b = b[size:]
for len(b) > 0 {
r, size := utf8.DecodeRune(b)
switch {
case asciiOnly && size != 1:
case r == '-' || isSpace(r):
buf.WriteRune('-')
case isAlphaNumeric(r):
buf.WriteRune(unicode.ToLower(r))
default:
}
b = b[size:]
}
}
if hook != nil {
@@ -83,19 +92,19 @@ func isSpace(r rune) bool {
var _ parser.IDs = (*idFactory)(nil)
type idFactory struct {
asciiOnly bool
vals map[string]struct{}
idType string
vals map[string]struct{}
}
func newIDFactory(idType string) *idFactory {
return &idFactory{
vals: make(map[string]struct{}),
asciiOnly: idType == goldmark_config.AutoHeadingIDTypeGitHubAscii,
vals: make(map[string]struct{}),
idType: idType,
}
}
func (ids *idFactory) Generate(value []byte, kind ast.NodeKind) []byte {
return sanitizeAnchorNameWithHook(value, ids.asciiOnly, func(buf *bytes.Buffer) {
return sanitizeAnchorNameWithHook(value, ids.idType, func(buf *bytes.Buffer) {
if buf.Len() == 0 {
if kind == ast.KindHeading {
buf.WriteString("heading")

33
markup/goldmark/autoid_test.go
View File

@@ -17,6 +17,8 @@ import (
"strings"
"testing"
"github.com/gohugoio/hugo/markup/goldmark/goldmark_config"
qt "github.com/frankban/quicktest"
)
@@ -69,9 +71,9 @@ under_score
expect := expectlines[i]
c.Run(input, func(c *qt.C) {
b := []byte(input)
got := string(sanitizeAnchorName(b, false))
got := string(sanitizeAnchorName(b, goldmark_config.AutoHeadingIDTypeGitHub))
c.Assert(got, qt.Equals, expect)
c.Assert(sanitizeAnchorNameString(input, false), qt.Equals, expect)
c.Assert(sanitizeAnchorNameString(input, goldmark_config.AutoHeadingIDTypeGitHub), qt.Equals, expect)
c.Assert(string(b), qt.Equals, input)
})
}
@@ -80,16 +82,21 @@ under_score
func TestSanitizeAnchorNameAsciiOnly(t *testing.T) {
c := qt.New(t)
c.Assert(sanitizeAnchorNameString("god is神真美好 good", true), qt.Equals, "god-is-good")
c.Assert(sanitizeAnchorNameString("Resumé", true), qt.Equals, "resume")
c.Assert(sanitizeAnchorNameString("god is神真美好 good", goldmark_config.AutoHeadingIDTypeGitHubAscii), qt.Equals, "god-is-good")
c.Assert(sanitizeAnchorNameString("Resumé", goldmark_config.AutoHeadingIDTypeGitHubAscii), qt.Equals, "resume")
}
func TestSanitizeAnchorNameBlackfriday(t *testing.T) {
c := qt.New(t)
c.Assert(sanitizeAnchorNameString("Let's try this, shall we?", goldmark_config.AutoHeadingIDTypeBlackfriday), qt.Equals, "let-s-try-this-shall-we")
}
func BenchmarkSanitizeAnchorName(b *testing.B) {
input := []byte("God is good: 神真美好")
b.ResetTimer()
for i := 0; i < b.N; i++ {
result := sanitizeAnchorName(input, false)
result := sanitizeAnchorName(input, goldmark_config.AutoHeadingIDTypeGitHub)
if len(result) != 24 {
b.Fatalf("got %d", len(result))
@@ -101,7 +108,7 @@ func BenchmarkSanitizeAnchorNameAsciiOnly(b *testing.B) {
input := []byte("God is good: 神真美好")
b.ResetTimer()
for i := 0; i < b.N; i++ {
result := sanitizeAnchorName(input, true)
result := sanitizeAnchorName(input, goldmark_config.AutoHeadingIDTypeGitHubAscii)
if len(result) != 12 {
b.Fatalf("got %d", len(result))
@@ -109,11 +116,23 @@ func BenchmarkSanitizeAnchorNameAsciiOnly(b *testing.B) {
}
}
func BenchmarkSanitizeAnchorNameBlackfriday(b *testing.B) {
input := []byte("God is good: 神真美好")
b.ResetTimer()
for i := 0; i < b.N; i++ {
result := sanitizeAnchorName(input, goldmark_config.AutoHeadingIDTypeBlackfriday)
if len(result) != 24 {
b.Fatalf("got %d", len(result))
}
}
}
func BenchmarkSanitizeAnchorNameString(b *testing.B) {
input := "God is good: 神真美好"
b.ResetTimer()
for i := 0; i < b.N; i++ {
result := sanitizeAnchorNameString(input, false)
result := sanitizeAnchorNameString(input, goldmark_config.AutoHeadingIDTypeGitHub)
if len(result) != 24 {
b.Fatalf("got %d", len(result))
}

3
markup/goldmark/convert.go
View File

@@ -29,7 +29,6 @@ import (
"github.com/gohugoio/hugo/hugofs"
"github.com/gohugoio/hugo/markup/converter"
"github.com/gohugoio/hugo/markup/goldmark/goldmark_config"
"github.com/gohugoio/hugo/markup/highlight"
"github.com/gohugoio/hugo/markup/tableofcontents"
"github.com/yuin/goldmark"
@@ -57,7 +56,7 @@ func (p provide) New(cfg converter.ProviderConfig) (converter.Provider, error) {
cfg: cfg,
md: md,
sanitizeAnchorName: func(s string) string {
return sanitizeAnchorNameString(s, cfg.MarkupConfig.Goldmark.Parser.AutoHeadingIDType == goldmark_config.AutoHeadingIDTypeGitHub)
return sanitizeAnchorNameString(s, cfg.MarkupConfig.Goldmark.Parser.AutoHeadingIDType)
},
}, nil
}), nil

15
markup/goldmark/convert_test.go
View File

@@ -178,6 +178,21 @@ func TestConvertAutoIDAsciiOnly(t *testing.T) {
c.Assert(got, qt.Contains, "<h2 id=\"god-is-good-\">")
}
func TestConvertAutoIDBlackfriday(t *testing.T) {
c := qt.New(t)
content := `
## Let's try this, shall we?
`
mconf := markup_config.Default
mconf.Goldmark.Parser.AutoHeadingIDType = goldmark_config.AutoHeadingIDTypeBlackfriday
b := convert(c, mconf, content)
got := string(b.Bytes())
c.Assert(got, qt.Contains, "<h2 id=\"let-s-try-this-shall-we\">")
}
func TestCodeFence(t *testing.T) {
c := qt.New(t)

1
markup/goldmark/goldmark_config/config.go
View File

@@ -17,6 +17,7 @@ package goldmark_config
const (
AutoHeadingIDTypeGitHub = "github"
AutoHeadingIDTypeGitHubAscii = "github-ascii"
AutoHeadingIDTypeBlackfriday = "blackfriday"
)
// DefaultConfig holds the default Goldmark configuration.