markup/goldmark: Add config options for the typographer extension

Note that the config per language part of this will be handled in #10602.

Updates #9772

5 files changed, 139 insertions, 12 deletions

diff --git a/markup/goldmark/convert.go b/markup/goldmark/convert.go
index 3c8dbb299..efcfb7142 100644
--- a/markup/goldmark/convert.go
+++ b/markup/goldmark/convert.go
@@ -20,6 +20,7 @@ import (
 	"github.com/gohugoio/hugo/identity"
 
 	"github.com/gohugoio/hugo/markup/goldmark/codeblocks"
+	"github.com/gohugoio/hugo/markup/goldmark/goldmark_config"
 	"github.com/gohugoio/hugo/markup/goldmark/images"
 	"github.com/gohugoio/hugo/markup/goldmark/internal/extensions/attributes"
 	"github.com/gohugoio/hugo/markup/goldmark/internal/render"
@@ -120,8 +121,11 @@ func newMarkdown(pcfg converter.ProviderConfig) goldmark.Markdown {
 		extensions = append(extensions, extension.TaskList)
 	}
 
-	if cfg.Extensions.Typographer {
-		extensions = append(extensions, extension.Typographer)
+	if !cfg.Extensions.Typographer.Disable {
+		t := extension.NewTypographer(
+			extension.WithTypographicSubstitutions(toTypographicPunctuationMap(cfg.Extensions.Typographer)),
+		)
+		extensions = append(extensions, t)
 	}
 
 	if cfg.Extensions.DefinitionList {
@@ -278,3 +282,21 @@ func (p *parserContext) TableOfContents() *tableofcontents.Fragments {
 	}
 	return nil
 }
+
+// Note: It's tempting to put this in the config package, but that doesn't work.
+// TODO(bep) create upstream issue.
+func toTypographicPunctuationMap(t goldmark_config.Typographer) map[extension.TypographicPunctuation][]byte {
+	return map[extension.TypographicPunctuation][]byte{
+		extension.LeftSingleQuote:  []byte(t.LeftSingleQuote),
+		extension.RightSingleQuote: []byte(t.RightSingleQuote),
+		extension.LeftDoubleQuote:  []byte(t.LeftDoubleQuote),
+		extension.RightDoubleQuote: []byte(t.RightDoubleQuote),
+		extension.EnDash:           []byte(t.EnDash),
+		extension.EmDash:           []byte(t.EmDash),
+		extension.Ellipsis:         []byte(t.Ellipsis),
+		extension.LeftAngleQuote:   []byte(t.LeftAngleQuote),
+		extension.RightAngleQuote:  []byte(t.RightAngleQuote),
+		extension.Apostrophe:       []byte(t.Apostrophe),
+	}
+
+}
diff --git a/markup/goldmark/convert_test.go b/markup/goldmark/convert_test.go
index 647ffce58..e92c651fc 100644
--- a/markup/goldmark/convert_test.go
+++ b/markup/goldmark/convert_test.go
@@ -499,3 +499,18 @@ LINE5
 		c.Assert(result, qt.Contains, "<span class=\"ln\">2</span><span class=\"cl\">LINE2\n</span></span>")
 	})
 }
+
+func TestTypographerConfig(t *testing.T) {
+	c := qt.New(t)
+
+	content := `
+A "quote" and 'another quote' and a "quote with a 'nested' quote" and a 'quote with a "nested" quote' and an ellipsis...
+`
+	mconf := markup_config.Default
+	mconf.Goldmark.Extensions.Typographer.LeftDoubleQuote = "&laquo;"
+	mconf.Goldmark.Extensions.Typographer.RightDoubleQuote = "&raquo;"
+	b := convert(c, mconf, content)
+	got := string(b.Bytes())
+
+	c.Assert(got, qt.Contains, "<p>A &laquo;quote&raquo; and &lsquo;another quote&rsquo; and a &laquo;quote with a &rsquo;nested&rsquo; quote&raquo; and a &lsquo;quote with a &laquo;nested&raquo; quote&rsquo; and an ellipsis&hellip;</p>\n")
+}
diff --git a/markup/goldmark/goldmark_config/config.go b/markup/goldmark/goldmark_config/config.go
index ff0b6bbef..f35427ac1 100644
--- a/markup/goldmark/goldmark_config/config.go
+++ b/markup/goldmark/goldmark_config/config.go
@@ -23,7 +23,19 @@ const (
 // DefaultConfig holds the default Goldmark configuration.
 var Default = Config{
 	Extensions: Extensions{
-		Typographer:     true,
+		Typographer: Typographer{
+			Disable:          false,
+			LeftSingleQuote:  "&lsquo;",
+			RightSingleQuote: "&rsquo;",
+			LeftDoubleQuote:  "&ldquo;",
+			RightDoubleQuote: "&rdquo;",
+			EnDash:           "&ndash;",
+			EmDash:           "&mdash;",
+			Ellipsis:         "&hellip;",
+			LeftAngleQuote:   "&laquo;",
+			RightAngleQuote:  "&raquo;",
+			Apostrophe:       "&rsquo;",
+		},
 		Footnote:        true,
 		DefinitionList:  true,
 		Table:           true,
@@ -54,7 +66,7 @@ type Config struct {
 }
 
 type Extensions struct {
-	Typographer    bool
+	Typographer    Typographer
 	Footnote       bool
 	DefinitionList bool
 
@@ -66,6 +78,33 @@ type Extensions struct {
 	TaskList        bool
 }
 
+// Typographer holds typographer configuration.
+type Typographer struct {
+	// Whether to disable typographer.
+	Disable bool
+
+	// Value used for left single quote.
+	LeftSingleQuote string
+	// Value used for right single quote.
+	RightSingleQuote string
+	// Value used for left double quote.
+	LeftDoubleQuote string
+	// Value used for right double quote.
+	RightDoubleQuote string
+	// Value used for en dash.
+	EnDash string
+	// Value used for em dash.
+	EmDash string
+	// Value used for ellipsis.
+	Ellipsis string
+	// Value used for left angle quote.
+	LeftAngleQuote string
+	// Value used for right angle quote.
+	RightAngleQuote string
+	// Value used for apostrophe.
+	Apostrophe string
+}
+
 type Renderer struct {
 	// Whether softline breaks should be rendered as '<br>'
 	HardWraps bool
diff --git a/markup/markup_config/config.go b/markup/markup_config/config.go
index e254ba7a0..60446b9bc 100644
--- a/markup/markup_config/config.go
+++ b/markup/markup_config/config.go
@@ -62,15 +62,32 @@ func Decode(cfg config.Provider) (conf Config, err error) {
 
 func normalizeConfig(m map[string]any) {
 	v, err := maps.GetNestedParam("goldmark.parser", ".", m)
-	if err != nil {
-		return
+	if err == nil {
+		vm := maps.ToStringMap(v)
+		// Changed from a bool in 0.81.0
+		if vv, found := vm["attribute"]; found {
+			if vvb, ok := vv.(bool); ok {
+				vm["attribute"] = goldmark_config.ParserAttribute{
+					Title: vvb,
+				}
+			}
+		}
 	}
-	vm := maps.ToStringMap(v)
-	// Changed from a bool in 0.81.0
-	if vv, found := vm["attribute"]; found {
-		if vvb, ok := vv.(bool); ok {
-			vm["attribute"] = goldmark_config.ParserAttribute{
-				Title: vvb,
+
+	// Changed from a bool in 0.112.0.
+	v, err = maps.GetNestedParam("goldmark.extensions", ".", m)
+	if err == nil {
+		vm := maps.ToStringMap(v)
+		const typographerKey = "typographer"
+		if vv, found := vm[typographerKey]; found {
+			if vvb, ok := vv.(bool); ok {
+				if !vvb {
+					vm[typographerKey] = goldmark_config.Typographer{
+						Disable: true,
+					}
+				} else {
+					delete(vm, typographerKey)
+				}
 			}
 		}
 	}
diff --git a/markup/markup_config/config_test.go b/markup/markup_config/config_test.go
index a320e6912..782cedbc9 100644
--- a/markup/markup_config/config_test.go
+++ b/markup/markup_config/config_test.go
@@ -52,4 +52,38 @@ func TestConfig(t *testing.T) {
 		c.Assert(conf.AsciidocExt.Extensions[0], qt.Equals, "asciidoctor-html5s")
 	})
 
+	c.Run("Decode legacy typographer", func(c *qt.C) {
+		c.Parallel()
+		v := config.New()
+
+		// typographer was changed from a bool to a struct in 0.112.0.
+		v.Set("markup", map[string]any{
+			"goldmark": map[string]any{
+				"extensions": map[string]any{
+					"typographer": false,
+				},
+			},
+		})
+
+		conf, err := Decode(v)
+
+		c.Assert(err, qt.IsNil)
+		c.Assert(conf.Goldmark.Extensions.Typographer.Disable, qt.Equals, true)
+
+		v.Set("markup", map[string]any{
+			"goldmark": map[string]any{
+				"extensions": map[string]any{
+					"typographer": true,
+				},
+			},
+		})
+
+		conf, err = Decode(v)
+
+		c.Assert(err, qt.IsNil)
+		c.Assert(conf.Goldmark.Extensions.Typographer.Disable, qt.Equals, false)
+		c.Assert(conf.Goldmark.Extensions.Typographer.Ellipsis, qt.Equals, "&hellip;")
+
+	})
+
 }