Add ContentTypes to config

This is an empty struct for now, but we will most likely expand on that.

```
[contentTypes]
  [contentTypes.'text/markdown']
```

The above means that only Markdown will be considered a content type. E.g. HTML will be treated as plain text.

Fixes #12274

Author: bep

common/hreflect/helpers.go

@@ -74,6 +74,16 @@ func IsTruthful(in any) bool {
 	}
 }
 
+// IsMap reports whether v is a map.
+func IsMap(v any) bool {
+	return reflect.ValueOf(v).Kind() == reflect.Map
+}
+
+// IsSlice reports whether v is a slice.
+func IsSlice(v any) bool {
+	return reflect.ValueOf(v).Kind() == reflect.Slice
+}
+
 var zeroType = reflect.TypeOf((*types.Zeroer)(nil)).Elem()
 
 // IsTruthfulValue returns whether the given value has a meaningful truth value.

config/allconfig/allconfig.go

@@ -128,6 +128,9 @@ type Config struct {
 	// <docsmeta>{"identifiers": ["markup"] }</docsmeta>
 	Markup markup_config.Config `mapstructure:"-"`
 
+	// ContentTypes are the media types that's considered content in Hugo.
+	ContentTypes *config.ConfigNamespace[map[string]media.ContentTypeConfig, media.ContentTypes] `mapstructure:"-"`
+
 	// The mediatypes configuration section maps the MIME type (a string) to a configuration object for that type.
 	// <docsmeta>{"identifiers": ["mediatypes"], "refs": ["types:media:type"] }</docsmeta>
 	MediaTypes *config.ConfigNamespace[map[string]media.MediaTypeConfig, media.Types] `mapstructure:"-"`
@@ -433,7 +436,6 @@ func (c *Config) CompileConfig(logger loggers.Logger) error {
 		IgnoredLogs:         ignoredLogIDs,
 		KindOutputFormats:   kindOutputFormats,
 		DefaultOutputFormat: defaultOutputFormat,
-		ContentTypes:        media.DefaultContentTypes.FromTypes(c.MediaTypes.Config),
 		CreateTitle:         helpers.GetTitleFunc(c.TitleCaseStyle),
 		IsUglyURLSection:    isUglyURL,
 		IgnoreFile:          ignoreFile,
@@ -471,7 +473,6 @@ type ConfigCompiled struct {
 	ServerInterface     string
 	KindOutputFormats   map[string]output.Formats
 	DefaultOutputFormat output.Format
-	ContentTypes        media.ContentTypes
 	DisabledKinds       map[string]bool
 	DisabledLanguages   map[string]bool
 	IgnoredLogs         map[string]bool
@@ -839,7 +840,7 @@ func (c *Configs) Init() error {
 	c.Languages = languages
 	c.LanguagesDefaultFirst = languagesDefaultFirst
 
-	c.ContentPathParser = &paths.PathParser{LanguageIndex: languagesDefaultFirst.AsIndexSet(), IsLangDisabled: c.Base.IsLangDisabled, IsContentExt: c.Base.C.ContentTypes.IsContentSuffix}
+	c.ContentPathParser = &paths.PathParser{LanguageIndex: languagesDefaultFirst.AsIndexSet(), IsLangDisabled: c.Base.IsLangDisabled, IsContentExt: c.Base.ContentTypes.Config.IsContentSuffix}
 
 	c.configLangs = make([]config.AllProvider, len(c.Languages))
 	for i, l := range c.LanguagesDefaultFirst {

config/allconfig/allconfig_integration_test.go

@@ -7,6 +7,7 @@ import (
 	qt "github.com/frankban/quicktest"
 	"github.com/gohugoio/hugo/config/allconfig"
 	"github.com/gohugoio/hugo/hugolib"
+	"github.com/gohugoio/hugo/media"
 )
 
 func TestDirsMount(t *testing.T) {
@@ -97,7 +98,7 @@ suffixes = ["html", "xhtml"]
 	b := hugolib.Test(t, files)
 
 	conf := b.H.Configs.Base
-	contentTypes := conf.C.ContentTypes
+	contentTypes := conf.ContentTypes.Config
 
 	b.Assert(contentTypes.HTML.Suffixes(), qt.DeepEquals, []string{"html", "xhtml"})
 	b.Assert(contentTypes.Markdown.Suffixes(), qt.DeepEquals, []string{"md", "mdown", "markdown"})
@@ -215,3 +216,21 @@ weight = 3
 	b := hugolib.Test(t, files)
 	b.Assert(b.H.Configs.LanguageConfigSlice[0].Title, qt.Equals, `TITLE_DE`)
 }
+
+func TestContentTypesDefault(t *testing.T) {
+	files := `
+-- hugo.toml --
+baseURL = "https://example.com"
+
+
+`
+
+	b := hugolib.Test(t, files)
+
+	ct := b.H.Configs.Base.ContentTypes
+	c := ct.Config
+	s := ct.SourceStructure.(map[string]media.ContentTypeConfig)
+
+	b.Assert(c.IsContentFile("foo.md"), qt.Equals, true)
+	b.Assert(len(s), qt.Equals, 6)
+}

config/allconfig/alldecoders.go

@@ -163,6 +163,15 @@ var allDecoderSetups = map[string]decodeWeight{
 			return err
 		},
 	},
+	"contenttypes": {
+		key:    "contenttypes",
+		weight: 100, // This needs to be decoded after media types.
+		decode: func(d decodeWeight, p decodeConfig) error {
+			var err error
+			p.c.ContentTypes, err = media.DecodeContentTypes(p.p.GetStringMap(d.key), p.c.MediaTypes.Config)
+			return err
+		},
+	},
 	"mediatypes": {
 		key: "mediatypes",
 		decode: func(d decodeWeight, p decodeConfig) error {

config/allconfig/configlanguage.go

@@ -145,7 +145,7 @@ func (c ConfigLanguage) NewIdentityManager(name string, opts ...identity.Manager
 }
 
 func (c ConfigLanguage) ContentTypes() config.ContentTypesProvider {
-	return c.config.C.ContentTypes
+	return c.config.ContentTypes.Config
 }
 
 // GetConfigSection is mostly used in tests. The switch statement isn't complete, but what's in use.

hugolib/content_map_test.go

@@ -501,3 +501,51 @@ func (n *testContentNode) resetBuildState() {
 
 func (n *testContentNode) MarkStale() {
 }
+
+// Issue 12274.
+func TestHTMLNotContent(t *testing.T) {
+	filesTemplate := `
+-- hugo.toml.temp --
+[contentTypes]
+[contentTypes."text/markdown"]
+# Emopty for now.
+-- hugo.yaml.temp --
+contentTypes:
+  text/markdown: {}
+-- hugo.json.temp --
+{
+  "contentTypes": {
+    "text/markdown": {}
+  }
+}
+-- content/p1/index.md --
+---
+title: p1
+---
+-- content/p1/a.html --
+<p>a</p>
+-- content/p1/b.html --
+<p>b</p>
+-- content/p1/c.html --
+<p>c</p>
+-- layouts/_default/single.html --
+|{{ (.Resources.Get "a.html").RelPermalink -}}
+|{{ (.Resources.Get "b.html").RelPermalink -}}
+|{{ (.Resources.Get "c.html").Publish }}
+`
+
+	for _, format := range []string{"toml", "yaml", "json"} {
+		format := format
+		t.Run(format, func(t *testing.T) {
+			t.Parallel()
+
+			files := strings.Replace(filesTemplate, format+".temp", format, 1)
+			b := Test(t, files)
+
+			b.AssertFileContent("public/p1/index.html", "|/p1/a.html|/p1/b.html|")
+			b.AssertFileContent("public/p1/a.html", "<p>a</p>")
+			b.AssertFileContent("public/p1/b.html", "<p>b</p>")
+			b.AssertFileContent("public/p1/c.html", "<p>c</p>")
+		})
+	}
+}

media/config.go

@@ -71,11 +71,15 @@ func init() {
 		EmacsOrgMode:     Builtin.EmacsOrgModeType,
 	}
 
-	DefaultContentTypes.init()
+	DefaultContentTypes.init(nil)
 }
 
 var DefaultContentTypes ContentTypes
 
+type ContentTypeConfig struct {
+	// Empty for now.
+}
+
 // ContentTypes holds the media types that are considered content in Hugo.
 type ContentTypes struct {
 	HTML             Type
@@ -85,13 +89,36 @@ type ContentTypes struct {
 	ReStructuredText Type
 	EmacsOrgMode     Type
 
+	types Types
+
 	// Created in init().
-	types        Types
 	extensionSet map[string]bool
 }
 
-func (t *ContentTypes) init() {
-	t.types = Types{t.HTML, t.Markdown, t.AsciiDoc, t.Pandoc, t.ReStructuredText, t.EmacsOrgMode}
+func (t *ContentTypes) init(types Types) {
+	sort.Slice(t.types, func(i, j int) bool {
+		return t.types[i].Type < t.types[j].Type
+	})
+
+	if tt, ok := types.GetByType(t.HTML.Type); ok {
+		t.HTML = tt
+	}
+	if tt, ok := types.GetByType(t.Markdown.Type); ok {
+		t.Markdown = tt
+	}
+	if tt, ok := types.GetByType(t.AsciiDoc.Type); ok {
+		t.AsciiDoc = tt
+	}
+	if tt, ok := types.GetByType(t.Pandoc.Type); ok {
+		t.Pandoc = tt
+	}
+	if tt, ok := types.GetByType(t.ReStructuredText.Type); ok {
+		t.ReStructuredText = tt
+	}
+	if tt, ok := types.GetByType(t.EmacsOrgMode.Type); ok {
+		t.EmacsOrgMode = tt
+	}
+
 	t.extensionSet = make(map[string]bool)
 	for _, mt := range t.types {
 		for _, suffix := range mt.Suffixes() {
@@ -135,32 +162,6 @@ func (t ContentTypes) Types() Types {
 	return t.types
 }
 
-// FromTypes creates a new ContentTypes updated with the values from the given Types.
-func (t ContentTypes) FromTypes(types Types) ContentTypes {
-	if tt, ok := types.GetByType(t.HTML.Type); ok {
-		t.HTML = tt
-	}
-	if tt, ok := types.GetByType(t.Markdown.Type); ok {
-		t.Markdown = tt
-	}
-	if tt, ok := types.GetByType(t.AsciiDoc.Type); ok {
-		t.AsciiDoc = tt
-	}
-	if tt, ok := types.GetByType(t.Pandoc.Type); ok {
-		t.Pandoc = tt
-	}
-	if tt, ok := types.GetByType(t.ReStructuredText.Type); ok {
-		t.ReStructuredText = tt
-	}
-	if tt, ok := types.GetByType(t.EmacsOrgMode.Type); ok {
-		t.EmacsOrgMode = tt
-	}
-
-	t.init()
-
-	return t
-}
-
 // Hold the configuration for a given media type.
 type MediaTypeConfig struct {
 	// The file suffixes used for this media type.
@@ -169,6 +170,58 @@ type MediaTypeConfig struct {
 	Delimiter string
 }
 
+var defaultContentTypesConfig = map[string]ContentTypeConfig{
+	Builtin.HTMLType.Type:             {},
+	Builtin.MarkdownType.Type:         {},
+	Builtin.AsciiDocType.Type:         {},
+	Builtin.PandocType.Type:           {},
+	Builtin.ReStructuredTextType.Type: {},
+	Builtin.EmacsOrgModeType.Type:     {},
+}
+
+// DecodeContentTypes decodes the given map of content types.
+func DecodeContentTypes(in map[string]any, types Types) (*config.ConfigNamespace[map[string]ContentTypeConfig, ContentTypes], error) {
+	buildConfig := func(v any) (ContentTypes, any, error) {
+		var s map[string]ContentTypeConfig
+		c := DefaultContentTypes
+		m, err := maps.ToStringMapE(v)
+		if err != nil {
+			return c, nil, err
+		}
+		if len(m) == 0 {
+			s = defaultContentTypesConfig
+		} else {
+			s = make(map[string]ContentTypeConfig)
+			m = maps.CleanConfigStringMap(m)
+			for k, v := range m {
+				var ctc ContentTypeConfig
+				if err := mapstructure.WeakDecode(v, &ctc); err != nil {
+					return c, nil, err
+				}
+				s[k] = ctc
+			}
+		}
+
+		for k := range s {
+			mediaType, found := types.GetByType(k)
+			if !found {
+				return c, nil, fmt.Errorf("unknown media type %q", k)
+			}
+			c.types = append(c.types, mediaType)
+		}
+
+		c.init(types)
+
+		return c, s, nil
+	}
+
+	ns, err := config.DecodeNamespace[map[string]ContentTypeConfig](in, buildConfig)
+	if err != nil {
+		return nil, fmt.Errorf("failed to decode media types: %w", err)
+	}
+	return ns, nil
+}
+
 // DecodeTypes decodes the given map of media types.
 func DecodeTypes(in map[string]any) (*config.ConfigNamespace[map[string]MediaTypeConfig, Types], error) {
 	buildConfig := func(v any) (Types, any, error) {
@@ -220,6 +273,6 @@ func DecodeTypes(in map[string]any) (*config.ConfigNamespace[map[string]MediaTyp
 // TODO(bep) get rid of this.
 var DefaultPathParser = &paths.PathParser{
 	IsContentExt: func(ext string) bool {
-		return DefaultContentTypes.IsContentSuffix(ext)
+		panic("not supported")
 	},
 }

media/mediaType_test.go

@@ -214,11 +214,3 @@ func BenchmarkTypeOps(b *testing.B) {
 
 	}
 }
-
-func TestIsContentFile(t *testing.T) {
-	c := qt.New(t)
-
-	c.Assert(DefaultContentTypes.IsContentFile(filepath.FromSlash("my/file.md")), qt.Equals, true)
-	c.Assert(DefaultContentTypes.IsContentFile(filepath.FromSlash("my/file.ad")), qt.Equals, true)
-	c.Assert(DefaultContentTypes.IsContentFile(filepath.FromSlash("textfile.txt")), qt.Equals, false)
-}

parser/lowercase_camel_json.go

@@ -107,7 +107,7 @@ func (c ReplacingJSONMarshaller) MarshalJSON() ([]byte, error) {
 		var removeZeroVAlues func(m map[string]any)
 		removeZeroVAlues = func(m map[string]any) {
 			for k, v := range m {
-				if !hreflect.IsTruthful(v) {
+				if !hreflect.IsMap(v) && !hreflect.IsTruthful(v) {
 					delete(m, k)
 				} else {
 					switch vv := v.(type) {

resources/page/page_nop.go

@@ -21,7 +21,6 @@ import (
 	"html/template"
 	"time"
 
-	"github.com/gohugoio/hugo/hugofs/files"
 	"github.com/gohugoio/hugo/markup/converter"
 	"github.com/gohugoio/hugo/markup/tableofcontents"
 
@@ -59,8 +58,6 @@ var (
 // PageNop implements Page, but does nothing.
 type nopPage int
 
-var noOpPathInfo = media.DefaultPathParser.Parse(files.ComponentFolderContent, "no-op.md")
-
 func (p *nopPage) Aliases() []string {
 	return nil
 }
@@ -338,7 +335,7 @@ func (p *nopPage) Path() string {
 }
 
 func (p *nopPage) PathInfo() *paths.Path {
-	return noOpPathInfo
+	return nil
 }
 
 func (p *nopPage) Permalink() string {

source/fileInfo.go

@@ -132,6 +132,7 @@ func (fi *File) p() *paths.Path {
 	return fi.fim.Meta().PathInfo.Unnormalized()
 }
 
+// Used in tests.
 func NewFileInfoFrom(path, filename string) *File {
 	meta := &hugofs.FileMeta{
 		Filename: filename,