Merge pull request 'Linguist support enhancements from gitea#29267 & bugfixes' (#2487) from algernon/forgejo:linguist/gitea-29267 into forgejo

Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/2487
Reviewed-by: Earl Warren <earl-warren@noreply.codeberg.org>

modules/git/repo_attribute.go

@@ -11,6 +11,7 @@ import (
 	"os"
 
 	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/optional"
 )
 
 // CheckAttributeOpts represents the possible options to CheckAttribute
@@ -322,3 +323,16 @@ func (repo *Repository) CheckAttributeReader(commitID string) (*CheckAttributeRe
 
 	return checker, deferable
 }
+
+// true if "set"/"true", false if "unset"/"false", none otherwise
+func attributeToBool(attr map[string]string, name string) optional.Option[bool] {
+	if value, has := attr[name]; has && value != "unspecified" {
+		switch value {
+		case "set", "true":
+			return optional.Some(true)
+		case "unset", "false":
+			return optional.Some(false)
+		}
+	}
+	return optional.None[bool]()
+}

modules/git/repo_language_stats.go

@@ -13,18 +13,6 @@ const (
 	bigFileSize   int64 = 1024 * 1024 // 1 MiB
 )
 
-type LinguistBoolAttrib struct {
-	Value string
-}
-
-func (attrib *LinguistBoolAttrib) IsTrue() bool {
-	return attrib.Value == "set" || attrib.Value == "true"
-}
-
-func (attrib *LinguistBoolAttrib) IsFalse() bool {
-	return attrib.Value == "unset" || attrib.Value == "false"
-}
-
 // mergeLanguageStats mergers language names with different cases. The name with most upper case letters is used.
 func mergeLanguageStats(stats map[string]int64) map[string]int64 {
 	names := map[string]struct {

modules/git/repo_language_stats_gogit.go

@@ -12,6 +12,7 @@ import (
 	"strings"
 
 	"code.gitea.io/gitea/modules/analyze"
+	"code.gitea.io/gitea/modules/optional"
 
 	"github.com/go-enry/go-enry/v2"
 	"github.com/go-git/go-git/v5"
@@ -53,31 +54,30 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, err
 	firstExcludedLanguage := ""
 	firstExcludedLanguageSize := int64(0)
 
+	isTrue := func(v optional.Option[bool]) bool {
+		return v.ValueOrDefault(false)
+	}
+	isFalse := func(v optional.Option[bool]) bool {
+		return !v.ValueOrDefault(true)
+	}
+
 	err = tree.Files().ForEach(func(f *object.File) error {
 		if f.Size == 0 {
 			return nil
 		}
 
-		isVendored := LinguistBoolAttrib{}
+		isVendored := optional.None[bool]()
-		isGenerated := LinguistBoolAttrib{}
+		isGenerated := optional.None[bool]()
-		isDocumentation := LinguistBoolAttrib{}
+		isDocumentation := optional.None[bool]()
-		isDetectable := LinguistBoolAttrib{}
+		isDetectable := optional.None[bool]()
 
 		if checker != nil {
 			attrs, err := checker.CheckPath(f.Name)
 			if err == nil {
-				if vendored, has := attrs["linguist-vendored"]; has {
+				isVendored = attributeToBool(attrs, "linguist-vendored")
-					isVendored = LinguistBoolAttrib{Value: vendored}
+				isGenerated = attributeToBool(attrs, "linguist-generated")
-				}
+				isDocumentation = attributeToBool(attrs, "linguist-documentation")
-				if generated, has := attrs["linguist-generated"]; has {
+				isDetectable = attributeToBool(attrs, "linguist-detectable")
-					isGenerated = LinguistBoolAttrib{Value: generated}
-				}
-				if documentation, has := attrs["linguist-documentation"]; has {
-					isDocumentation = LinguistBoolAttrib{Value: documentation}
-				}
-				if detectable, has := attrs["linguist-detectable"]; has {
-					isDetectable = LinguistBoolAttrib{Value: detectable}
-				}
 				if language, has := attrs["linguist-language"]; has && language != "unspecified" && language != "" {
 					// group languages, such as Pug -> HTML; SCSS -> CSS
 					group := enry.GetLanguageGroup(language)
@@ -108,11 +108,11 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, err
 			}
 		}
 
-		if isDetectable.IsFalse() || isVendored.IsTrue() || isDocumentation.IsTrue() ||
+		if isFalse(isDetectable) || isTrue(isVendored) || isTrue(isDocumentation) ||
-			(!isVendored.IsFalse() && analyze.IsVendor(f.Name)) ||
+			(!isFalse(isVendored) && analyze.IsVendor(f.Name)) ||
 			enry.IsDotFile(f.Name) ||
 			enry.IsConfiguration(f.Name) ||
-			(!isDocumentation.IsFalse() && enry.IsDocumentation(f.Name)) {
+			(!isFalse(isDocumentation) && enry.IsDocumentation(f.Name)) {
 			return nil
 		}
 
@@ -121,7 +121,7 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, err
 		if f.Size <= bigFileSize {
 			content, _ = readFile(f, fileSizeLimit)
 		}
-		if !isGenerated.IsTrue() && enry.IsGenerated(f.Name, content) {
+		if !isTrue(isGenerated) && enry.IsGenerated(f.Name, content) {
 			return nil
 		}
 
@@ -138,21 +138,22 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, err
 		}
 
 		included, checked := includedLanguage[language]
+		langType := enry.GetLanguageType(language)
 		if !checked {
-			langtype := enry.GetLanguageType(language)
+			included = langType == enry.Programming || langType == enry.Markup
-			included = langtype == enry.Programming || langtype == enry.Markup
+			if !included && (isTrue(isDetectable) || (langType == enry.Prose && isFalse(isDocumentation))) {
-			if !included {
+				included = true
-				if isDetectable.IsTrue() {
-					included = true
-				} else {
-					return nil
-				}
 			}
 			includedLanguage[language] = included
 		}
 		if included {
 			sizes[language] += f.Size
 		} else if len(sizes) == 0 && (firstExcludedLanguage == "" || firstExcludedLanguage == language) {
+			// Only consider Programming or Markup languages as fallback
+			if !(langType == enry.Programming || langType == enry.Markup) {
+				return nil
+			}
+
 			firstExcludedLanguage = language
 			firstExcludedLanguageSize += f.Size
 		}

modules/git/repo_language_stats_nogogit.go

@@ -15,6 +15,7 @@ import (
 
 	"code.gitea.io/gitea/modules/analyze"
 	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/optional"
 
 	"github.com/go-enry/go-enry/v2"
 )
@@ -77,6 +78,13 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, err
 	firstExcludedLanguage := ""
 	firstExcludedLanguageSize := int64(0)
 
+	isTrue := func(v optional.Option[bool]) bool {
+		return v.ValueOrDefault(false)
+	}
+	isFalse := func(v optional.Option[bool]) bool {
+		return !v.ValueOrDefault(true)
+	}
+
 	for _, f := range entries {
 		select {
 		case <-repo.Ctx.Done():
@@ -91,26 +99,18 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, err
 			continue
 		}
 
-		isVendored := LinguistBoolAttrib{}
+		isVendored := optional.None[bool]()
-		isGenerated := LinguistBoolAttrib{}
+		isGenerated := optional.None[bool]()
-		isDocumentation := LinguistBoolAttrib{}
+		isDocumentation := optional.None[bool]()
-		isDetectable := LinguistBoolAttrib{}
+		isDetectable := optional.None[bool]()
 
 		if checker != nil {
 			attrs, err := checker.CheckPath(f.Name())
 			if err == nil {
-				if vendored, has := attrs["linguist-vendored"]; has {
+				isVendored = attributeToBool(attrs, "linguist-vendored")
-					isVendored = LinguistBoolAttrib{Value: vendored}
+				isGenerated = attributeToBool(attrs, "linguist-generated")
-				}
+				isDocumentation = attributeToBool(attrs, "linguist-documentation")
-				if generated, has := attrs["linguist-generated"]; has {
+				isDetectable = attributeToBool(attrs, "linguist-detectable")
-					isGenerated = LinguistBoolAttrib{Value: generated}
-				}
-				if documentation, has := attrs["linguist-documentation"]; has {
-					isDocumentation = LinguistBoolAttrib{Value: documentation}
-				}
-				if detectable, has := attrs["linguist-detectable"]; has {
-					isDetectable = LinguistBoolAttrib{Value: detectable}
-				}
 				if language, has := attrs["linguist-language"]; has && language != "unspecified" && language != "" {
 					// group languages, such as Pug -> HTML; SCSS -> CSS
 					group := enry.GetLanguageGroup(language)
@@ -142,11 +142,11 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, err
 			}
 		}
 
-		if isDetectable.IsFalse() || isVendored.IsTrue() || isDocumentation.IsTrue() ||
+		if isFalse(isDetectable) || isTrue(isVendored) || isTrue(isDocumentation) ||
-			(!isVendored.IsFalse() && analyze.IsVendor(f.Name())) ||
+			(!isFalse(isVendored) && analyze.IsVendor(f.Name())) ||
 			enry.IsDotFile(f.Name()) ||
 			enry.IsConfiguration(f.Name()) ||
-			(!isDocumentation.IsFalse() && enry.IsDocumentation(f.Name())) {
+			(!isFalse(isDocumentation) && enry.IsDocumentation(f.Name())) {
 			continue
 		}
 
@@ -179,7 +179,7 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, err
 				return nil, err
 			}
 		}
-		if !isGenerated.IsTrue() && enry.IsGenerated(f.Name(), content) {
+		if !isTrue(isGenerated) && enry.IsGenerated(f.Name(), content) {
 			continue
 		}
 
@@ -197,25 +197,24 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, err
 		}
 
 		included, checked := includedLanguage[language]
+		langType := enry.GetLanguageType(language)
 		if !checked {
-			langType := enry.GetLanguageType(language)
 			included = langType == enry.Programming || langType == enry.Markup
-			if !included {
+			if !included && (isTrue(isDetectable) || (langType == enry.Prose && isFalse(isDocumentation))) {
-				if isDetectable.IsTrue() {
+				included = true
-					included = true
-				} else {
-					continue
-				}
 			}
 			includedLanguage[language] = included
 		}
 		if included {
 			sizes[language] += f.Size()
 		} else if len(sizes) == 0 && (firstExcludedLanguage == "" || firstExcludedLanguage == language) {
+			// Only consider Programming or Markup languages as fallback
+			if !(langType == enry.Programming || langType == enry.Markup) {
+				continue
+			}
 			firstExcludedLanguage = language
 			firstExcludedLanguageSize += f.Size()
 		}
-		continue
 	}
 
 	// If there are no included languages add the first excluded language

tests/integration/linguist_test.go

@@ -251,5 +251,18 @@ func TestLinguistSupport(t *testing.T) {
 				assertFileLanguage(t, "/blame/branch/main/foo.c", "Bash")
 			})
 		})
+
+		// 10. Marking a file as non-documentation
+		t.Run("linguist-documentation=false", func(t *testing.T) {
+			defer tests.PrintCurrentTest(t)()
+
+			repo, sha, f := prep(t, "README.md linguist-documentation=false\n")
+			defer f()
+
+			langs := getFreshLanguageStats(t, repo, sha)
+			assert.Len(t, langs, 2)
+			assert.Equal(t, "Markdown", langs[0].Language)
+			assert.Equal(t, "C", langs[1].Language)
+		})
 	})
 }