feat: support custom scopes with engine: tree-stitter

errata-ai · Jan 10, 2025 · eeaa7e3 · eeaa7e3
1 parent f65c0bf
commit eeaa7e3
Show file tree

Hide file tree

Showing 25 changed files with 82 additions and 46 deletions.
diff --git a/internal/check/definition.go b/internal/check/definition.go
@@ -408,9 +408,12 @@ func checkScopes(scopes []string, path string) error {
 				fmt.Sprintf("scope '%v' is no longer supported; use 'raw' instead.", scope),
 				"scope",
 				path)
-		} else if !core.StringInSlice(scope, allowedScopes) {
+		}
+
+		// No spaces
+		if strings.Contains(scope, " ") {
 			return core.NewE201FromTarget(
-				fmt.Sprintf("'%v' is not a valid scope; must be one of %v", scope, allowedScopes),
+				fmt.Sprintf("scope '%v' contains spaces.", scope),
 				"scope",
 				path)
 		}

diff --git a/internal/lint/code.go b/internal/lint/code.go
@@ -13,20 +13,19 @@ import (
 	"github.com/errata-ai/vale/v3/internal/nlp"
 )
 
-func updateQueries(f *core.File, blueprints map[string]*core.Blueprint) ([]string, error) {
+func updateQueries(f *core.File, blueprints map[string]*core.Blueprint) ([]core.Scope, error) {
+	var found []core.Scope
+
 	for syntax, blueprint := range blueprints {
 		sec, err := glob.Compile(syntax)
 		if err != nil {
 			return nil, err
 		} else if sec.Match(f.Path) {
-			found := []string{}
-			for _, query := range blueprint.Scopes {
-				found = append(found, query.Expr)
-			}
-			return found, nil
+			found = blueprint.Scopes
 		}
 	}
-	return []string{}, nil
+
+	return found, nil
 }
 
 func (l *Linter) lintCode(f *core.File) error {
@@ -52,6 +51,7 @@ func (l *Linter) lintCode(f *core.File) error {
 
 	last := 0
 	for _, comment := range comments {
+		l.SetMetaScope(comment.Scope)
 		if core.StringInSlice("comment", ignored) {
 			continue
 		} else if core.StringInSlice(comment.Scope, ignored) {

diff --git a/internal/lint/code/c.go b/internal/lint/code/c.go
@@ -3,14 +3,17 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/c"
 )
 
 func C() *Language {
 	return &Language{
-		Delims:  regexp.MustCompile(`//|/\*|\*/`),
-		Parser:  c.GetLanguage(),
-		Queries: []string{`(comment)+ @comment`},
+		Delims: regexp.MustCompile(`//|/\*|\*/`),
+		Parser: c.GetLanguage(),
+		Queries: []core.Scope{
+			{Name: "", Expr: "(comment) @comment", Type: ""},
+		},
 		Padding: cStyle,
 	}
 }
diff --git a/internal/lint/code/comments.go b/internal/lint/code/comments.go
@@ -110,11 +110,11 @@ func GetComments(source []byte, lang *Language) ([]Comment, error) {
 	engine := NewQueryEngine(tree, lang)
 
 	for _, query := range lang.Queries {
-		q, qErr := sitter.NewQuery([]byte(query), lang.Parser)
+		q, qErr := sitter.NewQuery([]byte(query.Expr), lang.Parser)
 		if qErr != nil {
 			return comments, qErr
 		}
-		comments = append(comments, engine.run(q, source)...)
+		comments = append(comments, engine.run(query.Name, q, source)...)
 	}
 
 	if len(lang.Queries) > 1 {

diff --git a/internal/lint/code/cpp.go b/internal/lint/code/cpp.go
@@ -3,14 +3,15 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/cpp"
 )
 
 func Cpp() *Language {
 	return &Language{
 		Delims:  regexp.MustCompile(`//|/\*!?|\*/`),
 		Parser:  cpp.GetLanguage(),
-		Queries: []string{`(comment)+ @comment`},
+		Queries: []core.Scope{{Name: "", Expr: "(comment) @comment", Type: ""}},
 		Padding: cStyle,
 	}
 }
diff --git a/internal/lint/code/css.go b/internal/lint/code/css.go
@@ -3,14 +3,15 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/css"
 )
 
 func CSS() *Language {
 	return &Language{
 		Delims:  regexp.MustCompile(`/\*!?|\*/`),
 		Parser:  css.GetLanguage(),
-		Queries: []string{`(comment)+ @comment`},
+		Queries: []core.Scope{{Name: "", Expr: "(comment) @comment", Type: ""}},
 		Padding: func(s string) int {
 			return computePadding(s, []string{"/*"})
 		},

diff --git a/internal/lint/code/go.go b/internal/lint/code/go.go
@@ -3,14 +3,15 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/golang"
 )
 
 func Go() *Language {
 	return &Language{
 		Delims:  regexp.MustCompile(`//|/\*|\*/`),
 		Parser:  golang.GetLanguage(),
-		Queries: []string{`(comment)+ @comment`},
+		Queries: []core.Scope{{Name: "", Expr: "(comment) @comment", Type: ""}},
 		Padding: cStyle,
 	}
 }
diff --git a/internal/lint/code/jl.go b/internal/lint/code/jl.go
@@ -3,16 +3,17 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/jdkato/go-tree-sitter-julia/julia"
 )
 
 func Julia() *Language {
 	return &Language{
 		Delims: regexp.MustCompile(`#|#=|=#`),
 		Parser: julia.GetLanguage(),
-		Queries: []string{
-			`(line_comment)+ @comment`,
-			`(block_comment)+ @comment`,
+		Queries: []core.Scope{
+			{Name: "", Expr: "(line_comment)+ @comment", Type: ""},
+			{Name: "", Expr: "(block_comment)+ @comment", Type: ""},
 		},
 		Padding: func(s string) int {
 			return computePadding(s, []string{"#", `#=`, `=#`})

diff --git a/internal/lint/code/js.go b/internal/lint/code/js.go
@@ -3,6 +3,7 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/javascript"
 )
 
@@ -11,7 +12,7 @@ func JavaScript() *Language {
 		Delims: regexp.MustCompile(`//|/\*\*?|\*/`),
 		Parser: javascript.GetLanguage(),
 		//Cutset:  " *",
-		Queries: []string{`(comment)+ @comment`},
+		Queries: []core.Scope{{Name: "", Expr: "(comment) @comment", Type: ""}},
 		Padding: cStyle,
 	}
 }
diff --git a/internal/lint/code/lang.go b/internal/lint/code/lang.go
@@ -16,7 +16,7 @@ type padding func(string) int
 type Language struct {
 	Delims  *regexp.Regexp
 	Parser  *sitter.Language
-	Queries []string
+	Queries []core.Scope
 	Cutset  string
 	Padding padding
 }

diff --git a/internal/lint/code/proto.go b/internal/lint/code/proto.go
@@ -3,14 +3,15 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/protobuf"
 )
 
 func Protobuf() *Language {
 	return &Language{
 		Delims:  regexp.MustCompile(`//|/\*|\*/`),
 		Parser:  protobuf.GetLanguage(),
-		Queries: []string{`(comment)+ @comment`},
+		Queries: []core.Scope{{Name: "", Expr: "(comment) @comment", Type: ""}},
 		Padding: cStyle,
 	}
 }
diff --git a/internal/lint/code/py.go b/internal/lint/code/py.go
@@ -3,26 +3,27 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/python"
 )
 
 func Python() *Language {
 	return &Language{
 		Delims: regexp.MustCompile(`#|"""|'''`),
 		Parser: python.GetLanguage(),
-		Queries: []string{
-			`(comment)+ @comment`,
-			// Function docstring
-			`((function_definition
+		Queries: []core.Scope{
+			{Name: "", Expr: `(comment)+ @comment`, Type: ""},
+			// Function docstrings
+			{Name: "", Expr: `((function_definition
   body: (block . (expression_statement (string) @docstring)))
- (#offset! @docstring 0 3 0 -3))`,
-			// Class docstring
-			`((class_definition
+ (#offset! @docstring 0 3 0 -3))`, Type: ""},
+			// Class docstrings
+			{Name: "", Expr: `((class_definition
   body: (block . (expression_statement (string) @docstring)))
- (#offset! @docstring 0 3 0 -3))`,
-			// Module docstring
-			`((module . (expression_statement (string) @docstring))
- (#offset! @docstring 0 3 0 -3))`,
+ (#offset! @docstring 0 3 0 -3))`, Type: ""},
+			// Module docstrings
+			{Name: "", Expr: `((module . (expression_statement (string) @docstring))
+ (#offset! @docstring 0 3 0 -3))`, Type: ""},
 		},
 		Padding: func(s string) int {
 			return computePadding(s, []string{"#", `"""`, "'''"})

diff --git a/internal/lint/code/query.go b/internal/lint/code/query.go
@@ -26,9 +26,13 @@ func NewQueryEngine(tree *sitter.Tree, lang *Language) *QueryEngine {
 	}
 }
 
-func (qe *QueryEngine) run(q *sitter.Query, source []byte) []Comment {
+func (qe *QueryEngine) run(meta string, q *sitter.Query, source []byte) []Comment {
 	var comments []Comment
 
+	if meta != "" {
+		meta = "." + meta
+	}
+
 	qc := sitter.NewQueryCursor()
 	qc.Exec(q, qe.tree.RootNode())
 
@@ -43,9 +47,9 @@ func (qe *QueryEngine) run(q *sitter.Query, source []byte) []Comment {
 			rText := c.Node.Content(source)
 			cText := qe.lang.Delims.ReplaceAllString(rText, "")
 
-			scope := "text.comment.line"
+			scope := "text.comment" + meta + ".line"
 			if strings.Count(cText, "\n") > 1 {
-				scope = "text.comment.block"
+				scope = "text.comment" + meta + ".block"
 
 				buf := bytes.Buffer{}
 				for _, line := range strings.Split(cText, "\n") {

diff --git a/internal/lint/code/rb.go b/internal/lint/code/rb.go
@@ -3,14 +3,15 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/ruby"
 )
 
 func Ruby() *Language {
 	return &Language{
 		Delims:  regexp.MustCompile(`#|=begin|=end`),
 		Parser:  ruby.GetLanguage(),
-		Queries: []string{`(comment)+ @comment`},
+		Queries: []core.Scope{{Name: "", Expr: "(comment) @comment", Type: ""}},
 		Padding: func(s string) int {
 			return computePadding(s, []string{"#", `=begin`, `=end`})
 		},

diff --git a/internal/lint/code/rs.go b/internal/lint/code/rs.go
@@ -3,14 +3,15 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/rust"
 )
 
 func Rust() *Language {
 	return &Language{
 		Delims:  regexp.MustCompile(`/{2,3}!?`),
 		Parser:  rust.GetLanguage(),
-		Queries: []string{`(line_comment)+ @comment`},
+		Queries: []core.Scope{{Name: "", Expr: `(line_comment)+ @comment`, Type: ""}},
 		Padding: func(s string) int {
 			return computePadding(s, []string{"//", "//!", "///"})
 		},

diff --git a/internal/lint/code/ts.go b/internal/lint/code/ts.go
@@ -3,14 +3,15 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/typescript/typescript"
 )
 
 func TypeScript() *Language {
 	return &Language{
 		Delims:  regexp.MustCompile(`//|/\*|\*/`),
 		Parser:  typescript.GetLanguage(),
-		Queries: []string{`(comment)+ @comment`},
+		Queries: []core.Scope{{Name: "", Expr: "(comment) @comment", Type: ""}},
 		Padding: cStyle,
 	}
 }
diff --git a/internal/lint/code/tsx.go b/internal/lint/code/tsx.go
@@ -3,14 +3,15 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/typescript/tsx"
 )
 
 func Tsx() *Language {
 	return &Language{
 		Delims:  regexp.MustCompile(`//|/\*|\*/`),
 		Parser:  tsx.GetLanguage(),
-		Queries: []string{`(comment)+ @comment`},
+		Queries: []core.Scope{{Name: "", Expr: "(comment) @comment", Type: ""}},
 		Padding: cStyle,
 	}
 }
diff --git a/internal/lint/code/yaml.go b/internal/lint/code/yaml.go
@@ -3,14 +3,15 @@ package code
 import (
 	"regexp"
 
+	"github.com/errata-ai/vale/v3/internal/core"
 	"github.com/smacker/go-tree-sitter/yaml"
 )
 
 func YAML() *Language {
 	return &Language{
 		Delims:  regexp.MustCompile(`#`),
 		Parser:  yaml.GetLanguage(),
-		Queries: []string{`(comment)+ @comment`},
+		Queries: []core.Scope{{Name: "", Expr: "(comment) @comment", Type: ""}},
 		Padding: func(s string) int {
 			return computePadding(s, []string{"#"})
 		},

diff --git a/internal/lint/fragment.go b/internal/lint/fragment.go
@@ -72,6 +72,7 @@ func (l *Linter) lintFragments(f *core.File) error {
 
 	last := 0
 	for _, comment := range comments {
+		l.SetMetaScope(comment.Scope)
 		f.SetText(comment.Text)
 
 		switch f.NormedExt {

diff --git a/testdata/features/blueprints.feature b/testdata/features/blueprints.feature
@@ -3,12 +3,13 @@ Feature: Blueprints
         When I test "blueprints"
         Then the output should contain exactly:
             """
-            API.yml:3:10:OpenAPI.Titles:'sample API' should be capitalized
+            API.yml:3:10:Scopes.Titles:'sample API' should be capitalized
             API.yml:4:25:Vale.Spelling:Did you really mean 'multiline'?
             API.yml:9:70:Vale.Spelling:Did you really mean 'serrver'?
             API.yml:13:17:Vale.Spelling:Did you really mean 'serrver'?
             API.yml:15:70:Vale.Spelling:Did you really mean 'serrver'?
             Rule.yml:3:39:Vale.Repetition:'can' is repeated!
+            test.py:1:3:Scopes.Code:'FIXME' should not be capitalized
             test.py:1:3:vale.Annotations:'FIXME' left in text
             test.py:11:3:vale.Annotations:'XXX' left in text
             test.py:13:16:vale.Annotations:'XXX' left in text

diff --git a/testdata/fixtures/blueprints/.vale.ini b/testdata/fixtures/blueprints/.vale.ini
@@ -1,13 +1,17 @@
 StylesPath = ../../styles
 MinAlertLevel = suggestion
 
+[formats]
+py = md
+
 [*.py]
 vale.Annotations = YES
+Scopes.Code = YES
 
 Blueprint = Python
 
 [*.yml]
-BasedOnStyles = Vale, OpenAPI
+BasedOnStyles = Vale, Scopes
 
 [API.yml]
 Blueprint = OpenAPI