Port formatter, enable in LSP

internal/ast/kind.go

@@ -423,4 +423,6 @@ const (
 	KindFirstContextualKeyword  = KindAbstractKeyword
 	KindLastContextualKeyword   = KindOfKeyword
 	KindComment                 = KindSingleLineCommentTrivia | KindMultiLineCommentTrivia
+	KindFirstTriviaToken        = KindSingleLineCommentTrivia
+	KindLastTriviaToken         = KindConflictMarkerTrivia
 )

internal/ast/utilities.go

@@ -126,6 +126,10 @@ func IsModifier(node *Node) bool {
 	return IsModifierKind(node.Kind)
 }
 
+func IsModifierLike(node *Node) bool {
+	return IsModifier(node) || IsDecorator(node)
+}
+
 func IsKeywordKind(token Kind) bool {
 	return KindFirstKeyword <= token && token <= KindLastKeyword
 }
@@ -3471,3 +3475,11 @@ func IsInitializedProperty(member *ClassElement) bool {
 	return member.Kind == KindPropertyDeclaration &&
 		member.Initializer() != nil
 }
+
+func IsTrivia(token Kind) bool {
+	return KindFirstTriviaToken <= token && token <= KindLastTriviaToken
+}
+
+func HasDecorators(node *Node) bool {
+	return HasSyntacticModifier(node, ModifierFlagsDecorator)
+}

internal/checker/checker.go

@@ -5654,7 +5654,7 @@ func (c *Checker) checkVarDeclaredNamesNotShadowed(node *ast.Node) {
 func (c *Checker) checkDecorators(node *ast.Node) {
 	// skip this check for nodes that cannot have decorators. These should have already had an error reported by
 	// checkGrammarModifiers.
-	if !ast.CanHaveDecorators(node) || !hasDecorators(node) || !nodeCanBeDecorated(c.legacyDecorators, node, node.Parent, node.Parent.Parent) {
+	if !ast.CanHaveDecorators(node) || !ast.HasDecorators(node) || !nodeCanBeDecorated(c.legacyDecorators, node, node.Parent, node.Parent.Parent) {
 		return
 	}
 	firstDecorator := core.Find(node.ModifierNodes(), ast.IsDecorator)
@@ -11618,7 +11618,7 @@ func (c *Checker) getThisTypeOfDeclaration(declaration *ast.Node) *Type {
 func (c *Checker) checkThisInStaticClassFieldInitializerInDecoratedClass(thisExpression *ast.Node, container *ast.Node) {
 	if ast.IsPropertyDeclaration(container) && ast.HasStaticModifier(container) && c.legacyDecorators {
 		initializer := container.Initializer()
-		if initializer != nil && initializer.Loc.ContainsInclusive(thisExpression.Pos()) && hasDecorators(container.Parent) {
+		if initializer != nil && initializer.Loc.ContainsInclusive(thisExpression.Pos()) && ast.HasDecorators(container.Parent) {
 			c.error(thisExpression, diagnostics.Cannot_use_this_in_a_static_property_initializer_of_a_decorated_class)
 		}
 	}
@@ -26786,7 +26786,7 @@ func (c *Checker) markLinkedReferences(location *ast.Node, hint ReferenceHint, p
 		if !c.compilerOptions.EmitDecoratorMetadata.IsTrue() {
 			return
 		}
-		if !ast.CanHaveDecorators(location) || !hasDecorators(location) || location.Modifiers() == nil || !nodeCanBeDecorated(c.legacyDecorators, location, location.Parent, location.Parent.Parent) {
+		if !ast.CanHaveDecorators(location) || !ast.HasDecorators(location) || location.Modifiers() == nil || !nodeCanBeDecorated(c.legacyDecorators, location, location.Parent, location.Parent.Parent) {
 			return
 		}
 

internal/checker/grammarchecks.go

@@ -233,7 +233,7 @@ func (c *Checker) checkGrammarModifiers(node *ast.Node /*Union[HasModifiers, Has
 				}
 			} else if c.legacyDecorators && (node.Kind == ast.KindGetAccessor || node.Kind == ast.KindSetAccessor) {
 				accessors := c.getAllAccessorDeclarationsForDeclaration(node)
-				if hasDecorators(accessors.firstAccessor) && node == accessors.secondAccessor {
+				if ast.HasDecorators(accessors.firstAccessor) && node == accessors.secondAccessor {
 					return c.grammarErrorOnFirstToken(node, diagnostics.Decorators_cannot_be_applied_to_multiple_get_Slashset_accessors_of_the_same_name)
 				}
 			}

internal/checker/utilities.go

@@ -83,10 +83,6 @@ func hasAsyncModifier(node *ast.Node) bool {
 	return ast.HasSyntacticModifier(node, ast.ModifierFlagsAsync)
 }
 
-func hasDecorators(node *ast.Node) bool {
-	return ast.HasSyntacticModifier(node, ast.ModifierFlagsDecorator)
-}
-
 func getSelectedModifierFlags(node *ast.Node, flags ast.ModifierFlags) ast.ModifierFlags {
 	return node.ModifierFlags() & flags
 }

internal/core/text.go

@@ -54,3 +54,13 @@ func (t TextRange) WithPos(pos int) TextRange {
 func (t TextRange) WithEnd(end int) TextRange {
 	return TextRange{pos: t.pos, end: TextPos(end)}
 }
+
+func (t TextRange) ContainedBy(t2 TextRange) bool {
+	return t2.pos <= t.pos && t2.end >= t.end
+}
+
+func (t TextRange) Overlaps(t2 TextRange) bool {
+	start := max(t.pos, t2.pos)
+	end := min(t.end, t2.end)
+	return start < end
+}

internal/core/textchange.go

@@ -0,0 +1,10 @@
+package core
+
+type TextChange struct {
+	TextRange
+	NewText string
+}
+
+func (t TextChange) ApplyTo(text string) string {
+	return text[:t.Pos()] + t.NewText + text[t.End():]
+}

internal/format/README.md

@@ -0,0 +1,26 @@
+# How does TypeScript formatting work?
+
+To format code you need to have a formatting context and a `SourceFile`.  The formatting context contains 
+all user settings like tab size, newline character, etc.
+
+The end result of formatting is represented by TextChange objects which hold the new string content, and 
+the text to replace it with. 
+
+## Internals
+
+Most of the exposed APIs internally are `Format*` and they all set up and configure `FormatSpan` which could be considered the root call for formatting. Span in this case refers to the range of 
+the sourcefile which should be formatted. 
+
+The formatSpan then uses a scanner (either with or without JSX support) which starts at the highest
+node the covers the span of text and recurses down through the node's children.
+
+As it recurses, `processNode` is called on the children setting the indentation is decided and passed 
+through into each of that node's children.
+
+The meat of formatting decisions is made via `processPair`, the pair here being the current node and the previous node. `processPair` which mutates the formatting context to represent the current place in the scanner and requests a set of rules which can be applied to the items via `createRulesMap`.
+
+There are a lot of rules, which you can find in [rules.ts](./rules.ts) each one has a left and right reference to nodes or token ranges and note of what action should be applied by the formatter.
+
+### Where is this used?
+
+The formatter is used mainly from any language service operation that inserts or modifies code.  The formatter is not exported publicly, and so all usage can only come through the language server.

internal/format/api.go

@@ -0,0 +1,161 @@
+package format
+
+import (
+	"context"
+	"unicode/utf8"
+
+	"github.com/microsoft/typescript-go/internal/ast"
+	"github.com/microsoft/typescript-go/internal/core"
+	"github.com/microsoft/typescript-go/internal/scanner"
+	"github.com/microsoft/typescript-go/internal/stringutil"
+)
+
+type FormatRequestKind int
+
+const (
+	FormatRequestKindFormatDocument FormatRequestKind = iota
+	FormatRequestKindFormatSelection
+	FormatRequestKindFormatOnEnter
+	FormatRequestKindFormatOnSemicolon
+	FormatRequestKindFormatOnOpeningCurlyBrace
+	FormatRequestKindFormatOnClosingCurlyBrace
+)
+
+type formatContextKey int
+
+const (
+	formatOptionsKey formatContextKey = iota
+	formatNewlineKey
+)
+
+func NewContext(ctx context.Context, options *FormatCodeSettings, newLine string) context.Context {
+	ctx = context.WithValue(ctx, formatOptionsKey, options)
+	ctx = context.WithValue(ctx, formatNewlineKey, newLine)
+	// In strada, the rules map was both globally cached *and* cached into the context, for some reason. We skip that here and just use the global one.
+	return ctx
+}
+
+func getNewLineOrDefaultFromContext(ctx context.Context) string { // TODO: Move into broader LS - more than just the formatter uses the newline editor setting/host new line
+	opt := ctx.Value(formatOptionsKey).(*FormatCodeSettings)
+	if opt != nil && len(opt.NewLineCharacter) > 0 {
+		return opt.NewLineCharacter
+	}
+	host := ctx.Value(formatNewlineKey).(string)
+	if len(host) > 0 {
+		return host
+	}
+	return "\n"
+}
+
+func FormatSpan(ctx context.Context, span core.TextRange, file *ast.SourceFile, kind FormatRequestKind) []core.TextChange {
+	// find the smallest node that fully wraps the range and compute the initial indentation for the node
+	enclosingNode := findEnclosingNode(span, file)
+	opts := ctx.Value(formatOptionsKey).(*FormatCodeSettings)
+
+	return newFormattingScanner(
+		file.Text(),
+		file.LanguageVariant,
+		getScanStartPosition(enclosingNode, span, file),
+		span.End(),
+		newFormatSpanWorker(
+			ctx,
+			span,
+			enclosingNode,
+			GetIndentationForNode(enclosingNode, &span, file, opts),
+			getOwnOrInheritedDelta(enclosingNode, opts, file),
+			kind,
+			prepareRangeContainsErrorFunction(file.Diagnostics(), span),
+			file,
+		),
+	)
+}
+
+func formatNodeLines(ctx context.Context, sourceFile *ast.SourceFile, node *ast.Node, requestKind FormatRequestKind) []core.TextChange {
+	if node == nil {
+		return nil
+	}
+	tokenStart := scanner.GetTokenPosOfNode(node, sourceFile, false)
+	lineStart := getLineStartPositionForPosition(tokenStart, sourceFile)
+	span := core.NewTextRange(lineStart, node.End())
+	return FormatSpan(ctx, span, sourceFile, requestKind)
+}
+
+func FormatDocument(ctx context.Context, sourceFile *ast.SourceFile) []core.TextChange {
+	return FormatSpan(ctx, core.NewTextRange(0, sourceFile.End()), sourceFile, FormatRequestKindFormatDocument)
+}
+
+func FormatSelection(ctx context.Context, sourceFile *ast.SourceFile, start int, end int) []core.TextChange {
+	return FormatSpan(ctx, core.NewTextRange(getLineStartPositionForPosition(start, sourceFile), end), sourceFile, FormatRequestKindFormatSelection)
+}
+
+func FormatOnOpeningCurly(ctx context.Context, sourceFile *ast.SourceFile, position int) []core.TextChange {
+	openingCurly := findImmediatelyPrecedingTokenOfKind(position, ast.KindOpenBraceToken, sourceFile)
+	if openingCurly == nil {
+		return nil
+	}
+	curlyBraceRange := openingCurly.Parent
+	outermostNode := findOutermostNodeWithinListLevel(curlyBraceRange)
+	/**
+	 * We limit the span to end at the opening curly to handle the case where
+	 * the brace matched to that just typed will be incorrect after further edits.
+	 * For example, we could type the opening curly for the following method
+	 * body without brace-matching activated:
+	 * ```
+	 * class C {
+	 *     foo()
+	 * }
+	 * ```
+	 * and we wouldn't want to move the closing brace.
+	 */
+	textRange := core.NewTextRange(getLineStartPositionForPosition(scanner.GetTokenPosOfNode(outermostNode, sourceFile, false), sourceFile), position)
+	return FormatSpan(ctx, textRange, sourceFile, FormatRequestKindFormatOnOpeningCurlyBrace)
+}
+
+func FormatOnClosingCurly(ctx context.Context, sourceFile *ast.SourceFile, position int) []core.TextChange {
+	precedingToken := findImmediatelyPrecedingTokenOfKind(position, ast.KindCloseBraceToken, sourceFile)
+	return formatNodeLines(ctx, sourceFile, findOutermostNodeWithinListLevel(precedingToken), FormatRequestKindFormatOnClosingCurlyBrace)
+}
+
+func FormatOnSemicolon(ctx context.Context, sourceFile *ast.SourceFile, position int) []core.TextChange {
+	semicolon := findImmediatelyPrecedingTokenOfKind(position, ast.KindSemicolonToken, sourceFile)
+	return formatNodeLines(ctx, sourceFile, findOutermostNodeWithinListLevel(semicolon), FormatRequestKindFormatOnSemicolon)
+}
+
+func FormatOnEnter(ctx context.Context, sourceFile *ast.SourceFile, position int) []core.TextChange {
+	line, _ := scanner.GetLineAndCharacterOfPosition(sourceFile, position)
+	if line == 0 {
+		return nil
+	}
+	// get start position for the previous line
+	startPos := int(scanner.GetLineStarts(sourceFile)[line-1])
+	// After the enter key, the cursor is now at a new line. The new line may or may not contain non-whitespace characters.
+	// If the new line has only whitespaces, we won't want to format this line, because that would remove the indentation as
+	// trailing whitespaces. So the end of the formatting span should be the later one between:
+	//  1. the end of the previous line
+	//  2. the last non-whitespace character in the current line
+	endOfFormatSpan := scanner.GetEndLinePosition(sourceFile, line)
+	for endOfFormatSpan > startPos {
+		ch, s := utf8.DecodeRuneInString(sourceFile.Text()[endOfFormatSpan:])
+		if s == 0 || stringutil.IsWhiteSpaceSingleLine(ch) { // on multibyte character keep backing up
+			endOfFormatSpan--
+			continue
+		}
+		break
+	}
+
+	// if the character at the end of the span is a line break, we shouldn't include it, because it indicates we don't want to
+	// touch the current line at all. Also, on some OSes the line break consists of two characters (\r\n), we should test if the
+	// previous character before the end of format span is line break character as well.
+	ch, _ := utf8.DecodeRuneInString(sourceFile.Text()[endOfFormatSpan:])
+	if stringutil.IsLineBreak(ch) {
+		endOfFormatSpan--
+	}
+
+	span := core.NewTextRange(
+		startPos,
+		// end value is exclusive so add 1 to the result
+		endOfFormatSpan+1,
+	)
+
+	return FormatSpan(ctx, span, sourceFile, FormatRequestKindFormatOnEnter)
+}