preallocate the line offset table

internal/ast/ast.go

@@ -1260,7 +1260,8 @@ type ImportRecord struct {
 }
 
 type AST struct {
-	HasLazyExport bool
+	ApproximateLineCount int32
+	HasLazyExport        bool
 
 	// This is a list of CommonJS features. When a file uses CommonJS features,
 	// it's not a candidate for "flat bundling" and must be wrapped in its own

internal/lexer/lexer.go

@@ -235,6 +235,7 @@ type Lexer struct {
 	current                         int
 	start                           int
 	end                             int
+	ApproximateNewlineCount         int
 	Token                           T
 	HasNewlineBefore                bool
 	HasPureCommentBefore            bool
@@ -2198,6 +2199,16 @@ func (lexer *Lexer) step() {
 		codePoint = -1
 	}
 
+	// Track the approximate number of newlines in the file so we can preallocate
+	// the line offset table in the printer for source maps. The line offset table
+	// is the #1 highest allocation in the heap profile, so this is worth doing.
+	// This count is approximate because it handles "\n" and "\r\n" (the common
+	// cases) but not "\r" or "\u2028" or "\u2029". Getting this wrong is harmless
+	// because it's only a preallocation. The array will just grow if it's too small.
+	if codePoint == '\n' {
+		lexer.ApproximateNewlineCount++
+	}
+
 	lexer.codePoint = codePoint
 	lexer.end = lexer.current
 	lexer.current += width

internal/parser/parser.go

@@ -8947,6 +8947,7 @@ func (p *parser) toAST(source logging.Source, parts []ast.Part, hashbang string,
 		TopLevelSymbolToParts:   p.topLevelSymbolToParts,
 		ExportStarImportRecords: p.exportStarImportRecords,
 		ImportRecords:           p.importRecords,
+		ApproximateLineCount:    int32(p.lexer.ApproximateNewlineCount) + 1,
 
 		// CommonJS features
 		HasTopLevelReturn: p.hasTopLevelReturn,

internal/printer/printer.go

@@ -550,14 +550,16 @@ func (p *printer) updateGeneratedLineAndColumn() {
 	p.lastGeneratedUpdate = len(p.js)
 }
 
-func generateLineOffsetTables(contents string) []lineOffsetTable {
-	var lineOffsetTables []lineOffsetTable
+func generateLineOffsetTables(contents string, approximateLineCount int32) []lineOffsetTable {
 	var columnsForNonASCII []int32
 	byteOffsetToFirstNonASCII := int32(0)
 	lineByteOffset := 0
 	columnByteOffset := 0
 	column := int32(0)
 
+	// Preallocate the top-level table using the approximate line count from the lexer
+	lineOffsetTables := make([]lineOffsetTable, 0, approximateLineCount)
+
 	for i, c := range contents {
 		// Mark the start of the next line
 		if column == 0 {
@@ -2870,6 +2872,7 @@ func createPrinter(
 	symbols ast.SymbolMap,
 	importRecords []ast.ImportRecord,
 	options PrintOptions,
+	approximateLineCount int32,
 ) *printer {
 	p := &printer{
 		symbols:            symbols,
@@ -2900,7 +2903,7 @@ func createPrinter(
 	// If we're writing out a source map, prepare a table of line start indices
 	// to do binary search on to figure out what line a given AST node came from
 	if options.SourceForSourceMap != nil {
-		p.lineOffsetTables = generateLineOffsetTables(options.SourceForSourceMap.Contents)
+		p.lineOffsetTables = generateLineOffsetTables(options.SourceForSourceMap.Contents, approximateLineCount)
 	}
 
 	return p
@@ -2918,7 +2921,7 @@ type PrintResult struct {
 }
 
 func Print(tree ast.AST, symbols ast.SymbolMap, options PrintOptions) PrintResult {
-	p := createPrinter(symbols, tree.ImportRecords, options)
+	p := createPrinter(symbols, tree.ImportRecords, options, tree.ApproximateLineCount)
 
 	for _, part := range tree.Parts {
 		for _, stmt := range part.Stmts {
@@ -2943,7 +2946,7 @@ func Print(tree ast.AST, symbols ast.SymbolMap, options PrintOptions) PrintResul
 }
 
 func PrintExpr(expr ast.Expr, symbols ast.SymbolMap, options PrintOptions) PrintResult {
-	p := createPrinter(symbols, nil, options)
+	p := createPrinter(symbols, nil, options, 0)
 
 	p.printExpr(expr, ast.LLowest, 0)