perf(csvpp,csvpputil): cache reflection type info and optimize YAML encoding

benchmark_test.go

@@ -332,6 +332,40 @@ func BenchmarkMarshal(b *testing.B) {
 	}
 }
 
+func BenchmarkUnmarshal_100Records(b *testing.B) {
+	var sb strings.Builder
+	sb.WriteString("name,age,phone[]\n")
+	for range 100 {
+		sb.WriteString("Alice,30,555-1234~555-5678\n")
+	}
+	inputBytes := []byte(sb.String())
+
+	for b.Loop() {
+		var people []BenchmarkPerson
+		if err := csvpp.Unmarshal(bytes.NewReader(inputBytes), &people); err != nil {
+			b.Fatal(err)
+		}
+	}
+}
+
+func BenchmarkMarshal_100Records(b *testing.B) {
+	people := make([]BenchmarkPerson, 100)
+	for i := range people {
+		people[i] = BenchmarkPerson{
+			Name:   "Alice",
+			Age:    30,
+			Phones: []string{"555-1234", "555-5678"},
+		}
+	}
+
+	for b.Loop() {
+		var buf bytes.Buffer
+		if err := csvpp.Marshal(&buf, people); err != nil {
+			b.Fatal(err)
+		}
+	}
+}
+
 // splitByRune Benchmark
 
 func BenchmarkSplitByRune(b *testing.B) {

csvpputil/benchmark_test.go

@@ -98,6 +98,21 @@ func BenchmarkYAMLArrayWriter_Write(b *testing.B) {
 	}
 }
 
+func BenchmarkYAMLArrayWriter_WriteWithCapacity(b *testing.B) {
+	b.ReportAllocs()
+	for b.Loop() {
+		w := csvpputil.NewYAMLArrayWriter(io.Discard, benchHeaders, csvpputil.WithYAMLCapacity(len(benchRecords)))
+		for _, record := range benchRecords {
+			if err := w.Write(record); err != nil {
+				b.Fatal(err)
+			}
+		}
+		if err := w.Close(); err != nil {
+			b.Fatal(err)
+		}
+	}
+}
+
 func BenchmarkJSONArrayWriter_SingleRecord(b *testing.B) {
 	b.ReportAllocs()
 	for b.Loop() {

csvpputil/yaml_array_writer.go

@@ -8,6 +8,19 @@ import (
 	"github.com/osamingo/go-csvpp"
 )
 
+// YAMLArrayWriterOption is a functional option for YAMLArrayWriter.
+type YAMLArrayWriterOption func(*YAMLArrayWriter)
+
+// WithYAMLCapacity pre-allocates the internal buffer for the expected number of records.
+// This reduces memory allocations when the approximate record count is known in advance.
+func WithYAMLCapacity(n int) YAMLArrayWriterOption {
+	return func(w *YAMLArrayWriter) {
+		if n > 0 {
+			w.records = make([]yaml.MapSlice, 0, n)
+		}
+	}
+}
+
 // YAMLArrayWriter writes CSV++ records as a YAML array.
 // Due to YAML's structure (go-yaml doesn't support streaming array elements),
 // records are buffered until Close.
@@ -19,11 +32,15 @@ type YAMLArrayWriter struct {
 }
 
 // NewYAMLArrayWriter creates a new YAMLArrayWriter that writes to w.
-func NewYAMLArrayWriter(w io.Writer, headers []*csvpp.ColumnHeader) *YAMLArrayWriter {
-	return &YAMLArrayWriter{
+func NewYAMLArrayWriter(w io.Writer, headers []*csvpp.ColumnHeader, opts ...YAMLArrayWriterOption) *YAMLArrayWriter {
+	writer := &YAMLArrayWriter{
 		w:       w,
 		headers: headers,
 	}
+	for _, opt := range opts {
+		opt(writer)
+	}
+	return writer
 }
 
 // Write adds a single record to the buffer.
@@ -57,31 +74,29 @@ func (w *YAMLArrayWriter) Close() error {
 // The output is a YAML array where each element is a record.
 func MarshalYAML(headers []*csvpp.ColumnHeader, records [][]*csvpp.Field) ([]byte, error) {
 	var buf bytes.Buffer
-	w := NewYAMLArrayWriter(&buf, headers)
-
-	for _, record := range records {
-		if err := w.Write(record); err != nil {
-			return nil, err
-		}
-	}
-
-	if err := w.Close(); err != nil {
+	if err := encodeYAMLRecords(&buf, headers, records); err != nil {
 		return nil, err
 	}
-
 	return buf.Bytes(), nil
 }
 
 // WriteYAML writes CSV++ records as a YAML array to the provided writer.
 // The output is a YAML array where each element is a record.
 func WriteYAML(w io.Writer, headers []*csvpp.ColumnHeader, records [][]*csvpp.Field) error {
-	writer := NewYAMLArrayWriter(w, headers)
+	return encodeYAMLRecords(w, headers, records)
+}
 
-	for _, record := range records {
-		if err := writer.Write(record); err != nil {
-			return err
-		}
+// encodeYAMLRecords builds the complete MapSlice array with exact allocation
+// and encodes it in one shot. This avoids the overhead of the YAMLArrayWriter's
+// per-record append growth.
+func encodeYAMLRecords(w io.Writer, headers []*csvpp.ColumnHeader, records [][]*csvpp.Field) error {
+	ms := make([]yaml.MapSlice, len(records))
+	for i, record := range records {
+		ms[i] = fieldsToMapSlice(headers, record)
 	}
-
-	return writer.Close()
+	enc := yaml.NewEncoder(w)
+	if err := enc.Encode(ms); err != nil {
+		return err
+	}
+	return enc.Close()
 }

csvpputil/yaml_array_writer_test.go

@@ -118,6 +118,77 @@ func TestYAMLArrayWriter_Write(t *testing.T) {
 	})
 }
 
+func TestYAMLArrayWriter_WriteWithCapacity(t *testing.T) {
+	t.Parallel()
+
+	headers := []*csvpp.ColumnHeader{
+		{Name: "name", Kind: csvpp.SimpleField},
+		{Name: "tags", Kind: csvpp.ArrayField},
+	}
+
+	t.Run("success: writer with capacity hint", func(t *testing.T) {
+		t.Parallel()
+
+		var buf bytes.Buffer
+		w := csvpputil.NewYAMLArrayWriter(&buf, headers, csvpputil.WithYAMLCapacity(2))
+
+		records := [][]*csvpp.Field{
+			{{Value: "Alice"}, {Values: []string{"go"}}},
+			{{Value: "Bob"}, {Values: []string{"rust", "python"}}},
+		}
+
+		for _, record := range records {
+			if err := w.Write(record); err != nil {
+				t.Fatalf("Write() error = %v", err)
+			}
+		}
+
+		if err := w.Close(); err != nil {
+			t.Fatalf("Close() error = %v", err)
+		}
+
+		var got []map[string]any
+		if err := yaml.Unmarshal(buf.Bytes(), &got); err != nil {
+			t.Fatalf("yaml.Unmarshal() error = %v", err)
+		}
+
+		want := []map[string]any{
+			{"name": "Alice", "tags": []any{"go"}},
+			{"name": "Bob", "tags": []any{"rust", "python"}},
+		}
+		if diff := cmp.Diff(want, got); diff != "" {
+			t.Errorf("output mismatch (-want +got):\n%s", diff)
+		}
+	})
+
+	t.Run("success: zero capacity is safe", func(t *testing.T) {
+		t.Parallel()
+
+		var buf bytes.Buffer
+		w := csvpputil.NewYAMLArrayWriter(&buf, headers, csvpputil.WithYAMLCapacity(0))
+
+		if err := w.Write([]*csvpp.Field{{Value: "Alice"}, {Values: []string{"go"}}}); err != nil {
+			t.Fatalf("Write() error = %v", err)
+		}
+
+		if err := w.Close(); err != nil {
+			t.Fatalf("Close() error = %v", err)
+		}
+
+		var got []map[string]any
+		if err := yaml.Unmarshal(buf.Bytes(), &got); err != nil {
+			t.Fatalf("yaml.Unmarshal() error = %v", err)
+		}
+
+		want := []map[string]any{
+			{"name": "Alice", "tags": []any{"go"}},
+		}
+		if diff := cmp.Diff(want, got); diff != "" {
+			t.Errorf("output mismatch (-want +got):\n%s", diff)
+		}
+	})
+}
+
 func TestMarshalYAML(t *testing.T) {
 	t.Parallel()
 

export_test.go

@@ -13,4 +13,5 @@ var (
 	FormatColumnHeader            = formatColumnHeader
 	FormatComponentList           = formatComponentList
 	ExtractTagName                = extractTagName
+	CachedTypeInfo                = cachedTypeInfo
 )

marshal.go

@@ -95,23 +95,23 @@ func MarshalWriter(w *Writer, src any) error {
 		return fmt.Errorf("csvpp: slice element must be a struct")
 	}
 
-	// Build headers
-	headers := buildHeaders(elemType)
-	w.SetHeaders(headers)
+	// Get cached type info
+	ti := cachedTypeInfo(elemType)
+	w.SetHeaders(ti.headers)
 
 	// Write headers
 	if err := w.WriteHeader(); err != nil {
 		return err
 	}
 
-	// Encode each element
+	// Encode each element using cached encode field mappings
 	for i := 0; i < srcVal.Len(); i++ {
 		elemVal := srcVal.Index(i)
 		if elemVal.Kind() == reflect.Pointer {
 			elemVal = elemVal.Elem()
 		}
 
-		record := encodeRecord(elemVal, headers)
+		record := encodeRecord(elemVal, ti.headers, ti.encodeFields)
 		if err := w.Write(record); err != nil {
 			return err
 		}
@@ -130,23 +130,14 @@ type fieldMapping struct {
 
 // buildFieldMap creates a mapping between struct fields and headers.
 func buildFieldMap(t reflect.Type, headers []*ColumnHeader) []fieldMapping {
+	ti := cachedTypeInfo(t)
 	var mappings []fieldMapping
 
-	for i := 0; i < t.NumField(); i++ {
-		field := t.Field(i)
-		tag := field.Tag.Get("csvpp")
-		if tag == "" || tag == "-" {
-			continue
-		}
-
-		// Extract column name from tag (first part is the column name)
-		tagName := extractTagName(tag)
-
-		// Find corresponding column in headers
+	for _, tn := range ti.tagNames {
 		for j, h := range headers {
-			if h.Name == tagName {
+			if h.Name == tn.tagName {
 				mappings = append(mappings, fieldMapping{
-					fieldIndex:  i,
+					fieldIndex:  tn.structIndex,
 					header:      h,
 					columnIndex: j,
 				})
@@ -170,31 +161,6 @@ func extractTagName(tag string) string {
 	return tag
 }
 
-// buildHeaders builds headers from a struct.
-func buildHeaders(t reflect.Type) []*ColumnHeader {
-	var headers []*ColumnHeader
-
-	for i := 0; i < t.NumField(); i++ {
-		field := t.Field(i)
-		tag := field.Tag.Get("csvpp")
-		if tag == "" || tag == "-" {
-			continue
-		}
-
-		h, err := parseColumnHeader(tag)
-		if err != nil {
-			// Treat as simple field if error
-			h = &ColumnHeader{
-				Name: tag,
-				Kind: SimpleField,
-			}
-		}
-		headers = append(headers, h)
-	}
-
-	return headers
-}
-
 // decodeRecord decodes a record into a struct.
 func decodeRecord(record []*Field, dst reflect.Value, mappings []fieldMapping) error {
 	for _, m := range mappings {
@@ -355,25 +321,17 @@ func decodeStructComponents(components []*Field, dst reflect.Value, headers []*C
 }
 
 // encodeRecord encodes a struct to a record.
-func encodeRecord(src reflect.Value, headers []*ColumnHeader) []*Field {
+func encodeRecord(src reflect.Value, headers []*ColumnHeader, encodeFields []encodeFieldInfo) []*Field {
 	fields := make([]*Field, 0, len(headers))
 
-	fieldIdx := 0
-	for i := 0; i < src.NumField(); i++ {
-		structField := src.Type().Field(i)
-		tag := structField.Tag.Get("csvpp")
-		if tag == "" || tag == "-" {
-			continue
-		}
-
-		if fieldIdx >= len(headers) {
+	for _, ef := range encodeFields {
+		if ef.headerIndex >= len(headers) {
 			break
 		}
 
-		field := src.Field(i)
-		f := encodeField(field, headers[fieldIdx])
+		field := src.Field(ef.structIndex)
+		f := encodeField(field, headers[ef.headerIndex])
 		fields = append(fields, f)
-		fieldIdx++
 	}
 
 	return fields