Merge pull request #7 from slash3b/misc_bug_fixes

do not copy whole input and misc

Author: slash3b

utfbom.go

@@ -19,6 +19,8 @@ var _ io.Reader = (*Reader)(nil)
 // ErrRead helps to trace error origin.
 var ErrRead = errors.New("utfbom: I/O error during BOM processing")
 
+const maxBOMLen = 4
+
 // Encoding is a character encoding standard.
 type Encoding int
 
@@ -59,6 +61,10 @@ const (
 //   - UTF-32 Big Endian (BOM: 0x00 0x00 0xfe 0xff)
 //   - UTF-32 Little Endian (BOM: 0xff 0xfe 0x00 0x00)
 func DetectEncoding[T ~string | ~[]byte](input T) Encoding {
+	if len(input) > maxBOMLen {
+		input = input[:maxBOMLen]
+	}
+
 	b := []byte(input)
 
 	if len(b) < 2 {
@@ -149,31 +155,28 @@ func (e Encoding) Bytes() []byte {
 // Trim removes the BOM prefix from the input.
 // Supports string or []byte inputs and returns the same type without the BOM.
 func Trim[T ~string | ~[]byte](input T) (T, Encoding) {
-	b := []byte(input)
-	enc := DetectEncoding(b)
+	enc := DetectEncoding(input)
 
 	if enc == Unknown {
 		return input, enc
 	}
 
-	return T(b[enc.Len():]), enc
+	return input[enc.Len():], enc
 }
 
 // Prepend adds the corresponding Byte Order Mark (BOM) for a given encoding
 // to the beginning of a string or byte slice.
-// If the provided encoding is Unknown, the input is returned unmodified.
+// The input is returned unmodified if enc is Unknown or if the input already has any BOM.
 func Prepend[T ~string | ~[]byte](input T, enc Encoding) T {
 	if enc == Unknown {
 		return input
 	}
 
-	b := []byte(input)
-
-	if DetectEncoding(b) != Unknown {
+	if DetectEncoding(input) != Unknown {
 		return input
 	}
 
-	return T(append(enc.Bytes(), b...))
+	return T(append(enc.Bytes(), []byte(input)...))
 }
 
 // Reader implements automatic BOM (Unicode Byte Order Mark) checking and
@@ -201,8 +204,6 @@ func NewReader(rd io.Reader) *Reader {
 // On the first call, it detects and removes any Byte Order Mark (BOM).
 // Subsequent calls delegate directly to the underlying Reader.
 func (r *Reader) Read(buf []byte) (int, error) {
-	const maxBOMLen = 4
-
 	if len(buf) == 0 {
 		return 0, nil
 	}

utfbom_test.go

@@ -4,6 +4,7 @@ import (
 	"bytes"
 	"encoding/csv"
 	"encoding/hex"
+	"errors"
 	"fmt"
 	"io"
 	"strings"
@@ -218,41 +219,23 @@ func TestEncoding_Len(t *testing.T) {
 	t.Parallel()
 
 	testCases := []struct {
+		name     string
 		enc      utfbom.Encoding
 		expected int
 	}{
-		{
-			enc:      utfbom.Unknown,
-			expected: 0,
-		},
-		{
-			enc:      utfbom.UTF8,
-			expected: 3,
-		},
-		{
-			enc:      utfbom.UTF16BigEndian,
-			expected: 2,
-		},
-		{
-			enc:      utfbom.UTF16LittleEndian,
-			expected: 2,
-		},
-		{
-			enc:      utfbom.UTF32BigEndian,
-			expected: 4,
-		},
-		{
-			enc:      utfbom.UTF32LittleEndian,
-			expected: 4,
-		},
-		{
-			enc:      999,
-			expected: 0,
-		},
+		{"Unknown", utfbom.Unknown, 0},
+		{"UTF8", utfbom.UTF8, 3},
+		{"UTF16BigEndian", utfbom.UTF16BigEndian, 2},
+		{"UTF16LittleEndian", utfbom.UTF16LittleEndian, 2},
+		{"UTF32BigEndian", utfbom.UTF32BigEndian, 4},
+		{"UTF32LittleEndian", utfbom.UTF32LittleEndian, 4},
+		{"InvalidEncoding", 999, 0},
 	}
 
 	for _, tc := range testCases {
-		be.Equal(t, tc.enc.Len(), tc.expected)
+		t.Run(tc.name, func(t *testing.T) {
+			be.Equal(t, tc.enc.Len(), tc.expected)
+		})
 	}
 }
 
@@ -313,16 +296,6 @@ func TestReader_StringWithoutBOM(t *testing.T) {
 	be.Err(t, iotest.TestReader(rd, []byte(nobomstring)), nil)
 }
 
-func TestReader_UsualReader(t *testing.T) {
-	t.Parallel()
-
-	bomPrefixedStringReader := strings.NewReader(teststring)
-
-	rd := utfbom.NewReader(bomPrefixedStringReader)
-
-	be.Err(t, iotest.TestReader(rd, []byte(teststring[3:])), nil)
-}
-
 func TestReader_OneByteReader(t *testing.T) {
 	t.Parallel()
 
@@ -570,6 +543,19 @@ func TestPrepend_TypeAliases(t *testing.T) {
 	})
 }
 
+// TestReader_UnderlyingReaderError verifies that when the underlying reader
+// returns a non-EOF error during BOM detection, it is wrapped with ErrRead.
+func TestReader_UnderlyingReaderError(t *testing.T) {
+	t.Parallel()
+
+	rd := utfbom.NewReader(iotest.ErrReader(errors.New("disk failure")))
+
+	buf := make([]byte, 10)
+	n, err := rd.Read(buf)
+	be.Equal(t, 0, n)
+	be.True(t, errors.Is(err, utfbom.ErrRead))
+}
+
 func TestNewReader_NilPanics(t *testing.T) {
 	t.Parallel()