v3 candidate

internal/leviathan/binsnoop/common.go

@@ -0,0 +1,127 @@
+package binsnoop
+
+import (
+	"encoding/binary"
+	"fmt"
+	"hash/crc32"
+	"os"
+	"time"
+)
+
+const MaxFileReadBytes int64 = 64 << 20
+
+// CRC32 calculates the CRC32 checksum using the IEEE polynomial.
+func CRC32(data []byte) uint32 {
+	return crc32.ChecksumIEEE(data)
+}
+
+// EpochToString converts a Unix timestamp to a formatted time string.
+func EpochToString(epoch uint32) string {
+	t := time.Unix(int64(epoch), 0).UTC()
+	return t.Format("2006-01-02 15:04:05")
+}
+
+// GetCString extracts a NULL-terminated C string from raw bytes.
+func GetCString(data []byte) string {
+	for i, b := range data {
+		if b == 0 {
+			s := string(data[:i])
+			// Validate UTF-8 printability
+			for _, c := range s {
+				if c == 0 {
+					return ""
+				}
+			}
+			return s
+		}
+	}
+	return string(data)
+}
+
+// IsASCIINumber returns true if the byte is an ASCII digit (0-9).
+func IsASCIINumber(b byte) bool {
+	return b >= '0' && b <= '9'
+}
+
+// IsPrintableASCII returns true if the byte is a printable ASCII character.
+func IsPrintableASCII(b byte) bool {
+	return b >= 0x0A && b <= 0x7E
+}
+
+// IsOffsetSafe validates that the next offset is within bounds and progressing forward.
+func IsOffsetSafe(availableData, nextOffset int, lastOffset *int) bool {
+	if lastOffset != nil {
+		if *lastOffset >= nextOffset {
+			return false
+		}
+	}
+	return nextOffset < availableData
+}
+
+// ReadFile reads a bounded file into memory for scanner APIs that require a
+// byte slice. Larger callers should use a streaming document path.
+func ReadFile(path string) ([]byte, error) {
+	return ReadFileLimited(path, MaxFileReadBytes)
+}
+
+func ReadFileLimited(path string, maxBytes int64) ([]byte, error) {
+	info, err := os.Stat(path)
+	if err != nil {
+		return nil, err
+	}
+	if info.IsDir() {
+		return nil, fmt.Errorf("binsnoop: path is a directory")
+	}
+	if maxBytes > 0 && info.Size() > maxBytes {
+		return nil, fmt.Errorf("binsnoop: file is too large: %d bytes exceeds %d bytes", info.Size(), maxBytes)
+	}
+	return os.ReadFile(path)
+}
+
+// LE16 reads a little-endian uint16 from data at the given offset.
+func LE16(data []byte, offset int) uint16 {
+	if offset+2 > len(data) {
+		return 0
+	}
+	return binary.LittleEndian.Uint16(data[offset:])
+}
+
+// BE16 reads a big-endian uint16 from data at the given offset.
+func BE16(data []byte, offset int) uint16 {
+	if offset+2 > len(data) {
+		return 0
+	}
+	return binary.BigEndian.Uint16(data[offset:])
+}
+
+// LE32 reads a little-endian uint32 from data at the given offset.
+func LE32(data []byte, offset int) uint32 {
+	if offset+4 > len(data) {
+		return 0
+	}
+	return binary.LittleEndian.Uint32(data[offset:])
+}
+
+// BE32 reads a big-endian uint32 from data at the given offset.
+func BE32(data []byte, offset int) uint32 {
+	if offset+4 > len(data) {
+		return 0
+	}
+	return binary.BigEndian.Uint32(data[offset:])
+}
+
+// LE64 reads a little-endian uint64 from data at the given offset.
+func LE64(data []byte, offset int) uint64 {
+	if offset+8 > len(data) {
+		return 0
+	}
+	return binary.LittleEndian.Uint64(data[offset:])
+}
+
+// BE64 reads a big-endian uint64 from data at the given offset.
+func BE64(data []byte, offset int) uint64 {
+	if offset+8 > len(data) {
+		return 0
+	}
+	return binary.BigEndian.Uint64(data[offset:])
+}