[scanner] 新增哈希接口，支持多种哈希实现；新增MD5哈希实现

cmd/start/start.go

@@ -76,15 +76,8 @@ var StartCmd = &cobra.Command{
 		logger.Log.Info("正在扫描文件系统...")
 
 		startTime := time.Now()
-		fileCfg := &config.FileScannerConfig{
-			IncludePaths:  cfg.Scanner.File.IncludePaths,
-			ExcludePaths:  cfg.Scanner.File.ExcludePaths,
-			FastHash:      cfg.Scanner.File.FastHash,
-			FastHashSize:  cfg.Scanner.File.FastHashSize,
-			FastHashChunk: cfg.Scanner.File.FastHashChunk,
-		}
+		fileScanner := file.NewScanner(cfg)
 
-		fileScanner := file.NewScanner(fileCfg)
 		files, err := fileScanner.Scan()
 		if err != nil {
 			logger.Log.Error("扫描文件系统失败", zap.Error(err))

config.yaml

@@ -9,6 +9,9 @@ audit:
   buffer_size: 1000
 
 scanner:
+  hash:
+    # algorithm: "sha256"
+    algorithm: "md5"
   file:
     include_paths:
       - /home

internal/config/config.go

@@ -24,6 +24,11 @@ type AuditConfig struct {
 
 type ScannerConfig struct {
 	File FileScannerConfig `yaml:"file"`
+	Hash hashConfig        `yaml:"hash"`
+}
+
+type hashConfig struct {
+	Algorithm string `yaml:"algorithm"`
 }
 
 type StorageConfig struct {

internal/config/utils.go

@@ -5,6 +5,8 @@ import (
 	"regexp"
 	"strconv"
 	"strings"
+
+	"sysmonitord/internal/scanner/hash"
 )
 
 func ParseSize(sizeStr string) (int64, error) {
@@ -44,3 +46,34 @@ func ParseSize(sizeStr string) (int64, error) {
 
 	return value * multiplier, nil
 }
+
+func (c *Config) GetHashAlgorithm() (hash.HashAlgorithm, error) {
+	algoName := c.Scanner.Hash.Algorithm
+
+	switch strings.ToLower(algoName) {
+	case "sha256":
+		return &hash.SHA256Algorithm{}, nil
+	case "md5":
+		return &hash.MD5Algorithm{}, nil
+	default:
+		return nil, fmt.Errorf("不支持的哈希算法: %s", algoName)
+	}
+}
+
+func (c *Config) GetFileScannerConfig() (*FileScannerConfig, error) {
+	return &c.Scanner.File, nil
+}
+
+func (c *Config) GetHashConfig() (*hash.Config, error) {
+	algo, err := c.GetHashAlgorithm()
+	if err != nil {
+		return nil, err
+	}
+
+	return &hash.Config{
+		UseFastHash: c.Scanner.File.FastHash,
+		Threshold:   c.Scanner.File.FastHashSize,
+		ChunkSize:   c.Scanner.File.FastHashChunk,
+		Algorithm:   algo,
+	}, nil
+}

internal/scanner/file/scanner.go

@@ -21,17 +21,17 @@ type FileInfo struct {
 }
 
 type Scanner struct {
-	cfg *config.FileScannerConfig
+	cfg *config.Config
 }
 
-func NewScanner(cfg *config.FileScannerConfig) *Scanner {
+func NewScanner(cfg *config.Config) *Scanner {
 	return &Scanner{
 		cfg: cfg,
 	}
 }
 
 func (s *Scanner) Scan() ([]FileInfo, error) {
-	targetPaths := s.cfg.IncludePaths
+	targetPaths := s.cfg.Scanner.File.IncludePaths
 	if len(targetPaths) == 0 {
 		targetPaths = []string{"/"}
 	}
@@ -65,10 +65,10 @@ func (s *Scanner) WalkFunc(result *[]FileInfo) fs.WalkDirFunc {
 			return nil
 		}
 
-		for _, exclude := range s.cfg.ExcludePaths {
+		for _, exclude := range s.cfg.Scanner.File.ExcludePaths {
 			if strings.HasPrefix(path, exclude) {
 				logger.Log.Debug("[scan]跳过路径", zap.String("path", path), zap.String("reason", "匹配排除路径"))
-				return fs.SkipDir
+				return nil
 			}
 		}
 
@@ -78,13 +78,14 @@ func (s *Scanner) WalkFunc(result *[]FileInfo) fs.WalkDirFunc {
 			return nil
 		}
 
-		if info.Size() != 0 {
-			hash, err := hash.SHA256(path, &hash.Config{
-				UseFastHash: s.cfg.FastHash,
-				Threshold:   s.cfg.FastHashSize,
-				ChunkSize:   s.cfg.FastHashChunk,
-			})
+		if info.Size() > 0 {
+			hashCfg, err := s.cfg.GetHashConfig()
+			if err != nil {
+				logger.Log.Debug("[scan]无法获取哈希配置", zap.String("path", path), zap.Error(err))
+				return nil
+			}
 
+			hash, err := hash.CalculateHash(path, hashCfg)
 			if err != nil {
 				logger.Log.Debug("[scan]无法计算文件哈希", zap.String("path", path), zap.Error(err))
 				return nil

internal/scanner/hash/hash.go

@@ -1,9 +1,11 @@
 package hash
 
 import (
+	"crypto/md5"
 	"crypto/sha256"
 	"encoding/binary"
 	"encoding/hex"
+	"hash"
 	"io"
 	"os"
 	"sysmonitord/pkg/logger"
@@ -11,13 +13,51 @@ import (
 	"go.uber.org/zap"
 )
 
+type HashAlgorithm interface {
+	Hash() hash.Hash
+	Name() string
+}
+
+// ==== SHA256 ====
+
+type SHA256Algorithm struct{}
+
+func (a *SHA256Algorithm) Hash() hash.Hash {
+	return sha256.New()
+}
+
+func (a *SHA256Algorithm) Name() string {
+	return "sha256"
+}
+
+// ==== MD5 ====
+
+type MD5Algorithm struct{}
+
+func (a *MD5Algorithm) Hash() hash.Hash {
+	return md5.New()
+}
+
+func (a *MD5Algorithm) Name() string {
+	return "md5"
+}
+
+// ==== xxHash64 ====
+
+// Todo: 添加 xxHash64 实现
+
+// ==== 配置结构体 ====
+
 type Config struct {
 	UseFastHash bool
 	Threshold   int64
 	ChunkSize   int64
+	Algorithm   HashAlgorithm
 }
 
-func SHA256(filePath string, cfg *Config) (string, error) {
+// ==== 计算文件哈希 ====
+
+func CalculateHash(filePath string, cfg *Config) (string, error) {
 	info, err := os.Stat(filePath)
 	if err != nil {
 		logger.Log.Warn("[hash]获取文件信息失败", zap.String("path", filePath), zap.Error(err))
@@ -26,18 +66,27 @@ func SHA256(filePath string, cfg *Config) (string, error) {
 
 	fileSize := info.Size()
 
-	if cfg != nil && cfg.UseFastHash && fileSize > cfg.Threshold {
+	if cfg.Algorithm == nil {
+		cfg.Algorithm = &SHA256Algorithm{}
+	}
+
+	logger.Log.Debug("[hash]计算文件哈希",
+		zap.String("path", filePath),
+		zap.Int64("fileSize", fileSize),
+		zap.String("Algorithm", cfg.Algorithm.Name()))
+
+	if cfg.UseFastHash && fileSize > cfg.Threshold {
 		logger.Log.Debug("[hash] 分层哈希...",
 			zap.String("path", filePath),
 			zap.Int64("fileSize", fileSize),
 		)
-		return calculateFastHash(filePath, fileSize, cfg.ChunkSize)
+		return calculateFast(filePath, fileSize, cfg)
 	}
 
-	return calculateFullHash(filePath)
+	return calculateFull(filePath, cfg)
 }
 
-func calculateFullHash(filePath string) (string, error) {
+func calculateFull(filePath string, cfg *Config) (string, error) {
 	file, err := os.Open(filePath)
 	if err != nil {
 		logger.Log.Warn("[scanner]打开文件失败", zap.String("path", filePath), zap.Error(err))
@@ -45,7 +94,7 @@ func calculateFullHash(filePath string) (string, error) {
 	}
 	defer file.Close()
 
-	hasher := sha256.New()
+	hasher := cfg.Algorithm.Hash()
 	if _, err := io.Copy(hasher, file); err != nil {
 		logger.Log.Error("[scanner]读取文件失败", zap.String("path", filePath), zap.Error(err))
 		return "", err
@@ -56,7 +105,7 @@ func calculateFullHash(filePath string) (string, error) {
 	return hashString, nil
 }
 
-func calculateFastHash(filePath string, fileSize int64, chunkSize int64) (string, error) {
+func calculateFast(filePath string, fileSize int64, cfg *Config) (string, error) {
 	file, err := os.Open(filePath)
 	if err != nil {
 		logger.Log.Warn("[scanner]打开文件失败", zap.String("path", filePath), zap.Error(err))
@@ -64,7 +113,8 @@ func calculateFastHash(filePath string, fileSize int64, chunkSize int64) (string
 	}
 	defer file.Close()
 
-	hasher := sha256.New()
+	hasher := cfg.Algorithm.Hash()
+	chunkSize := cfg.ChunkSize
 
 	if _, err := io.CopyN(hasher, file, chunkSize); err != nil {
 		if err != io.EOF {

internal/scanner/process/process.go

@@ -58,7 +58,7 @@ func ScanAllProcesses(hashCfg *hash.Config) ([]ProcessInfo, error) {
 
 		if exePath != "" {
 			if _, err := os.Stat(exePath); err == nil {
-				fileHash, err := hash.SHA256(exePath, hashCfg)
+				fileHash, err := hash.CalculateHash(exePath, hashCfg)
 				if err == nil {
 					info.FileHash = fileHash
 				} else {