|
| 1 | +package transformers |
| 2 | + |
| 3 | +import ( |
| 4 | + "bufio" |
| 5 | + "encoding/json" |
| 6 | + "errors" |
| 7 | + "fmt" |
| 8 | + "os" |
| 9 | + "regexp" |
| 10 | + "strings" |
| 11 | + "time" |
| 12 | + |
| 13 | + "github.com/dmachard/go-dnscollector/dnsutils" |
| 14 | + "github.com/dmachard/go-dnscollector/pkgconfig" |
| 15 | + "github.com/dmachard/go-logger" |
| 16 | + "github.com/hashicorp/golang-lru/v2/expirable" |
| 17 | +) |
| 18 | + |
| 19 | +type NewDomainTracker struct { |
| 20 | + ttl time.Duration // Time window to consider a domain as "new" |
| 21 | + cache *expirable.LRU[string, struct{}] // Expirable LRU Cache |
| 22 | + whitelist map[string]*regexp.Regexp // Whitelisted domains |
| 23 | + persistencePath string |
| 24 | + logInfo func(msg string, v ...interface{}) |
| 25 | + logError func(msg string, v ...interface{}) |
| 26 | +} |
| 27 | + |
| 28 | +func NewNewDomainTracker(ttl time.Duration, maxSize int, whitelist map[string]*regexp.Regexp, persistencePath string, logInfo, logError func(msg string, v ...interface{})) (*NewDomainTracker, error) { |
| 29 | + |
| 30 | + if ttl <= 0 { |
| 31 | + return nil, fmt.Errorf("invalid TTL value: %v", ttl) |
| 32 | + } |
| 33 | + |
| 34 | + cache := expirable.NewLRU[string, struct{}](maxSize, nil, ttl) |
| 35 | + |
| 36 | + tracker := &NewDomainTracker{ |
| 37 | + ttl: ttl, |
| 38 | + cache: cache, |
| 39 | + whitelist: whitelist, |
| 40 | + persistencePath: persistencePath, |
| 41 | + logInfo: logInfo, |
| 42 | + logError: logError, |
| 43 | + } |
| 44 | + // Load cache state from disk if persistence is enabled |
| 45 | + if persistencePath != "" { |
| 46 | + if err := tracker.loadCacheFromDisk(); err != nil { |
| 47 | + return nil, fmt.Errorf("failed to load cache state: %w", err) |
| 48 | + } |
| 49 | + } |
| 50 | + |
| 51 | + return tracker, nil |
| 52 | +} |
| 53 | + |
| 54 | +func (ndt *NewDomainTracker) isWhitelisted(domain string) bool { |
| 55 | + for _, d := range ndt.whitelist { |
| 56 | + if d.MatchString(domain) { |
| 57 | + return true |
| 58 | + } |
| 59 | + } |
| 60 | + return false |
| 61 | +} |
| 62 | + |
| 63 | +func (ndt *NewDomainTracker) IsNewDomain(domain string) bool { |
| 64 | + // Check if the domain is whitelisted |
| 65 | + if ndt.isWhitelisted(domain) { |
| 66 | + return false |
| 67 | + } |
| 68 | + |
| 69 | + // Check if the domain exists in the cache |
| 70 | + if _, exists := ndt.cache.Get(domain); exists { |
| 71 | + // Domain was recently seen, not new |
| 72 | + return false |
| 73 | + } |
| 74 | + |
| 75 | + // Otherwise, mark the domain as new |
| 76 | + ndt.cache.Add(domain, struct{}{}) |
| 77 | + return true |
| 78 | +} |
| 79 | + |
| 80 | +func (ndt *NewDomainTracker) SaveCacheToDisk() error { |
| 81 | + keys := ndt.cache.Keys() |
| 82 | + data, err := json.Marshal(keys) |
| 83 | + if err != nil { |
| 84 | + return err |
| 85 | + } |
| 86 | + |
| 87 | + return os.WriteFile(ndt.persistencePath, data, 0644) |
| 88 | +} |
| 89 | + |
| 90 | +// loadCacheFromDisk loads the cache state from a file |
| 91 | +func (ndt *NewDomainTracker) loadCacheFromDisk() error { |
| 92 | + if ndt.persistencePath == "" { |
| 93 | + return errors.New("persistence filepath not set") |
| 94 | + } |
| 95 | + |
| 96 | + data, err := os.ReadFile(ndt.persistencePath) |
| 97 | + if err != nil { |
| 98 | + if os.IsNotExist(err) { |
| 99 | + return nil // File does not exist, no previous state to load |
| 100 | + } |
| 101 | + return err |
| 102 | + } |
| 103 | + |
| 104 | + var keys []string |
| 105 | + if err := json.Unmarshal(data, &keys); err != nil { |
| 106 | + return err |
| 107 | + } |
| 108 | + |
| 109 | + for _, key := range keys { |
| 110 | + ndt.cache.Add(key, struct{}{}) |
| 111 | + } |
| 112 | + |
| 113 | + return nil |
| 114 | +} |
| 115 | + |
| 116 | +// NewDomainTransform is the Transformer for DNS messages |
| 117 | +type NewDomainTrackerTransform struct { |
| 118 | + GenericTransformer |
| 119 | + domainTracker *NewDomainTracker |
| 120 | + listDomainsRegex map[string]*regexp.Regexp |
| 121 | +} |
| 122 | + |
| 123 | +// NewNewDomainTransform creates a new instance of the transformer |
| 124 | +func NewNewDomainTrackerTransform(config *pkgconfig.ConfigTransformers, logger *logger.Logger, name string, instance int, nextWorkers []chan dnsutils.DNSMessage) *NewDomainTrackerTransform { |
| 125 | + t := &NewDomainTrackerTransform{GenericTransformer: NewTransformer(config, logger, "new-domain-tracker", name, instance, nextWorkers)} |
| 126 | + t.listDomainsRegex = make(map[string]*regexp.Regexp) |
| 127 | + return t |
| 128 | +} |
| 129 | + |
| 130 | +// ReloadConfig reloads the configuration |
| 131 | +func (t *NewDomainTrackerTransform) ReloadConfig(config *pkgconfig.ConfigTransformers) { |
| 132 | + t.GenericTransformer.ReloadConfig(config) |
| 133 | + ttl := time.Duration(config.NewDomainTracker.TTL) * time.Second |
| 134 | + t.domainTracker.ttl = ttl |
| 135 | + t.LogInfo("new-domain-transformer configuration reloaded") |
| 136 | +} |
| 137 | + |
| 138 | +func (t *NewDomainTrackerTransform) GetTransforms() ([]Subtransform, error) { |
| 139 | + subtransforms := []Subtransform{} |
| 140 | + if t.config.NewDomainTracker.Enable { |
| 141 | + // init whitelist |
| 142 | + if err := t.LoadWhiteDomainsList(); err != nil { |
| 143 | + return nil, err |
| 144 | + } |
| 145 | + |
| 146 | + // Initialize the domain tracker |
| 147 | + ttl := time.Duration(t.config.NewDomainTracker.TTL) * time.Second |
| 148 | + maxSize := t.config.NewDomainTracker.CacheSize |
| 149 | + tracker, err := NewNewDomainTracker(ttl, maxSize, t.listDomainsRegex, t.config.NewDomainTracker.PersistenceFile, t.LogInfo, t.LogError) |
| 150 | + if err != nil { |
| 151 | + return nil, err |
| 152 | + } |
| 153 | + t.domainTracker = tracker |
| 154 | + |
| 155 | + subtransforms = append(subtransforms, Subtransform{name: "new-domain-tracker:detect", processFunc: t.trackNewDomain}) |
| 156 | + } |
| 157 | + return subtransforms, nil |
| 158 | +} |
| 159 | + |
| 160 | +func (t *NewDomainTrackerTransform) LoadWhiteDomainsList() error { |
| 161 | + // before to start, reset all maps |
| 162 | + for key := range t.listDomainsRegex { |
| 163 | + delete(t.listDomainsRegex, key) |
| 164 | + } |
| 165 | + |
| 166 | + if len(t.config.NewDomainTracker.WhiteDomainsFile) > 0 { |
| 167 | + file, err := os.Open(t.config.NewDomainTracker.WhiteDomainsFile) |
| 168 | + if err != nil { |
| 169 | + return fmt.Errorf("unable to open regex list file: %w", err) |
| 170 | + } else { |
| 171 | + |
| 172 | + scanner := bufio.NewScanner(file) |
| 173 | + for scanner.Scan() { |
| 174 | + domain := strings.ToLower(scanner.Text()) |
| 175 | + t.listDomainsRegex[domain] = regexp.MustCompile(domain) |
| 176 | + } |
| 177 | + t.LogInfo("loaded with %d domains in the whitelist", len(t.listDomainsRegex)) |
| 178 | + } |
| 179 | + } |
| 180 | + return nil |
| 181 | +} |
| 182 | + |
| 183 | +// Process processes DNS messages and detects newly observed domains |
| 184 | +func (t *NewDomainTrackerTransform) trackNewDomain(dm *dnsutils.DNSMessage) (int, error) { |
| 185 | + // Log a warning if the cache is full (before adding the new domain) |
| 186 | + if t.domainTracker.cache.Len() == t.config.NewDomainTracker.CacheSize { |
| 187 | + return ReturnError, fmt.Errorf("LRU cache is full. Consider increasing cache-size to avoid frequent evictions") |
| 188 | + } |
| 189 | + |
| 190 | + // Check if the domain is newly observed |
| 191 | + if t.domainTracker.IsNewDomain(dm.DNS.Qname) { |
| 192 | + return ReturnKeep, nil |
| 193 | + } |
| 194 | + return ReturnDrop, nil |
| 195 | +} |
| 196 | + |
| 197 | +func (t *NewDomainTrackerTransform) Reset() { |
| 198 | + if len(t.domainTracker.persistencePath) != 0 { |
| 199 | + if err := t.domainTracker.SaveCacheToDisk(); err != nil { |
| 200 | + t.LogError("failed to save cache state: %v", err) |
| 201 | + } |
| 202 | + t.LogInfo("cache content saved on disk with success") |
| 203 | + } |
| 204 | +} |
0 commit comments