Highly-opinionated (ex-bullshit-free) MTPROTO proxy for Telegram. If you use v1.0 or upgrade broke you proxy, please read the chapter Version 2
Nelze vybrat více než 25 témat Téma musí začínat písmenem nebo číslem, může obsahovat pomlčky („-“) a může být dlouhé až 35 znaků.

firehol.go 7.9KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371
  1. package ipblocklist
  2. import (
  3. "bufio"
  4. "context"
  5. "fmt"
  6. "io"
  7. "net"
  8. "net/http"
  9. "net/url"
  10. "os"
  11. "regexp"
  12. "strings"
  13. "sync"
  14. "time"
  15. "github.com/9seconds/mtg/v2/mtglib"
  16. "github.com/kentik/patricia"
  17. "github.com/kentik/patricia/bool_tree"
  18. "github.com/panjf2000/ants/v2"
  19. )
  20. const (
  21. fireholIPv4DefaultCIDR = 32
  22. fireholIPv6DefaultCIDR = 128
  23. )
  24. var fireholRegexpComment = regexp.MustCompile(`\s*#.*?$`)
  25. // Firehol is IPBlocklist which uses lists from FireHOL:
  26. // https://iplists.firehol.org/
  27. //
  28. // It can use both local files and remote URLs. This is not necessary
  29. // that blocklists should be taken from this website, we expect only
  30. // compatible formats here.
  31. //
  32. // Example of the format:
  33. //
  34. // # this is a comment
  35. // # to ignore
  36. // 127.0.0.1 # you can specify an IP
  37. // 10.0.0.0/8 # or cidr
  38. type Firehol struct {
  39. ctx context.Context
  40. ctxCancel context.CancelFunc
  41. logger mtglib.Logger
  42. rwMutex sync.RWMutex
  43. remoteURLs []string
  44. localFiles []string
  45. httpClient *http.Client
  46. workerPool *ants.Pool
  47. treeV4 *bool_tree.TreeV4
  48. treeV6 *bool_tree.TreeV6
  49. }
  50. // Shutdown stop a background update process.
  51. func (f *Firehol) Shutdown() {
  52. f.ctxCancel()
  53. }
  54. // Contains is given IP list can be found in FireHOL blocklists.
  55. func (f *Firehol) Contains(ip net.IP) bool {
  56. if ip == nil {
  57. return true
  58. }
  59. f.rwMutex.RLock()
  60. defer f.rwMutex.RUnlock()
  61. if ip4 := ip.To4(); ip4 != nil {
  62. return f.containsIPv4(ip4)
  63. }
  64. return f.containsIPv6(ip.To16())
  65. }
  66. // Run starts a background update process.
  67. //
  68. // This is a blocking method so you probably want to run it in a
  69. // goroutine.
  70. func (f *Firehol) Run(updateEach time.Duration) {
  71. if updateEach == 0 {
  72. updateEach = DefaultFireholUpdateEach
  73. }
  74. ticker := time.NewTicker(updateEach)
  75. defer func() {
  76. ticker.Stop()
  77. select {
  78. case <-ticker.C:
  79. default:
  80. }
  81. }()
  82. if err := f.update(); err != nil {
  83. f.logger.WarningError("cannot update blocklist", err)
  84. } else {
  85. f.logger.Info("blocklist was updated")
  86. }
  87. for {
  88. select {
  89. case <-f.ctx.Done():
  90. return
  91. case <-ticker.C:
  92. if err := f.update(); err != nil {
  93. f.logger.WarningError("cannot update blocklist", err)
  94. } else {
  95. f.logger.Info("blocklist was updated")
  96. }
  97. }
  98. }
  99. }
  100. func (f *Firehol) containsIPv4(addr net.IP) bool {
  101. ip := patricia.NewIPv4AddressFromBytes(addr, 32) // nolint: gomnd
  102. if ok, _, err := f.treeV4.FindDeepestTag(ip); ok && err == nil {
  103. return true
  104. }
  105. return false
  106. }
  107. func (f *Firehol) containsIPv6(addr net.IP) bool {
  108. ip := patricia.NewIPv6Address(addr, 128) // nolint: gomnd
  109. if ok, _, err := f.treeV6.FindDeepestTag(ip); ok && err == nil {
  110. return true
  111. }
  112. return false
  113. }
  114. func (f *Firehol) update() error { // nolint: funlen, cyclop
  115. ctx, cancel := context.WithCancel(f.ctx)
  116. defer cancel()
  117. wg := &sync.WaitGroup{}
  118. wg.Add(len(f.remoteURLs) + len(f.localFiles))
  119. treeMutex := &sync.Mutex{}
  120. v4tree := bool_tree.NewTreeV4()
  121. v6tree := bool_tree.NewTreeV6()
  122. errorChan := make(chan error, 1)
  123. defer close(errorChan)
  124. for _, v := range f.localFiles {
  125. go func(filename string) {
  126. defer wg.Done()
  127. if err := f.updateLocalFile(ctx, filename, treeMutex, v4tree, v6tree); err != nil {
  128. cancel()
  129. f.logger.BindStr("filename", filename).WarningError("cannot update", err)
  130. select {
  131. case errorChan <- err:
  132. default:
  133. }
  134. }
  135. }(v)
  136. }
  137. for _, v := range f.remoteURLs {
  138. value := v
  139. f.workerPool.Submit(func() { // nolint: errcheck
  140. defer wg.Done()
  141. if err := f.updateRemoteURL(ctx, value, treeMutex, v4tree, v6tree); err != nil {
  142. cancel()
  143. f.logger.BindStr("url", value).WarningError("cannot update", err)
  144. select {
  145. case errorChan <- err:
  146. default:
  147. }
  148. }
  149. })
  150. }
  151. wg.Wait()
  152. select {
  153. case err := <-errorChan:
  154. return fmt.Errorf("cannot update trees: %w", err)
  155. default:
  156. }
  157. f.rwMutex.Lock()
  158. defer f.rwMutex.Unlock()
  159. f.treeV4 = v4tree
  160. f.treeV6 = v6tree
  161. return nil
  162. }
  163. func (f *Firehol) updateLocalFile(ctx context.Context, filename string,
  164. mutex sync.Locker,
  165. v4tree *bool_tree.TreeV4, v6tree *bool_tree.TreeV6) error {
  166. filefp, err := os.Open(filename)
  167. if err != nil {
  168. return fmt.Errorf("cannot open file: %w", err)
  169. }
  170. go func(ctx context.Context, closer io.Closer) {
  171. <-ctx.Done()
  172. closer.Close()
  173. }(ctx, filefp)
  174. defer filefp.Close()
  175. return f.updateTrees(mutex, filefp, v4tree, v6tree)
  176. }
  177. func (f *Firehol) updateRemoteURL(ctx context.Context, url string,
  178. mutex sync.Locker,
  179. v4tree *bool_tree.TreeV4, v6tree *bool_tree.TreeV6) error {
  180. req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
  181. if err != nil {
  182. return fmt.Errorf("cannot build a request: %w", err)
  183. }
  184. resp, err := f.httpClient.Do(req) // nolint: bodyclose
  185. if err != nil {
  186. return fmt.Errorf("cannot request a remote URL %s: %w", url, err)
  187. }
  188. go func(ctx context.Context, closer io.Closer) {
  189. <-ctx.Done()
  190. closer.Close()
  191. }(ctx, resp.Body)
  192. defer func(rc io.ReadCloser) {
  193. io.Copy(io.Discard, rc) // nolint: errcheck
  194. rc.Close()
  195. }(resp.Body)
  196. return f.updateTrees(mutex, resp.Body, v4tree, v6tree)
  197. }
  198. func (f *Firehol) updateTrees(mutex sync.Locker,
  199. reader io.Reader,
  200. v4tree *bool_tree.TreeV4,
  201. v6tree *bool_tree.TreeV6) error {
  202. scanner := bufio.NewScanner(reader)
  203. for scanner.Scan() {
  204. text := scanner.Text()
  205. text = fireholRegexpComment.ReplaceAllLiteralString(text, "")
  206. text = strings.TrimSpace(text)
  207. if text == "" {
  208. continue
  209. }
  210. ip, cidr, err := f.updateParseLine(text)
  211. if err != nil {
  212. return fmt.Errorf("cannot parse a line: %w", err)
  213. }
  214. if err := f.updateAddToTrees(ip, cidr, mutex, v4tree, v6tree); err != nil {
  215. return fmt.Errorf("cannot add a node to the tree: %w", err)
  216. }
  217. }
  218. if scanner.Err() != nil {
  219. return fmt.Errorf("cannot parse a response: %w", scanner.Err())
  220. }
  221. return nil
  222. }
  223. func (f *Firehol) updateParseLine(text string) (net.IP, uint, error) {
  224. _, ipnet, err := net.ParseCIDR(text)
  225. if err != nil {
  226. ipaddr := net.ParseIP(text)
  227. if ipaddr == nil {
  228. return nil, 0, fmt.Errorf("incorrect ip address %s", text)
  229. }
  230. ip4 := ipaddr.To4()
  231. if ip4 != nil {
  232. return ip4, fireholIPv4DefaultCIDR, nil
  233. }
  234. return ipaddr.To16(), fireholIPv6DefaultCIDR, nil
  235. }
  236. ones, _ := ipnet.Mask.Size()
  237. return ipnet.IP, uint(ones), nil
  238. }
  239. func (f *Firehol) updateAddToTrees(ip net.IP, cidr uint,
  240. mutex sync.Locker,
  241. v4tree *bool_tree.TreeV4, v6tree *bool_tree.TreeV6) error {
  242. mutex.Lock()
  243. defer mutex.Unlock()
  244. if ip.To4() != nil {
  245. addr := patricia.NewIPv4AddressFromBytes(ip, cidr)
  246. if _, _, err := v4tree.Set(addr, true); err != nil {
  247. return err // nolint: wrapcheck
  248. }
  249. } else {
  250. addr := patricia.NewIPv6Address(ip, cidr)
  251. if _, _, err := v6tree.Set(addr, true); err != nil {
  252. return err // nolint: wrapcheck
  253. }
  254. }
  255. return nil
  256. }
  257. // NewFirehol creates a new instance of FireHOL IP blocklist.
  258. //
  259. // This method does not start an update process so please execute Run
  260. // when it is necessary.
  261. func NewFirehol(logger mtglib.Logger, network mtglib.Network,
  262. downloadConcurrency uint,
  263. remoteURLs []string,
  264. localFiles []string) (*Firehol, error) {
  265. for _, v := range remoteURLs {
  266. parsed, err := url.Parse(v)
  267. if err != nil {
  268. return nil, fmt.Errorf("incorrect url %s: %w", v, err)
  269. }
  270. switch parsed.Scheme {
  271. case "http", "https":
  272. default:
  273. return nil, fmt.Errorf("unsupported url %s", v)
  274. }
  275. }
  276. for _, v := range localFiles {
  277. if stat, err := os.Stat(v); os.IsNotExist(err) || stat.IsDir() || stat.Mode().Perm()&0o400 == 0 {
  278. return nil, fmt.Errorf("%s is not a readable file", v)
  279. }
  280. }
  281. if downloadConcurrency == 0 {
  282. downloadConcurrency = DefaultFireholDownloadConcurrency
  283. }
  284. workerPool, _ := ants.NewPool(int(downloadConcurrency))
  285. ctx, cancel := context.WithCancel(context.Background())
  286. return &Firehol{
  287. ctx: ctx,
  288. ctxCancel: cancel,
  289. logger: logger.Named("firehol"),
  290. httpClient: network.MakeHTTPClient(nil),
  291. treeV4: bool_tree.NewTreeV4(),
  292. treeV6: bool_tree.NewTreeV6(),
  293. workerPool: workerPool,
  294. remoteURLs: remoteURLs,
  295. localFiles: localFiles,
  296. }, nil
  297. }