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
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

firehol.go 6.7KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340
  1. package ipblocklist
  2. import (
  3. "bufio"
  4. "context"
  5. "fmt"
  6. "io"
  7. "io/ioutil"
  8. "net"
  9. "net/http"
  10. "net/url"
  11. "os"
  12. "regexp"
  13. "strings"
  14. "sync"
  15. "time"
  16. "github.com/9seconds/mtg/v2/mtglib"
  17. "github.com/kentik/patricia"
  18. "github.com/kentik/patricia/bool_tree"
  19. "github.com/panjf2000/ants"
  20. )
  21. const (
  22. fireholIPv4DefaultCIDR = 32
  23. fireholIPv6DefaultCIDR = 128
  24. )
  25. var fireholRegexpComment = regexp.MustCompile(`\s*#.*?$`)
  26. type Firehol struct {
  27. ctx context.Context
  28. ctxCancel context.CancelFunc
  29. logger mtglib.Logger
  30. rwMutex sync.RWMutex
  31. remoteURLs []string
  32. localFiles []string
  33. httpClient *http.Client
  34. workerPool *ants.Pool
  35. treeV4 *bool_tree.TreeV4
  36. treeV6 *bool_tree.TreeV6
  37. }
  38. func (f *Firehol) Contains(ip net.IP) bool {
  39. if ip == nil {
  40. return true
  41. }
  42. ip4 := ip.To4()
  43. f.rwMutex.RLock()
  44. defer f.rwMutex.RUnlock()
  45. if ip4 != nil {
  46. return f.containsIPv4(ip4)
  47. }
  48. return f.containsIPv6(ip.To16())
  49. }
  50. func (f *Firehol) containsIPv4(addr net.IP) bool {
  51. ip := patricia.NewIPv4AddressFromBytes(addr, 32)
  52. if ok, _, err := f.treeV4.FindDeepestTag(ip); ok && err == nil {
  53. return true
  54. }
  55. return false
  56. }
  57. func (f *Firehol) containsIPv6(addr net.IP) bool {
  58. ip := patricia.NewIPv6Address(addr, 128)
  59. if ok, _, err := f.treeV6.FindDeepestTag(ip); ok && err == nil {
  60. return true
  61. }
  62. return false
  63. }
  64. func (f *Firehol) Run(updateEach time.Duration) {
  65. ticker := time.NewTicker(updateEach)
  66. defer func() {
  67. ticker.Stop()
  68. select {
  69. case <-ticker.C:
  70. default:
  71. }
  72. }()
  73. if err := f.update(); err != nil {
  74. f.logger.WarningError("cannot update blocklist", err)
  75. }
  76. for {
  77. select {
  78. case <-f.ctx.Done():
  79. return
  80. case <-ticker.C:
  81. if err := f.update(); err != nil {
  82. f.logger.WarningError("cannot update blocklist", err)
  83. }
  84. }
  85. }
  86. }
  87. func (f *Firehol) Shutdown() {
  88. f.ctxCancel()
  89. }
  90. func (f *Firehol) update() error { // nolint: funlen, cyclop
  91. ctx, cancel := context.WithCancel(f.ctx)
  92. defer cancel()
  93. wg := &sync.WaitGroup{}
  94. wg.Add(len(f.remoteURLs) + len(f.localFiles))
  95. treeMutex := &sync.Mutex{}
  96. v4tree := bool_tree.NewTreeV4()
  97. v6tree := bool_tree.NewTreeV6()
  98. errorChan := make(chan error, 1)
  99. defer close(errorChan)
  100. for _, v := range f.localFiles {
  101. go func(filename string) {
  102. defer wg.Done()
  103. if err := f.updateLocalFile(ctx, filename, treeMutex, v4tree, v6tree); err != nil {
  104. cancel()
  105. f.logger.BindStr("filename", filename).WarningError("cannot update", err)
  106. select {
  107. case errorChan <- err:
  108. default:
  109. }
  110. }
  111. }(v)
  112. }
  113. for _, v := range f.remoteURLs {
  114. value := v
  115. f.workerPool.Submit(func() { // nolint: errcheck
  116. defer wg.Done()
  117. if err := f.updateRemoteURL(ctx, value, treeMutex, v4tree, v6tree); err != nil {
  118. cancel()
  119. f.logger.BindStr("url", value).WarningError("cannot update", err)
  120. select {
  121. case errorChan <- err:
  122. default:
  123. }
  124. }
  125. })
  126. }
  127. wg.Wait()
  128. select {
  129. case err := <-errorChan:
  130. return fmt.Errorf("cannot update trees: %w", err)
  131. default:
  132. }
  133. f.rwMutex.Lock()
  134. defer f.rwMutex.Unlock()
  135. f.treeV4 = v4tree
  136. f.treeV6 = v6tree
  137. return nil
  138. }
  139. func (f *Firehol) updateLocalFile(ctx context.Context, filename string,
  140. mutex sync.Locker,
  141. v4tree *bool_tree.TreeV4, v6tree *bool_tree.TreeV6) error {
  142. filefp, err := os.Open(filename)
  143. if err != nil {
  144. return fmt.Errorf("cannot open file: %w", err)
  145. }
  146. defer filefp.Close()
  147. return f.updateTrees(ctx, mutex, filefp, v4tree, v6tree)
  148. }
  149. func (f *Firehol) updateRemoteURL(ctx context.Context, url string,
  150. mutex sync.Locker,
  151. v4tree *bool_tree.TreeV4, v6tree *bool_tree.TreeV6) error {
  152. req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
  153. if err != nil {
  154. return fmt.Errorf("cannot build a request: %w", err)
  155. }
  156. resp, err := f.httpClient.Do(req)
  157. if err != nil {
  158. return fmt.Errorf("cannot request a remote URL %s: %w", url, err)
  159. }
  160. defer func() {
  161. io.Copy(ioutil.Discard, resp.Body) // nolint: errcheck
  162. resp.Body.Close()
  163. }()
  164. return f.updateTrees(ctx, mutex, resp.Body, v4tree, v6tree)
  165. }
  166. func (f *Firehol) updateTrees(ctx context.Context,
  167. mutex sync.Locker,
  168. reader io.Reader,
  169. v4tree *bool_tree.TreeV4,
  170. v6tree *bool_tree.TreeV6) error {
  171. scanner := bufio.NewScanner(reader)
  172. for scanner.Scan() {
  173. select {
  174. case <-ctx.Done():
  175. return ctx.Err()
  176. default:
  177. }
  178. text := scanner.Text()
  179. text = fireholRegexpComment.ReplaceAllLiteralString(text, "")
  180. text = strings.TrimSpace(text)
  181. if text == "" {
  182. continue
  183. }
  184. ip, cidr, err := f.updateParseLine(text)
  185. if err != nil {
  186. return fmt.Errorf("cannot parse a line: %w", err)
  187. }
  188. if err := f.updateAddToTrees(ip, cidr, mutex, v4tree, v6tree); err != nil {
  189. return fmt.Errorf("cannot add a node to the tree: %w", err)
  190. }
  191. }
  192. if scanner.Err() != nil {
  193. return fmt.Errorf("cannot parse a response: %w", scanner.Err())
  194. }
  195. return nil
  196. }
  197. func (f *Firehol) updateParseLine(text string) (net.IP, uint, error) {
  198. _, ipnet, err := net.ParseCIDR(text)
  199. if err != nil {
  200. ipaddr := net.ParseIP(text)
  201. if ipaddr == nil {
  202. return nil, 0, fmt.Errorf("incorrect ip address %s", text)
  203. }
  204. ip4 := ipaddr.To4()
  205. if ip4 != nil {
  206. return ip4, fireholIPv4DefaultCIDR, nil
  207. }
  208. return ipaddr.To16(), fireholIPv6DefaultCIDR, nil
  209. }
  210. ones, _ := ipnet.Mask.Size()
  211. return ipnet.IP, uint(ones), nil
  212. }
  213. func (f *Firehol) updateAddToTrees(ip net.IP, cidr uint,
  214. mutex sync.Locker,
  215. v4tree *bool_tree.TreeV4, v6tree *bool_tree.TreeV6) error {
  216. mutex.Lock()
  217. defer mutex.Unlock()
  218. if ip.To4() != nil {
  219. addr := patricia.NewIPv4AddressFromBytes(ip, cidr)
  220. if _, _, err := v4tree.Set(addr, true); err != nil {
  221. return err // nolint: wrapcheck
  222. }
  223. } else {
  224. addr := patricia.NewIPv6Address(ip, cidr)
  225. if _, _, err := v6tree.Set(addr, true); err != nil {
  226. return err // nolint: wrapcheck
  227. }
  228. }
  229. return nil
  230. }
  231. func NewFirehol(logger mtglib.Logger, network mtglib.Network,
  232. downloadConcurrency uint,
  233. remoteURLs []string,
  234. localFiles []string) (*Firehol, error) {
  235. for _, v := range remoteURLs {
  236. parsed, err := url.Parse(v)
  237. if err != nil {
  238. return nil, fmt.Errorf("incorrect url %s: %w", v, err)
  239. }
  240. switch parsed.Scheme {
  241. case "http", "https":
  242. default:
  243. return nil, fmt.Errorf("unsupported url %s", v)
  244. }
  245. }
  246. for _, v := range localFiles {
  247. if stat, err := os.Stat(v); os.IsNotExist(err) || stat.IsDir() || stat.Mode().Perm()&0o400 == 0 {
  248. return nil, fmt.Errorf("%s is not a readable file", v)
  249. }
  250. }
  251. if downloadConcurrency == 0 {
  252. downloadConcurrency = 1
  253. }
  254. workerPool, _ := ants.NewPool(int(downloadConcurrency))
  255. ctx, cancel := context.WithCancel(context.Background())
  256. return &Firehol{
  257. ctx: ctx,
  258. ctxCancel: cancel,
  259. logger: logger.Named("firehol"),
  260. httpClient: network.MakeHTTPClient(nil),
  261. treeV4: bool_tree.NewTreeV4(),
  262. treeV6: bool_tree.NewTreeV6(),
  263. workerPool: workerPool,
  264. remoteURLs: remoteURLs,
  265. localFiles: localFiles,
  266. }, nil
  267. }