psinet.go 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496
  1. /*
  2. * Copyright (c) 2016, Psiphon Inc.
  3. * All rights reserved.
  4. *
  5. * This program is free software: you can redistribute it and/or modify
  6. * it under the terms of the GNU General Public License as published by
  7. * the Free Software Foundation, either version 3 of the License, or
  8. * (at your option) any later version.
  9. *
  10. * This program is distributed in the hope that it will be useful,
  11. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13. * GNU General Public License for more details.
  14. *
  15. * You should have received a copy of the GNU General Public License
  16. * along with this program. If not, see <http://www.gnu.org/licenses/>.
  17. *
  18. */
  19. // Package psiphon/server/psinet implements psinet database services. The psinet
  20. // database is a JSON-format file containing information about the Psiphon network,
  21. // including sponsors, home pages, stats regexes, available upgrades, and other
  22. // servers for discovery. This package also implements the Psiphon discovery algorithm.
  23. package psinet
  24. import (
  25. "encoding/hex"
  26. "encoding/json"
  27. "fmt"
  28. "math"
  29. "math/rand"
  30. "strconv"
  31. "strings"
  32. "time"
  33. "github.com/Psiphon-Labs/psiphon-tunnel-core/psiphon/common"
  34. )
  35. // Database serves Psiphon API data requests. It's safe for
  36. // concurrent usage. The Reload function supports hot reloading
  37. // of Psiphon network data while the server is running.
  38. type Database struct {
  39. common.ReloadableFile
  40. Hosts map[string]Host `json:"hosts"`
  41. Servers []Server `json:"servers"`
  42. Sponsors map[string]Sponsor `json:"sponsors"`
  43. Versions map[string][]ClientVersion `json:"client_versions"`
  44. }
  45. type Host struct {
  46. DatacenterName string `json:"datacenter_name"`
  47. Id string `json:"id"`
  48. IpAddress string `json:"ip_address"`
  49. IsTCS bool `json:"is_TCS"`
  50. MeekCookieEncryptionPublicKey string `json:"meek_cookie_encryption_public_key"`
  51. MeekServerObfuscatedKey string `json:"meek_server_obfuscated_key"`
  52. MeekServerPort int `json:"meek_server_port"`
  53. Region string `json:"region"`
  54. }
  55. type Server struct {
  56. AlternateSshObfuscatedPorts []string `json:"alternate_ssh_obfuscated_ports"`
  57. Capabilities map[string]bool `json:"capabilities"`
  58. DiscoveryDateRange []string `json:"discovery_date_range"`
  59. EgressIpAddress string `json:"egress_ip_address"`
  60. HostId string `json:"host_id"`
  61. Id string `json:"id"`
  62. InternalIpAddress string `json:"internal_ip_address"`
  63. IpAddress string `json:"ip_address"`
  64. IsEmbedded bool `json:"is_embedded"`
  65. IsPermanent bool `json:"is_permanent"`
  66. PropogationChannelId string `json:"propagation_channel_id"`
  67. SshHostKey string `json:"ssh_host_key"`
  68. SshObfuscatedKey string `json:"ssh_obfuscated_key"`
  69. SshObfuscatedPort int `json:"ssh_obfuscated_port"`
  70. SshPassword string `json:"ssh_password"`
  71. SshPort string `json:"ssh_port"`
  72. SshUsername string `json:"ssh_username"`
  73. WebServerCertificate string `json:"web_server_certificate"`
  74. WebServerPort string `json:"web_server_port"`
  75. WebServerSecret string `json:"web_server_secret"`
  76. }
  77. type Sponsor struct {
  78. Banner string
  79. HomePages map[string][]HomePage `json:"home_pages"`
  80. HttpsRequestRegexes []HttpsRequestRegex `json:"https_request_regexes"`
  81. Id string `json:"id"`
  82. MobileHomePages map[string][]HomePage `json:"mobile_home_pages"`
  83. Name string `json:"name"`
  84. PageViewRegexes []PageViewRegex `json:"page_view_regexes"`
  85. WebsiteBanner string `json:"website_banner"`
  86. WebsiteBannerLink string `json:"website_banner_link"`
  87. }
  88. type ClientVersion struct {
  89. Version string `json:"version"`
  90. }
  91. type HomePage struct {
  92. Region string `json:"region"`
  93. Url string `json:"url"`
  94. }
  95. type HttpsRequestRegex struct {
  96. Regex string `json:"regex"`
  97. Replace string `json:"replace"`
  98. }
  99. type MobileHomePage struct {
  100. Region string `json:"region"`
  101. Url string `json:"url"`
  102. }
  103. type PageViewRegex struct {
  104. Regex string `json:"regex"`
  105. Replace string `json:"replace"`
  106. }
  107. // NewDatabase initializes a Database, calling Reload on the specified
  108. // filename.
  109. func NewDatabase(filename string) (*Database, error) {
  110. database := &Database{}
  111. database.ReloadableFile = common.NewReloadableFile(
  112. filename,
  113. func(fileContent []byte) error {
  114. err := json.Unmarshal(fileContent, &database)
  115. if err != nil {
  116. // On error, state remains the same
  117. // (Unmarshal first validates the provided
  118. // JOSN and then populates the interface)
  119. return common.ContextError(err)
  120. }
  121. return nil
  122. })
  123. _, err := database.Reload()
  124. if err != nil {
  125. return nil, common.ContextError(err)
  126. }
  127. return database, nil
  128. }
  129. // GetRandomHomepage returns a random home page from a set of home pages
  130. // for the specified sponsor, region, and platform.
  131. func (db *Database) GetRandomHomepage(sponsorID, clientRegion string, isMobilePlatform bool) []string {
  132. homepages := db.GetHomepages(sponsorID, clientRegion, isMobilePlatform)
  133. if len(homepages) > 0 {
  134. index := rand.Intn(len(homepages))
  135. return homepages[index : index+1]
  136. }
  137. return homepages
  138. }
  139. // GetHomepages returns a list of home pages for the specified sponsor,
  140. // region, and platform.
  141. func (db *Database) GetHomepages(sponsorID, clientRegion string, isMobilePlatform bool) []string {
  142. db.ReloadableFile.RLock()
  143. defer db.ReloadableFile.RUnlock()
  144. sponsorHomePages := make([]string, 0)
  145. // Sponsor id does not exist: fail gracefully
  146. sponsor, ok := db.Sponsors[sponsorID]
  147. if !ok {
  148. return sponsorHomePages
  149. }
  150. homePages := sponsor.HomePages
  151. if isMobilePlatform {
  152. if sponsor.MobileHomePages != nil {
  153. homePages = sponsor.MobileHomePages
  154. }
  155. }
  156. // Case: lookup succeeded and corresponding homepages found for region
  157. homePagesByRegion, ok := homePages[clientRegion]
  158. if ok {
  159. for _, homePage := range homePagesByRegion {
  160. sponsorHomePages = append(sponsorHomePages, strings.Replace(homePage.Url, "client_region=XX", "client_region="+clientRegion, 1))
  161. }
  162. }
  163. // Case: lookup failed or no corresponding homepages found for region --> use default
  164. if len(sponsorHomePages) == 0 {
  165. defaultHomePages, ok := homePages["None"]
  166. if ok {
  167. for _, homePage := range defaultHomePages {
  168. // client_region query parameter substitution
  169. sponsorHomePages = append(sponsorHomePages, strings.Replace(homePage.Url, "client_region=XX", "client_region="+clientRegion, 1))
  170. }
  171. }
  172. }
  173. return sponsorHomePages
  174. }
  175. // GetUpgradeClientVersion returns a new client version when an upgrade is
  176. // indicated for the specified client current version. The result is "" when
  177. // no upgrade is available. Caller should normalize clientPlatform.
  178. func (db *Database) GetUpgradeClientVersion(clientVersion, clientPlatform string) string {
  179. db.ReloadableFile.RLock()
  180. defer db.ReloadableFile.RUnlock()
  181. // Check lastest version number against client version number
  182. clientVersions, ok := db.Versions[clientPlatform]
  183. if !ok {
  184. return ""
  185. }
  186. if len(clientVersions) == 0 {
  187. return ""
  188. }
  189. // NOTE: Assumes versions list is in ascending version order
  190. lastVersion := clientVersions[len(clientVersions)-1].Version
  191. lastVersionInt, err := strconv.Atoi(lastVersion)
  192. if err != nil {
  193. return ""
  194. }
  195. clientVersionInt, err := strconv.Atoi(clientVersion)
  196. if err != nil {
  197. return ""
  198. }
  199. // Return latest version if upgrade needed
  200. if lastVersionInt > clientVersionInt {
  201. return lastVersion
  202. }
  203. return ""
  204. }
  205. // GetHttpsRequestRegexes returns bytes transferred stats regexes for the
  206. // specified sponsor. The result is nil when an unknown sponsorID is provided.
  207. func (db *Database) GetHttpsRequestRegexes(sponsorID string) []map[string]string {
  208. db.ReloadableFile.RLock()
  209. defer db.ReloadableFile.RUnlock()
  210. regexes := make([]map[string]string, 0)
  211. for i := range db.Sponsors[sponsorID].HttpsRequestRegexes {
  212. regex := make(map[string]string)
  213. regex["replace"] = db.Sponsors[sponsorID].HttpsRequestRegexes[i].Replace
  214. regex["regex"] = db.Sponsors[sponsorID].HttpsRequestRegexes[i].Regex
  215. regexes = append(regexes, regex)
  216. }
  217. return regexes
  218. }
  219. // DiscoverServers selects new encoded server entries to be "discovered" by
  220. // the client, using the discoveryValue as the input into the discovery algorithm.
  221. // The server list (db.Servers) loaded from JSON is stored as an array instead of
  222. // a map to ensure servers are discovered deterministically. Each iteration over a
  223. // map in go is seeded with a random value which causes non-deterministic ordering.
  224. func (db *Database) DiscoverServers(discoveryValue int) []string {
  225. db.ReloadableFile.RLock()
  226. defer db.ReloadableFile.RUnlock()
  227. var servers []Server
  228. discoveryDate := time.Now().UTC()
  229. candidateServers := make([]Server, 0)
  230. for _, server := range db.Servers {
  231. var start time.Time
  232. var end time.Time
  233. var err error
  234. // All servers that are discoverable on this day are eligable for discovery
  235. if len(server.DiscoveryDateRange) != 0 {
  236. start, err = time.Parse("2006-01-02T15:04:05", server.DiscoveryDateRange[0])
  237. if err != nil {
  238. continue
  239. }
  240. end, err = time.Parse("2006-01-02T15:04:05", server.DiscoveryDateRange[1])
  241. if err != nil {
  242. continue
  243. }
  244. if discoveryDate.After(start) && discoveryDate.Before(end) {
  245. candidateServers = append(candidateServers, server)
  246. }
  247. }
  248. }
  249. servers = selectServers(candidateServers, discoveryValue)
  250. encodedServerEntries := make([]string, 0)
  251. for _, server := range servers {
  252. encodedServerEntries = append(encodedServerEntries, db.getEncodedServerEntry(server))
  253. }
  254. return encodedServerEntries
  255. }
  256. // Combine client IP address and time-of-day strategies to give out different
  257. // discovery servers to different clients. The aim is to achieve defense against
  258. // enumerability. We also want to achieve a degree of load balancing clients
  259. // and these strategies are expected to have reasonably random distribution,
  260. // even for a cluster of users coming from the same network.
  261. //
  262. // We only select one server: multiple results makes enumeration easier; the
  263. // strategies have a built-in load balancing effect; and date range discoverability
  264. // means a client will actually learn more servers later even if they happen to
  265. // always pick the same result at this point.
  266. //
  267. // This is a blended strategy: as long as there are enough servers to pick from,
  268. // both aspects determine which server is selected. IP address is given the
  269. // priority: if there are only a couple of servers, for example, IP address alone
  270. // determines the outcome.
  271. func selectServers(servers []Server, discoveryValue int) []Server {
  272. TIME_GRANULARITY := 3600
  273. if len(servers) == 0 {
  274. return nil
  275. }
  276. // Current time truncated to an hour
  277. timeInSeconds := int(time.Now().Unix())
  278. timeStrategyValue := timeInSeconds / TIME_GRANULARITY
  279. // Divide servers into buckets. The bucket count is chosen such that the number
  280. // of buckets and the number of items in each bucket are close (using sqrt).
  281. // IP address selects the bucket, time selects the item in the bucket.
  282. // NOTE: this code assumes that the range of possible timeStrategyValues
  283. // and discoveryValues are sufficient to index to all bucket items.
  284. bucketCount := calculateBucketCount(len(servers))
  285. buckets := bucketizeServerList(servers, bucketCount)
  286. bucket := buckets[discoveryValue%len(buckets)]
  287. server := bucket[timeStrategyValue%len(bucket)]
  288. serverList := make([]Server, 1)
  289. serverList[0] = server
  290. return serverList
  291. }
  292. // Number of buckets such that first strategy picks among about the same number
  293. // of choices as the second strategy. Gives an edge to the "outer" strategy.
  294. func calculateBucketCount(length int) int {
  295. return int(math.Ceil(math.Sqrt(float64(length))))
  296. }
  297. // Create bucketCount buckets.
  298. // Each bucket will be of size division or divison-1.
  299. func bucketizeServerList(servers []Server, bucketCount int) [][]Server {
  300. division := float64(len(servers)) / float64(bucketCount)
  301. buckets := make([][]Server, bucketCount)
  302. var currentBucketIndex int = 0
  303. var serverIndex int = 0
  304. for _, server := range servers {
  305. bucketEndIndex := int(math.Floor(division * (float64(currentBucketIndex) + 1)))
  306. buckets[currentBucketIndex] = append(buckets[currentBucketIndex], server)
  307. serverIndex++
  308. if serverIndex > bucketEndIndex {
  309. currentBucketIndex++
  310. }
  311. }
  312. return buckets
  313. }
  314. // Return hex encoded server entry string for comsumption by client.
  315. // Newer clients ignore the legacy fields and only utilize the extended (new) config.
  316. func (db *Database) getEncodedServerEntry(server Server) string {
  317. host, hostExists := db.Hosts[server.HostId]
  318. if !hostExists {
  319. return ""
  320. }
  321. // TCS web server certificate has PEM headers and newlines, so strip those now
  322. // for legacy format compatibility
  323. webServerCertificate := server.WebServerCertificate
  324. if host.IsTCS {
  325. splitCert := strings.Split(server.WebServerCertificate, "\n")
  326. if len(splitCert) <= 2 {
  327. webServerCertificate = ""
  328. } else {
  329. webServerCertificate = strings.Join(splitCert[1:len(splitCert)-2], "")
  330. }
  331. }
  332. // Double-check that we're not giving our blank server credentials
  333. if len(server.IpAddress) <= 1 || len(server.WebServerPort) <= 1 || len(server.WebServerSecret) <= 1 || len(webServerCertificate) <= 1 {
  334. return ""
  335. }
  336. // Extended (new) entry fields are in a JSON string
  337. var extendedConfig struct {
  338. IpAddress string `json:"ipAddress"`
  339. WebServerPort string `json:"webServerPort"` // not an int
  340. WebServerSecret string `json:"webServerSecret"`
  341. WebServerCertificate string `json:"webServerCertificate"`
  342. SshPort int `json:"sshPort"`
  343. SshUsername string `json:"sshUsername"`
  344. SshPassword string `json:"sshPassword"`
  345. SshHostKey string `json:"sshHostKey"`
  346. SshObfuscatedPort int `json:"sshObfuscatedPort"`
  347. SshObfuscatedKey string `json:"sshObfuscatedKey"`
  348. Capabilities []string `json:"capabilities"`
  349. Region string `json:"region"`
  350. MeekServerPort int `json:"meekServerPort"`
  351. MeekCookieEncryptionPublicKey string `json:"meekCookieEncryptionPublicKey"`
  352. MeekObfuscatedKey string `json:"meekObfuscatedKey"`
  353. }
  354. // NOTE: also putting original values in extended config for easier parsing by new clients
  355. extendedConfig.IpAddress = server.IpAddress
  356. extendedConfig.WebServerPort = server.WebServerPort
  357. extendedConfig.WebServerSecret = server.WebServerSecret
  358. extendedConfig.WebServerCertificate = webServerCertificate
  359. sshPort, err := strconv.Atoi(server.SshPort)
  360. if err != nil {
  361. extendedConfig.SshPort = 0
  362. } else {
  363. extendedConfig.SshPort = sshPort
  364. }
  365. extendedConfig.SshUsername = server.SshUsername
  366. extendedConfig.SshPassword = server.SshPassword
  367. sshHostKeyType, sshHostKey := parseSshKeyString(server.SshHostKey)
  368. if strings.Compare(sshHostKeyType, "ssh-rsa") == 0 {
  369. extendedConfig.SshHostKey = sshHostKey
  370. } else {
  371. extendedConfig.SshHostKey = ""
  372. }
  373. extendedConfig.SshObfuscatedPort = server.SshObfuscatedPort
  374. // Use the latest alternate port unless tunneling through meek
  375. if len(server.AlternateSshObfuscatedPorts) > 0 && !server.Capabilities["UNFRONTED-MEEK"] {
  376. port, err := strconv.Atoi(server.AlternateSshObfuscatedPorts[len(server.AlternateSshObfuscatedPorts)-1])
  377. if err == nil {
  378. extendedConfig.SshObfuscatedPort = port
  379. }
  380. }
  381. extendedConfig.SshObfuscatedKey = server.SshObfuscatedKey
  382. extendedConfig.Region = host.Region
  383. extendedConfig.MeekCookieEncryptionPublicKey = host.MeekCookieEncryptionPublicKey
  384. extendedConfig.MeekServerPort = host.MeekServerPort
  385. extendedConfig.MeekObfuscatedKey = host.MeekServerObfuscatedKey
  386. serverCapabilities := make(map[string]bool, 0)
  387. for capability, enabled := range server.Capabilities {
  388. serverCapabilities[capability] = enabled
  389. }
  390. if serverCapabilities["UNFRONTED-MEEK"] && host.MeekServerPort == 443 {
  391. serverCapabilities["UNFRONTED-MEEK"] = false
  392. serverCapabilities["UNFRONTED-MEEK-HTTPS"] = true
  393. }
  394. for capability, enabled := range serverCapabilities {
  395. if enabled == true {
  396. extendedConfig.Capabilities = append(extendedConfig.Capabilities, capability)
  397. }
  398. }
  399. jsonDump, err := json.Marshal(extendedConfig)
  400. if err != nil {
  401. return ""
  402. }
  403. // Legacy format + extended (new) config
  404. prefixString := fmt.Sprintf("%s %s %s %s ", server.IpAddress, server.WebServerPort, server.WebServerSecret, webServerCertificate)
  405. return hex.EncodeToString(append([]byte(prefixString)[:], []byte(jsonDump)[:]...))
  406. }
  407. // Parse string of format "ssh-key-type ssh-key".
  408. func parseSshKeyString(sshKeyString string) (keyType string, key string) {
  409. sshKeyArr := strings.Split(sshKeyString, " ")
  410. if len(sshKeyArr) != 2 {
  411. return "", ""
  412. }
  413. return sshKeyArr[0], sshKeyArr[1]
  414. }