psinet.go 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549
  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 psinet implements psinet database services. The psinet database is a
  20. // JSON-format file containing information about the Psiphon network, including
  21. // sponsors, home pages, stats regexes, available upgrades, and other servers for
  22. // 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. DefaultSponsorID string `json:"default_sponsor_id"`
  45. }
  46. type Host struct {
  47. DatacenterName string `json:"datacenter_name"`
  48. Id string `json:"id"`
  49. IpAddress string `json:"ip_address"`
  50. IsTCS bool `json:"is_TCS"`
  51. MeekCookieEncryptionPublicKey string `json:"meek_cookie_encryption_public_key"`
  52. MeekServerObfuscatedKey string `json:"meek_server_obfuscated_key"`
  53. MeekServerPort int `json:"meek_server_port"`
  54. TacticsRequestPublicKey string `json:"tactics_request_public_key"`
  55. TacticsRequestObfuscatedKey string `json:"tactics_request_obfuscated_key"`
  56. Region string `json:"region"`
  57. }
  58. type Server struct {
  59. AlternateSshObfuscatedPorts []string `json:"alternate_ssh_obfuscated_ports"`
  60. Capabilities map[string]bool `json:"capabilities"`
  61. DiscoveryDateRange []string `json:"discovery_date_range"`
  62. EgressIpAddress string `json:"egress_ip_address"`
  63. HostId string `json:"host_id"`
  64. Id string `json:"id"`
  65. InternalIpAddress string `json:"internal_ip_address"`
  66. IpAddress string `json:"ip_address"`
  67. IsEmbedded bool `json:"is_embedded"`
  68. IsPermanent bool `json:"is_permanent"`
  69. PropogationChannelId string `json:"propagation_channel_id"`
  70. SshHostKey string `json:"ssh_host_key"`
  71. SshObfuscatedKey string `json:"ssh_obfuscated_key"`
  72. SshObfuscatedPort int `json:"ssh_obfuscated_port"`
  73. SshObfuscatedQUICPort int `json:"ssh_obfuscated_quic_port"`
  74. SshObfuscatedTapdancePort int `json:"ssh_obfuscated_tapdance_port"`
  75. SshPassword string `json:"ssh_password"`
  76. SshPort string `json:"ssh_port"`
  77. SshUsername string `json:"ssh_username"`
  78. WebServerCertificate string `json:"web_server_certificate"`
  79. WebServerPort string `json:"web_server_port"`
  80. WebServerSecret string `json:"web_server_secret"`
  81. ConfigurationVersion int `json:"configuration_version"`
  82. }
  83. type Sponsor struct {
  84. Banner string
  85. HomePages map[string][]HomePage `json:"home_pages"`
  86. HttpsRequestRegexes []HttpsRequestRegex `json:"https_request_regexes"`
  87. Id string `json:"id"`
  88. MobileHomePages map[string][]HomePage `json:"mobile_home_pages"`
  89. Name string `json:"name"`
  90. PageViewRegexes []PageViewRegex `json:"page_view_regexes"`
  91. WebsiteBanner string `json:"website_banner"`
  92. WebsiteBannerLink string `json:"website_banner_link"`
  93. }
  94. type ClientVersion struct {
  95. Version string `json:"version"`
  96. }
  97. type HomePage struct {
  98. Region string `json:"region"`
  99. Url string `json:"url"`
  100. }
  101. type HttpsRequestRegex struct {
  102. Regex string `json:"regex"`
  103. Replace string `json:"replace"`
  104. }
  105. type MobileHomePage struct {
  106. Region string `json:"region"`
  107. Url string `json:"url"`
  108. }
  109. type PageViewRegex struct {
  110. Regex string `json:"regex"`
  111. Replace string `json:"replace"`
  112. }
  113. // NewDatabase initializes a Database, calling Reload on the specified
  114. // filename.
  115. func NewDatabase(filename string) (*Database, error) {
  116. database := &Database{}
  117. database.ReloadableFile = common.NewReloadableFile(
  118. filename,
  119. func(fileContent []byte) error {
  120. var newDatabase Database
  121. err := json.Unmarshal(fileContent, &newDatabase)
  122. if err != nil {
  123. return common.ContextError(err)
  124. }
  125. // Note: an unmarshal directly into &database would fail
  126. // to reset to zero value fields not present in the JSON.
  127. database.Hosts = newDatabase.Hosts
  128. database.Servers = newDatabase.Servers
  129. database.Sponsors = newDatabase.Sponsors
  130. database.Versions = newDatabase.Versions
  131. database.DefaultSponsorID = newDatabase.DefaultSponsorID
  132. return nil
  133. })
  134. _, err := database.Reload()
  135. if err != nil {
  136. return nil, common.ContextError(err)
  137. }
  138. return database, nil
  139. }
  140. // GetRandomizedHomepages returns a randomly ordered list of home pages
  141. // for the specified sponsor, region, and platform.
  142. func (db *Database) GetRandomizedHomepages(sponsorID, clientRegion string, isMobilePlatform bool) []string {
  143. homepages := db.GetHomepages(sponsorID, clientRegion, isMobilePlatform)
  144. if len(homepages) > 1 {
  145. shuffledHomepages := make([]string, len(homepages))
  146. perm := rand.Perm(len(homepages))
  147. for i, v := range perm {
  148. shuffledHomepages[v] = homepages[i]
  149. }
  150. return shuffledHomepages
  151. }
  152. return homepages
  153. }
  154. // GetHomepages returns a list of home pages for the specified sponsor,
  155. // region, and platform.
  156. func (db *Database) GetHomepages(sponsorID, clientRegion string, isMobilePlatform bool) []string {
  157. db.ReloadableFile.RLock()
  158. defer db.ReloadableFile.RUnlock()
  159. sponsorHomePages := make([]string, 0)
  160. // Sponsor id does not exist: fail gracefully
  161. sponsor, ok := db.Sponsors[sponsorID]
  162. if !ok {
  163. sponsor, ok = db.Sponsors[db.DefaultSponsorID]
  164. if !ok {
  165. return sponsorHomePages
  166. }
  167. }
  168. homePages := sponsor.HomePages
  169. if isMobilePlatform {
  170. if len(sponsor.MobileHomePages) > 0 {
  171. homePages = sponsor.MobileHomePages
  172. }
  173. }
  174. // Case: lookup succeeded and corresponding homepages found for region
  175. homePagesByRegion, ok := homePages[clientRegion]
  176. if ok {
  177. for _, homePage := range homePagesByRegion {
  178. sponsorHomePages = append(sponsorHomePages, strings.Replace(homePage.Url, "client_region=XX", "client_region="+clientRegion, 1))
  179. }
  180. }
  181. // Case: lookup failed or no corresponding homepages found for region --> use default
  182. if len(sponsorHomePages) == 0 {
  183. defaultHomePages, ok := homePages["None"]
  184. if ok {
  185. for _, homePage := range defaultHomePages {
  186. // client_region query parameter substitution
  187. sponsorHomePages = append(sponsorHomePages, strings.Replace(homePage.Url, "client_region=XX", "client_region="+clientRegion, 1))
  188. }
  189. }
  190. }
  191. return sponsorHomePages
  192. }
  193. // GetUpgradeClientVersion returns a new client version when an upgrade is
  194. // indicated for the specified client current version. The result is "" when
  195. // no upgrade is available. Caller should normalize clientPlatform.
  196. func (db *Database) GetUpgradeClientVersion(clientVersion, clientPlatform string) string {
  197. db.ReloadableFile.RLock()
  198. defer db.ReloadableFile.RUnlock()
  199. // Check lastest version number against client version number
  200. clientVersions, ok := db.Versions[clientPlatform]
  201. if !ok {
  202. return ""
  203. }
  204. if len(clientVersions) == 0 {
  205. return ""
  206. }
  207. // NOTE: Assumes versions list is in ascending version order
  208. lastVersion := clientVersions[len(clientVersions)-1].Version
  209. lastVersionInt, err := strconv.Atoi(lastVersion)
  210. if err != nil {
  211. return ""
  212. }
  213. clientVersionInt, err := strconv.Atoi(clientVersion)
  214. if err != nil {
  215. return ""
  216. }
  217. // Return latest version if upgrade needed
  218. if lastVersionInt > clientVersionInt {
  219. return lastVersion
  220. }
  221. return ""
  222. }
  223. // GetHttpsRequestRegexes returns bytes transferred stats regexes for the
  224. // specified sponsor.
  225. func (db *Database) GetHttpsRequestRegexes(sponsorID string) []map[string]string {
  226. db.ReloadableFile.RLock()
  227. defer db.ReloadableFile.RUnlock()
  228. regexes := make([]map[string]string, 0)
  229. sponsor, ok := db.Sponsors[sponsorID]
  230. if !ok {
  231. sponsor, _ = db.Sponsors[db.DefaultSponsorID]
  232. }
  233. // If neither sponsorID or DefaultSponsorID were found, sponsor will be the
  234. // zero value of the map, an empty Sponsor struct.
  235. for _, sponsorRegex := range sponsor.HttpsRequestRegexes {
  236. regex := make(map[string]string)
  237. regex["replace"] = sponsorRegex.Replace
  238. regex["regex"] = sponsorRegex.Regex
  239. regexes = append(regexes, regex)
  240. }
  241. return regexes
  242. }
  243. // DiscoverServers selects new encoded server entries to be "discovered" by
  244. // the client, using the discoveryValue -- a function of the client's IP
  245. // address -- as the input into the discovery algorithm.
  246. // The server list (db.Servers) loaded from JSON is stored as an array instead of
  247. // a map to ensure servers are discovered deterministically. Each iteration over a
  248. // map in go is seeded with a random value which causes non-deterministic ordering.
  249. func (db *Database) DiscoverServers(discoveryValue int) []string {
  250. db.ReloadableFile.RLock()
  251. defer db.ReloadableFile.RUnlock()
  252. var servers []Server
  253. discoveryDate := time.Now().UTC()
  254. candidateServers := make([]Server, 0)
  255. for _, server := range db.Servers {
  256. var start time.Time
  257. var end time.Time
  258. var err error
  259. // All servers that are discoverable on this day are eligible for discovery
  260. if len(server.DiscoveryDateRange) != 0 {
  261. start, err = time.Parse("2006-01-02T15:04:05", server.DiscoveryDateRange[0])
  262. if err != nil {
  263. continue
  264. }
  265. end, err = time.Parse("2006-01-02T15:04:05", server.DiscoveryDateRange[1])
  266. if err != nil {
  267. continue
  268. }
  269. if discoveryDate.After(start) && discoveryDate.Before(end) {
  270. candidateServers = append(candidateServers, server)
  271. }
  272. }
  273. }
  274. timeInSeconds := int(discoveryDate.Unix())
  275. servers = selectServers(candidateServers, timeInSeconds, discoveryValue)
  276. encodedServerEntries := make([]string, 0)
  277. for _, server := range servers {
  278. encodedServerEntries = append(encodedServerEntries, db.getEncodedServerEntry(server))
  279. }
  280. return encodedServerEntries
  281. }
  282. // Combine client IP address and time-of-day strategies to give out different
  283. // discovery servers to different clients. The aim is to achieve defense against
  284. // enumerability. We also want to achieve a degree of load balancing clients
  285. // and these strategies are expected to have reasonably random distribution,
  286. // even for a cluster of users coming from the same network.
  287. //
  288. // We only select one server: multiple results makes enumeration easier; the
  289. // strategies have a built-in load balancing effect; and date range discoverability
  290. // means a client will actually learn more servers later even if they happen to
  291. // always pick the same result at this point.
  292. //
  293. // This is a blended strategy: as long as there are enough servers to pick from,
  294. // both aspects determine which server is selected. IP address is given the
  295. // priority: if there are only a couple of servers, for example, IP address alone
  296. // determines the outcome.
  297. func selectServers(servers []Server, timeInSeconds, discoveryValue int) []Server {
  298. TIME_GRANULARITY := 3600
  299. if len(servers) == 0 {
  300. return nil
  301. }
  302. // Time truncated to an hour
  303. timeStrategyValue := timeInSeconds / TIME_GRANULARITY
  304. // Divide servers into buckets. The bucket count is chosen such that the number
  305. // of buckets and the number of items in each bucket are close (using sqrt).
  306. // IP address selects the bucket, time selects the item in the bucket.
  307. // NOTE: this code assumes that the range of possible timeStrategyValues
  308. // and discoveryValues are sufficient to index to all bucket items.
  309. bucketCount := calculateBucketCount(len(servers))
  310. buckets := bucketizeServerList(servers, bucketCount)
  311. if len(buckets) == 0 {
  312. return nil
  313. }
  314. bucket := buckets[discoveryValue%len(buckets)]
  315. if len(bucket) == 0 {
  316. return nil
  317. }
  318. server := bucket[timeStrategyValue%len(bucket)]
  319. serverList := make([]Server, 1)
  320. serverList[0] = server
  321. return serverList
  322. }
  323. // Number of buckets such that first strategy picks among about the same number
  324. // of choices as the second strategy. Gives an edge to the "outer" strategy.
  325. func calculateBucketCount(length int) int {
  326. return int(math.Ceil(math.Sqrt(float64(length))))
  327. }
  328. // bucketizeServerList creates nearly equal sized slices of the input list.
  329. func bucketizeServerList(servers []Server, bucketCount int) [][]Server {
  330. // This code creates the same partitions as legacy servers:
  331. // https://bitbucket.org/psiphon/psiphon-circumvention-system/src/03bc1a7e51e7c85a816e370bb3a6c755fd9c6fee/Automation/psi_ops_discovery.py
  332. //
  333. // Both use the same algorithm from:
  334. // http://stackoverflow.com/questions/2659900/python-slicing-a-list-into-n-nearly-equal-length-partitions
  335. // TODO: this partition is constant for fixed Database content, so it could
  336. // be done once and cached in the Database ReloadableFile reloadAction.
  337. buckets := make([][]Server, bucketCount)
  338. division := float64(len(servers)) / float64(bucketCount)
  339. for i := 0; i < bucketCount; i++ {
  340. start := int((division * float64(i)) + 0.5)
  341. end := int((division * (float64(i) + 1)) + 0.5)
  342. buckets[i] = servers[start:end]
  343. }
  344. return buckets
  345. }
  346. // Return hex encoded server entry string for comsumption by client.
  347. // Newer clients ignore the legacy fields and only utilize the extended (new) config.
  348. func (db *Database) getEncodedServerEntry(server Server) string {
  349. host, hostExists := db.Hosts[server.HostId]
  350. if !hostExists {
  351. return ""
  352. }
  353. // TCS web server certificate has PEM headers and newlines, so strip those now
  354. // for legacy format compatibility
  355. webServerCertificate := server.WebServerCertificate
  356. if host.IsTCS {
  357. splitCert := strings.Split(server.WebServerCertificate, "\n")
  358. if len(splitCert) <= 2 {
  359. webServerCertificate = ""
  360. } else {
  361. webServerCertificate = strings.Join(splitCert[1:len(splitCert)-2], "")
  362. }
  363. }
  364. // Double-check that we're not giving our blank server credentials
  365. if len(server.IpAddress) <= 1 || len(server.WebServerPort) <= 1 || len(server.WebServerSecret) <= 1 || len(webServerCertificate) <= 1 {
  366. return ""
  367. }
  368. // Extended (new) entry fields are in a JSON string
  369. var extendedConfig struct {
  370. IpAddress string `json:"ipAddress"`
  371. WebServerPort string `json:"webServerPort"` // not an int
  372. WebServerSecret string `json:"webServerSecret"`
  373. WebServerCertificate string `json:"webServerCertificate"`
  374. SshPort int `json:"sshPort"`
  375. SshUsername string `json:"sshUsername"`
  376. SshPassword string `json:"sshPassword"`
  377. SshHostKey string `json:"sshHostKey"`
  378. SshObfuscatedPort int `json:"sshObfuscatedPort"`
  379. SshObfuscatedQUICPort int `json:"sshObfuscatedQUICPort"`
  380. SshObfuscatedTapdancePort int `json:"sshObfuscatedTapdancePort"`
  381. SshObfuscatedKey string `json:"sshObfuscatedKey"`
  382. Capabilities []string `json:"capabilities"`
  383. Region string `json:"region"`
  384. MeekServerPort int `json:"meekServerPort"`
  385. MeekCookieEncryptionPublicKey string `json:"meekCookieEncryptionPublicKey"`
  386. MeekObfuscatedKey string `json:"meekObfuscatedKey"`
  387. TacticsRequestPublicKey string `json:"tacticsRequestPublicKey"`
  388. TacticsRequestObfuscatedKey string `json:"tacticsRequestObfuscatedKey"`
  389. ConfigurationVersion int `json:"configurationVersion"`
  390. }
  391. // NOTE: also putting original values in extended config for easier parsing by new clients
  392. extendedConfig.IpAddress = server.IpAddress
  393. extendedConfig.WebServerPort = server.WebServerPort
  394. extendedConfig.WebServerSecret = server.WebServerSecret
  395. extendedConfig.WebServerCertificate = webServerCertificate
  396. sshPort, err := strconv.Atoi(server.SshPort)
  397. if err != nil {
  398. extendedConfig.SshPort = 0
  399. } else {
  400. extendedConfig.SshPort = sshPort
  401. }
  402. extendedConfig.SshUsername = server.SshUsername
  403. extendedConfig.SshPassword = server.SshPassword
  404. sshHostKeyType, sshHostKey := parseSshKeyString(server.SshHostKey)
  405. if strings.Compare(sshHostKeyType, "ssh-rsa") == 0 {
  406. extendedConfig.SshHostKey = sshHostKey
  407. } else {
  408. extendedConfig.SshHostKey = ""
  409. }
  410. extendedConfig.SshObfuscatedPort = server.SshObfuscatedPort
  411. // Use the latest alternate port unless tunneling through meek
  412. if len(server.AlternateSshObfuscatedPorts) > 0 && !server.Capabilities["UNFRONTED-MEEK"] {
  413. port, err := strconv.Atoi(server.AlternateSshObfuscatedPorts[len(server.AlternateSshObfuscatedPorts)-1])
  414. if err == nil {
  415. extendedConfig.SshObfuscatedPort = port
  416. }
  417. }
  418. extendedConfig.SshObfuscatedQUICPort = server.SshObfuscatedQUICPort
  419. extendedConfig.SshObfuscatedTapdancePort = server.SshObfuscatedTapdancePort
  420. extendedConfig.SshObfuscatedKey = server.SshObfuscatedKey
  421. extendedConfig.Region = host.Region
  422. extendedConfig.MeekCookieEncryptionPublicKey = host.MeekCookieEncryptionPublicKey
  423. extendedConfig.MeekServerPort = host.MeekServerPort
  424. extendedConfig.MeekObfuscatedKey = host.MeekServerObfuscatedKey
  425. extendedConfig.TacticsRequestPublicKey = host.TacticsRequestPublicKey
  426. extendedConfig.TacticsRequestObfuscatedKey = host.TacticsRequestObfuscatedKey
  427. serverCapabilities := make(map[string]bool, 0)
  428. for capability, enabled := range server.Capabilities {
  429. serverCapabilities[capability] = enabled
  430. }
  431. if serverCapabilities["UNFRONTED-MEEK"] && host.MeekServerPort == 443 {
  432. serverCapabilities["UNFRONTED-MEEK"] = false
  433. serverCapabilities["UNFRONTED-MEEK-HTTPS"] = true
  434. }
  435. for capability, enabled := range serverCapabilities {
  436. if enabled == true {
  437. extendedConfig.Capabilities = append(extendedConfig.Capabilities, capability)
  438. }
  439. }
  440. extendedConfig.ConfigurationVersion = server.ConfigurationVersion
  441. jsonDump, err := json.Marshal(extendedConfig)
  442. if err != nil {
  443. return ""
  444. }
  445. // Legacy format + extended (new) config
  446. prefixString := fmt.Sprintf("%s %s %s %s ", server.IpAddress, server.WebServerPort, server.WebServerSecret, webServerCertificate)
  447. return hex.EncodeToString(append([]byte(prefixString)[:], []byte(jsonDump)[:]...))
  448. }
  449. // Parse string of format "ssh-key-type ssh-key".
  450. func parseSshKeyString(sshKeyString string) (keyType string, key string) {
  451. sshKeyArr := strings.Split(sshKeyString, " ")
  452. if len(sshKeyArr) != 2 {
  453. return "", ""
  454. }
  455. return sshKeyArr[0], sshKeyArr[1]
  456. }