psinet.go 16 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490
  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. "io/ioutil"
  29. "math"
  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(filename string) error {
  114. psinetJSON, err := ioutil.ReadFile(filename)
  115. if err != nil {
  116. // On error, state remains the same
  117. return common.ContextError(err)
  118. }
  119. err = json.Unmarshal(psinetJSON, &database)
  120. if err != nil {
  121. // On error, state remains the same
  122. // (Unmarshal first validates the provided
  123. // JOSN and then populates the interface)
  124. return common.ContextError(err)
  125. }
  126. return nil
  127. })
  128. _, err := database.Reload()
  129. if err != nil {
  130. return nil, common.ContextError(err)
  131. }
  132. return database, nil
  133. }
  134. // GetHomepages returns a list of home pages for the specified sponsor,
  135. // region, and platform.
  136. func (db *Database) GetHomepages(sponsorID, clientRegion string, isMobilePlatform bool) []string {
  137. db.ReloadableFile.RLock()
  138. defer db.ReloadableFile.RUnlock()
  139. sponsorHomePages := make([]string, 0)
  140. // Sponsor id does not exist: fail gracefully
  141. sponsor, ok := db.Sponsors[sponsorID]
  142. if !ok {
  143. return nil
  144. }
  145. homePages := sponsor.HomePages
  146. if isMobilePlatform {
  147. if sponsor.MobileHomePages != nil {
  148. homePages = sponsor.MobileHomePages
  149. }
  150. }
  151. // Case: lookup succeeded and corresponding homepages found for region
  152. homePagesByRegion, ok := homePages[clientRegion]
  153. if ok {
  154. for _, homePage := range homePagesByRegion {
  155. sponsorHomePages = append(sponsorHomePages, strings.Replace(homePage.Url, "client_region=XX", "client_region="+clientRegion, 1))
  156. }
  157. }
  158. // Case: lookup failed or no corresponding homepages found for region --> use default
  159. if sponsorHomePages == nil {
  160. defaultHomePages, ok := homePages["None"]
  161. if ok {
  162. for _, homePage := range defaultHomePages {
  163. // client_region query parameter substitution
  164. sponsorHomePages = append(sponsorHomePages, strings.Replace(homePage.Url, "client_region=XX", "client_region="+clientRegion, 1))
  165. }
  166. }
  167. }
  168. return sponsorHomePages
  169. }
  170. // GetUpgradeClientVersion returns a new client version when an upgrade is
  171. // indicated for the specified client current version. The result is "" when
  172. // no upgrade is available. Caller should normalize clientPlatform.
  173. func (db *Database) GetUpgradeClientVersion(clientVersion, clientPlatform string) string {
  174. db.ReloadableFile.RLock()
  175. defer db.ReloadableFile.RUnlock()
  176. // Check lastest version number against client version number
  177. clientVersions, ok := db.Versions[clientPlatform]
  178. if !ok {
  179. return ""
  180. }
  181. if len(clientVersions) == 0 {
  182. return ""
  183. }
  184. // NOTE: Assumes versions list is in ascending version order
  185. lastVersion := clientVersions[len(clientVersions)-1].Version
  186. lastVersionInt, err := strconv.Atoi(lastVersion)
  187. if err != nil {
  188. return ""
  189. }
  190. clientVersionInt, err := strconv.Atoi(clientVersion)
  191. if err != nil {
  192. return ""
  193. }
  194. // Return latest version if upgrade needed
  195. if lastVersionInt > clientVersionInt {
  196. return lastVersion
  197. }
  198. return ""
  199. }
  200. // GetHttpsRequestRegexes returns bytes transferred stats regexes for the
  201. // specified sponsor. The result is nil when an unknown sponsorID is provided.
  202. func (db *Database) GetHttpsRequestRegexes(sponsorID string) []map[string]string {
  203. db.ReloadableFile.RLock()
  204. defer db.ReloadableFile.RUnlock()
  205. regexes := make([]map[string]string, 0)
  206. for i := range db.Sponsors[sponsorID].HttpsRequestRegexes {
  207. regex := make(map[string]string)
  208. regex["replace"] = db.Sponsors[sponsorID].HttpsRequestRegexes[i].Replace
  209. regex["regex"] = db.Sponsors[sponsorID].HttpsRequestRegexes[i].Regex
  210. regexes = append(regexes, regex)
  211. }
  212. return regexes
  213. }
  214. // DiscoverServers selects new encoded server entries to be "discovered" by
  215. // the client, using the discoveryValue as the input into the discovery algorithm.
  216. // The server list (db.Servers) loaded from JSON is stored as an array instead of
  217. // a map to ensure servers are discovered deterministically. Each iteration over a
  218. // map in go is seeded with a random value which causes non-deterministic ordering.
  219. func (db *Database) DiscoverServers(discoveryValue int) []string {
  220. db.ReloadableFile.RLock()
  221. defer db.ReloadableFile.RUnlock()
  222. var servers []Server
  223. discoveryDate := time.Now().UTC()
  224. candidateServers := make([]Server, 0)
  225. for _, server := range db.Servers {
  226. var start time.Time
  227. var end time.Time
  228. var err error
  229. // All servers that are discoverable on this day are eligable for discovery
  230. if len(server.DiscoveryDateRange) != 0 {
  231. start, err = time.Parse("2006-01-02T15:04:05", server.DiscoveryDateRange[0])
  232. if err != nil {
  233. continue
  234. }
  235. end, err = time.Parse("2006-01-02T15:04:05", server.DiscoveryDateRange[1])
  236. if err != nil {
  237. continue
  238. }
  239. if discoveryDate.After(start) && discoveryDate.Before(end) {
  240. candidateServers = append(candidateServers, server)
  241. }
  242. }
  243. }
  244. servers = selectServers(candidateServers, discoveryValue)
  245. encodedServerEntries := make([]string, 0)
  246. for _, server := range servers {
  247. encodedServerEntries = append(encodedServerEntries, db.getEncodedServerEntry(server))
  248. }
  249. return encodedServerEntries
  250. }
  251. // Combine client IP address and time-of-day strategies to give out different
  252. // discovery servers to different clients. The aim is to achieve defense against
  253. // enumerability. We also want to achieve a degree of load balancing clients
  254. // and these strategies are expected to have reasonably random distribution,
  255. // even for a cluster of users coming from the same network.
  256. //
  257. // We only select one server: multiple results makes enumeration easier; the
  258. // strategies have a built-in load balancing effect; and date range discoverability
  259. // means a client will actually learn more servers later even if they happen to
  260. // always pick the same result at this point.
  261. //
  262. // This is a blended strategy: as long as there are enough servers to pick from,
  263. // both aspects determine which server is selected. IP address is given the
  264. // priority: if there are only a couple of servers, for example, IP address alone
  265. // determines the outcome.
  266. func selectServers(servers []Server, discoveryValue int) []Server {
  267. TIME_GRANULARITY := 3600
  268. if len(servers) == 0 {
  269. return nil
  270. }
  271. // Current time truncated to an hour
  272. timeInSeconds := int(time.Now().Unix())
  273. timeStrategyValue := timeInSeconds / TIME_GRANULARITY
  274. // Divide servers into buckets. The bucket count is chosen such that the number
  275. // of buckets and the number of items in each bucket are close (using sqrt).
  276. // IP address selects the bucket, time selects the item in the bucket.
  277. // NOTE: this code assumes that the range of possible timeStrategyValues
  278. // and discoveryValues are sufficient to index to all bucket items.
  279. bucketCount := calculateBucketCount(len(servers))
  280. buckets := bucketizeServerList(servers, bucketCount)
  281. bucket := buckets[discoveryValue%len(buckets)]
  282. server := bucket[timeStrategyValue%len(bucket)]
  283. serverList := make([]Server, 1)
  284. serverList[0] = server
  285. return serverList
  286. }
  287. // Number of buckets such that first strategy picks among about the same number
  288. // of choices as the second strategy. Gives an edge to the "outer" strategy.
  289. func calculateBucketCount(length int) int {
  290. return int(math.Ceil(math.Sqrt(float64(length))))
  291. }
  292. // Create bucketCount buckets.
  293. // Each bucket will be of size division or divison-1.
  294. func bucketizeServerList(servers []Server, bucketCount int) [][]Server {
  295. division := float64(len(servers)) / float64(bucketCount)
  296. buckets := make([][]Server, bucketCount)
  297. var currentBucketIndex int = 0
  298. var serverIndex int = 0
  299. for _, server := range servers {
  300. bucketEndIndex := int(math.Floor(division * (float64(currentBucketIndex) + 1)))
  301. buckets[currentBucketIndex] = append(buckets[currentBucketIndex], server)
  302. serverIndex++
  303. if serverIndex > bucketEndIndex {
  304. currentBucketIndex++
  305. }
  306. }
  307. return buckets
  308. }
  309. // Return hex encoded server entry string for comsumption by client.
  310. // Newer clients ignore the legacy fields and only utilize the extended (new) config.
  311. func (db *Database) getEncodedServerEntry(server Server) string {
  312. host, hostExists := db.Hosts[server.HostId]
  313. if !hostExists {
  314. return ""
  315. }
  316. // TCS web server certificate has PEM headers and newlines, so strip those now
  317. // for legacy format compatibility
  318. webServerCertificate := server.WebServerCertificate
  319. if host.IsTCS {
  320. splitCert := strings.Split(server.WebServerCertificate, "\n")
  321. if len(splitCert) <= 2 {
  322. webServerCertificate = ""
  323. } else {
  324. webServerCertificate = strings.Join(splitCert[1:len(splitCert)-2], "")
  325. }
  326. }
  327. // Double-check that we're not giving our blank server credentials
  328. if len(server.IpAddress) <= 1 || len(server.WebServerPort) <= 1 || len(server.WebServerSecret) <= 1 || len(webServerCertificate) <= 1 {
  329. return ""
  330. }
  331. // Extended (new) entry fields are in a JSON string
  332. var extendedConfig struct {
  333. IpAddress string
  334. WebServerPort string
  335. WebServerSecret string
  336. WebServerCertificate string
  337. SshPort int
  338. SshUsername string
  339. SshPassword string
  340. SshHostKey string
  341. SshObfuscatedPort int
  342. SshObfuscatedKey string
  343. Region string
  344. MeekCookieEncryptionPublicKey string
  345. MeekObfuscatedKey string
  346. MeekServerPort int
  347. capabilities []string
  348. }
  349. // NOTE: also putting original values in extended config for easier parsing by new clients
  350. extendedConfig.IpAddress = server.IpAddress
  351. extendedConfig.WebServerPort = server.WebServerPort
  352. extendedConfig.WebServerSecret = server.WebServerSecret
  353. extendedConfig.WebServerCertificate = webServerCertificate
  354. sshPort, err := strconv.Atoi(server.SshPort)
  355. if err != nil {
  356. extendedConfig.SshPort = 0
  357. } else {
  358. extendedConfig.SshPort = sshPort
  359. }
  360. extendedConfig.SshUsername = server.SshUsername
  361. extendedConfig.SshPassword = server.SshPassword
  362. sshHostKeyType, sshHostKey := parseSshKeyString(server.SshHostKey)
  363. if strings.Compare(sshHostKeyType, "ssh-rsa") == 0 {
  364. extendedConfig.SshHostKey = sshHostKey
  365. } else {
  366. extendedConfig.SshHostKey = ""
  367. }
  368. extendedConfig.SshObfuscatedPort = server.SshObfuscatedPort
  369. // Use the latest alternate port unless tunneling through meek
  370. if len(server.AlternateSshObfuscatedPorts) > 0 && !server.Capabilities["UNFRONTED-MEEK"] {
  371. port, err := strconv.Atoi(server.AlternateSshObfuscatedPorts[len(server.AlternateSshObfuscatedPorts)-1])
  372. if err == nil {
  373. extendedConfig.SshObfuscatedPort = port
  374. }
  375. }
  376. extendedConfig.SshObfuscatedKey = server.SshObfuscatedKey
  377. extendedConfig.Region = host.Region
  378. extendedConfig.MeekCookieEncryptionPublicKey = host.MeekCookieEncryptionPublicKey
  379. extendedConfig.MeekServerPort = host.MeekServerPort
  380. extendedConfig.MeekObfuscatedKey = host.MeekServerObfuscatedKey
  381. serverCapabilities := make(map[string]bool, 0)
  382. for capability, enabled := range server.Capabilities {
  383. serverCapabilities[capability] = enabled
  384. }
  385. if serverCapabilities["UNFRONTED-MEEK"] && host.MeekServerPort == 443 {
  386. serverCapabilities["UNFRONTED-MEEK"] = false
  387. serverCapabilities["UNFRONTED-MEEK-HTTPS"] = true
  388. }
  389. for capability, enabled := range serverCapabilities {
  390. if enabled == true {
  391. extendedConfig.capabilities = append(extendedConfig.capabilities, capability)
  392. }
  393. }
  394. jsonDump, err := json.Marshal(extendedConfig)
  395. if err != nil {
  396. return ""
  397. }
  398. // Legacy format + extended (new) config
  399. prefixString := fmt.Sprintf("%s %s %s %s ", server.IpAddress, server.WebServerPort, server.WebServerSecret, webServerCertificate)
  400. return hex.EncodeToString(append([]byte(prefixString)[:], []byte(jsonDump)[:]...))
  401. }
  402. // Parse string of format "ssh-key-type ssh-key".
  403. func parseSshKeyString(sshKeyString string) (keyType string, key string) {
  404. sshKeyArr := strings.Split(sshKeyString, " ")
  405. if len(sshKeyArr) != 2 {
  406. return "", ""
  407. }
  408. return sshKeyArr[0], sshKeyArr[1]
  409. }