psinet.go 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394
  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/json"
  26. "math"
  27. "math/rand"
  28. "strconv"
  29. "strings"
  30. "time"
  31. "github.com/Psiphon-Labs/psiphon-tunnel-core/psiphon/common"
  32. )
  33. const (
  34. MAX_DATABASE_AGE_FOR_SERVER_ENTRY_VALIDITY = 48 * time.Hour
  35. )
  36. // Database serves Psiphon API data requests. It's safe for
  37. // concurrent usage. The Reload function supports hot reloading
  38. // of Psiphon network data while the server is running.
  39. type Database struct {
  40. common.ReloadableFile
  41. Sponsors map[string]*Sponsor `json:"sponsors"`
  42. Versions map[string][]ClientVersion `json:"client_versions"`
  43. DefaultSponsorID string `json:"default_sponsor_id"`
  44. ValidServerEntryTags map[string]bool `json:"valid_server_entry_tags"`
  45. DiscoveryServers []*DiscoveryServer `json:"discovery_servers`
  46. fileModTime time.Time
  47. }
  48. type DiscoveryServer struct {
  49. DiscoveryDateRange []time.Time `json:"discovery_date_range"`
  50. EncodedServerEntry string `json:"encoded_server_entry"`
  51. }
  52. type Sponsor struct {
  53. Banner string
  54. HomePages map[string][]HomePage `json:"home_pages"`
  55. HttpsRequestRegexes []HttpsRequestRegex `json:"https_request_regexes"`
  56. Id string `json:"id"`
  57. MobileHomePages map[string][]HomePage `json:"mobile_home_pages"`
  58. Name string `json:"name"`
  59. PageViewRegexes []PageViewRegex `json:"page_view_regexes"`
  60. WebsiteBanner string `json:"website_banner"`
  61. WebsiteBannerLink string `json:"website_banner_link"`
  62. }
  63. type ClientVersion struct {
  64. Version string `json:"version"`
  65. }
  66. type HomePage struct {
  67. Region string `json:"region"`
  68. Url string `json:"url"`
  69. }
  70. type HttpsRequestRegex struct {
  71. Regex string `json:"regex"`
  72. Replace string `json:"replace"`
  73. }
  74. type MobileHomePage struct {
  75. Region string `json:"region"`
  76. Url string `json:"url"`
  77. }
  78. type PageViewRegex struct {
  79. Regex string `json:"regex"`
  80. Replace string `json:"replace"`
  81. }
  82. // NewDatabase initializes a Database, calling Reload on the specified
  83. // filename.
  84. func NewDatabase(filename string) (*Database, error) {
  85. database := &Database{}
  86. database.ReloadableFile = common.NewReloadableFile(
  87. filename,
  88. true,
  89. func(fileContent []byte, fileModTime time.Time) error {
  90. var newDatabase Database
  91. err := json.Unmarshal(fileContent, &newDatabase)
  92. if err != nil {
  93. return common.ContextError(err)
  94. }
  95. // Note: an unmarshal directly into &database would fail
  96. // to reset to zero value fields not present in the JSON.
  97. database.Sponsors = newDatabase.Sponsors
  98. database.Versions = newDatabase.Versions
  99. database.DefaultSponsorID = newDatabase.DefaultSponsorID
  100. database.ValidServerEntryTags = newDatabase.ValidServerEntryTags
  101. database.DiscoveryServers = newDatabase.DiscoveryServers
  102. database.fileModTime = fileModTime
  103. return nil
  104. })
  105. _, err := database.Reload()
  106. if err != nil {
  107. return nil, common.ContextError(err)
  108. }
  109. return database, nil
  110. }
  111. // GetRandomizedHomepages returns a randomly ordered list of home pages
  112. // for the specified sponsor, region, and platform.
  113. func (db *Database) GetRandomizedHomepages(sponsorID, clientRegion string, isMobilePlatform bool) []string {
  114. homepages := db.GetHomepages(sponsorID, clientRegion, isMobilePlatform)
  115. if len(homepages) > 1 {
  116. shuffledHomepages := make([]string, len(homepages))
  117. perm := rand.Perm(len(homepages))
  118. for i, v := range perm {
  119. shuffledHomepages[v] = homepages[i]
  120. }
  121. return shuffledHomepages
  122. }
  123. return homepages
  124. }
  125. // GetHomepages returns a list of home pages for the specified sponsor,
  126. // region, and platform.
  127. func (db *Database) GetHomepages(sponsorID, clientRegion string, isMobilePlatform bool) []string {
  128. db.ReloadableFile.RLock()
  129. defer db.ReloadableFile.RUnlock()
  130. sponsorHomePages := make([]string, 0)
  131. // Sponsor id does not exist: fail gracefully
  132. sponsor, ok := db.Sponsors[sponsorID]
  133. if !ok {
  134. sponsor, ok = db.Sponsors[db.DefaultSponsorID]
  135. if !ok {
  136. return sponsorHomePages
  137. }
  138. }
  139. homePages := sponsor.HomePages
  140. if isMobilePlatform {
  141. if len(sponsor.MobileHomePages) > 0 {
  142. homePages = sponsor.MobileHomePages
  143. }
  144. }
  145. // Case: lookup succeeded and corresponding homepages found for region
  146. homePagesByRegion, ok := homePages[clientRegion]
  147. if ok {
  148. for _, homePage := range homePagesByRegion {
  149. sponsorHomePages = append(sponsorHomePages, strings.Replace(homePage.Url, "client_region=XX", "client_region="+clientRegion, 1))
  150. }
  151. }
  152. // Case: lookup failed or no corresponding homepages found for region --> use default
  153. if len(sponsorHomePages) == 0 {
  154. defaultHomePages, ok := homePages["None"]
  155. if ok {
  156. for _, homePage := range defaultHomePages {
  157. // client_region query parameter substitution
  158. sponsorHomePages = append(sponsorHomePages, strings.Replace(homePage.Url, "client_region=XX", "client_region="+clientRegion, 1))
  159. }
  160. }
  161. }
  162. return sponsorHomePages
  163. }
  164. // GetUpgradeClientVersion returns a new client version when an upgrade is
  165. // indicated for the specified client current version. The result is "" when
  166. // no upgrade is available. Caller should normalize clientPlatform.
  167. func (db *Database) GetUpgradeClientVersion(clientVersion, clientPlatform string) string {
  168. db.ReloadableFile.RLock()
  169. defer db.ReloadableFile.RUnlock()
  170. // Check lastest version number against client version number
  171. clientVersions, ok := db.Versions[clientPlatform]
  172. if !ok {
  173. return ""
  174. }
  175. if len(clientVersions) == 0 {
  176. return ""
  177. }
  178. // NOTE: Assumes versions list is in ascending version order
  179. lastVersion := clientVersions[len(clientVersions)-1].Version
  180. lastVersionInt, err := strconv.Atoi(lastVersion)
  181. if err != nil {
  182. return ""
  183. }
  184. clientVersionInt, err := strconv.Atoi(clientVersion)
  185. if err != nil {
  186. return ""
  187. }
  188. // Return latest version if upgrade needed
  189. if lastVersionInt > clientVersionInt {
  190. return lastVersion
  191. }
  192. return ""
  193. }
  194. // GetHttpsRequestRegexes returns bytes transferred stats regexes for the
  195. // specified sponsor.
  196. func (db *Database) GetHttpsRequestRegexes(sponsorID string) []map[string]string {
  197. db.ReloadableFile.RLock()
  198. defer db.ReloadableFile.RUnlock()
  199. regexes := make([]map[string]string, 0)
  200. sponsor, ok := db.Sponsors[sponsorID]
  201. if !ok {
  202. sponsor, _ = db.Sponsors[db.DefaultSponsorID]
  203. }
  204. // If neither sponsorID or DefaultSponsorID were found, sponsor will be the
  205. // zero value of the map, an empty Sponsor struct.
  206. for _, sponsorRegex := range sponsor.HttpsRequestRegexes {
  207. regex := make(map[string]string)
  208. regex["replace"] = sponsorRegex.Replace
  209. regex["regex"] = sponsorRegex.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 -- a function of the client's IP
  216. // address -- as the input into the discovery algorithm.
  217. func (db *Database) DiscoverServers(discoveryValue int) []string {
  218. db.ReloadableFile.RLock()
  219. defer db.ReloadableFile.RUnlock()
  220. var servers []*DiscoveryServer
  221. discoveryDate := time.Now().UTC()
  222. candidateServers := make([]*DiscoveryServer, 0)
  223. for _, server := range db.DiscoveryServers {
  224. // All servers that are discoverable on this day are eligible for discovery
  225. if len(server.DiscoveryDateRange) == 2 &&
  226. discoveryDate.After(server.DiscoveryDateRange[0]) &&
  227. discoveryDate.Before(server.DiscoveryDateRange[1]) {
  228. candidateServers = append(candidateServers, server)
  229. }
  230. }
  231. timeInSeconds := int(discoveryDate.Unix())
  232. servers = selectServers(candidateServers, timeInSeconds, discoveryValue)
  233. encodedServerEntries := make([]string, 0)
  234. for _, server := range servers {
  235. encodedServerEntries = append(encodedServerEntries, server.EncodedServerEntry)
  236. }
  237. return encodedServerEntries
  238. }
  239. // Combine client IP address and time-of-day strategies to give out different
  240. // discovery servers to different clients. The aim is to achieve defense against
  241. // enumerability. We also want to achieve a degree of load balancing clients
  242. // and these strategies are expected to have reasonably random distribution,
  243. // even for a cluster of users coming from the same network.
  244. //
  245. // We only select one server: multiple results makes enumeration easier; the
  246. // strategies have a built-in load balancing effect; and date range discoverability
  247. // means a client will actually learn more servers later even if they happen to
  248. // always pick the same result at this point.
  249. //
  250. // This is a blended strategy: as long as there are enough servers to pick from,
  251. // both aspects determine which server is selected. IP address is given the
  252. // priority: if there are only a couple of servers, for example, IP address alone
  253. // determines the outcome.
  254. func selectServers(
  255. servers []*DiscoveryServer, timeInSeconds, discoveryValue int) []*DiscoveryServer {
  256. TIME_GRANULARITY := 3600
  257. if len(servers) == 0 {
  258. return nil
  259. }
  260. // Time truncated to an hour
  261. timeStrategyValue := timeInSeconds / TIME_GRANULARITY
  262. // Divide servers into buckets. The bucket count is chosen such that the number
  263. // of buckets and the number of items in each bucket are close (using sqrt).
  264. // IP address selects the bucket, time selects the item in the bucket.
  265. // NOTE: this code assumes that the range of possible timeStrategyValues
  266. // and discoveryValues are sufficient to index to all bucket items.
  267. bucketCount := calculateBucketCount(len(servers))
  268. buckets := bucketizeServerList(servers, bucketCount)
  269. if len(buckets) == 0 {
  270. return nil
  271. }
  272. bucket := buckets[discoveryValue%len(buckets)]
  273. if len(bucket) == 0 {
  274. return nil
  275. }
  276. server := bucket[timeStrategyValue%len(bucket)]
  277. serverList := make([]*DiscoveryServer, 1)
  278. serverList[0] = server
  279. return serverList
  280. }
  281. // Number of buckets such that first strategy picks among about the same number
  282. // of choices as the second strategy. Gives an edge to the "outer" strategy.
  283. func calculateBucketCount(length int) int {
  284. return int(math.Ceil(math.Sqrt(float64(length))))
  285. }
  286. // bucketizeServerList creates nearly equal sized slices of the input list.
  287. func bucketizeServerList(servers []*DiscoveryServer, bucketCount int) [][]*DiscoveryServer {
  288. // This code creates the same partitions as legacy servers:
  289. // https://bitbucket.org/psiphon/psiphon-circumvention-system/src/03bc1a7e51e7c85a816e370bb3a6c755fd9c6fee/Automation/psi_ops_discovery.py
  290. //
  291. // Both use the same algorithm from:
  292. // http://stackoverflow.com/questions/2659900/python-slicing-a-list-into-n-nearly-equal-length-partitions
  293. // TODO: this partition is constant for fixed Database content, so it could
  294. // be done once and cached in the Database ReloadableFile reloadAction.
  295. buckets := make([][]*DiscoveryServer, bucketCount)
  296. division := float64(len(servers)) / float64(bucketCount)
  297. for i := 0; i < bucketCount; i++ {
  298. start := int((division * float64(i)) + 0.5)
  299. end := int((division * (float64(i) + 1)) + 0.5)
  300. buckets[i] = servers[start:end]
  301. }
  302. return buckets
  303. }
  304. // IsValidServerEntryTag checks if the specified server entry tag is valid.
  305. func (db *Database) IsValidServerEntryTag(serverEntryTag string) bool {
  306. db.ReloadableFile.RLock()
  307. defer db.ReloadableFile.RUnlock()
  308. // Default to "valid" if the valid list is unexpectedly empty or stale. This
  309. // helps prevent premature client-side server-entry pruning when there is an
  310. // issue with updating the database.
  311. if len(db.ValidServerEntryTags) == 0 ||
  312. db.fileModTime.Add(MAX_DATABASE_AGE_FOR_SERVER_ENTRY_VALIDITY).Before(time.Now()) {
  313. return true
  314. }
  315. // The tag must be in the map and have the value "true".
  316. return db.ValidServerEntryTags[serverEntryTag]
  317. }