backfill.go 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378
  1. package main
  2. import (
  3. "context"
  4. "crypto/sha256"
  5. "encoding/base64"
  6. "fmt"
  7. "sort"
  8. "github.com/bwmarrin/discordgo"
  9. "github.com/rs/zerolog"
  10. "maunium.net/go/mautrix"
  11. "maunium.net/go/mautrix/bridge/bridgeconfig"
  12. "maunium.net/go/mautrix/event"
  13. "maunium.net/go/mautrix/id"
  14. "go.mau.fi/mautrix-discord/database"
  15. )
  16. func (portal *Portal) forwardBackfillInitial(source *User, thread *Thread) {
  17. defer portal.forwardBackfillLock.Unlock()
  18. // This should only be called from CreateMatrixRoom which locks forwardBackfillLock before creating the room.
  19. if portal.forwardBackfillLock.TryLock() {
  20. panic("forwardBackfillInitial() called without locking forwardBackfillLock")
  21. }
  22. limit := portal.bridge.Config.Bridge.Backfill.Limits.Initial.Channel
  23. if portal.GuildID == "" {
  24. limit = portal.bridge.Config.Bridge.Backfill.Limits.Initial.DM
  25. if thread != nil {
  26. limit = portal.bridge.Config.Bridge.Backfill.Limits.Initial.Thread
  27. thread.initialBackfillAttempted = true
  28. }
  29. }
  30. if limit == 0 {
  31. return
  32. }
  33. with := portal.log.With().
  34. Str("action", "initial backfill").
  35. Str("room_id", portal.MXID.String()).
  36. Int("limit", limit)
  37. if thread != nil {
  38. with = with.Str("thread_id", thread.ID)
  39. }
  40. log := with.Logger()
  41. portal.backfillLimited(log, source, limit, "", thread)
  42. }
  43. func (portal *Portal) ForwardBackfillMissed(source *User, serverLastMessageID string, thread *Thread) {
  44. if portal.MXID == "" {
  45. return
  46. }
  47. limit := portal.bridge.Config.Bridge.Backfill.Limits.Missed.Channel
  48. if portal.GuildID == "" {
  49. limit = portal.bridge.Config.Bridge.Backfill.Limits.Missed.DM
  50. if thread != nil {
  51. limit = portal.bridge.Config.Bridge.Backfill.Limits.Missed.Thread
  52. }
  53. }
  54. if limit == 0 {
  55. return
  56. }
  57. with := portal.log.With().
  58. Str("action", "missed event backfill").
  59. Str("room_id", portal.MXID.String()).
  60. Int("limit", limit)
  61. if thread != nil {
  62. with = with.Str("thread_id", thread.ID)
  63. }
  64. log := with.Logger()
  65. portal.forwardBackfillLock.Lock()
  66. defer portal.forwardBackfillLock.Unlock()
  67. var lastMessage *database.Message
  68. if thread != nil {
  69. lastMessage = portal.bridge.DB.Message.GetLastInThread(portal.Key, thread.ID)
  70. } else {
  71. lastMessage = portal.bridge.DB.Message.GetLast(portal.Key)
  72. }
  73. if lastMessage == nil || serverLastMessageID == "" {
  74. log.Debug().Msg("Not backfilling, no last message in database or no last message in metadata")
  75. return
  76. } else if !shouldBackfill(lastMessage.DiscordID, serverLastMessageID) {
  77. log.Debug().
  78. Str("last_bridged_message", lastMessage.DiscordID).
  79. Str("last_server_message", serverLastMessageID).
  80. Msg("Not backfilling, last message in database is newer than last message in metadata")
  81. return
  82. }
  83. log.Debug().
  84. Str("last_bridged_message", lastMessage.DiscordID).
  85. Str("last_server_message", serverLastMessageID).
  86. Msg("Backfilling missed messages")
  87. if limit < 0 {
  88. portal.backfillUnlimitedMissed(log, source, lastMessage.DiscordID, thread)
  89. } else {
  90. portal.backfillLimited(log, source, limit, lastMessage.DiscordID, thread)
  91. }
  92. }
  93. const messageFetchChunkSize = 50
  94. func (portal *Portal) collectBackfillMessages(log zerolog.Logger, source *User, limit int, until string, thread *Thread) ([]*discordgo.Message, bool, error) {
  95. var messages []*discordgo.Message
  96. var before string
  97. var foundAll bool
  98. protoChannelID := portal.Key.ChannelID
  99. if thread != nil {
  100. protoChannelID = thread.ID
  101. }
  102. for {
  103. log.Debug().Str("before_id", before).Msg("Fetching messages for backfill")
  104. newMessages, err := source.Session.ChannelMessages(protoChannelID, messageFetchChunkSize, before, "", "")
  105. if err != nil {
  106. return nil, false, err
  107. }
  108. if until != "" {
  109. for i, msg := range newMessages {
  110. if compareMessageIDs(msg.ID, until) <= 0 {
  111. log.Debug().
  112. Str("message_id", msg.ID).
  113. Str("until_id", until).
  114. Msg("Found message that was already bridged")
  115. newMessages = newMessages[:i]
  116. foundAll = true
  117. break
  118. }
  119. }
  120. }
  121. messages = append(messages, newMessages...)
  122. log.Debug().Int("count", len(newMessages)).Msg("Added messages to backfill collection")
  123. if len(newMessages) < messageFetchChunkSize || len(messages) >= limit {
  124. break
  125. }
  126. before = newMessages[len(newMessages)-1].ID
  127. }
  128. if len(messages) > limit {
  129. foundAll = false
  130. messages = messages[:limit]
  131. }
  132. return messages, foundAll, nil
  133. }
  134. func (portal *Portal) backfillLimited(log zerolog.Logger, source *User, limit int, after string, thread *Thread) {
  135. messages, foundAll, err := portal.collectBackfillMessages(log, source, limit, after, thread)
  136. if err != nil {
  137. log.Err(err).Msg("Error collecting messages to forward backfill")
  138. return
  139. }
  140. log.Info().
  141. Int("count", len(messages)).
  142. Bool("found_all", foundAll).
  143. Msg("Collected messages to backfill")
  144. sort.Sort(MessageSlice(messages))
  145. if !foundAll && after != "" {
  146. _, err = portal.sendMatrixMessage(portal.MainIntent(), event.EventMessage, &event.MessageEventContent{
  147. MsgType: event.MsgNotice,
  148. Body: "Some messages may have been missed here while the bridge was offline.",
  149. }, nil, 0)
  150. if err != nil {
  151. log.Warn().Err(err).Msg("Failed to send missed message warning")
  152. } else {
  153. log.Debug().Msg("Sent warning about possibly missed messages")
  154. }
  155. }
  156. portal.sendBackfillBatch(log, source, messages, thread)
  157. }
  158. func (portal *Portal) backfillUnlimitedMissed(log zerolog.Logger, source *User, after string, thread *Thread) {
  159. protoChannelID := portal.Key.ChannelID
  160. if thread != nil {
  161. protoChannelID = thread.ID
  162. }
  163. for {
  164. log.Debug().Str("after_id", after).Msg("Fetching chunk of messages to backfill")
  165. messages, err := source.Session.ChannelMessages(protoChannelID, messageFetchChunkSize, "", after, "")
  166. if err != nil {
  167. log.Err(err).Msg("Error fetching chunk of messages to forward backfill")
  168. return
  169. }
  170. log.Debug().Int("count", len(messages)).Msg("Fetched chunk of messages to backfill")
  171. sort.Sort(MessageSlice(messages))
  172. portal.sendBackfillBatch(log, source, messages, thread)
  173. if len(messages) < messageFetchChunkSize {
  174. // Assume that was all the missing messages
  175. log.Debug().Msg("Chunk had less than 50 messages, stopping backfill")
  176. return
  177. }
  178. after = messages[len(messages)-1].ID
  179. }
  180. }
  181. func (portal *Portal) sendBackfillBatch(log zerolog.Logger, source *User, messages []*discordgo.Message, thread *Thread) {
  182. if portal.bridge.Config.Homeserver.Software == bridgeconfig.SoftwareHungry {
  183. log.Debug().Msg("Using hungryserv, sending messages with batch send endpoint")
  184. portal.forwardBatchSend(log, source, messages, thread)
  185. } else {
  186. log.Debug().Msg("Not using hungryserv, sending messages one by one")
  187. for _, msg := range messages {
  188. portal.handleDiscordMessageCreate(source, msg, thread)
  189. }
  190. }
  191. }
  192. func (portal *Portal) forwardBatchSend(log zerolog.Logger, source *User, messages []*discordgo.Message, thread *Thread) {
  193. evts, metas, dbMessages := portal.convertMessageBatch(log, source, messages, thread)
  194. if len(evts) == 0 {
  195. log.Warn().Msg("Didn't get any events to backfill")
  196. return
  197. }
  198. log.Info().Int("events", len(evts)).Msg("Converted messages to backfill")
  199. resp, err := portal.MainIntent().BatchSend(portal.MXID, &mautrix.ReqBatchSend{
  200. BeeperNewMessages: true,
  201. Events: evts,
  202. })
  203. if err != nil {
  204. log.Err(err).Msg("Error sending backfill batch")
  205. return
  206. }
  207. for i, evtID := range resp.EventIDs {
  208. dbMessages[i].MXID = evtID
  209. if metas[i] != nil && metas[i].Flags == discordgo.MessageFlagsHasThread {
  210. // TODO proper context
  211. ctx := log.WithContext(context.Background())
  212. portal.bridge.threadFound(ctx, source, &dbMessages[i], metas[i].ID, metas[i].Thread)
  213. }
  214. }
  215. portal.bridge.DB.Message.MassInsert(portal.Key, dbMessages)
  216. log.Info().Msg("Inserted backfilled batch to database")
  217. }
  218. func (portal *Portal) convertMessageBatch(log zerolog.Logger, source *User, messages []*discordgo.Message, thread *Thread) ([]*event.Event, []*discordgo.Message, []database.Message) {
  219. var discordThreadID string
  220. var threadRootEvent, lastThreadEvent id.EventID
  221. if thread != nil {
  222. discordThreadID = thread.ID
  223. threadRootEvent = thread.RootMXID
  224. lastThreadEvent = threadRootEvent
  225. lastInThread := portal.bridge.DB.Message.GetLastInThread(portal.Key, thread.ID)
  226. if lastInThread != nil {
  227. lastThreadEvent = lastInThread.MXID
  228. }
  229. }
  230. evts := make([]*event.Event, 0, len(messages))
  231. dbMessages := make([]database.Message, 0, len(messages))
  232. metas := make([]*discordgo.Message, 0, len(messages))
  233. ctx := context.Background()
  234. for _, msg := range messages {
  235. for _, mention := range msg.Mentions {
  236. puppet := portal.bridge.GetPuppetByID(mention.ID)
  237. puppet.UpdateInfo(nil, mention, "")
  238. }
  239. puppet := portal.bridge.GetPuppetByID(msg.Author.ID)
  240. puppet.UpdateInfo(source, msg.Author, msg.WebhookID)
  241. intent := puppet.IntentFor(portal)
  242. replyTo := portal.getReplyTarget(source, discordThreadID, msg.MessageReference, msg.Embeds, true)
  243. mentions := portal.convertDiscordMentions(msg, false)
  244. ts, _ := discordgo.SnowflakeTimestamp(msg.ID)
  245. log := log.With().
  246. Str("message_id", msg.ID).
  247. Int("message_type", int(msg.Type)).
  248. Str("author_id", msg.Author.ID).
  249. Logger()
  250. parts := portal.convertDiscordMessage(log.WithContext(ctx), puppet, intent, msg)
  251. for i, part := range parts {
  252. if (replyTo != nil || threadRootEvent != "") && part.Content.RelatesTo == nil {
  253. part.Content.RelatesTo = &event.RelatesTo{}
  254. }
  255. if threadRootEvent != "" {
  256. part.Content.RelatesTo.SetThread(threadRootEvent, lastThreadEvent)
  257. }
  258. if replyTo != nil {
  259. part.Content.RelatesTo.SetReplyTo(replyTo.EventID)
  260. // Only set reply for first event
  261. replyTo = nil
  262. }
  263. part.Content.Mentions = mentions
  264. // Only set mentions for first event, but keep empty object for rest
  265. mentions = &event.Mentions{}
  266. partName := part.AttachmentID
  267. // Always use blank part name for first part so that replies and other things
  268. // can reference it without knowing about attachments.
  269. if i == 0 {
  270. partName = ""
  271. }
  272. evt := &event.Event{
  273. ID: portal.deterministicEventID(msg.ID, partName),
  274. Type: part.Type,
  275. Sender: intent.UserID,
  276. Timestamp: ts.UnixMilli(),
  277. Content: event.Content{
  278. Parsed: part.Content,
  279. Raw: part.Extra,
  280. },
  281. }
  282. var err error
  283. evt.Type, err = portal.encrypt(intent, &evt.Content, evt.Type)
  284. if err != nil {
  285. log.Err(err).Msg("Failed to encrypt event")
  286. continue
  287. }
  288. intent.AddDoublePuppetValue(&evt.Content)
  289. evts = append(evts, evt)
  290. dbMessages = append(dbMessages, database.Message{
  291. Channel: portal.Key,
  292. DiscordID: msg.ID,
  293. SenderID: msg.Author.ID,
  294. Timestamp: ts,
  295. AttachmentID: part.AttachmentID,
  296. SenderMXID: intent.UserID,
  297. })
  298. if i == 0 {
  299. metas = append(metas, msg)
  300. } else {
  301. metas = append(metas, nil)
  302. }
  303. lastThreadEvent = evt.ID
  304. }
  305. }
  306. return evts, metas, dbMessages
  307. }
  308. func (portal *Portal) deterministicEventID(messageID, partName string) id.EventID {
  309. data := fmt.Sprintf("%s/discord/%s/%s", portal.MXID, messageID, partName)
  310. sum := sha256.Sum256([]byte(data))
  311. return id.EventID(fmt.Sprintf("$%s:discord.com", base64.RawURLEncoding.EncodeToString(sum[:])))
  312. }
  313. // compareMessageIDs compares two Discord message IDs.
  314. //
  315. // If the first ID is lower, -1 is returned.
  316. // If the second ID is lower, 1 is returned.
  317. // If the IDs are equal, 0 is returned.
  318. func compareMessageIDs(id1, id2 string) int {
  319. if id1 == id2 {
  320. return 0
  321. }
  322. if len(id1) < len(id2) {
  323. return -1
  324. } else if len(id2) < len(id1) {
  325. return 1
  326. }
  327. if id1 < id2 {
  328. return -1
  329. }
  330. return 1
  331. }
  332. func shouldBackfill(latestBridgedIDStr, latestIDFromServerStr string) bool {
  333. return compareMessageIDs(latestBridgedIDStr, latestIDFromServerStr) == -1
  334. }
  335. type MessageSlice []*discordgo.Message
  336. var _ sort.Interface = (MessageSlice)(nil)
  337. func (a MessageSlice) Len() int {
  338. return len(a)
  339. }
  340. func (a MessageSlice) Swap(i, j int) {
  341. a[i], a[j] = a[j], a[i]
  342. }
  343. func (a MessageSlice) Less(i, j int) bool {
  344. return compareMessageIDs(a[i].ID, a[j].ID) == -1
  345. }