backfill.go 9.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308
  1. package main
  2. import (
  3. "crypto/sha256"
  4. "encoding/base64"
  5. "fmt"
  6. "sort"
  7. "github.com/bwmarrin/discordgo"
  8. "github.com/rs/zerolog"
  9. "maunium.net/go/mautrix"
  10. "maunium.net/go/mautrix/bridge/bridgeconfig"
  11. "maunium.net/go/mautrix/event"
  12. "maunium.net/go/mautrix/id"
  13. "go.mau.fi/mautrix-discord/database"
  14. )
  15. func (portal *Portal) forwardBackfillInitial(source *User) {
  16. defer portal.forwardBackfillLock.Unlock()
  17. // This should only be called from CreateMatrixRoom which locks forwardBackfillLock before creating the room.
  18. if portal.forwardBackfillLock.TryLock() {
  19. panic("forwardBackfillInitial() called without locking forwardBackfillLock")
  20. }
  21. limit := portal.bridge.Config.Bridge.Backfill.Limits.Initial.Channel
  22. if portal.GuildID == "" {
  23. limit = portal.bridge.Config.Bridge.Backfill.Limits.Initial.DM
  24. }
  25. if limit == 0 {
  26. return
  27. }
  28. log := portal.zlog.With().
  29. Str("action", "initial backfill").
  30. Str("room_id", portal.MXID.String()).
  31. Int("limit", limit).
  32. Logger()
  33. portal.backfillLimited(log, source, limit, "")
  34. }
  35. func (portal *Portal) ForwardBackfillMissed(source *User, meta *discordgo.Channel) {
  36. if portal.MXID == "" {
  37. return
  38. }
  39. limit := portal.bridge.Config.Bridge.Backfill.Limits.Missed.Channel
  40. if portal.GuildID == "" {
  41. limit = portal.bridge.Config.Bridge.Backfill.Limits.Missed.DM
  42. }
  43. if limit == 0 {
  44. return
  45. }
  46. log := portal.zlog.With().
  47. Str("action", "missed event backfill").
  48. Str("room_id", portal.MXID.String()).
  49. Int("limit", limit).
  50. Logger()
  51. portal.forwardBackfillLock.Lock()
  52. defer portal.forwardBackfillLock.Unlock()
  53. lastMessage := portal.bridge.DB.Message.GetLast(portal.Key)
  54. if lastMessage == nil || meta.LastMessageID == "" {
  55. log.Debug().Msg("Not backfilling, no last message in database or no last message in metadata")
  56. return
  57. } else if !shouldBackfill(lastMessage.DiscordID, meta.LastMessageID) {
  58. log.Debug().
  59. Str("last_bridged_message", lastMessage.DiscordID).
  60. Str("last_server_message", meta.LastMessageID).
  61. Msg("Not backfilling, last message in database is newer than last message in metadata")
  62. return
  63. }
  64. log.Debug().
  65. Str("last_bridged_message", lastMessage.DiscordID).
  66. Str("last_server_message", meta.LastMessageID).
  67. Msg("Backfilling missed messages")
  68. if limit < 0 {
  69. portal.backfillUnlimitedMissed(log, source, lastMessage.DiscordID)
  70. } else {
  71. portal.backfillLimited(log, source, limit, lastMessage.DiscordID)
  72. }
  73. }
  74. const messageFetchChunkSize = 50
  75. func (portal *Portal) collectBackfillMessages(log zerolog.Logger, source *User, limit int, until string) ([]*discordgo.Message, bool, error) {
  76. var messages []*discordgo.Message
  77. var before string
  78. var foundAll bool
  79. for {
  80. log.Debug().Str("before_id", before).Msg("Fetching messages for backfill")
  81. newMessages, err := source.Session.ChannelMessages(portal.Key.ChannelID, messageFetchChunkSize, before, "", "")
  82. if err != nil {
  83. return nil, false, err
  84. }
  85. if until != "" {
  86. for i, msg := range newMessages {
  87. if compareMessageIDs(msg.ID, until) <= 0 {
  88. log.Debug().
  89. Str("message_id", msg.ID).
  90. Str("until_id", until).
  91. Msg("Found message that was already bridged")
  92. newMessages = newMessages[:i]
  93. foundAll = true
  94. break
  95. }
  96. }
  97. }
  98. messages = append(messages, newMessages...)
  99. log.Debug().Int("count", len(newMessages)).Msg("Added messages to backfill collection")
  100. if len(newMessages) < messageFetchChunkSize || len(messages) >= limit {
  101. break
  102. }
  103. before = newMessages[len(newMessages)-1].ID
  104. }
  105. if len(messages) > limit {
  106. foundAll = false
  107. messages = messages[:limit]
  108. }
  109. return messages, foundAll, nil
  110. }
  111. func (portal *Portal) backfillLimited(log zerolog.Logger, source *User, limit int, after string) {
  112. messages, foundAll, err := portal.collectBackfillMessages(log, source, limit, after)
  113. if err != nil {
  114. log.Err(err).Msg("Error collecting messages to forward backfill")
  115. return
  116. }
  117. log.Info().
  118. Int("count", len(messages)).
  119. Bool("found_all", foundAll).
  120. Msg("Collected messages to backfill")
  121. sort.Sort(MessageSlice(messages))
  122. if !foundAll {
  123. _, err = portal.sendMatrixMessage(portal.MainIntent(), event.EventMessage, &event.MessageEventContent{
  124. MsgType: event.MsgNotice,
  125. Body: "Some messages may have been missed here while the bridge was offline.",
  126. }, nil, 0)
  127. if err != nil {
  128. log.Warn().Err(err).Msg("Failed to send missed message warning")
  129. } else {
  130. log.Debug().Msg("Sent warning about possibly missed messages")
  131. }
  132. }
  133. portal.sendBackfillBatch(log, source, messages)
  134. }
  135. func (portal *Portal) backfillUnlimitedMissed(log zerolog.Logger, source *User, after string) {
  136. for {
  137. log.Debug().Str("after_id", after).Msg("Fetching chunk of messages to backfill")
  138. messages, err := source.Session.ChannelMessages(portal.Key.ChannelID, messageFetchChunkSize, "", after, "")
  139. if err != nil {
  140. log.Err(err).Msg("Error fetching chunk of messages to forward backfill")
  141. return
  142. }
  143. log.Debug().Int("count", len(messages)).Msg("Fetched chunk of messages to backfill")
  144. sort.Sort(MessageSlice(messages))
  145. portal.sendBackfillBatch(log, source, messages)
  146. if len(messages) < messageFetchChunkSize {
  147. // Assume that was all the missing messages
  148. log.Debug().Msg("Chunk had less than 50 messages, stopping backfill")
  149. return
  150. }
  151. after = messages[len(messages)-1].ID
  152. }
  153. }
  154. func (portal *Portal) sendBackfillBatch(log zerolog.Logger, source *User, messages []*discordgo.Message) {
  155. if portal.bridge.Config.Homeserver.Software == bridgeconfig.SoftwareHungry {
  156. log.Debug().Msg("Using hungryserv, sending messages with batch send endpoint")
  157. portal.forwardBatchSend(log, source, messages)
  158. } else {
  159. log.Debug().Msg("Not using hungryserv, sending messages one by one")
  160. for _, msg := range messages {
  161. portal.handleDiscordMessageCreate(source, msg, nil)
  162. }
  163. }
  164. }
  165. func (portal *Portal) forwardBatchSend(log zerolog.Logger, source *User, messages []*discordgo.Message) {
  166. evts, dbMessages := portal.convertMessageBatch(log, source, messages)
  167. if len(evts) == 0 {
  168. log.Warn().Msg("Didn't get any events to backfill")
  169. return
  170. }
  171. log.Info().Int("events", len(evts)).Msg("Converted messages to backfill")
  172. resp, err := portal.MainIntent().BatchSend(portal.MXID, &mautrix.ReqBatchSend{
  173. BeeperNewMessages: true,
  174. Events: evts,
  175. })
  176. if err != nil {
  177. log.Err(err).Msg("Error sending backfill batch")
  178. return
  179. }
  180. for i, evtID := range resp.EventIDs {
  181. dbMessages[i].MXID = evtID
  182. }
  183. portal.bridge.DB.Message.MassInsert(portal.Key, dbMessages)
  184. log.Info().Msg("Inserted backfilled batch to database")
  185. }
  186. func (portal *Portal) convertMessageBatch(log zerolog.Logger, source *User, messages []*discordgo.Message) ([]*event.Event, []database.Message) {
  187. evts := make([]*event.Event, 0, len(messages))
  188. dbMessages := make([]database.Message, 0, len(messages))
  189. for _, msg := range messages {
  190. for _, mention := range msg.Mentions {
  191. puppet := portal.bridge.GetPuppetByID(mention.ID)
  192. puppet.UpdateInfo(nil, mention)
  193. }
  194. puppet := portal.bridge.GetPuppetByID(msg.Author.ID)
  195. puppet.UpdateInfo(source, msg.Author)
  196. intent := puppet.IntentFor(portal)
  197. replyTo := portal.getReplyTarget(source, msg.MessageReference, true)
  198. ts, _ := discordgo.SnowflakeTimestamp(msg.ID)
  199. parts := portal.convertDiscordMessage(intent, msg)
  200. for i, part := range parts {
  201. if replyTo != nil {
  202. part.Content.RelatesTo = &event.RelatesTo{InReplyTo: replyTo}
  203. // Only set reply for first event
  204. replyTo = nil
  205. }
  206. partName := part.AttachmentID
  207. // Always use blank part name for first part so that replies and other things
  208. // can reference it without knowing about attachments.
  209. if i == 0 {
  210. partName = ""
  211. }
  212. evt := &event.Event{
  213. ID: portal.deterministicEventID(msg.ID, partName),
  214. Type: part.Type,
  215. Sender: intent.UserID,
  216. Timestamp: ts.UnixMilli(),
  217. Content: event.Content{
  218. Parsed: part.Content,
  219. Raw: part.Extra,
  220. },
  221. }
  222. var err error
  223. evt.Type, err = portal.encrypt(intent, &evt.Content, evt.Type)
  224. if err != nil {
  225. log.Err(err).Msg("Failed to encrypt event")
  226. continue
  227. }
  228. intent.AddDoublePuppetValue(&evt.Content)
  229. evts = append(evts, evt)
  230. dbMessages = append(dbMessages, database.Message{
  231. Channel: portal.Key,
  232. DiscordID: msg.ID,
  233. SenderID: msg.Author.ID,
  234. Timestamp: ts,
  235. AttachmentID: part.AttachmentID,
  236. })
  237. }
  238. }
  239. return evts, dbMessages
  240. }
  241. func (portal *Portal) deterministicEventID(messageID, partName string) id.EventID {
  242. data := fmt.Sprintf("%s/discord/%s/%s", portal.MXID, messageID, partName)
  243. sum := sha256.Sum256([]byte(data))
  244. return id.EventID(fmt.Sprintf("$%s:discord.com", base64.RawURLEncoding.EncodeToString(sum[:])))
  245. }
  246. // compareMessageIDs compares two Discord message IDs.
  247. //
  248. // If the first ID is lower, -1 is returned.
  249. // If the second ID is lower, 1 is returned.
  250. // If the IDs are equal, 0 is returned.
  251. func compareMessageIDs(id1, id2 string) int {
  252. if id1 == id2 {
  253. return 0
  254. }
  255. if len(id1) < len(id2) {
  256. return -1
  257. } else if len(id2) < len(id1) {
  258. return 1
  259. }
  260. if id1 < id2 {
  261. return -1
  262. }
  263. return 1
  264. }
  265. func shouldBackfill(latestBridgedIDStr, latestIDFromServerStr string) bool {
  266. return compareMessageIDs(latestBridgedIDStr, latestIDFromServerStr) == -1
  267. }
  268. type MessageSlice []*discordgo.Message
  269. var _ sort.Interface = (MessageSlice)(nil)
  270. func (a MessageSlice) Len() int {
  271. return len(a)
  272. }
  273. func (a MessageSlice) Swap(i, j int) {
  274. a[i], a[j] = a[j], a[i]
  275. }
  276. func (a MessageSlice) Less(i, j int) bool {
  277. return compareMessageIDs(a[i].ID, a[j].ID) == -1
  278. }