CI / Test (ubuntu-latest) (push) Canceled after 0s
CI / Test (windows-latest) (push) Canceled after 0s
CI / Storage tests on PostgreSQL (push) Canceled after 0s
CI / Complement (federation off) (push) Canceled after 0s
CI / golangci-lint (push) Canceled after 0s
CI / Smoke test the binary (push) Canceled after 0s
Implements PLAN.md Phase 4: the read model. /sync, /messages and /context, the filters that narrow them, and the streams a response carries that are not room events — typing, receipts, presence and to-device messages. With this a client can be used against Warp, which is milestone M1. The one architectural choice is that the syncapi has no copy of the rooms: it reads them through the roomserver. Dendrite gives its syncapi its own tables and feeds them from the output stream, which is right for a design where the two can be separate processes, and Dendrite's could be. Warp is one process by decision, and there the copy buys nothing and costs the hardest class of bug a homeserver has — a read model that disagrees with the room, where a client is shown a message that was rejected and the only way to find out is to compare two tables. So the roomserver grew a read model (stream.go, roomsync.go, visibility.go) rather than exporting its store: reading an event means parsing it under its room's version, and a second copy of the version registry, the parser and the client projection would be a second place for them to disagree. syncapi: stream tokens naming a position in each of five independent streams, so a client that has seen receipt 7 but only event 4 can say so in one opaque string. A token with fewer positions than this build writes is accepted and the rest read as zero, which is what lets a token stored before a stream existed go on working. The PDU position comes from MAX(stream_ordering) and not from the counter row the other four use: the counter is advanced inside the writing transaction, so a token taken from it would name a position whose event no reader can see yet, and the event would be skipped for good once that transaction committed. Long polling goes through a notifier that wakes by room and by user rather than globally — an event wakes the listeners of its room, a membership event also wakes the user it names, so an invite reaches a client that is not in the room yet. A wake-up is only ever a hint: the listener re-reads the streams and decides for itself what changed, which is what makes it safe for the notifier to hold no state about what each client has seen. The listener registers before the read, so an event landing in between leaves the channel signalled and the sync reads again rather than blocking on a change it just missed. EventPersisted must not block, because it is called with the room's write lock still held. "Something happened" is not "a token moved". A stream can move without this user being concerned by it, and returning on every such move would turn a long poll back into a busy loop. The same trap caught presence twice over: a client says it is still online by syncing, so ?set_presence= makes every sync a presence write, and a stream that moved each time would wake every peer, which would sync, which would move it again. SetPresence therefore reports whether anything a peer would notice actually changed, and a write that only refreshes the last-active timestamp leaves the stream alone. Timeline windows are named by their two ends and a direction, Low exclusive and High inclusive in both directions, rather than by "from" and "to" as the wire format spells them. Each of those would otherwise be the low bound in one direction and the high bound in the other, which is exactly the off-by-one every pagination bug is made of; the handlers translate once and nothing below them thinks about it again. A sync's timeline is read backwards and reported forwards, because a client wants the newest events of the window it missed — a forward read with a limit hands it the oldest and then calls the rest a gap. visibility: history visibility is now a span rather than a yes or no, which is what Phase 3 deferred here. A user who has left a room is served the room as it was when they left — their state reads, their members and their timeline — instead of being refused, and is bounded above at their own membership event so a kicked client cannot go on streaming. Their span is found from their own member events, so a user who joined, left and rejoined sees their latest span. ?at= on /members is honoured for the same reason. What is left is that the span is computed from the room's current history_visibility where the spec judges each event by the one in force when it was sent; that is a state read per event and belongs with open question 1. Typing is held in memory and nowhere else: notifications expire in seconds, are worthless across a restart, and arrive one per keystroke-burst per room, so persisting them would be the busiest table in the database holding the least valuable data in it. Receipts only ever advance, compared on the position of the event acknowledged rather than on arrival, because two of a user's clients read at different speeds and the slower one must not drag the marker backwards. To-device queues are per device and drained by the next sync rather than by the one that delivered them, so a client that crashes in between gets the batch again — the right way round, since a lost key share leaves a message permanently undecryptable and a duplicate is ignored. A wildcard addressee is expanded when sent, not stored, because "every device" means the ones that existed then. Presence is off by default, which answers PLAN.md open question 3. It is the noisiest thing in Matrix and the one genuinely fan-out notification here; a private homeserver gains little from it. With it off the endpoints refuse rather than disappearing, because a client that gets a 404 from /presence concludes the server is too old to have it and stops asking. storage: one migration per driver adding the receipt, to-device, presence and filter tables with a counter row each, and a stream position on the existing account data table rather than a parallel one — there is exactly one current value per (user, room, type), and a second table would be a second thing to keep in step. Account data is reported in an event list but is not an event, so it gets its own response type: a room event with an empty sender and no timestamp would make a client decide what that meant. Deliberately missing, each named in docs/sync.md: notification counts (they need push rules, Phase 8), device lists (the keyserver, Phase 7), anything remote (Phase 5), threads (Phase 10). device_lists is sent as two empty lists because a client reads a missing one as malformed; the notification counts are left out entirely, because a zero there is a claim and an absent field is not. Tested through HTTP rather than against the builder, because what is worth pinning down is the response a client sees — which section a room lands in, whether a token handed back can be handed in again, whether paging terminates and loses nothing. The storage tests run against both SQL drivers; the notifier has its own tests for the concurrent case. A live server was driven end to end, which is what caught the two bugs the unit tests had missed: a backwards /messages page must be in reverse chronological order, and `start` must name where the page began rather than position zero when ?from= was omitted. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
327 lines
10 KiB
Go
327 lines
10 KiB
Go
package roomserver
|
|
|
|
import (
|
|
"context"
|
|
"strings"
|
|
|
|
"github.com/dariuszjaniak/warp/server/internal/matrix/event"
|
|
"github.com/dariuszjaniak/warp/server/internal/matrix/roomversion"
|
|
"github.com/dariuszjaniak/warp/server/internal/storage"
|
|
)
|
|
|
|
// RoomSyncRequest asks what one user sees of one room between two
|
|
// positions.
|
|
//
|
|
// It is the read model's largest operation and the one /sync is built on.
|
|
// It is a single call rather than a timeline read plus two state reads
|
|
// because the three answers have to agree with each other: which state the
|
|
// response reports depends on where the timeline turned out to start, and
|
|
// working that out in the caller would mean the caller deciding what the
|
|
// timeline was — which is the question it asked.
|
|
type RoomSyncRequest struct {
|
|
RoomID string
|
|
|
|
// User is the full user ID syncing.
|
|
User string
|
|
|
|
// Since is the position the user has already seen, exclusive. Zero is
|
|
// an initial sync.
|
|
Since int64
|
|
|
|
// To is the top of this sync, inclusive. It is fixed by the caller for
|
|
// every room in one response, so that a token names one moment rather
|
|
// than a different one per room.
|
|
To int64
|
|
|
|
// TimelineLimit caps the events in the timeline.
|
|
TimelineLimit int
|
|
|
|
// TimelineFilter and StateFilter narrow the two halves of the
|
|
// response. The state filter is applied in this package rather than in
|
|
// SQL: a state read is a set of tuples, not a window of a stream.
|
|
TimelineFilter storage.TimelineFilter
|
|
StateFilter storage.TimelineFilter
|
|
|
|
// FullState asks for the room's whole state even on an incremental
|
|
// sync, which is what ?full_state=true means.
|
|
FullState bool
|
|
|
|
// LazyLoadMembers leaves the member events out of the state except for
|
|
// the senders who appear in the timeline.
|
|
LazyLoadMembers bool
|
|
}
|
|
|
|
// RoomSync is what one user sees of one room between two positions.
|
|
type RoomSync struct {
|
|
// Membership is the user's current membership in the room.
|
|
Membership string
|
|
|
|
Timeline Timeline
|
|
|
|
// State is the room's state as of the start of the timeline on an
|
|
// initial sync, and what changed between Since and the start of the
|
|
// timeline on an incremental one. For an incremental sync with no gap
|
|
// it is empty: every state change in the window is in the timeline
|
|
// itself.
|
|
State []ClientEvent
|
|
|
|
// Summary is the room's member counts and heroes.
|
|
Summary RoomSyncSummary
|
|
|
|
// Visible is the span of the stream the user may read, which the
|
|
// caller needs in order to bound the ephemeral streams the same way.
|
|
Visible VisibleRange
|
|
}
|
|
|
|
// RoomSync reads one room for one user.
|
|
func (r *API) RoomSync(ctx context.Context, req RoomSyncRequest) (RoomSync, error) {
|
|
room, rules, err := r.roomAndRules(ctx, req.RoomID)
|
|
if err != nil {
|
|
return RoomSync{}, err
|
|
}
|
|
summary, err := r.store.RoomSummary(ctx, room.ID)
|
|
if err != nil {
|
|
return RoomSync{}, err
|
|
}
|
|
visible, err := r.visibleRange(ctx, room.ID, rules, req.User, summary.HistoryVisibility)
|
|
if err != nil {
|
|
return RoomSync{}, err
|
|
}
|
|
if err := visible.refusal(room.ID, req.User); err != nil {
|
|
return RoomSync{}, err
|
|
}
|
|
|
|
out := RoomSync{Membership: visible.Membership, Visible: visible}
|
|
|
|
// The timeline is read backwards even though it is reported forwards.
|
|
// A client wants the *newest* events of the window it has missed — the
|
|
// bottom of the room, which is what it is looking at — and a forward
|
|
// read with a limit would hand it the oldest instead and then call the
|
|
// rest a gap.
|
|
out.Timeline, err = r.timeline(ctx, rules, visible, TimelineRequest{
|
|
RoomID: req.RoomID,
|
|
User: req.User,
|
|
Low: req.Since,
|
|
High: req.To,
|
|
Direction: storage.DirectionBackward,
|
|
Limit: req.TimelineLimit,
|
|
Filter: req.TimelineFilter,
|
|
})
|
|
if err != nil {
|
|
return RoomSync{}, err
|
|
}
|
|
|
|
if out.State, err = r.syncState(ctx, rules, visible, req, out.Timeline); err != nil {
|
|
return RoomSync{}, err
|
|
}
|
|
if out.Summary, err = r.RoomSummaryFor(ctx, req.RoomID, req.User); err != nil {
|
|
return RoomSync{}, err
|
|
}
|
|
return out, nil
|
|
}
|
|
|
|
// syncState decides which state a room's half of a sync response reports.
|
|
//
|
|
// The rule the spec sets is that a client which applies `state` and then
|
|
// `timeline` ends up with the room's state at the end of the timeline. That
|
|
// gives three cases:
|
|
//
|
|
// - An initial sync, or ?full_state=true: the whole state as of the start
|
|
// of the timeline.
|
|
// - An incremental sync whose timeline reaches back to `since`: nothing.
|
|
// Every state change in between is a timeline event the client is
|
|
// already being given.
|
|
// - An incremental sync whose timeline was cut short — a gap: what
|
|
// changed between `since` and where the timeline starts. Those changes
|
|
// are not in the timeline and the client would otherwise never see
|
|
// them.
|
|
func (r *API) syncState(ctx context.Context, rules roomversion.Rules, visible VisibleRange, req RoomSyncRequest, timeline Timeline) ([]ClientEvent, error) {
|
|
statePos, haveState := statePosition(timeline, req.To, visible)
|
|
|
|
switch {
|
|
case !haveState:
|
|
// The timeline reaches back to the room's create event, so there
|
|
// is nothing before it to report.
|
|
return nil, nil
|
|
|
|
case req.FullState || req.Since == 0:
|
|
state, err := r.stateAt(ctx, req.RoomID, rules, statePos)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
return r.shapeState(ctx, rules, req, timeline, state)
|
|
|
|
case !timeline.Limited:
|
|
return nil, nil
|
|
|
|
default:
|
|
before, err := r.stateAt(ctx, req.RoomID, rules, req.Since)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
after, err := r.stateAt(ctx, req.RoomID, rules, statePos)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
return r.shapeState(ctx, rules, req, timeline, stateDelta(before, after))
|
|
}
|
|
}
|
|
|
|
// statePosition is the position a room's state is reported as of, and
|
|
// whether there is any state to report at all.
|
|
//
|
|
// The state belongs at the start of the timeline, so that applying it and
|
|
// then the timeline lands on the room's current state. A room with no
|
|
// timeline events in this window is reported as of the top of the sync
|
|
// instead: the state may still have been read for the first time.
|
|
func statePosition(timeline Timeline, to int64, visible VisibleRange) (int64, bool) {
|
|
if len(timeline.Events) > 0 {
|
|
return timeline.Start, timeline.Start > 0
|
|
}
|
|
if visible.To != 0 && (to == 0 || to > visible.To) {
|
|
to = visible.To
|
|
}
|
|
return to, to > 0
|
|
}
|
|
|
|
// shapeState applies the state filter and lazy loading to a state set.
|
|
func (r *API) shapeState(ctx context.Context, rules roomversion.Rules, req RoomSyncRequest, timeline Timeline, state []ClientEvent) ([]ClientEvent, error) {
|
|
state = filterClientEvents(state, req.StateFilter)
|
|
if !req.LazyLoadMembers {
|
|
return state, nil
|
|
}
|
|
|
|
// Lazy loading: drop the member events and put back only the ones the
|
|
// client needs to render the timeline it is being given.
|
|
//
|
|
// The spec also lets a server leave out members it has already sent
|
|
// this device, which would need per-device bookkeeping of what each
|
|
// one has seen. Warp sends them every time instead: it is the
|
|
// behaviour `include_redundant_members: true` asks for, so no client
|
|
// can be surprised by it, and it costs one state read rather than a
|
|
// table.
|
|
kept := make([]ClientEvent, 0, len(state))
|
|
for _, e := range state {
|
|
if e.Type != event.TypeMember {
|
|
kept = append(kept, e)
|
|
}
|
|
}
|
|
|
|
senders := timelineSenders(timeline)
|
|
members, err := r.memberEvents(ctx, req.RoomID, rules, senders)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
return append(kept, filterClientEvents(members, req.StateFilter)...), nil
|
|
}
|
|
|
|
// timelineSenders lists the distinct senders of a timeline, in the order
|
|
// they first appear.
|
|
func timelineSenders(timeline Timeline) []string {
|
|
seen := make(map[string]struct{}, len(timeline.Events))
|
|
out := make([]string, 0, len(timeline.Events))
|
|
for _, e := range timeline.Events {
|
|
if _, dup := seen[e.Sender]; dup {
|
|
continue
|
|
}
|
|
seen[e.Sender] = struct{}{}
|
|
out = append(out, e.Sender)
|
|
}
|
|
return out
|
|
}
|
|
|
|
// stateKey identifies one piece of state.
|
|
type stateKeyTuple struct {
|
|
eventType string
|
|
stateKey string
|
|
}
|
|
|
|
// tupleOf is the state tuple of a client event, and whether it has one.
|
|
func tupleOf(e ClientEvent) (stateKeyTuple, bool) {
|
|
if e.StateKey == nil {
|
|
return stateKeyTuple{}, false
|
|
}
|
|
return stateKeyTuple{eventType: e.Type, stateKey: *e.StateKey}, true
|
|
}
|
|
|
|
// stateDelta is what changed between two state sets: every tuple of `after`
|
|
// whose event is not the one `before` had.
|
|
//
|
|
// A tuple that `before` had and `after` does not is left out. State is only
|
|
// ever replaced, never removed — there is no way to unset a piece of room
|
|
// state — so such a tuple can only come from state resolution dropping an
|
|
// event, and there is nothing a client could be sent that would express it.
|
|
func stateDelta(before, after []ClientEvent) []ClientEvent {
|
|
had := make(map[stateKeyTuple]string, len(before))
|
|
for _, e := range before {
|
|
if tuple, ok := tupleOf(e); ok {
|
|
had[tuple] = e.EventID
|
|
}
|
|
}
|
|
|
|
out := make([]ClientEvent, 0, len(after))
|
|
for _, e := range after {
|
|
tuple, ok := tupleOf(e)
|
|
if !ok {
|
|
continue
|
|
}
|
|
if had[tuple] == e.EventID {
|
|
continue
|
|
}
|
|
out = append(out, e)
|
|
}
|
|
return out
|
|
}
|
|
|
|
// filterClientEvents applies the part of an event filter that is a
|
|
// comparison on the event itself.
|
|
//
|
|
// It is the same filter the store applies to a timeline read, applied in Go
|
|
// because a state read is a set of tuples rather than a window of a stream:
|
|
// there is no index to push it down to.
|
|
func filterClientEvents(events []ClientEvent, filter storage.TimelineFilter) []ClientEvent {
|
|
if filter.IsEmpty() {
|
|
return events
|
|
}
|
|
out := make([]ClientEvent, 0, len(events))
|
|
for _, e := range events {
|
|
if matchesFilter(e, filter) {
|
|
out = append(out, e)
|
|
}
|
|
}
|
|
return out
|
|
}
|
|
|
|
// matchesFilter reports whether one event passes a filter. Exclusions win
|
|
// over inclusions, which is what the spec requires.
|
|
func matchesFilter(e ClientEvent, filter storage.TimelineFilter) bool {
|
|
switch {
|
|
case matchesAny(e.Type, filter.NotTypes):
|
|
return false
|
|
case matchesAny(e.Sender, filter.NotSenders):
|
|
return false
|
|
case len(filter.Types) > 0 && !matchesAny(e.Type, filter.Types):
|
|
return false
|
|
case len(filter.Senders) > 0 && !matchesAny(e.Sender, filter.Senders):
|
|
return false
|
|
}
|
|
return true
|
|
}
|
|
|
|
// matchesAny reports whether value matches one of the filter patterns, each
|
|
// of which may end in '*' as the spec's wildcard.
|
|
func matchesAny(value string, patterns []string) bool {
|
|
for _, pattern := range patterns {
|
|
if prefix, isWildcard := strings.CutSuffix(pattern, "*"); isWildcard {
|
|
if strings.HasPrefix(value, prefix) {
|
|
return true
|
|
}
|
|
continue
|
|
}
|
|
if value == pattern {
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
}
|