Struct Service 

pub struct Service {
    mutex: MutexMap<OwnedUserId, ()>,
    services: Arc<OnceServices>,
    profilechangeid_userid: Arc<Map>,
    useridprofilekey_value: Arc<Map>,
}

Fields

mutex: MutexMap<OwnedUserId, ()>

services: Arc<OnceServices>

profilechangeid_userid: Arc<Map>

useridprofilekey_value: Arc<Map>

Implementations

impl Service

pub async fn mirror_remote_profile( &self, user_id: &UserId, ) -> Result<SmallVec<[ProfileFieldName; 1]>>

Replaces a remote user’s cached profile with the one their server serves.

Unlike fetch_remote_profile, which only adds and overwrites, a cached field missing from the response is removed, so a value the remote user has since deleted stops reaching clients. Returns the names of the removed fields.

impl Service

pub(super) async fn mirror_profile( &self, user_id: &UserId, response: Response, ) -> Result<SmallVec<[ProfileFieldName; 1]>>

Stores a profile response as the user’s complete cached profile.

Every returned field is written and every cached field the response omits is deleted in one logged write under the profile lock, so no concurrent write interleaves and connected clients see the removals.

impl Service

pub async fn update_all_rooms( &self, user_id: &UserId, profile_values: &[(ProfileFieldName, Option<Value>)], propagation: Propagation, )

impl Service

async fn update_room( &self, user_id: &UserId, room_id: &RoomId, profile_values: &[(ProfileFieldName, Option<Value>)], propagation: Propagation, current_displayname: Option<&str>, current_avatar_url: Option<&MxcUri>, ) -> Result

impl Service

pub async fn set_displayname( &self, user_id: &UserId, displayname: Option<&str>, propagation: Option<Propagation>, ) -> Result

Sets a new displayname or removes it if displayname is None. You still need to notify all rooms of this change.

impl Service

pub async fn displayname(&self, user_id: &UserId) -> Result<String>

Returns the displayname of a user on this homeserver.

impl Service

pub async fn set_avatar_url( &self, user_id: &UserId, avatar_url: Option<&MxcUri>, propagation: Option<Propagation>, ) -> Result

Sets a new avatar_url or removes it if avatar_url is None.

impl Service

pub async fn avatar_url(&self, user_id: &UserId) -> Result<OwnedMxcUri>

Get the avatar_url of a user.

impl Service

pub async fn set_timezone( &self, user_id: &UserId, timezone: Option<&str>, propagation: Option<Propagation>, ) -> Result

Sets a new timezone or removes it if timezone is None.

impl Service

pub async fn timezone(&self, user_id: &UserId) -> Result<String>

Get the timezone of a user.

impl Service

pub fn all_profile_keys( &self, user_id: &UserId, ) -> impl Stream<Item = ProfileFieldValue> + Send

Streams every stored profile field.

A field whose stored value no longer parses is dropped; try_all_profile_keys surfaces it instead.

impl Service

pub fn try_all_profile_keys( &self, user_id: &UserId, ) -> impl Stream<Item = Result<ProfileFieldValue>> + Send

Streams every stored profile field, surfacing storage and decoding errors.

A stored value that no longer parses as a profile field is an error item rather than a dropped one, so a caller validating the whole profile can refuse to proceed on it.

impl Service

pub fn profile_field_names( &self, user_id: &UserId, ) -> impl Stream<Item = ProfileFieldName> + Send

Streams the names of the fields a user’s profile holds.

The name rides the key, so a caller after names alone skips deserializing every value the way all_profile_keys must.

impl Service

pub fn try_profile_field_names( &self, user_id: &UserId, ) -> impl Stream<Item = Result<ProfileFieldName>> + Send

Enumerates current profile field names, preserving storage and decoding errors.

Sync bases must finish this inventory before acknowledging profile delivery.

impl Service

pub async fn clear_profile_keys(&self, user_id: &UserId) -> Result

Clears every stored profile field and propagates canonical removals.

Member events are rewritten only for a local user, before the removals and their change rows commit together under the profile lock. A preparation error leaves storage unchanged but does not undo emitted member events.

impl Service

pub async fn set_profile_keys( &self, user_id: &UserId, profile_values: &[(ProfileFieldName, Option<Value>)], propagation: Option<Propagation>, ) -> Result

Sets profile field values, removing a field whose value is None.

Member events are rewritten first, then the stored fields and their change rows commit together under the profile lock. A preparation error leaves storage unchanged but does not undo emitted member events.

impl Service

async fn set_profile_keys_locked( &self, _profile_lock: &MutexMapGuard<OwnedUserId, ()>, user_id: &UserId, profile_values: &[(ProfileFieldName, Option<Value>)], propagation: Option<Propagation>, ) -> Result

Sets profile field values under a profile lock the caller already holds.

A caller that must read the stored profile and write it back without an interleaved writer takes the lock once and passes it here.

impl Service

async fn changed_fields<'a>( &self, user_id: &UserId, profile_values: &'a [(ProfileFieldName, Option<Value>)], ) -> Result<SmallVec<[&'a str; 1]>>

Names the fields whose stored value the write would actually change.

Only these are logged for the user’s rooms, or at all for a remote user. A write restoring what is already stored is news to nobody else, and the on-demand remote refresh reissues every field on every lookup of a remote profile, so logging those would multiply the log by the request rate rather than the change rate.

impl Service

async fn publish_update<'a, L, T, S>( &self, user_id: &UserId, logged: L, changed: &[T], txn: Txn, rooms: impl FnOnce() -> S + Send, ) -> Result

where L: IntoIterator<Item: AsRef<str>> + Clone, T: AsRef<str> + Sync, S: Stream<Item = Result<&'a RoomId>> + Send,

Commits staged profile fields together with their change rows.

Every logged field gets a row under the user’s own prefix; every changed field also gets one under each room they are joined to. The row names the field and not only the user because a removal is otherwise unreportable: a reader re-reading the live profile cannot tell a cleared field from one that was never set. A room scan error discards the whole batch, so no row for that count is ever visible.

impl Service

fn stage_update<I>( &self, txn: Txn, scope: &str, user_id: &UserId, count: u64, fields: I, ) -> Txn

where I: IntoIterator<Item: AsRef<str>>,

impl Service

pub fn profile_changed<'a>( &'a self, user_id: &'a UserId, from: u64, to: Option<u64>, ) -> impl Stream<Item = ProfileChange<'a>> + Send + 'a

Streams the profile fields the user wrote themselves, restatements included.

The range is half-open on the low side, so a caller passes the sync token it already delivered. An absent to leaves the walk unbounded above. Storage and decoding failures are logged and skipped.

impl Service

pub fn try_profile_changed<'a>( &'a self, user_id: &'a UserId, from: u64, to: Option<u64>, ) -> impl Stream<Item = Result<ProfileChange<'a>>> + Send + 'a

Streams the user’s logged profile fields, surfacing read failures.

The range excludes from and includes to, when provided. Fields and user IDs borrow the cursor and must be consumed or owned before it advances.

impl Service

pub fn room_profile_changed<'a>( &'a self, room_id: &'a RoomId, from: u64, to: Option<u64>, ) -> impl Stream<Item = ProfileChange<'a>> + Send + 'a

Streams the profile fields any member of the room changed.

The range works as it does for a single user. A member appears once per field they changed, however many of the caller’s rooms they share. Storage and decoding failures are logged and skipped.

impl Service

pub fn try_room_profile_changed<'a>( &'a self, room_id: &'a RoomId, from: u64, to: Option<u64>, ) -> impl Stream<Item = Result<ProfileChange<'a>>> + Send + 'a

Streams a room’s logged profile fields, surfacing read failures.

The range excludes from and includes to, when provided. Fields and user IDs borrow the cursor and must be consumed or owned before it advances.

impl Service

fn profile_changed_user_or_room<'a>( &'a self, user_or_room_id: &'a str, from: u64, to: Option<u64>, ) -> impl Stream<Item = Result<ProfileChange<'a>>> + Send + 'a

impl Service

pub async fn profile_key<T>( &self, user_id: &UserId, profile_key: &ProfileFieldName, ) -> Result<T>

where T: for<'de> Deserialize<'de> + Send,

Gets a specific user profile key

impl Service

pub async fn fill_content( &self, user_id: &UserId, content: RoomMemberEventContent, ) -> RoomMemberEventContent

Fill membership content with the user’s display name and avatar.

Profile lookup failures clear the corresponding fields. All other content fields are preserved.

impl Service

pub async fn fetch_remote_profile(&self, user_id: &UserId) -> Result

impl Service

async fn enforce_profile_size( &self, user_id: &UserId, key: &str, value: &Value, ) -> Result

Reject a prospective profile write exceeding the MSC4133 64 KiB cap.

The candidate value replaces the stored field for the size calculation. A stored field that no longer parses is logged and left out of the total, so it cannot block writes to the other fields; removals skip this check.

Trait Implementations

impl Service for Service

fn build(args: &Args<'_>) -> Result<Arc<Self>>

Implement the construction of the service instance. Services are generally singletons so expect this to only be called once for a service type. Note that it may be called again after a server reload, but the prior instance will have been dropped first. Failure will shutdown the server with an error.

fn name(&self) -> &str

Return the name of the service. i.e. crate::service::make_name(std::module_path!())

fn worker<'async_trait>( self: Arc<Self>, ) -> Pin<Box<dyn Future<Output = Result> + Send + 'async_trait>>

where Self: 'async_trait,

Implement the service’s worker loop. The service manager spawns a task and calls this function after all services have been built.

fn interrupt<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = ()> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Interrupt the service. This is sent to initiate a graceful shutdown. The service worker should return from its work loop.

fn clear_cache<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = ()> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Clear any caches or similar runtime state.

fn memory_usage<'life0, 'life1, 'async_trait>( &'life0 self, _out: &'life1 mut (dyn Write + Send), ) -> Pin<Box<dyn Future<Output = Result> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Memory usage report in a markdown string.

fn unconstrained(&self) -> bool

Return true if the service worker opts out of the tokio cooperative budgeting. This can reduce tail latency at the risk of event loop starvation.