use std::{collections::BTreeMap, fmt, hash::Hash, iter};
pub use matrix_sdk_common::deserialized_responses::*;
use once_cell::sync::Lazy;
use regex::Regex;
use ruma::{
events::{
room::{
member::{MembershipState, RoomMemberEvent, RoomMemberEventContent},
power_levels::{RoomPowerLevels, RoomPowerLevelsEventContent},
},
AnyStrippedStateEvent, AnySyncStateEvent, AnySyncTimelineEvent, EventContentFromType,
PossiblyRedactedStateEventContent, RedactContent, RedactedStateEventContent,
StateEventContent, StaticStateEventContent, StrippedStateEvent, SyncStateEvent,
},
serde::Raw,
EventId, MilliSecondsSinceUnixEpoch, OwnedEventId, OwnedRoomId, OwnedUserId, UserId,
};
use serde::Serialize;
use unicode_normalization::UnicodeNormalization;
#[derive(Clone, Debug)]
#[non_exhaustive]
pub struct AmbiguityChange {
pub member_id: OwnedUserId,
pub member_ambiguous: bool,
pub disambiguated_member: Option<OwnedUserId>,
pub ambiguated_member: Option<OwnedUserId>,
}
impl AmbiguityChange {
pub fn user_ids(&self) -> impl Iterator<Item = &UserId> {
iter::once(&*self.member_id)
.chain(self.disambiguated_member.as_deref())
.chain(self.ambiguated_member.as_deref())
}
}
#[derive(Clone, Debug, Default)]
#[non_exhaustive]
pub struct AmbiguityChanges {
pub changes: BTreeMap<OwnedRoomId, BTreeMap<OwnedEventId, AmbiguityChange>>,
}
static MXID_REGEX: Lazy<Regex> = Lazy::new(|| {
Regex::new(DisplayName::MXID_PATTERN)
.expect("We should be able to create a regex from our static MXID pattern")
});
static LEFT_TO_RIGHT_REGEX: Lazy<Regex> = Lazy::new(|| {
Regex::new(DisplayName::LEFT_TO_RIGHT_PATTERN)
.expect("We should be able to create a regex from our static left-to-right pattern")
});
static HIDDEN_CHARACTERS_REGEX: Lazy<Regex> = Lazy::new(|| {
Regex::new(DisplayName::HIDDEN_CHARACTERS_PATTERN)
.expect("We should be able to create a regex from our static hidden characters pattern")
});
static I_REGEX: Lazy<Regex> = Lazy::new(|| {
Regex::new("[i]").expect("We should be able to create a regex from our uppercase I pattern")
});
static ZERO_REGEX: Lazy<Regex> = Lazy::new(|| {
Regex::new("[0]").expect("We should be able to create a regex from our zero pattern")
});
static DOT_REGEX: Lazy<Regex> = Lazy::new(|| {
Regex::new("[.\u{1d16d}]").expect("We should be able to create a regex from our dot pattern")
});
#[derive(Debug, Clone, Eq)]
pub struct DisplayName {
raw: String,
decancered: Option<String>,
}
impl Hash for DisplayName {
fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
if let Some(decancered) = &self.decancered {
decancered.hash(state);
} else {
self.raw.hash(state);
}
}
}
impl PartialEq for DisplayName {
fn eq(&self, other: &Self) -> bool {
match (self.decancered.as_deref(), other.decancered.as_deref()) {
(None, None) => self.raw == other.raw,
(None, Some(_)) | (Some(_), None) => false,
(Some(this), Some(other)) => this == other,
}
}
}
impl DisplayName {
const MXID_PATTERN: &'static str = "@.+[:.].+";
const LEFT_TO_RIGHT_PATTERN: &'static str = "[\u{202a}-\u{202f}\u{200e}\u{200f}]";
const HIDDEN_CHARACTERS_PATTERN: &'static str =
"[\u{2000}-\u{200D}\u{300}-\u{036f}\u{2062}-\u{2063}\u{2800}\u{061c}\u{feff}]";
pub fn new(raw: &str) -> Self {
let normalized = raw.nfd().collect::<String>();
let replaced = DOT_REGEX.replace_all(&normalized, ":");
let replaced = HIDDEN_CHARACTERS_REGEX.replace_all(&replaced, "");
let decancered = decancer::cure!(&replaced).ok().map(|cured| {
let removed_left_to_right = LEFT_TO_RIGHT_REGEX.replace_all(cured.as_ref(), "");
let replaced = I_REGEX.replace_all(&removed_left_to_right, "l");
let replaced = DOT_REGEX.replace_all(&replaced, ":");
let replaced = ZERO_REGEX.replace_all(&replaced, "o");
replaced.to_string()
});
Self { raw: raw.to_owned(), decancered }
}
pub fn is_inherently_ambiguous(&self) -> bool {
self.looks_like_an_mxid() || self.has_hidden_characters() || self.decancered.is_none()
}
pub fn as_raw_str(&self) -> &str {
&self.raw
}
pub fn as_normalized_str(&self) -> Option<&str> {
self.decancered.as_deref()
}
fn has_hidden_characters(&self) -> bool {
HIDDEN_CHARACTERS_REGEX.is_match(&self.raw)
}
fn looks_like_an_mxid(&self) -> bool {
self.decancered
.as_deref()
.map(|d| MXID_REGEX.is_match(d))
.unwrap_or_else(|| MXID_REGEX.is_match(&self.raw))
}
}
#[derive(Clone, Debug, Default)]
pub struct MembersResponse {
pub chunk: Vec<RoomMemberEvent>,
pub ambiguity_changes: AmbiguityChanges,
}
#[derive(Clone, Debug, Serialize)]
#[serde(untagged)]
pub enum RawAnySyncOrStrippedTimelineEvent {
Sync(Raw<AnySyncTimelineEvent>),
Stripped(Raw<AnyStrippedStateEvent>),
}
#[derive(Clone, Debug, Serialize)]
#[serde(untagged)]
pub enum RawAnySyncOrStrippedState {
Sync(Raw<AnySyncStateEvent>),
Stripped(Raw<AnyStrippedStateEvent>),
}
impl RawAnySyncOrStrippedState {
pub fn deserialize(&self) -> serde_json::Result<AnySyncOrStrippedState> {
match self {
Self::Sync(raw) => Ok(AnySyncOrStrippedState::Sync(raw.deserialize()?)),
Self::Stripped(raw) => Ok(AnySyncOrStrippedState::Stripped(raw.deserialize()?)),
}
}
pub fn cast<C>(self) -> RawSyncOrStrippedState<C>
where
C: StaticStateEventContent + RedactContent,
C::Redacted: RedactedStateEventContent,
{
match self {
Self::Sync(raw) => RawSyncOrStrippedState::Sync(raw.cast()),
Self::Stripped(raw) => RawSyncOrStrippedState::Stripped(raw.cast()),
}
}
}
#[derive(Clone, Debug)]
pub enum AnySyncOrStrippedState {
Sync(AnySyncStateEvent),
Stripped(AnyStrippedStateEvent),
}
impl AnySyncOrStrippedState {
pub fn as_sync(&self) -> Option<&AnySyncStateEvent> {
match self {
Self::Sync(ev) => Some(ev),
Self::Stripped(_) => None,
}
}
pub fn as_stripped(&self) -> Option<&AnyStrippedStateEvent> {
match self {
Self::Sync(_) => None,
Self::Stripped(ev) => Some(ev),
}
}
}
#[derive(Clone, Debug, Serialize)]
#[serde(untagged)]
pub enum RawSyncOrStrippedState<C>
where
C: StaticStateEventContent + RedactContent,
C::Redacted: RedactedStateEventContent,
{
Sync(Raw<SyncStateEvent<C>>),
Stripped(Raw<StrippedStateEvent<C::PossiblyRedacted>>),
}
impl<C> RawSyncOrStrippedState<C>
where
C: StaticStateEventContent + RedactContent,
C::Redacted: RedactedStateEventContent + fmt::Debug + Clone,
{
pub fn deserialize(&self) -> serde_json::Result<SyncOrStrippedState<C>>
where
C: StaticStateEventContent + EventContentFromType + RedactContent,
C::Redacted: RedactedStateEventContent<StateKey = C::StateKey> + EventContentFromType,
C::PossiblyRedacted: PossiblyRedactedStateEventContent + EventContentFromType,
{
match self {
Self::Sync(ev) => Ok(SyncOrStrippedState::Sync(ev.deserialize()?)),
Self::Stripped(ev) => Ok(SyncOrStrippedState::Stripped(ev.deserialize()?)),
}
}
}
pub type RawMemberEvent = RawSyncOrStrippedState<RoomMemberEventContent>;
#[derive(Clone, Debug)]
pub enum SyncOrStrippedState<C>
where
C: StaticStateEventContent + RedactContent,
C::Redacted: RedactedStateEventContent + fmt::Debug + Clone,
{
Sync(SyncStateEvent<C>),
Stripped(StrippedStateEvent<C::PossiblyRedacted>),
}
impl<C> SyncOrStrippedState<C>
where
C: StaticStateEventContent + RedactContent,
C::Redacted: RedactedStateEventContent<StateKey = C::StateKey> + fmt::Debug + Clone,
C::PossiblyRedacted: PossiblyRedactedStateEventContent<StateKey = C::StateKey>,
{
pub fn as_sync(&self) -> Option<&SyncStateEvent<C>> {
match self {
Self::Sync(ev) => Some(ev),
Self::Stripped(_) => None,
}
}
pub fn as_stripped(&self) -> Option<&StrippedStateEvent<C::PossiblyRedacted>> {
match self {
Self::Sync(_) => None,
Self::Stripped(ev) => Some(ev),
}
}
pub fn sender(&self) -> &UserId {
match self {
Self::Sync(e) => e.sender(),
Self::Stripped(e) => &e.sender,
}
}
pub fn event_id(&self) -> Option<&EventId> {
match self {
Self::Sync(e) => Some(e.event_id()),
Self::Stripped(_) => None,
}
}
pub fn origin_server_ts(&self) -> Option<MilliSecondsSinceUnixEpoch> {
match self {
Self::Sync(e) => Some(e.origin_server_ts()),
Self::Stripped(_) => None,
}
}
pub fn state_key(&self) -> &C::StateKey {
match self {
Self::Sync(e) => e.state_key(),
Self::Stripped(e) => &e.state_key,
}
}
}
impl<C> SyncOrStrippedState<C>
where
C: StaticStateEventContent<PossiblyRedacted = C>
+ RedactContent
+ PossiblyRedactedStateEventContent,
C::Redacted: RedactedStateEventContent<StateKey = <C as StateEventContent>::StateKey>
+ fmt::Debug
+ Clone,
{
pub fn original_content(&self) -> Option<&C> {
match self {
Self::Sync(e) => e.as_original().map(|e| &e.content),
Self::Stripped(e) => Some(&e.content),
}
}
}
pub type MemberEvent = SyncOrStrippedState<RoomMemberEventContent>;
impl MemberEvent {
pub fn membership(&self) -> &MembershipState {
match self {
MemberEvent::Sync(e) => e.membership(),
MemberEvent::Stripped(e) => &e.content.membership,
}
}
pub fn user_id(&self) -> &UserId {
self.state_key()
}
pub fn display_name(&self) -> DisplayName {
DisplayName::new(
self.original_content()
.and_then(|c| c.displayname.as_deref())
.unwrap_or_else(|| self.user_id().localpart()),
)
}
}
impl SyncOrStrippedState<RoomPowerLevelsEventContent> {
pub fn power_levels(&self) -> RoomPowerLevels {
match self {
Self::Sync(e) => e.power_levels(),
Self::Stripped(e) => e.power_levels(),
}
}
}
#[cfg(test)]
mod test {
macro_rules! assert_display_name_eq {
($left:expr, $right:expr $(, $desc:expr)?) => {{
let left = crate::deserialized_responses::DisplayName::new($left);
let right = crate::deserialized_responses::DisplayName::new($right);
similar_asserts::assert_eq!(
left,
right
$(, $desc)?
);
}};
}
macro_rules! assert_display_name_ne {
($left:expr, $right:expr $(, $desc:expr)?) => {{
let left = crate::deserialized_responses::DisplayName::new($left);
let right = crate::deserialized_responses::DisplayName::new($right);
assert_ne!(
left,
right
$(, $desc)?
);
}};
}
macro_rules! assert_ambiguous {
($name:expr) => {
let name = crate::deserialized_responses::DisplayName::new($name);
assert!(
name.is_inherently_ambiguous(),
"The display {:?} should be considered amgibuous",
name
);
};
}
macro_rules! assert_not_ambiguous {
($name:expr) => {
let name = crate::deserialized_responses::DisplayName::new($name);
assert!(
!name.is_inherently_ambiguous(),
"The display {:?} should not be considered amgibuous",
name
);
};
}
#[test]
fn test_display_name_inherently_ambiguous() {
assert_not_ambiguous!("Alice");
assert_not_ambiguous!("Carol");
assert_not_ambiguous!("Car0l");
assert_not_ambiguous!("Ivan");
assert_not_ambiguous!("๐ฎ๐ถ๐ฝ๐ถ๐๐๐ถ๐ฝ๐๐ถ");
assert_not_ambiguous!("โโโโโขโกโโโโ");
assert_not_ambiguous!("๐
๐ฐ๐ท๐ฐ๐
๐
๐ฐ๐ท๐ป๐ฐ");
assert_not_ambiguous!("๏ผณ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ");
assert_not_ambiguous!("\u{202e}alharsahas");
assert_ambiguous!("Saฬดhasrahla");
assert_ambiguous!("Sahas\u{200D}rahla");
}
#[test]
fn test_display_name_equality_capitalization() {
assert_display_name_eq!("Alice", "alice");
}
#[test]
fn test_display_name_equality_different_names() {
assert_display_name_ne!("Alice", "Carol");
}
#[test]
fn test_display_name_equality_capital_l() {
assert_display_name_eq!("Hello", "HeIlo");
}
#[test]
fn test_display_name_equality_confusable_zero() {
assert_display_name_eq!("Carol", "Car0l");
}
#[test]
fn test_display_name_equality_cyrilic() {
assert_display_name_eq!("alice", "ะฐlice");
}
#[test]
fn test_display_name_equality_scriptures() {
assert_display_name_eq!("Sahasrahla", "๐ฎ๐ถ๐ฝ๐ถ๐๐๐ถ๐ฝ๐๐ถ");
}
#[test]
fn test_display_name_equality_frakturs() {
assert_display_name_eq!("Sahasrahla", "๐๐๐ฅ๐๐ฐ๐ฏ๐๐ฅ๐ฉ๐");
}
#[test]
fn test_display_name_equality_circled() {
assert_display_name_eq!("Sahasrahla", "โโโโโขโกโโโโ");
}
#[test]
fn test_display_name_equality_squared() {
assert_display_name_eq!("Sahasrahla", "๐
๐ฐ๐ท๐ฐ๐
๐
๐ฐ๐ท๐ป๐ฐ");
}
#[test]
fn test_display_name_equality_big_unicode() {
assert_display_name_eq!("Sahasrahla", "๏ผณ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ");
}
#[test]
fn test_display_name_equality_left_to_right() {
assert_display_name_eq!("Sahasrahla", "\u{202e}alharsahas");
}
#[test]
fn test_display_name_equality_diacritical() {
assert_display_name_eq!("Sahasrahla", "Saฬดhasrahla");
}
#[test]
fn test_display_name_equality_zero_width_joiner() {
assert_display_name_eq!("Sahasrahla", "Sahas\u{200B}rahla");
}
#[test]
fn test_display_name_equality_zero_width_space() {
assert_display_name_eq!("Sahasrahla", "Sahas\u{200D}rahla");
}
#[test]
fn test_display_name_equality_ligatures() {
assert_display_name_eq!("ff", "\u{FB00}");
}
#[test]
fn test_display_name_confusable_mxid_colon() {
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{0589}domain.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{05c3}domain.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{0703}domain.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{0a83}domain.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{16ec}domain.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{205a}domain.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{2236}domain.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{fe13}domain.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{fe52}domain.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{fe30}domain.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{ff1a}domain.tld");
assert_ambiguous!("@mxid\u{0589}domain.tld");
assert_ambiguous!("@mxid\u{05c3}domain.tld");
assert_ambiguous!("@mxid\u{0703}domain.tld");
assert_ambiguous!("@mxid\u{0a83}domain.tld");
assert_ambiguous!("@mxid\u{16ec}domain.tld");
assert_ambiguous!("@mxid\u{205a}domain.tld");
assert_ambiguous!("@mxid\u{2236}domain.tld");
assert_ambiguous!("@mxid\u{fe13}domain.tld");
assert_ambiguous!("@mxid\u{fe52}domain.tld");
assert_ambiguous!("@mxid\u{fe30}domain.tld");
assert_ambiguous!("@mxid\u{ff1a}domain.tld");
}
#[test]
fn test_display_name_confusable_mxid_dot() {
assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{0701}tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{0702}tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{2024}tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{fe52}tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{ff0e}tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{1d16d}tld");
assert_ambiguous!("@mxid:domain\u{0701}tld");
assert_ambiguous!("@mxid:domain\u{0702}tld");
assert_ambiguous!("@mxid:domain\u{2024}tld");
assert_ambiguous!("@mxid:domain\u{fe52}tld");
assert_ambiguous!("@mxid:domain\u{ff0e}tld");
assert_ambiguous!("@mxid:domain\u{1d16d}tld");
}
#[test]
fn test_display_name_confusable_mxid_replacing_a() {
assert_display_name_eq!("@mxid:domain.tld", "@mxid:dom\u{1d44e}in.tld");
assert_display_name_eq!("@mxid:domain.tld", "@mxid:dom\u{0430}in.tld");
assert_ambiguous!("@mxid:dom\u{1d44e}in.tld");
assert_ambiguous!("@mxid:dom\u{0430}in.tld");
}
#[test]
fn test_display_name_confusable_mxid_replacing_l() {
assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain.tId");
assert_display_name_eq!("mxid:domain.tld", "mxid:domain.t\u{217c}d");
assert_display_name_eq!("mxid:domain.tld", "mxid:domain.t\u{ff4c}d");
assert_display_name_eq!("mxid:domain.tld", "mxid:domain.t\u{1d5f9}d");
assert_display_name_eq!("mxid:domain.tld", "mxid:domain.t\u{1d695}d");
assert_display_name_eq!("mxid:domain.tld", "mxid:domain.t\u{2223}d");
assert_ambiguous!("@mxid:domain.tId");
assert_ambiguous!("@mxid:domain.t\u{217c}d");
assert_ambiguous!("@mxid:domain.t\u{ff4c}d");
assert_ambiguous!("@mxid:domain.t\u{1d5f9}d");
assert_ambiguous!("@mxid:domain.t\u{1d695}d");
assert_ambiguous!("@mxid:domain.t\u{2223}d");
}
}