1use std::{collections::BTreeMap, fmt, hash::Hash, iter};
18
19pub use matrix_sdk_common::deserialized_responses::*;
20use once_cell::sync::Lazy;
21use regex::Regex;
22use ruma::{
23 events::{
24 room::{
25 member::{MembershipState, RoomMemberEvent, RoomMemberEventContent},
26 power_levels::{RoomPowerLevels, RoomPowerLevelsEventContent},
27 },
28 AnyStrippedStateEvent, AnySyncStateEvent, AnySyncTimelineEvent, EventContentFromType,
29 PossiblyRedactedStateEventContent, RedactContent, RedactedStateEventContent,
30 StateEventContent, StaticStateEventContent, StrippedStateEvent, SyncStateEvent,
31 },
32 serde::Raw,
33 EventId, MilliSecondsSinceUnixEpoch, OwnedEventId, OwnedRoomId, OwnedUserId, UInt, UserId,
34};
35use serde::Serialize;
36use unicode_normalization::UnicodeNormalization;
37
38#[derive(Clone, Debug)]
41#[non_exhaustive]
42pub struct AmbiguityChange {
43 pub member_id: OwnedUserId,
46 pub member_ambiguous: bool,
49 pub disambiguated_member: Option<OwnedUserId>,
51 pub ambiguated_member: Option<OwnedUserId>,
53}
54
55impl AmbiguityChange {
56 pub fn user_ids(&self) -> impl Iterator<Item = &UserId> {
58 iter::once(&*self.member_id)
59 .chain(self.disambiguated_member.as_deref())
60 .chain(self.ambiguated_member.as_deref())
61 }
62}
63
64#[derive(Clone, Debug, Default)]
66#[non_exhaustive]
67pub struct AmbiguityChanges {
68 pub changes: BTreeMap<OwnedRoomId, BTreeMap<OwnedEventId, AmbiguityChange>>,
71}
72
73static MXID_REGEX: Lazy<Regex> = Lazy::new(|| {
74 Regex::new(DisplayName::MXID_PATTERN)
75 .expect("We should be able to create a regex from our static MXID pattern")
76});
77static LEFT_TO_RIGHT_REGEX: Lazy<Regex> = Lazy::new(|| {
78 Regex::new(DisplayName::LEFT_TO_RIGHT_PATTERN)
79 .expect("We should be able to create a regex from our static left-to-right pattern")
80});
81static HIDDEN_CHARACTERS_REGEX: Lazy<Regex> = Lazy::new(|| {
82 Regex::new(DisplayName::HIDDEN_CHARACTERS_PATTERN)
83 .expect("We should be able to create a regex from our static hidden characters pattern")
84});
85
86static I_REGEX: Lazy<Regex> = Lazy::new(|| {
91 Regex::new("[i]").expect("We should be able to create a regex from our uppercase I pattern")
92});
93
94static ZERO_REGEX: Lazy<Regex> = Lazy::new(|| {
99 Regex::new("[0]").expect("We should be able to create a regex from our zero pattern")
100});
101
102static DOT_REGEX: Lazy<Regex> = Lazy::new(|| {
107 Regex::new("[.\u{1d16d}]").expect("We should be able to create a regex from our dot pattern")
108});
109
110#[derive(Debug, Clone, Eq)]
136pub struct DisplayName {
137 raw: String,
138 decancered: Option<String>,
139}
140
141impl Hash for DisplayName {
142 fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
143 if let Some(decancered) = &self.decancered {
144 decancered.hash(state);
145 } else {
146 self.raw.hash(state);
147 }
148 }
149}
150
151impl PartialEq for DisplayName {
152 fn eq(&self, other: &Self) -> bool {
153 match (self.decancered.as_deref(), other.decancered.as_deref()) {
154 (None, None) => self.raw == other.raw,
155 (None, Some(_)) | (Some(_), None) => false,
156 (Some(this), Some(other)) => this == other,
157 }
158 }
159}
160
161impl DisplayName {
162 const MXID_PATTERN: &'static str = "@.+[:.].+";
164
165 const LEFT_TO_RIGHT_PATTERN: &'static str = "[\u{202a}-\u{202f}\u{200e}\u{200f}]";
169
170 const HIDDEN_CHARACTERS_PATTERN: &'static str =
180 "[\u{2000}-\u{200D}\u{300}-\u{036f}\u{2062}-\u{2063}\u{2800}\u{061c}\u{feff}]";
181
182 pub fn new(raw: &str) -> Self {
191 let normalized = raw.nfd().collect::<String>();
192 let replaced = DOT_REGEX.replace_all(&normalized, ":");
193 let replaced = HIDDEN_CHARACTERS_REGEX.replace_all(&replaced, "");
194
195 let decancered = decancer::cure!(&replaced).ok().map(|cured| {
196 let removed_left_to_right = LEFT_TO_RIGHT_REGEX.replace_all(cured.as_ref(), "");
197 let replaced = I_REGEX.replace_all(&removed_left_to_right, "l");
198 let replaced = DOT_REGEX.replace_all(&replaced, ":");
201 let replaced = ZERO_REGEX.replace_all(&replaced, "o");
202
203 replaced.to_string()
204 });
205
206 Self { raw: raw.to_owned(), decancered }
207 }
208
209 pub fn is_inherently_ambiguous(&self) -> bool {
214 self.looks_like_an_mxid() || self.has_hidden_characters() || self.decancered.is_none()
216 }
217
218 pub fn as_raw_str(&self) -> &str {
221 &self.raw
222 }
223
224 pub fn as_normalized_str(&self) -> Option<&str> {
230 self.decancered.as_deref()
231 }
232
233 fn has_hidden_characters(&self) -> bool {
234 HIDDEN_CHARACTERS_REGEX.is_match(&self.raw)
235 }
236
237 fn looks_like_an_mxid(&self) -> bool {
238 self.decancered
239 .as_deref()
240 .map(|d| MXID_REGEX.is_match(d))
241 .unwrap_or_else(|| MXID_REGEX.is_match(&self.raw))
242 }
243}
244
245#[derive(Clone, Debug, Default)]
249pub struct MembersResponse {
250 pub chunk: Vec<RoomMemberEvent>,
252 pub ambiguity_changes: AmbiguityChanges,
254}
255
256#[derive(Clone, Debug, Serialize)]
258#[serde(untagged)]
259pub enum RawAnySyncOrStrippedTimelineEvent {
260 Sync(Raw<AnySyncTimelineEvent>),
262 Stripped(Raw<AnyStrippedStateEvent>),
264}
265
266#[derive(Clone, Debug, Serialize)]
268#[serde(untagged)]
269pub enum RawAnySyncOrStrippedState {
270 Sync(Raw<AnySyncStateEvent>),
272 Stripped(Raw<AnyStrippedStateEvent>),
274}
275
276impl RawAnySyncOrStrippedState {
277 pub fn deserialize(&self) -> serde_json::Result<AnySyncOrStrippedState> {
279 match self {
280 Self::Sync(raw) => Ok(AnySyncOrStrippedState::Sync(Box::new(raw.deserialize()?))),
281 Self::Stripped(raw) => {
282 Ok(AnySyncOrStrippedState::Stripped(Box::new(raw.deserialize()?)))
283 }
284 }
285 }
286
287 pub fn cast<C>(self) -> RawSyncOrStrippedState<C>
290 where
291 C: StaticStateEventContent + RedactContent,
292 C::Redacted: RedactedStateEventContent,
293 {
294 match self {
295 Self::Sync(raw) => RawSyncOrStrippedState::Sync(raw.cast()),
296 Self::Stripped(raw) => RawSyncOrStrippedState::Stripped(raw.cast()),
297 }
298 }
299}
300
301#[derive(Clone, Debug)]
303pub enum AnySyncOrStrippedState {
304 Sync(Box<AnySyncStateEvent>),
309 Stripped(Box<AnyStrippedStateEvent>),
314}
315
316impl AnySyncOrStrippedState {
317 pub fn as_sync(&self) -> Option<&AnySyncStateEvent> {
320 match self {
321 Self::Sync(ev) => Some(ev),
322 Self::Stripped(_) => None,
323 }
324 }
325
326 pub fn as_stripped(&self) -> Option<&AnyStrippedStateEvent> {
329 match self {
330 Self::Sync(_) => None,
331 Self::Stripped(ev) => Some(ev),
332 }
333 }
334}
335
336#[derive(Clone, Debug, Serialize)]
338#[serde(untagged)]
339pub enum RawSyncOrStrippedState<C>
340where
341 C: StaticStateEventContent + RedactContent,
342 C::Redacted: RedactedStateEventContent,
343{
344 Sync(Raw<SyncStateEvent<C>>),
346 Stripped(Raw<StrippedStateEvent<C::PossiblyRedacted>>),
348}
349
350impl<C> RawSyncOrStrippedState<C>
351where
352 C: StaticStateEventContent + RedactContent,
353 C::Redacted: RedactedStateEventContent + fmt::Debug + Clone,
354{
355 pub fn deserialize(&self) -> serde_json::Result<SyncOrStrippedState<C>>
357 where
358 C: StaticStateEventContent + EventContentFromType + RedactContent,
359 C::Redacted: RedactedStateEventContent<StateKey = C::StateKey> + EventContentFromType,
360 C::PossiblyRedacted: PossiblyRedactedStateEventContent + EventContentFromType,
361 {
362 match self {
363 Self::Sync(ev) => Ok(SyncOrStrippedState::Sync(ev.deserialize()?)),
364 Self::Stripped(ev) => Ok(SyncOrStrippedState::Stripped(ev.deserialize()?)),
365 }
366 }
367}
368
369pub type RawMemberEvent = RawSyncOrStrippedState<RoomMemberEventContent>;
371
372#[derive(Clone, Debug)]
374pub enum SyncOrStrippedState<C>
375where
376 C: StaticStateEventContent + RedactContent,
377 C::Redacted: RedactedStateEventContent + fmt::Debug + Clone,
378{
379 Sync(SyncStateEvent<C>),
381 Stripped(StrippedStateEvent<C::PossiblyRedacted>),
383}
384
385impl<C> SyncOrStrippedState<C>
386where
387 C: StaticStateEventContent + RedactContent,
388 C::Redacted: RedactedStateEventContent<StateKey = C::StateKey> + fmt::Debug + Clone,
389 C::PossiblyRedacted: PossiblyRedactedStateEventContent<StateKey = C::StateKey>,
390{
391 pub fn as_sync(&self) -> Option<&SyncStateEvent<C>> {
393 match self {
394 Self::Sync(ev) => Some(ev),
395 Self::Stripped(_) => None,
396 }
397 }
398
399 pub fn as_stripped(&self) -> Option<&StrippedStateEvent<C::PossiblyRedacted>> {
402 match self {
403 Self::Sync(_) => None,
404 Self::Stripped(ev) => Some(ev),
405 }
406 }
407
408 pub fn sender(&self) -> &UserId {
410 match self {
411 Self::Sync(e) => e.sender(),
412 Self::Stripped(e) => &e.sender,
413 }
414 }
415
416 pub fn event_id(&self) -> Option<&EventId> {
418 match self {
419 Self::Sync(e) => Some(e.event_id()),
420 Self::Stripped(_) => None,
421 }
422 }
423
424 pub fn origin_server_ts(&self) -> Option<MilliSecondsSinceUnixEpoch> {
426 match self {
427 Self::Sync(e) => Some(e.origin_server_ts()),
428 Self::Stripped(_) => None,
429 }
430 }
431
432 pub fn state_key(&self) -> &C::StateKey {
434 match self {
435 Self::Sync(e) => e.state_key(),
436 Self::Stripped(e) => &e.state_key,
437 }
438 }
439}
440
441impl<C> SyncOrStrippedState<C>
442where
443 C: StaticStateEventContent<PossiblyRedacted = C>
444 + RedactContent
445 + PossiblyRedactedStateEventContent,
446 C::Redacted: RedactedStateEventContent<StateKey = <C as StateEventContent>::StateKey>
447 + fmt::Debug
448 + Clone,
449{
450 pub fn original_content(&self) -> Option<&C> {
452 match self {
453 Self::Sync(e) => e.as_original().map(|e| &e.content),
454 Self::Stripped(e) => Some(&e.content),
455 }
456 }
457}
458
459pub type MemberEvent = SyncOrStrippedState<RoomMemberEventContent>;
461
462impl MemberEvent {
463 pub fn membership(&self) -> &MembershipState {
465 match self {
466 MemberEvent::Sync(e) => e.membership(),
467 MemberEvent::Stripped(e) => &e.content.membership,
468 }
469 }
470
471 pub fn user_id(&self) -> &UserId {
473 self.state_key()
474 }
475
476 pub fn display_name(&self) -> DisplayName {
481 DisplayName::new(
482 self.original_content()
483 .and_then(|c| c.displayname.as_deref())
484 .unwrap_or_else(|| self.user_id().localpart()),
485 )
486 }
487
488 pub fn reason(&self) -> Option<&str> {
490 match self {
491 MemberEvent::Sync(SyncStateEvent::Original(c)) => c.content.reason.as_deref(),
492 MemberEvent::Stripped(e) => e.content.reason.as_deref(),
493 _ => None,
494 }
495 }
496
497 pub fn timestamp(&self) -> Option<UInt> {
499 match self {
500 MemberEvent::Sync(SyncStateEvent::Original(c)) => Some(c.origin_server_ts.0),
501 _ => None,
502 }
503 }
504}
505
506impl SyncOrStrippedState<RoomPowerLevelsEventContent> {
507 pub fn power_levels(&self) -> RoomPowerLevels {
509 match self {
510 Self::Sync(e) => e.power_levels(),
511 Self::Stripped(e) => e.power_levels(),
512 }
513 }
514}
515
516#[cfg(test)]
517mod test {
518 macro_rules! assert_display_name_eq {
519 ($left:expr, $right:expr $(, $desc:expr)?) => {{
520 let left = crate::deserialized_responses::DisplayName::new($left);
521 let right = crate::deserialized_responses::DisplayName::new($right);
522
523 similar_asserts::assert_eq!(
524 left,
525 right
526 $(, $desc)?
527 );
528 }};
529 }
530
531 macro_rules! assert_display_name_ne {
532 ($left:expr, $right:expr $(, $desc:expr)?) => {{
533 let left = crate::deserialized_responses::DisplayName::new($left);
534 let right = crate::deserialized_responses::DisplayName::new($right);
535
536 assert_ne!(
537 left,
538 right
539 $(, $desc)?
540 );
541 }};
542 }
543
544 macro_rules! assert_ambiguous {
545 ($name:expr) => {
546 let name = crate::deserialized_responses::DisplayName::new($name);
547
548 assert!(
549 name.is_inherently_ambiguous(),
550 "The display {:?} should be considered amgibuous",
551 name
552 );
553 };
554 }
555
556 macro_rules! assert_not_ambiguous {
557 ($name:expr) => {
558 let name = crate::deserialized_responses::DisplayName::new($name);
559
560 assert!(
561 !name.is_inherently_ambiguous(),
562 "The display {:?} should not be considered amgibuous",
563 name
564 );
565 };
566 }
567
568 #[test]
569 fn test_display_name_inherently_ambiguous() {
570 assert_not_ambiguous!("Alice");
573 assert_not_ambiguous!("Carol");
574 assert_not_ambiguous!("Car0l");
575 assert_not_ambiguous!("Ivan");
576 assert_not_ambiguous!("๐ฎ๐ถ๐ฝ๐ถ๐๐๐ถ๐ฝ๐๐ถ");
577 assert_not_ambiguous!("โโโโโขโกโโโโ");
578 assert_not_ambiguous!("๐
๐ฐ๐ท๐ฐ๐
๐
๐ฐ๐ท๐ป๐ฐ");
579 assert_not_ambiguous!("๏ผณ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ");
580 assert_not_ambiguous!("\u{202e}alharsahas");
582
583 assert_ambiguous!("Saฬดhasrahla");
585 assert_ambiguous!("Sahas\u{200D}rahla");
586 }
587
588 #[test]
589 fn test_display_name_equality_capitalization() {
590 assert_display_name_eq!("Alice", "alice");
592 }
593
594 #[test]
595 fn test_display_name_equality_different_names() {
596 assert_display_name_ne!("Alice", "Carol");
598 }
599
600 #[test]
601 fn test_display_name_equality_capital_l() {
602 assert_display_name_eq!("Hello", "HeIlo");
604 }
605
606 #[test]
607 fn test_display_name_equality_confusable_zero() {
608 assert_display_name_eq!("Carol", "Car0l");
610 }
611
612 #[test]
613 fn test_display_name_equality_cyrillic() {
614 assert_display_name_eq!("alice", "ะฐlice");
616 }
617
618 #[test]
619 fn test_display_name_equality_scriptures() {
620 assert_display_name_eq!("Sahasrahla", "๐ฎ๐ถ๐ฝ๐ถ๐๐๐ถ๐ฝ๐๐ถ");
622 }
623
624 #[test]
625 fn test_display_name_equality_frakturs() {
626 assert_display_name_eq!("Sahasrahla", "๐๐๐ฅ๐๐ฐ๐ฏ๐๐ฅ๐ฉ๐");
628 }
629
630 #[test]
631 fn test_display_name_equality_circled() {
632 assert_display_name_eq!("Sahasrahla", "โโโโโขโกโโโโ");
634 }
635
636 #[test]
637 fn test_display_name_equality_squared() {
638 assert_display_name_eq!("Sahasrahla", "๐
๐ฐ๐ท๐ฐ๐
๐
๐ฐ๐ท๐ป๐ฐ");
640 }
641
642 #[test]
643 fn test_display_name_equality_big_unicode() {
644 assert_display_name_eq!("Sahasrahla", "๏ผณ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ๏ฝ");
646 }
647
648 #[test]
649 fn test_display_name_equality_left_to_right() {
650 assert_display_name_eq!("Sahasrahla", "\u{202e}alharsahas");
652 }
653
654 #[test]
655 fn test_display_name_equality_diacritical() {
656 assert_display_name_eq!("Sahasrahla", "Saฬดhasrahla");
658 }
659
660 #[test]
661 fn test_display_name_equality_zero_width_joiner() {
662 assert_display_name_eq!("Sahasrahla", "Sahas\u{200B}rahla");
664 }
665
666 #[test]
667 fn test_display_name_equality_zero_width_space() {
668 assert_display_name_eq!("Sahasrahla", "Sahas\u{200D}rahla");
670 }
671
672 #[test]
673 fn test_display_name_equality_ligatures() {
674 assert_display_name_eq!("ff", "\u{FB00}");
676 }
677
678 #[test]
679 fn test_display_name_confusable_mxid_colon() {
680 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{0589}domain.tld");
681 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{05c3}domain.tld");
682 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{0703}domain.tld");
683 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{0a83}domain.tld");
684 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{16ec}domain.tld");
685 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{205a}domain.tld");
686 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{2236}domain.tld");
687 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{fe13}domain.tld");
688 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{fe52}domain.tld");
689 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{fe30}domain.tld");
690 assert_display_name_eq!("@mxid:domain.tld", "@mxid\u{ff1a}domain.tld");
691
692 assert_ambiguous!("@mxid\u{0589}domain.tld");
694 assert_ambiguous!("@mxid\u{05c3}domain.tld");
695 assert_ambiguous!("@mxid\u{0703}domain.tld");
696 assert_ambiguous!("@mxid\u{0a83}domain.tld");
697 assert_ambiguous!("@mxid\u{16ec}domain.tld");
698 assert_ambiguous!("@mxid\u{205a}domain.tld");
699 assert_ambiguous!("@mxid\u{2236}domain.tld");
700 assert_ambiguous!("@mxid\u{fe13}domain.tld");
701 assert_ambiguous!("@mxid\u{fe52}domain.tld");
702 assert_ambiguous!("@mxid\u{fe30}domain.tld");
703 assert_ambiguous!("@mxid\u{ff1a}domain.tld");
704 }
705
706 #[test]
707 fn test_display_name_confusable_mxid_dot() {
708 assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{0701}tld");
709 assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{0702}tld");
710 assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{2024}tld");
711 assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{fe52}tld");
712 assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{ff0e}tld");
713 assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain\u{1d16d}tld");
714
715 assert_ambiguous!("@mxid:domain\u{0701}tld");
717 assert_ambiguous!("@mxid:domain\u{0702}tld");
718 assert_ambiguous!("@mxid:domain\u{2024}tld");
719 assert_ambiguous!("@mxid:domain\u{fe52}tld");
720 assert_ambiguous!("@mxid:domain\u{ff0e}tld");
721 assert_ambiguous!("@mxid:domain\u{1d16d}tld");
722 }
723
724 #[test]
725 fn test_display_name_confusable_mxid_replacing_a() {
726 assert_display_name_eq!("@mxid:domain.tld", "@mxid:dom\u{1d44e}in.tld");
727 assert_display_name_eq!("@mxid:domain.tld", "@mxid:dom\u{0430}in.tld");
728
729 assert_ambiguous!("@mxid:dom\u{1d44e}in.tld");
731 assert_ambiguous!("@mxid:dom\u{0430}in.tld");
732 }
733
734 #[test]
735 fn test_display_name_confusable_mxid_replacing_l() {
736 assert_display_name_eq!("@mxid:domain.tld", "@mxid:domain.tId");
737 assert_display_name_eq!("mxid:domain.tld", "mxid:domain.t\u{217c}d");
738 assert_display_name_eq!("mxid:domain.tld", "mxid:domain.t\u{ff4c}d");
739 assert_display_name_eq!("mxid:domain.tld", "mxid:domain.t\u{1d5f9}d");
740 assert_display_name_eq!("mxid:domain.tld", "mxid:domain.t\u{1d695}d");
741 assert_display_name_eq!("mxid:domain.tld", "mxid:domain.t\u{2223}d");
742
743 assert_ambiguous!("@mxid:domain.tId");
745 assert_ambiguous!("@mxid:domain.t\u{217c}d");
746 assert_ambiguous!("@mxid:domain.t\u{ff4c}d");
747 assert_ambiguous!("@mxid:domain.t\u{1d5f9}d");
748 assert_ambiguous!("@mxid:domain.t\u{1d695}d");
749 assert_ambiguous!("@mxid:domain.t\u{2223}d");
750 }
751}