Implement nick normalization and mentioning

CHANGELOG.md

@@ -19,6 +19,8 @@ Procedure when bumping the version number:
 - `euphoxide::Emoji` for finding, replacing and removing colon-delimited emoji in text
 - `euphoxide::api::Time::new`
 - `euphoxide::nick::hue`
+- `euphoxide::nick::mention`
+- `euphoxide::nick::normalize`
 - Debug logging using the `log` crate
 - `testbot_instance` example using the new `euphoxide::bot::instance::Instance`
 - VSCode project settings

Cargo.toml

@@ -8,6 +8,7 @@ bot = ["dep:async-trait", "dep:clap", "dep:cookie"]
 
 [dependencies]
 async-trait = { version = "0.1.63", optional = true }
+caseless = "0.2.1"
 cookie = { version = "0.16.2", optional = true }
 futures-util = { version = "0.3.25", default-features = false, features = ["sink"] }
 log = "0.4.17"
@@ -17,6 +18,7 @@ time = { version = "0.3.17", features = ["serde"] }
 tokio = { version = "1.23.0", features = ["time", "sync", "macros", "rt"] }
 tokio-stream = "0.1.11"
 tokio-tungstenite = { version = "0.18.0", features = ["rustls-tls-native-roots"] }
+unicode-normalization = "0.1.22"
 
 [dependencies.clap]
 version = "4.1.3"

src/nick.rs

@@ -1,5 +1,8 @@
 //! Nick-related utility functions.
 
+use caseless::Caseless;
+use unicode_normalization::UnicodeNormalization;
+
 use crate::emoji::Emoji;
 
 /// Does not remove emoji.
@@ -45,7 +48,60 @@ pub fn hue_without_removing_emoji(nick: &str) -> u8 {
 /// This is a reimplementation of [euphoria's nick hue hashing algorithm][0]. It
 /// should always return the same value as the official client's implementation.
 ///
-/// [0]: https://github.com/euphoria-io/heim/blob/master/client/lib/hueHash.js
+/// [0]: https://github.com/euphoria-io/heim/blob/978c921063e6b06012fc8d16d9fbf1b3a0be1191/client/lib/hueHash.js
 pub fn hue(emoji: &Emoji, nick: &str) -> u8 {
     hue_without_removing_emoji(&emoji.remove(nick))
 }
+
+fn delimits_mention(c: char) -> bool {
+    match c {
+        ',' | '.' | '!' | '?' | ';' | '&' | '<' | '>' | '\'' | '"' => true,
+        _ => c.is_whitespace(),
+    }
+}
+
+/// Normalize a nick to a form that can be compared against other nicks.
+///
+/// This normalization is less aggressive than the nick hue normalization. It is
+/// also less aggressive than the normalization used by the euphoria client to
+/// determine who is pinged by a mention. This means that it will not compute
+/// the same normal form for all pairs of nicks that ping each other in the
+/// euphoria client.
+///
+/// A nick and its mention form calculated via [`mention`] will always evaluate
+/// to the same normal form.
+///
+/// The steps performed are as follows:
+///
+/// 1. Remove all whitespace characters as well as the characters
+///    `,`, `.`, `!`, `?`, `;`, `&`, `<`, `>`, `'`, `"`.
+///    All of these are used by the [frontend to delimit mentions][0].
+///    The `>` character was confirmed to delimit mentions via experimentation.
+///
+/// 2. Convert to NFKC
+/// 3. Case fold
+///
+/// Steps 2 and 3 are meant to be an alternative to the NKFC_Casefold derived
+/// property that's easier to implement, even though it may be incorrect in some
+/// edge cases.
+///
+/// [0]: https://github.com/euphoria-io/heim/blob/978c921063e6b06012fc8d16d9fbf1b3a0be1191/client/lib/stores/chat.js#L14
+pub fn normalize(nick: &str) -> String {
+    nick.chars()
+        .filter(|c| !delimits_mention(*c)) // Step 1
+        .nfkc() // Step 2
+        .default_case_fold() // Step 3
+        .collect()
+}
+
+/// Compute a mentionable version of a nick while remaining as close to the
+/// original as possible.
+///
+/// The return value of this function appended to an `@` character will
+/// highlight as a mention in the official euphoria client. It should ping any
+/// people using the original nick. It might also ping other people.
+///
+/// This function performs step 1 of [`normalize`].
+pub fn mention(nick: &str) -> String {
+    nick.replace(delimits_mention, "")
+}