Use `codex::ModifierSet` (#6159)

Co-authored-by: Laurenz <laurmaedje@gmail.com>

6 files changed, 63 insertions, 95 deletions

diff --git a/Cargo.lock b/Cargo.lock
index b699d245..49e5203f 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -413,8 +413,7 @@ dependencies = [
 [[package]]
 name = "codex"
 version = "0.1.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "724d27a0ee38b700e5e164350e79aba601a0db673ac47fce1cb74c3e38864036"
+source = "git+https://github.com/typst/codex?rev=56eb2172fc0670f4c1c8b79a63d11f9354e5babe#56eb2172fc0670f4c1c8b79a63d11f9354e5babe"
 
 [[package]]
 name = "color-print"
@@ -2921,6 +2920,7 @@ name = "typst-docs"
 version = "0.13.1"
 dependencies = [
  "clap",
+ "codex",
  "ecow",
  "heck",
  "pulldown-cmark",
diff --git a/Cargo.toml b/Cargo.toml
index b548245f..03141cbb 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -47,7 +47,7 @@ clap = { version = "4.4", features = ["derive", "env", "wrap_help"] }
 clap_complete = "4.2.1"
 clap_mangen = "0.2.10"
 codespan-reporting = "0.11"
-codex = "0.1.1"
+codex = { git = "https://github.com/typst/codex", rev = "56eb217" }
 color-print = "0.3.6"
 comemo = "0.4"
 csv = "1"
diff --git a/crates/typst-library/src/foundations/symbol.rs b/crates/typst-library/src/foundations/symbol.rs
index 50fcfb40..0f503edd 100644
--- a/crates/typst-library/src/foundations/symbol.rs
+++ b/crates/typst-library/src/foundations/symbol.rs
@@ -1,8 +1,8 @@
-use std::cmp::Reverse;
 use std::collections::{BTreeSet, HashMap};
 use std::fmt::{self, Debug, Display, Formatter, Write};
 use std::sync::Arc;
 
+use codex::ModifierSet;
 use ecow::{eco_format, EcoString};
 use serde::{Serialize, Serializer};
 use typst_syntax::{is_ident, Span, Spanned};
@@ -54,18 +54,18 @@ enum Repr {
     /// A native symbol that has no named variant.
     Single(char),
     /// A native symbol with multiple named variants.
-    Complex(&'static [(&'static str, char)]),
+    Complex(&'static [(ModifierSet<&'static str>, char)]),
     /// A symbol with multiple named variants, where some modifiers may have
     /// been applied. Also used for symbols defined at runtime by the user with
     /// no modifier applied.
-    Modified(Arc<(List, EcoString)>),
+    Modified(Arc<(List, ModifierSet<EcoString>)>),
 }
 
 /// A collection of symbols.
 #[derive(Clone, Eq, PartialEq, Hash)]
 enum List {
-    Static(&'static [(&'static str, char)]),
-    Runtime(Box<[(EcoString, char)]>),
+    Static(&'static [(ModifierSet<&'static str>, char)]),
+    Runtime(Box<[(ModifierSet<EcoString>, char)]>),
 }
 
 impl Symbol {
@@ -76,24 +76,26 @@ impl Symbol {
 
     /// Create a symbol with a static variant list.
     #[track_caller]
-    pub const fn list(list: &'static [(&'static str, char)]) -> Self {
+    pub const fn list(list: &'static [(ModifierSet<&'static str>, char)]) -> Self {
         debug_assert!(!list.is_empty());
         Self(Repr::Complex(list))
     }
 
     /// Create a symbol with a runtime variant list.
     #[track_caller]
-    pub fn runtime(list: Box<[(EcoString, char)]>) -> Self {
+    pub fn runtime(list: Box<[(ModifierSet<EcoString>, char)]>) -> Self {
         debug_assert!(!list.is_empty());
-        Self(Repr::Modified(Arc::new((List::Runtime(list), EcoString::new()))))
+        Self(Repr::Modified(Arc::new((List::Runtime(list), ModifierSet::default()))))
     }
 
     /// Get the symbol's character.
     pub fn get(&self) -> char {
         match &self.0 {
             Repr::Single(c) => *c,
-            Repr::Complex(_) => find(self.variants(), "").unwrap(),
-            Repr::Modified(arc) => find(self.variants(), &arc.1).unwrap(),
+            Repr::Complex(_) => ModifierSet::<&'static str>::default()
+                .best_match_in(self.variants())
+                .unwrap(),
+            Repr::Modified(arc) => arc.1.best_match_in(self.variants()).unwrap(),
         }
     }
 
@@ -128,16 +130,14 @@ impl Symbol {
     /// Apply a modifier to the symbol.
     pub fn modified(mut self, modifier: &str) -> StrResult<Self> {
         if let Repr::Complex(list) = self.0 {
-            self.0 = Repr::Modified(Arc::new((List::Static(list), EcoString::new())));
+            self.0 =
+                Repr::Modified(Arc::new((List::Static(list), ModifierSet::default())));
         }
 
         if let Repr::Modified(arc) = &mut self.0 {
             let (list, modifiers) = Arc::make_mut(arc);
-            if !modifiers.is_empty() {
-                modifiers.push('.');
-            }
-            modifiers.push_str(modifier);
-            if find(list.variants(), modifiers).is_some() {
+            modifiers.insert_raw(modifier);
+            if modifiers.best_match_in(list.variants()).is_some() {
                 return Ok(self);
             }
         }
@@ -146,7 +146,7 @@ impl Symbol {
     }
 
     /// The characters that are covered by this symbol.
-    pub fn variants(&self) -> impl Iterator<Item = (&str, char)> {
+    pub fn variants(&self) -> impl Iterator<Item = (ModifierSet<&str>, char)> {
         match &self.0 {
             Repr::Single(c) => Variants::Single(Some(*c).into_iter()),
             Repr::Complex(list) => Variants::Static(list.iter()),
@@ -156,17 +156,15 @@ impl Symbol {
 
     /// Possible modifiers.
     pub fn modifiers(&self) -> impl Iterator<Item = &str> + '_ {
-        let mut set = BTreeSet::new();
         let modifiers = match &self.0 {
-            Repr::Modified(arc) => arc.1.as_str(),
-            _ => "",
+            Repr::Modified(arc) => arc.1.as_deref(),
+            _ => ModifierSet::default(),
         };
-        for modifier in self.variants().flat_map(|(name, _)| name.split('.')) {
-            if !modifier.is_empty() && !contained(modifiers, modifier) {
-                set.insert(modifier);
-            }
-        }
-        set.into_iter()
+        self.variants()
+            .flat_map(|(m, _)| m)
+            .filter(|modifier| !modifier.is_empty() && !modifiers.contains(modifier))
+            .collect::<BTreeSet<_>>()
+            .into_iter()
     }
 }
 
@@ -256,7 +254,10 @@ impl Symbol {
             seen.insert(hash, i);
         }
 
-        let list = variants.into_iter().map(|s| (s.v.0, s.v.1)).collect();
+        let list = variants
+            .into_iter()
+            .map(|s| (ModifierSet::from_raw_dotted(s.v.0), s.v.1))
+            .collect();
         Ok(Symbol::runtime(list))
     }
 }
@@ -291,14 +292,23 @@ impl crate::foundations::Repr for Symbol {
         match &self.0 {
             Repr::Single(c) => eco_format!("symbol(\"{}\")", *c),
             Repr::Complex(variants) => {
-                eco_format!("symbol{}", repr_variants(variants.iter().copied(), ""))
+                eco_format!(
+                    "symbol{}",
+                    repr_variants(variants.iter().copied(), ModifierSet::default())
+                )
             }
             Repr::Modified(arc) => {
                 let (list, modifiers) = arc.as_ref();
                 if modifiers.is_empty() {
-                    eco_format!("symbol{}", repr_variants(list.variants(), ""))
+                    eco_format!(
+                        "symbol{}",
+                        repr_variants(list.variants(), ModifierSet::default())
+                    )
                 } else {
-                    eco_format!("symbol{}", repr_variants(list.variants(), modifiers))
+                    eco_format!(
+                        "symbol{}",
+                        repr_variants(list.variants(), modifiers.as_deref())
+                    )
                 }
             }
         }
@@ -306,24 +316,24 @@ impl crate::foundations::Repr for Symbol {
 }
 
 fn repr_variants<'a>(
-    variants: impl Iterator<Item = (&'a str, char)>,
-    applied_modifiers: &str,
+    variants: impl Iterator<Item = (ModifierSet<&'a str>, char)>,
+    applied_modifiers: ModifierSet<&str>,
 ) -> String {
     crate::foundations::repr::pretty_array_like(
         &variants
-            .filter(|(variant, _)| {
+            .filter(|(modifiers, _)| {
                 // Only keep variants that can still be accessed, i.e., variants
                 // that contain all applied modifiers.
-                parts(applied_modifiers).all(|am| variant.split('.').any(|m| m == am))
+                applied_modifiers.iter().all(|am| modifiers.contains(am))
             })
-            .map(|(variant, c)| {
-                let trimmed_variant = variant
-                    .split('.')
-                    .filter(|&m| parts(applied_modifiers).all(|am| m != am));
-                if trimmed_variant.clone().all(|m| m.is_empty()) {
+            .map(|(modifiers, c)| {
+                let trimmed_modifiers =
+                    modifiers.into_iter().filter(|&m| !applied_modifiers.contains(m));
+                if trimmed_modifiers.clone().all(|m| m.is_empty()) {
                     eco_format!("\"{c}\"")
                 } else {
-                    let trimmed_modifiers = trimmed_variant.collect::<Vec<_>>().join(".");
+                    let trimmed_modifiers =
+                        trimmed_modifiers.collect::<Vec<_>>().join(".");
                     eco_format!("(\"{}\", \"{}\")", trimmed_modifiers, c)
                 }
             })
@@ -369,67 +379,22 @@ cast! {
 /// Iterator over variants.
 enum Variants<'a> {
     Single(std::option::IntoIter<char>),
-    Static(std::slice::Iter<'static, (&'static str, char)>),
-    Runtime(std::slice::Iter<'a, (EcoString, char)>),
+    Static(std::slice::Iter<'static, (ModifierSet<&'static str>, char)>),
+    Runtime(std::slice::Iter<'a, (ModifierSet<EcoString>, char)>),
 }
 
 impl<'a> Iterator for Variants<'a> {
-    type Item = (&'a str, char);
+    type Item = (ModifierSet<&'a str>, char);
 
     fn next(&mut self) -> Option<Self::Item> {
         match self {
-            Self::Single(iter) => Some(("", iter.next()?)),
+            Self::Single(iter) => Some((ModifierSet::default(), iter.next()?)),
             Self::Static(list) => list.next().copied(),
-            Self::Runtime(list) => list.next().map(|(s, c)| (s.as_str(), *c)),
+            Self::Runtime(list) => list.next().map(|(m, c)| (m.as_deref(), *c)),
         }
     }
 }
 
-/// Find the best symbol from the list.
-fn find<'a>(
-    variants: impl Iterator<Item = (&'a str, char)>,
-    modifiers: &str,
-) -> Option<char> {
-    let mut best = None;
-    let mut best_score = None;
-
-    // Find the best table entry with this name.
-    'outer: for candidate in variants {
-        for modifier in parts(modifiers) {
-            if !contained(candidate.0, modifier) {
-                continue 'outer;
-            }
-        }
-
-        let mut matching = 0;
-        let mut total = 0;
-        for modifier in parts(candidate.0) {
-            if contained(modifiers, modifier) {
-                matching += 1;
-            }
-            total += 1;
-        }
-
-        let score = (matching, Reverse(total));
-        if best_score.is_none_or(|b| score > b) {
-            best = Some(candidate.1);
-            best_score = Some(score);
-        }
-    }
-
-    best
-}
-
-/// Split a modifier list into its parts.
-fn parts(modifiers: &str) -> impl Iterator<Item = &str> {
-    modifiers.split('.').filter(|s| !s.is_empty())
-}
-
-/// Whether the modifier string contains the modifier `m`.
-fn contained(modifiers: &str, m: &str) -> bool {
-    parts(modifiers).any(|part| part == m)
-}
-
 /// A single character.
 #[elem(Repr, PlainText)]
 pub struct SymbolElem {
diff --git a/docs/Cargo.toml b/docs/Cargo.toml
index acc55175..64e26128 100644
--- a/docs/Cargo.toml
+++ b/docs/Cargo.toml
@@ -22,6 +22,7 @@ typst-utils = { workspace = true }
 typst-assets = { workspace = true, features = ["fonts"] }
 typst-dev-assets = { workspace = true }
 clap = { workspace = true, optional = true }
+codex = { workspace = true }
 ecow = { workspace = true }
 heck = { workspace = true }
 pulldown-cmark = { workspace = true }
diff --git a/docs/src/lib.rs b/docs/src/lib.rs
index d848c59b..b81f0dc6 100644
--- a/docs/src/lib.rs
+++ b/docs/src/lib.rs
@@ -712,11 +712,11 @@ fn symbols_model(resolver: &dyn Resolver, group: &GroupData) -> SymbolsModel {
     let mut list = vec![];
     for (name, binding) in group.module().scope().iter() {
         let Value::Symbol(symbol) = binding.read() else { continue };
-        let complete = |variant: &str| {
+        let complete = |variant: codex::ModifierSet<&str>| {
             if variant.is_empty() {
                 name.clone()
             } else {
-                eco_format!("{}.{}", name, variant)
+                eco_format!("{}.{}", name, variant.as_str())
             }
         };
 
diff --git a/tests/suite/symbols/symbol.typ b/tests/suite/symbols/symbol.typ
index 5bc2cafa..9fdda929 100644
--- a/tests/suite/symbols/symbol.typ
+++ b/tests/suite/symbols/symbol.typ
@@ -104,9 +104,11 @@
     ("long", "⟹"),
     ("long.bar", "⟾"),
     ("not", "⇏"),
+    ("struck", "⤃"),
     ("l", "⇔"),
     ("l.long", "⟺"),
     ("l.not", "⇎"),
+    ("l.struck", "⤄"),
   )
   ```.text,
 )