fix: various improvements

Cargo.lock

@@ -1427,6 +1427,7 @@ dependencies = [
  "tar",
  "time",
  "tokio",
+ "toml",
  "tower-http",
  "tracing",
  "tracing-subscriber",

Cargo.toml

@@ -28,3 +28,4 @@ typst-utils = "0.15.1"
 tree-house = "0.4.0"
 tree-house-bindings = { version = "0.3.2", features = ["ropey"] }
 ropey = "1.6.1"
+toml = "0.8.23"

flake.nix

@@ -46,6 +46,14 @@
               cp -r ./* $out/
             '';
           };
+        packages.themes = pkgs.stdenv.mkDerivation {
+          name = "themes";
+          phases = [ "installPhase" ];
+          installPhase = ''
+            mkdir -p $out
+            cp ${helix}/runtime/themes/*.toml $out/
+          '';
+        };
         packages.karnaugh = pkgs.rustPlatform.buildRustPackage (final: {
           pname = "karnaugh";
           version = "0.1.5";

src/main.rs

@@ -26,6 +26,7 @@ use tracing_subscriber::FmtSubscriber;
 use crate::typst::TypstContext;
 
 mod syntax;
+mod themes;
 mod typst;
 
 #[derive(Clone)]
@@ -64,10 +65,9 @@ struct AppConfig {
     #[arg(short = 'g', long, value_name = "DIR", default_value = None)]
     syntax_root: Option<PathBuf>,
 
-    /// Grammars to load, separated by commas.
-    /// Useful to avoid the memory required to load all languages.
-    #[arg(short = 'L', long = "lang", value_name = "NAME", value_delimiter = ',')]
-    languages: Vec<String>,
+    /// The directory where Helix theme `.toml` files are stored.
+    #[arg(short = 't', long, value_name = "DIR", default_value = None)]
+    themes_root: Option<PathBuf>,
 
     /// The log level to use
     #[arg(short = 'l', long, value_name = "LEVEL", default_value_t = Level::INFO)]
@@ -105,6 +105,7 @@ async fn main() {
         .init();
 
     syntax::init(&state.config);
+    themes::init(&state.config);
 
     // TODO: Replace with config option
     let mut favicon = state.config.get_full_assets_path();

src/syntax.rs

@@ -1,10 +1,10 @@
 use std::{
     borrow::Cow,
-    collections::{HashMap, HashSet},
+    collections::{BTreeSet, HashMap},
     error::Error,
     fs,
-    path::Path,
-    sync::OnceLock,
+    path::{Path, PathBuf},
+    sync::{Arc, OnceLock, RwLock},
     time::Duration,
 };
 
@@ -24,48 +24,18 @@ use typst_html::{HtmlAttr, HtmlElem, attr, tag};
 
 use crate::AppConfig;
 
-static REGISTRY: OnceLock<LanguageRegistry> = OnceLock::new();
+/// Where grammars are loaded from, captured at `init`.
+static SYNTAX_ROOT: OnceLock<PathBuf> = OnceLock::new();
 
-/// Class names used in highlighted code.
-const HIGHLIGHT_NAMES: &[&str] = &[
-    "attribute",
-    "comment",
-    "comment.block",
-    "comment.line",
-    "constant",
-    "constant.builtin",
-    "constructor",
-    "function",
-    "function.builtin",
-    "function.macro",
-    "function.method",
-    "keyword",
-    "keyword.control",
-    "keyword.operator",
-    "label",
-    "namespace",
-    "number",
-    "operator",
-    "property",
-    "punctuation",
-    "punctuation.bracket",
-    "punctuation.delimiter",
-    "special",
-    "string",
-    "string.special",
-    "tag",
-    "type",
-    "type.builtin",
-    "variable",
-    "variable.builtin",
-    "variable.parameter",
-];
+/// LanguageRegistries keyed by ScopeSet
+static REGISTRIES: OnceLock<RwLock<HashMap<BTreeSet<String>, Arc<LanguageRegistry>>>> =
+    OnceLock::new();
 
-/// Get the longest matching highlight name for a given scope.
-fn highlight_for_scope(scope: &str) -> Option<Highlight> {
+/// Get the index of the longest prefix of `scope` present in `scopes`.
+fn highlight_for_scope(scope: &str, scopes: &[String]) -> Option<Highlight> {
     let mut candidate = scope;
     loop {
-        if let Some(i) = HIGHLIGHT_NAMES.iter().position(|&n| n == candidate) {
+        if let Some(i) = scopes.iter().position(|n| n == candidate) {
             return Some(Highlight::new(i as u32));
         }
         match candidate.rfind('.') {
@@ -76,49 +46,89 @@ fn highlight_for_scope(scope: &str) -> Option<Highlight> {
 }
 
 struct LanguageRegistry {
-    by_name: HashMap<String, Language>,
-    configs: Vec<LanguageConfig>,
+    syntax_root: PathBuf,
+    /// The scope vocabulary.
+    scopes: Vec<String>,
+    /// Name to Language mapping, Option to allow storing None for failed
+    /// configuration and preventing reloading.
+    by_name: RwLock<HashMap<String, Option<Language>>>,
+    configs: RwLock<Vec<&'static LanguageConfig>>,
 }
 
 impl LanguageRegistry {
-    /// Loads grammars from `syntax_root`. When `languages` is non-empty, only
-    /// grammars with those names are loaded; otherwise every grammar found is.
-    fn new(syntax_root: &Path, languages: &[String]) -> Self {
-        let filter: Option<HashSet<&str>> =
-            (!languages.is_empty()).then(|| languages.iter().map(String::as_str).collect());
-
-        let mut by_name = HashMap::new();
-        let mut configs = Vec::new();
-
-        for entry in fs::read_dir(syntax_root).into_iter().flatten().flatten() {
-            let name = entry.file_name().to_string_lossy().into_owned();
-            if filter.as_ref().is_some_and(|f| !f.contains(name.as_str())) {
-                continue;
+    fn new(syntax_root: PathBuf, scopes: Vec<String>) -> Self {
+        Self {
+            syntax_root,
+            scopes,
+            by_name: RwLock::new(HashMap::new()),
+            configs: RwLock::new(Vec::new()),
         }
-            let so_path = entry.path().join(format!("{name}.so"));
-            if !so_path.exists() {
-                continue;
     }
 
-            match load_language(&name, &entry.path(), syntax_root) {
-                Ok(config) => {
-                    by_name.insert(name, Language::new(configs.len() as u32));
+    /// The `Language` for `name`, loading and configuring its grammar on first
+    /// request. Returns `None` if the grammar is missing or fails to load.
+    fn language(&self, name: &str) -> Option<Language> {
+        if let Some(cached) = self.by_name.read().unwrap().get(name) {
+            return *cached;
+        }
+
+        let loaded = self.load(name);
+
+        let mut by_name = self.by_name.write().unwrap();
+        if let Some(cached) = by_name.get(name) {
+            return *cached;
+        }
+        let language = loaded.map(|config| {
+            let mut configs = self.configs.write().unwrap();
             configs.push(config);
+            Language::new((configs.len() - 1) as u32)
+        });
+        by_name.insert(name.to_owned(), language);
+        language
+    }
+
+    /// Loads and configures the grammar named `name`, leaking the config to a
+    /// `'static` borrow. `None` if the `.so` is absent or parsing fails.
+    fn load(&self, name: &str) -> Option<&'static LanguageConfig> {
+        let lang_dir = self.syntax_root.join(name);
+        if !lang_dir.join(format!("{name}.so")).exists() {
+            warn!("No grammar found for language {name}");
+            return None;
+        }
+        match load_language(name, &lang_dir, &self.syntax_root) {
+            Ok(config) => {
+                config.configure(|scope| highlight_for_scope(scope, &self.scopes));
+                Some(Box::leak(Box::new(config)))
+            }
+            Err(e) => {
+                warn!("Failed to load grammar {name}: {e}");
+                None
+            }
         }
-                Err(e) => warn!("Failed to load grammar {name}: {e}"),
     }
 }
 
-        // Warn about explicitly requested grammars that were not found.
-        for lang in languages {
-            if !by_name.contains_key(lang) {
-                warn!("Requested grammar {lang} was not found in the syntax root");
-            }
+/// Fetch or build the registry configured for `scopes`.
+fn registry_for(scopes: BTreeSet<String>) -> Option<Arc<LanguageRegistry>> {
+    let registries = REGISTRIES.get()?;
+
+    // Fast path: already built.
+    if let Some(registry) = registries.read().unwrap().get(&scopes) {
+        return Some(registry.clone());
     }
 
-        info!("Loaded {} syntax highlighting grammars", configs.len());
-        Self { by_name, configs }
-    }
+    let syntax_root = SYNTAX_ROOT.get()?;
+    let ordered = scopes.iter().cloned().collect();
+    let registry = Arc::new(LanguageRegistry::new(syntax_root.clone(), ordered));
+
+    Some(
+        registries
+            .write()
+            .unwrap()
+            .entry(scopes)
+            .or_insert(registry)
+            .clone(),
+    )
 }
 
 fn load_language(
@@ -141,7 +151,6 @@ fn load_language(
         &read("injections"),
         &read("locals"),
     )?;
-    config.configure(highlight_for_scope);
 
     Ok(config)
 }
@@ -149,17 +158,17 @@ fn load_language(
 impl LanguageLoader for LanguageRegistry {
     fn language_for_marker(&self, marker: InjectionLanguageMarker<'_>) -> Option<Language> {
         match marker {
-            InjectionLanguageMarker::Name(name) => self.by_name.get(name).copied(),
+            InjectionLanguageMarker::Name(name) => self.language(name),
             InjectionLanguageMarker::Match(text) => {
                 let name: Cow<str> = text.into();
-                self.by_name.get(name.as_ref()).copied()
+                self.language(name.as_ref())
             }
             _ => None,
         }
     }
 
     fn get_config(&self, lang: Language) -> Option<&LanguageConfig> {
-        self.configs.get(lang.idx())
+        self.configs.read().unwrap().get(lang.idx()).copied()
     }
 }
 
@@ -167,7 +176,8 @@ pub fn init(config: &AppConfig) {
     match &config.syntax_root {
         Some(syntax_root) => {
             info!("Loading syntax files from {}", syntax_root.display());
-            REGISTRY.get_or_init(|| LanguageRegistry::new(syntax_root, &config.languages));
+            SYNTAX_ROOT.get_or_init(|| syntax_root.clone());
+            REGISTRIES.get_or_init(|| RwLock::new(HashMap::new()));
         }
         None => {
             warn!(
@@ -190,8 +200,15 @@ fn highlight(
     #[named]
     #[default(None)]
     lang: Option<String>,
+    #[named]
+    #[default(None)]
+    light: Option<String>,
+    #[named]
+    #[default(None)]
+    dark: Option<String>,
 ) -> Content {
-    let highlighted = highlight_code(&text, lang.as_deref()).unwrap_or(TextElem::packed(text));
+    let highlighted = highlight_code(&text, lang.as_deref(), light.as_deref(), dark.as_deref())
+        .unwrap_or_else(|| TextElem::packed(text));
 
     if block {
         HtmlElem::new(tag::pre)
@@ -207,14 +224,25 @@ fn highlight(
 }
 
 /// Highlights `source` as `lang`.
-fn highlight_code(source: &str, lang: Option<&str>) -> Option<Content> {
-    let registry = REGISTRY.get()?;
-    let language = *registry.by_name.get(lang?)?;
+fn highlight_code(
+    source: &str,
+    lang: Option<&str>,
+    light: Option<&str>,
+    dark: Option<&str>,
+) -> Option<Content> {
+    let (lang, light, dark) = (lang?, light?, dark?);
+    let Some(scopes) = crate::themes::shared_scopes(light, dark) else {
+        warn!("Unknown theme in pair (light={light}, dark={dark})");
+        return None;
+    };
+    let registry = registry_for(scopes)?;
+    let language = registry.language(lang)?;
 
     let rope = Rope::from_str(source);
     let slice = rope.slice(..);
-    let syntax = Syntax::new(slice, language, Duration::from_millis(500), registry).ok()?;
-    let mut hl = Highlighter::new(&syntax, slice, registry, 0..source.len() as u32);
+    let loader = registry.as_ref();
+    let syntax = Syntax::new(slice, language, Duration::from_secs(3), loader).ok()?;
+    let mut hl = Highlighter::new(&syntax, slice, loader, 0..source.len() as u32);
 
     let src = source.as_bytes();
     let src_len = source.len() as u32;
@@ -251,7 +279,7 @@ fn highlight_code(source: &str, lang: Option<&str>) -> Option<Content> {
             HighlightEvent::Push => classes.clone(),
         };
         classes = highlights.fold(base, |mut acc, h| {
-            let name = HIGHLIGHT_NAMES[h.idx()].replace('.', "-");
+            let name = registry.scopes[h.idx()].replace('.', "-");
             if !acc.is_empty() {
                 acc.push(' ');
             }

src/themes.rs

@@ -0,0 +1,336 @@
+use std::{
+    collections::{BTreeSet, HashMap, HashSet},
+    fs,
+    path::Path,
+    sync::OnceLock,
+};
+
+use toml::{Table, Value};
+use tracing::{info, warn};
+use typst::foundations::{Func, NativeFunc, func};
+
+use crate::AppConfig;
+
+static REGISTRY: OnceLock<ThemeRegistry> = OnceLock::new();
+
+#[derive(Debug)]
+struct ThemeRegistry {
+    themes: HashMap<String, Theme>,
+}
+
+impl ThemeRegistry {
+    fn new(theme_root: &Path) -> Self {
+        let mut helix_themes = HashMap::new();
+        for entry in fs::read_dir(theme_root).into_iter().flatten().flatten() {
+            let name = entry
+                .path()
+                .file_prefix()
+                .unwrap()
+                .to_string_lossy()
+                .into_owned();
+            if let Ok(text) = fs::read_to_string(entry.path())
+                && let Ok(toml) = toml::from_str::<Table>(&text)
+            {
+                helix_themes.insert(name, toml);
+            }
+        }
+
+        // resolve inheritance
+        while helix_themes
+            .values()
+            .any(|table| table.contains_key("inherits"))
+        {
+            let old_themes = helix_themes.clone();
+            old_themes
+                .keys()
+                .filter(|name| old_themes.get(*name).unwrap().contains_key("inherits"))
+                .for_each(|name| {
+                    let parent_name = old_themes
+                        .get(name)
+                        .unwrap()
+                        .get("inherits")
+                        .unwrap()
+                        .as_str()
+                        .unwrap();
+                    let parent = old_themes.get(parent_name).unwrap();
+                    if !parent.contains_key("inherits") {
+                        let table = helix_themes.get_mut(name).unwrap();
+                        parent.into_iter().for_each(|(k, v)| {
+                            if k == "palette" {
+                                let parent_palette = v.as_table();
+                                let child_palette = table
+                                    .entry(k.clone())
+                                    .or_insert_with(|| Value::Table(Table::new()))
+                                    .as_table_mut();
+                                if let (Some(parent_palette), Some(child_palette)) =
+                                    (parent_palette, child_palette)
+                                {
+                                    for (pk, pv) in parent_palette {
+                                        child_palette
+                                            .entry(pk.clone())
+                                            .or_insert_with(|| pv.clone());
+                                    }
+                                }
+                            } else if !table.contains_key(k) {
+                                table.insert(k.clone(), v.clone());
+                            }
+                        });
+                        table.remove("inherits");
+                    }
+                });
+        }
+
+        let themes = helix_themes
+            .into_iter()
+            .flat_map(|(name, table)| match Theme::from_toml(table) {
+                Some(theme) => Some((name, theme)),
+                None => {
+                    warn!("Theme {name} is missing an `ui.background`, skipping it");
+                    None
+                }
+            })
+            .collect();
+
+        Self { themes }
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+struct Color(String);
+
+impl Color {
+    fn from_str(color: &str) -> Option<Self> {
+        if color.starts_with('#') {
+            Some(Self(color.into()))
+        } else {
+            None
+        }
+    }
+
+    fn from_palette(color: &str, palette: &HashMap<String, Color>) -> Option<Self> {
+        if color.starts_with('#') {
+            Self::from_str(color)
+        } else if let Some(color) = palette.get(color) {
+            Some(color.clone())
+        } else {
+            // Neither a hex literal nor a palette entry; assume a named CSS
+            // colour and pass it through verbatim.
+            warn!("Unresolved colour {color}, passing it through as a literal");
+            Some(Self(color.into()))
+        }
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+struct Style {
+    fg: Option<Color>,
+    bold: bool,
+    italic: bool,
+    underlined: bool,
+}
+
+impl Style {
+    fn from_toml(style: Value, palette: &HashMap<String, Color>) -> Option<Self> {
+        let resolve = |s: &str| Color::from_palette(s, palette);
+        match style {
+            Value::String(s) => Some(Self {
+                fg: resolve(&s),
+                bold: false,
+                italic: false,
+                underlined: false,
+            }),
+            Value::Table(table) => {
+                let fg = table.get("fg").and_then(Value::as_str).and_then(resolve);
+                let modifiers: HashSet<&str> = table
+                    .get("modifiers")
+                    .and_then(Value::as_array)
+                    .into_iter()
+                    .flatten()
+                    .flat_map(Value::as_str)
+                    .collect();
+                Some(Self {
+                    fg,
+                    bold: modifiers.contains("bold"),
+                    italic: modifiers.contains("italic"),
+                    underlined: modifiers.contains("underlined"),
+                })
+            }
+            _ => None,
+        }
+    }
+}
+
+#[derive(Debug)]
+struct Theme {
+    foreground: Option<Color>,
+    background: Option<Color>,
+    styles: HashMap<String, Style>,
+}
+
+static IGNORED_SCOPES: &[&str] = &["warning", "error", "info", "hint"];
+
+impl Theme {
+    fn from_toml(theme: Table) -> Option<Self> {
+        let palette = theme
+            .get("palette")
+            .and_then(Value::as_table)
+            .cloned()
+            .map(|table| {
+                table
+                    .into_iter()
+                    .flat_map(|(k, v)| v.as_str().and_then(Color::from_str).map(|v| (k, v)))
+                    .collect::<HashMap<String, Color>>()
+            })
+            .unwrap_or_default();
+
+        let resolve = |c: &str| Color::from_palette(c, &palette);
+
+        let ui_background = theme.get("ui.background").and_then(Value::as_table)?;
+        let foreground = ui_background
+            .get("fg")
+            .and_then(Value::as_str)
+            .and_then(resolve);
+        let background = ui_background
+            .get("bg")
+            .and_then(Value::as_str)
+            .and_then(resolve);
+
+        let styles = theme
+            .into_iter()
+            .filter(|(k, _)| {
+                !k.starts_with("ui.")
+                    && !k.starts_with("diagnostic.")
+                    && !IGNORED_SCOPES.contains(&k.as_str())
+            })
+            .flat_map(|(k, v)| Style::from_toml(v, &palette).map(|v| (k, v)))
+            .collect();
+
+        Some(Self {
+            foreground,
+            background,
+            styles,
+        })
+    }
+}
+
+pub fn init(config: &AppConfig) {
+    match &config.themes_root {
+        Some(themes_root) => {
+            info!("Loading themes from {}", themes_root.display());
+            REGISTRY.get_or_init(|| ThemeRegistry::new(themes_root));
+        }
+        None => {
+            warn!("No themes root has been configured, theme-based colouring will not be possible.")
+        }
+    }
+}
+
+/// The intersection of scopes between two themes.
+pub fn shared_scopes(light: &str, dark: &str) -> Option<BTreeSet<String>> {
+    let registry = REGISTRY.get()?;
+    let light = registry.themes.get(light)?;
+    let dark = registry.themes.get(dark)?;
+    Some(
+        light
+            .styles
+            .keys()
+            .filter(|k| dark.styles.contains_key(*k))
+            .cloned()
+            .collect(),
+    )
+}
+
+pub fn theme_css_func() -> Func {
+    theme_css::func()
+}
+
+fn scope_to_selector(scope: &str) -> String {
+    format!(".hl-{}", scope.replace('.', "-"))
+}
+
+fn style_declarations(style: &Style) -> String {
+    let mut parts = Vec::new();
+    if let Some(Color(hex)) = &style.fg {
+        parts.push(format!("color:{hex}"));
+    }
+    parts.push(format!(
+        "font-weight:{}",
+        if style.bold { "bold" } else { "normal" }
+    ));
+    parts.push(format!(
+        "font-style:{}",
+        if style.italic { "italic" } else { "normal" }
+    ));
+    parts.push(format!(
+        "text-decoration:{}",
+        if style.underlined {
+            "underline"
+        } else {
+            "none"
+        }
+    ));
+    parts.join(";")
+}
+
+fn generate_css(scopes: &[&str], theme: &Theme) -> String {
+    let mut rules = Vec::new();
+
+    // The theme's editor foreground/background colour the code block itself.
+    let mut block_style = Vec::new();
+    if let Some(Color(hex)) = &theme.foreground {
+        block_style.push(format!("color:{hex}"));
+    }
+    if let Some(Color(hex)) = &theme.background {
+        block_style.push(format!("background:{hex}"));
+    }
+    if !block_style.is_empty() {
+        rules.push(format!(".code-block{{{}}}", block_style.join(";")));
+    }
+
+    let mut groups: HashMap<&Style, Vec<&str>> = HashMap::new();
+    for &scope in scopes {
+        if let Some(style) = theme.styles.get(scope) {
+            groups.entry(style).or_default().push(scope);
+        }
+    }
+
+    let mut group_rules: Vec<String> = groups
+        .iter()
+        .map(|(style, scopes)| {
+            let selectors = scopes
+                .iter()
+                .map(|s| scope_to_selector(s))
+                .collect::<Vec<_>>()
+                .join(",");
+            format!("{}{{{}}}", selectors, style_declarations(style))
+        })
+        .collect();
+    // Sort so the emitted stylesheet is deterministic across runs.
+    group_rules.sort();
+    rules.extend(group_rules);
+
+    rules.join("\n")
+}
+
+#[func]
+fn theme_css(light: String, dark: String) -> String {
+    let Some(registry) = REGISTRY.get() else {
+        return String::new();
+    };
+    let Some(light_theme) = registry.themes.get(&light) else {
+        warn!("Unknown light theme: {light}");
+        return String::new();
+    };
+    let Some(dark_theme) = registry.themes.get(&dark) else {
+        warn!("Unknown dark theme: {dark}");
+        return String::new();
+    };
+
+    let shared = shared_scopes(&light, &dark).unwrap_or_default();
+    let shared: Vec<&str> = shared.iter().map(String::as_str).collect();
+
+    let light_css = generate_css(&shared, light_theme);
+    let dark_css = generate_css(&shared, dark_theme);
+
+    format!("{light_css}\n@media (prefers-color-scheme:dark){{{dark_css}}}")
+}

src/typst.rs

@@ -30,6 +30,12 @@ impl TypstContext {
                 Value::Func(crate::syntax::highlight_func()),
             );
         }
+        if config.themes_root.is_some() {
+            inputs.insert(
+                "theme_css".into(),
+                Value::Func(crate::themes::theme_css_func()),
+            );
+        }
         let library = Library::builder()
             .with_inputs(inputs)
             .with_features([typst::Feature::Html].into_iter().collect())