feat(core): make checkbox lexing honor a configurable listsyms palette

Add a ListSyms type (vimwiki's g:vimwiki_listsyms) and thread it through
the lexer so checkbox glyphs are recognized from the configured palette
rather than the hardcoded ' .oOX'. Glyphs map to the existing five-bucket
CheckboxState, so the HTML renderer and AST consumers are unchanged.
VimwikiSyntax::parse_with_listsyms opts into a custom palette; the trait
parse() keeps the default.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

crates/nuwiki-core/src/lib.rs

@@ -5,5 +5,6 @@
 
 pub mod ast;
 pub mod date;
+pub mod listsyms;
 pub mod render;
 pub mod syntax;

crates/nuwiki-core/src/listsyms.rs

@@ -0,0 +1,170 @@
+//! Configurable checkbox progress palette (vimwiki's `g:vimwiki_listsyms`).
+//!
+//! A palette is a progression of glyphs from "empty" (index 0) to "done"
+//! (the last index). The default `" .oOX"` is the stock vimwiki palette.
+//! A separate `rejected` glyph (`-`) is fixed by convention.
+//!
+//! The palette maps glyphs to a normalised five-bucket [`CheckboxState`] for
+//! the AST (so the HTML renderer keeps its fixed `done0..done4` classes) while
+//! still letting the LSP cycle/toggle commands operate on the real glyphs.
+
+use crate::ast::CheckboxState;
+
+const REJECTED: char = '-';
+
+/// An ordered checkbox progress palette. Always holds at least two glyphs.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ListSyms {
+    progression: Vec<char>,
+}
+
+impl Default for ListSyms {
+    fn default() -> Self {
+        Self {
+            progression: " .oOX".chars().collect(),
+        }
+    }
+}
+
+impl ListSyms {
+    /// Build a palette from a glyph string. Falls back to the default when
+    /// fewer than two glyphs are supplied (a one-symbol palette can't express
+    /// both "empty" and "done").
+    pub fn new(s: &str) -> Self {
+        let progression: Vec<char> = s.chars().collect();
+        if progression.len() < 2 {
+            Self::default()
+        } else {
+            Self { progression }
+        }
+    }
+
+    pub fn len(&self) -> usize {
+        self.progression.len()
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.progression.is_empty()
+    }
+
+    pub fn empty_glyph(&self) -> char {
+        self.progression[0]
+    }
+
+    pub fn done_glyph(&self) -> char {
+        *self.progression.last().unwrap()
+    }
+
+    pub fn rejected_glyph(&self) -> char {
+        REJECTED
+    }
+
+    pub fn glyph_at(&self, index: usize) -> char {
+        self.progression[index.min(self.progression.len() - 1)]
+    }
+
+    pub fn index_of(&self, c: char) -> Option<usize> {
+        self.progression.iter().position(|&g| g == c)
+    }
+
+    /// Progress rate (0..=100) for a glyph index.
+    pub fn rate(&self, index: usize) -> i32 {
+        let last = (self.progression.len() - 1) as i32;
+        (index as i32) * 100 / last
+    }
+
+    /// Normalised five-bucket [`CheckboxState`] for a glyph, or `None` when the
+    /// glyph is not part of this palette (and is not the rejected marker).
+    pub fn state_of(&self, c: char) -> Option<CheckboxState> {
+        if c == REJECTED {
+            return Some(CheckboxState::Rejected);
+        }
+        let index = self.index_of(c)?;
+        let rate = self.rate(index);
+        // Round the rate to the nearest 25% bucket.
+        Some(match (rate + 12) / 25 {
+            0 => CheckboxState::Empty,
+            1 => CheckboxState::Quarter,
+            2 => CheckboxState::Half,
+            3 => CheckboxState::ThreeQuarters,
+            _ => CheckboxState::Done,
+        })
+    }
+
+    /// Snap a rate (`-1` = rejected, else 0..=100) to a glyph. Intermediate
+    /// rates distribute over the `n - 2` non-terminal glyphs using vimwiki's
+    /// `ceil` rule, matching the stock five-symbol behaviour.
+    pub fn glyph_for_rate(&self, rate: i32) -> char {
+        if rate < 0 {
+            return REJECTED;
+        }
+        if rate <= 0 {
+            return self.empty_glyph();
+        }
+        if rate >= 100 {
+            return self.done_glyph();
+        }
+        let mids = (self.progression.len().saturating_sub(2)).max(1) as f64;
+        let index = ((rate as f64) / 100.0 * mids).ceil() as usize;
+        self.glyph_at(index)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn default_palette_matches_stock_vimwiki() {
+        let s = ListSyms::default();
+        assert_eq!(s.len(), 5);
+        assert_eq!(s.empty_glyph(), ' ');
+        assert_eq!(s.done_glyph(), 'X');
+        assert_eq!(s.rejected_glyph(), '-');
+        assert_eq!(s.rate(0), 0);
+        assert_eq!(s.rate(1), 25);
+        assert_eq!(s.rate(2), 50);
+        assert_eq!(s.rate(3), 75);
+        assert_eq!(s.rate(4), 100);
+    }
+
+    #[test]
+    fn default_state_of_round_trips_each_glyph() {
+        let s = ListSyms::default();
+        assert_eq!(s.state_of(' '), Some(CheckboxState::Empty));
+        assert_eq!(s.state_of('.'), Some(CheckboxState::Quarter));
+        assert_eq!(s.state_of('o'), Some(CheckboxState::Half));
+        assert_eq!(s.state_of('O'), Some(CheckboxState::ThreeQuarters));
+        assert_eq!(s.state_of('X'), Some(CheckboxState::Done));
+        assert_eq!(s.state_of('-'), Some(CheckboxState::Rejected));
+        assert_eq!(s.state_of('?'), None);
+    }
+
+    #[test]
+    fn default_glyph_for_rate_matches_legacy_buckets() {
+        let s = ListSyms::default();
+        assert_eq!(s.glyph_for_rate(-1), '-');
+        assert_eq!(s.glyph_for_rate(0), ' ');
+        assert_eq!(s.glyph_for_rate(25), '.');
+        assert_eq!(s.glyph_for_rate(50), 'o');
+        assert_eq!(s.glyph_for_rate(75), 'O');
+        assert_eq!(s.glyph_for_rate(100), 'X');
+    }
+
+    #[test]
+    fn short_input_falls_back_to_default() {
+        assert_eq!(ListSyms::new(""), ListSyms::default());
+        assert_eq!(ListSyms::new("x"), ListSyms::default());
+    }
+
+    #[test]
+    fn custom_palette_buckets_to_nearest_state() {
+        // 3 glyphs → rates 0 / 50 / 100.
+        let s = ListSyms::new(" x✓");
+        assert_eq!(s.state_of(' '), Some(CheckboxState::Empty));
+        assert_eq!(s.state_of('x'), Some(CheckboxState::Half));
+        assert_eq!(s.state_of('✓'), Some(CheckboxState::Done));
+        assert_eq!(s.glyph_for_rate(50), 'x');
+        assert_eq!(s.glyph_for_rate(100), '✓');
+    }
+}

crates/nuwiki-core/src/syntax/vimwiki/lexer.rs

@@ -21,6 +21,7 @@
 use std::collections::HashMap;
 
 use crate::ast::{CheckboxState, Keyword, ListSymbol, Position, Span};
+use crate::listsyms::ListSyms;
 use crate::syntax::{Lexer, TokenStream};
 
 /// A single vimwiki token. The `kind` carries the variant, `span` carries
@@ -134,11 +135,20 @@ pub enum VimwikiTokenKind {
 }
 
 #[derive(Debug, Default, Clone)]
-pub struct VimwikiLexer;
+pub struct VimwikiLexer {
+    listsyms: ListSyms,
+}
 
 impl VimwikiLexer {
     pub fn new() -> Self {
-        Self
+        Self::default()
+    }
+
+    /// Recognise checkbox glyphs from a custom palette
+    /// (vimwiki's `g:vimwiki_listsyms`) instead of the default `" .oOX"`.
+    pub fn with_listsyms(mut self, listsyms: ListSyms) -> Self {
+        self.listsyms = listsyms;
+        self
     }
 }
 
@@ -146,7 +156,7 @@ impl Lexer for VimwikiLexer {
     type Token = VimwikiToken;
 
     fn lex(&self, text: &str) -> TokenStream<VimwikiToken> {
-        let mut state = LexState::new(text);
+        let mut state = LexState::new(text, &self.listsyms);
         state.run();
         TokenStream::from_vec(state.tokens)
     }
@@ -170,16 +180,18 @@ struct LexState<'src> {
     line: u32,
     tokens: Vec<VimwikiToken>,
     mode: BlockMode,
+    listsyms: &'src ListSyms,
 }
 
 impl<'src> LexState<'src> {
-    fn new(src: &'src str) -> Self {
+    fn new(src: &'src str, listsyms: &'src ListSyms) -> Self {
         Self {
             src,
             line_start_offset: 0,
             line: 0,
             tokens: Vec::new(),
             mode: BlockMode::Normal,
+            listsyms,
         }
     }
 
@@ -704,7 +716,7 @@ impl<'src> LexState<'src> {
 
         // Optional checkbox `[ ]` / `[.]` / `[o]` / `[O]` / `[X]` / `[-]`,
         // followed by a space.
-        if let Some((cb, cb_len)) = checkbox_at(after) {
+        if let Some((cb, cb_len)) = checkbox_at(after, self.listsyms) {
             let cb_start = indent + cursor as u32;
             let cb_end = cb_start + cb_len as u32;
             self.emit(VimwikiTokenKind::Checkbox(cb), cb_start, cb_end);
@@ -1230,19 +1242,18 @@ fn parse_sep_alignment(cell: &str) -> crate::ast::TableAlign {
     }
 }
 
-fn checkbox_at(s: &str) -> Option<(CheckboxState, usize)> {
+/// Recognise a `[g]` checkbox at the start of `s`, where `g` is a single
-    let b = s.as_bytes();
+/// glyph drawn from `listsyms` (or the rejected marker). Returns the bucketed
-    if b.len() < 3 || b[0] != b'[' || b[2] != b']' {
+/// state and the byte length consumed (including the brackets).
+fn checkbox_at(s: &str, listsyms: &ListSyms) -> Option<(CheckboxState, usize)> {
+    let rest = s.strip_prefix('[')?;
+    let mut chars = rest.char_indices();
+    let (_, glyph) = chars.next()?;
+    let (close_off, ']') = chars.next()? else {
         return None;
-    }
-    let cb = match b[1] {
-        b' ' => CheckboxState::Empty,
-        b'.' => CheckboxState::Quarter,
-        b'o' => CheckboxState::Half,
-        b'O' => CheckboxState::ThreeQuarters,
-        b'X' => CheckboxState::Done,
-        b'-' => CheckboxState::Rejected,
-        _ => return None,
     };
-    Some((cb, 3))
+    let state = listsyms.state_of(glyph)?;
+    // `[` + glyph + `]`: 1 + glyph.len_utf8() + 1, i.e. the close bracket
+    // offset within `rest` plus one for the opening bracket and one for `]`.
+    Some((state, 1 + close_off + 1))
 }

crates/nuwiki-core/src/syntax/vimwiki/mod.rs

@@ -7,6 +7,7 @@ pub use lexer::{VimwikiLexer, VimwikiToken, VimwikiTokenKind};
 pub use parser::VimwikiParser;
 
 use crate::ast::DocumentNode;
+use crate::listsyms::ListSyms;
 use crate::syntax::{Lexer, Parser, SyntaxPlugin};
 
 /// SyntaxPlugin facade: chains `VimwikiLexer` + `VimwikiParser` so the
@@ -18,6 +19,15 @@ impl VimwikiSyntax {
     pub fn new() -> Self {
         Self
     }
+
+    /// Parse with a custom checkbox palette (vimwiki's `g:vimwiki_listsyms`).
+    /// The trait-level [`SyntaxPlugin::parse`] uses [`ListSyms::default`].
+    pub fn parse_with_listsyms(&self, text: &str, listsyms: &ListSyms) -> DocumentNode {
+        let tokens = VimwikiLexer::new()
+            .with_listsyms(listsyms.clone())
+            .lex(text);
+        VimwikiParser::new().parse(tokens)
+    }
 }
 
 impl SyntaxPlugin for VimwikiSyntax {

crates/nuwiki-core/tests/lexer.rs

@@ -309,6 +309,45 @@ fn checkbox_states() {
     }
 }
 
+#[test]
+fn checkbox_states_honor_custom_palette() {
+    use nuwiki_core::listsyms::ListSyms;
+
+    let lex_with = |text: &str, syms: ListSyms| -> Vec<VimwikiTokenKind> {
+        VimwikiLexer::new()
+            .with_listsyms(syms)
+            .lex(text)
+            .into_iter()
+            .map(|t| t.kind)
+            .collect()
+    };
+
+    // 3-glyph palette ` x✓`: empty / half / done, plus the fixed rejected `-`.
+    let syms = ListSyms::new(" x✓");
+    let cases: &[(&str, CheckboxState)] = &[
+        ("- [ ] a\n", CheckboxState::Empty),
+        ("- [x] a\n", CheckboxState::Half),
+        ("- [✓] a\n", CheckboxState::Done),
+        ("- [-] a\n", CheckboxState::Rejected),
+    ];
+    for (line, expected) in cases {
+        let lexed = lex_with(line, syms.clone());
+        assert!(
+            matches!(lexed[1], Checkbox(s) if s == *expected),
+            "input {line:?} expected {expected:?}, got {:?}",
+            lexed[1]
+        );
+    }
+
+    // Default-palette glyphs are not checkboxes under this palette: `[o]`
+    // is plain inline text, so no Checkbox token is emitted.
+    let lexed = lex_with("- [o] a\n", syms);
+    assert!(
+        !lexed.iter().any(|k| matches!(k, Checkbox(_))),
+        "expected no checkbox token, got {lexed:?}"
+    );
+}
+
 // ===== Blockquotes =====
 
 #[test]