feat: initial nuwiki Rust workspace (core, lsp, ls)

crates/nuwiki-core/src/ast/block.rs

@@ -0,0 +1,192 @@
+//! Block-level AST nodes and their supporting types.
+
+use super::inline::InlineNode;
+use super::span::Span;
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum BlockNode {
+    Heading(HeadingNode),
+    Paragraph(ParagraphNode),
+    HorizontalRule(HorizontalRuleNode),
+    Blockquote(BlockquoteNode),
+    Preformatted(PreformattedNode),
+    MathBlock(MathBlockNode),
+    List(ListNode),
+    DefinitionList(DefinitionListNode),
+    Table(TableNode),
+    Comment(CommentNode),
+    /// Vimwiki tag line — `:tag1:tag2:`. The placement
+    /// rule (file / heading / standalone) is captured in `TagScope` so
+    /// LSP handlers and renderers can treat the three differently.
+    Tag(TagNode),
+    Error(ErrorNode),
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct HeadingNode {
+    pub span: Span,
+    /// 1..=6 — invariant enforced by the parser, not the type.
+    pub level: u8,
+    pub children: Vec<InlineNode>,
+    pub centered: bool,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ParagraphNode {
+    pub span: Span,
+    pub children: Vec<InlineNode>,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct HorizontalRuleNode {
+    pub span: Span,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct BlockquoteNode {
+    pub span: Span,
+    pub children: Vec<BlockNode>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct PreformattedNode {
+    pub span: Span,
+    pub content: String,
+    pub language: Option<String>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct MathBlockNode {
+    pub span: Span,
+    pub content: String,
+    pub environment: Option<String>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ListNode {
+    pub span: Span,
+    pub ordered: bool,
+    pub symbol: ListSymbol,
+    pub items: Vec<ListItemNode>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ListItemNode {
+    pub span: Span,
+    pub symbol: ListSymbol,
+    pub level: usize,
+    pub checkbox: Option<CheckboxState>,
+    pub children: Vec<InlineNode>,
+    pub sublist: Option<ListNode>,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum ListSymbol {
+    Dash,
+    Star,
+    Hash,
+    Numeric,
+    NumericParen,
+    AlphaParen,
+    AlphaUpperParen,
+    RomanParen,
+    RomanUpperParen,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum CheckboxState {
+    Empty,
+    Quarter,
+    Half,
+    ThreeQuarters,
+    Done,
+    Rejected,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct DefinitionListNode {
+    pub span: Span,
+    pub items: Vec<DefinitionItemNode>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct DefinitionItemNode {
+    pub span: Span,
+    pub term: Option<Vec<InlineNode>>,
+    pub definitions: Vec<Vec<InlineNode>>,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum TableAlign {
+    Default,
+    Left,
+    Right,
+    Center,
+}
+
+impl Default for TableAlign {
+    fn default() -> Self {
+        Self::Default
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct TableNode {
+    pub span: Span,
+    pub rows: Vec<TableRowNode>,
+    pub has_header: bool,
+    /// One entry per column, taken from a Markdown-style header
+    /// separator row (`|:--|--:|:--:|`). Empty when no alignment was
+    /// specified; cells past the end inherit `Default`.
+    pub alignments: Vec<TableAlign>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct TableRowNode {
+    pub span: Span,
+    pub cells: Vec<TableCellNode>,
+    pub is_header: bool,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct TableCellNode {
+    pub span: Span,
+    pub children: Vec<InlineNode>,
+    pub col_span: bool,
+    pub row_span: bool,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct CommentNode {
+    pub span: Span,
+    pub content: String,
+}
+
+/// Where a `TagNode` falls on the page, matching vimwiki's placement rules.
+///
+/// - `File`        — line 0 or 1 of the document; tags the entire page.
+/// - `Heading(i)`  — within two lines below the i-th `HeadingNode` in the
+///                   document's `children`; tags that heading.
+/// - `Standalone`  — anywhere else; the tag is its own anchor on the source
+///                   line.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum TagScope {
+    File,
+    Heading(usize),
+    Standalone,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct TagNode {
+    pub span: Span,
+    pub tags: Vec<String>,
+    pub scope: TagScope,
+}
+
+/// Resilient parse-failure node — emitted instead of aborting the document.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ErrorNode {
+    pub span: Span,
+    pub raw: String,
+    pub message: String,
+}

crates/nuwiki-core/src/ast/inline.rs

@@ -0,0 +1,188 @@
+//! Inline AST nodes.
+//!
+//! The link-related variants (`WikiLink`, `ExternalLink`,
+//! `Transclusion`, `RawUrl`) wrap structs defined in `super::link`.
+
+use super::link::{ExternalLinkNode, RawUrlNode, TransclusionNode, WikiLinkNode};
+use super::span::Span;
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum Keyword {
+    Todo,
+    Done,
+    Started,
+    Fixme,
+    Fixed,
+    Xxx,
+    Stopped,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum InlineNode {
+    Text(TextNode),
+    Bold(BoldNode),
+    Italic(ItalicNode),
+    BoldItalic(BoldItalicNode),
+    Strikethrough(StrikethroughNode),
+    Code(CodeNode),
+    Superscript(SuperscriptNode),
+    Subscript(SubscriptNode),
+    MathInline(MathInlineNode),
+    Keyword(KeywordNode),
+    Color(ColorNode),
+    WikiLink(WikiLinkNode),
+    ExternalLink(ExternalLinkNode),
+    Transclusion(TransclusionNode),
+    RawUrl(RawUrlNode),
+    /// A soft line break joining two content lines within a paragraph or
+    /// list item. Consumers treat it as whitespace; the HTML renderer emits
+    /// a space (vimwiki `*_ignore_newline = 1`, the default) or `<br />`
+    /// (`= 0`).
+    SoftBreak(SoftBreakNode),
+}
+
+impl InlineNode {
+    /// The source [`Span`] this node covers. Every variant wraps a node with
+    /// its own `span` field; this is the one place that maps variant → span.
+    pub fn span(&self) -> Span {
+        match self {
+            InlineNode::Text(n) => n.span,
+            InlineNode::Bold(n) => n.span,
+            InlineNode::Italic(n) => n.span,
+            InlineNode::BoldItalic(n) => n.span,
+            InlineNode::Strikethrough(n) => n.span,
+            InlineNode::Code(n) => n.span,
+            InlineNode::Superscript(n) => n.span,
+            InlineNode::Subscript(n) => n.span,
+            InlineNode::MathInline(n) => n.span,
+            InlineNode::Keyword(n) => n.span,
+            InlineNode::Color(n) => n.span,
+            InlineNode::WikiLink(n) => n.span,
+            InlineNode::ExternalLink(n) => n.span,
+            InlineNode::Transclusion(n) => n.span,
+            InlineNode::RawUrl(n) => n.span,
+            InlineNode::SoftBreak(n) => n.span,
+        }
+    }
+}
+
+/// Concatenate the plain-text content of a run of inlines, descending into
+/// formatting containers. Links/external links use their description, falling
+/// back to the target path / URL. This is the canonical heading/anchor text —
+/// the HTML renderer uses it for heading `id`s and the LSP for anchor matching,
+/// so the two never drift.
+pub fn inline_text(inlines: &[InlineNode]) -> String {
+    let mut out = String::new();
+    push_inline_text(inlines, &mut out);
+    out.trim().to_string()
+}
+
+fn push_inline_text(inlines: &[InlineNode], out: &mut String) {
+    for n in inlines {
+        match n {
+            InlineNode::Text(t) => out.push_str(&t.content),
+            InlineNode::Bold(b) => push_inline_text(&b.children, out),
+            InlineNode::Italic(i) => push_inline_text(&i.children, out),
+            InlineNode::BoldItalic(bi) => push_inline_text(&bi.children, out),
+            InlineNode::Strikethrough(s) => push_inline_text(&s.children, out),
+            InlineNode::Code(c) => out.push_str(&c.content),
+            InlineNode::Superscript(s) => push_inline_text(&s.children, out),
+            InlineNode::Subscript(s) => push_inline_text(&s.children, out),
+            InlineNode::Color(c) => push_inline_text(&c.children, out),
+            InlineNode::WikiLink(w) => match &w.description {
+                Some(d) => push_inline_text(d, out),
+                None => {
+                    if let Some(p) = &w.target.path {
+                        out.push_str(p);
+                    }
+                }
+            },
+            InlineNode::ExternalLink(e) => match &e.description {
+                Some(d) => push_inline_text(d, out),
+                None => out.push_str(&e.url),
+            },
+            _ => {}
+        }
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct TextNode {
+    pub span: Span,
+    pub content: String,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct SoftBreakNode {
+    pub span: Span,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct BoldNode {
+    pub span: Span,
+    pub children: Vec<InlineNode>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ItalicNode {
+    pub span: Span,
+    pub children: Vec<InlineNode>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct BoldItalicNode {
+    pub span: Span,
+    pub children: Vec<InlineNode>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct StrikethroughNode {
+    pub span: Span,
+    pub children: Vec<InlineNode>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct CodeNode {
+    pub span: Span,
+    pub content: String,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct SuperscriptNode {
+    pub span: Span,
+    pub children: Vec<InlineNode>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct SubscriptNode {
+    pub span: Span,
+    pub children: Vec<InlineNode>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct MathInlineNode {
+    pub span: Span,
+    pub content: String,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct KeywordNode {
+    pub span: Span,
+    pub keyword: Keyword,
+}
+
+/// A named colour span (`color` is a `color_dic` key, `children` the wrapped
+/// inline content).
+///
+/// This is an **extension point**, not a node the vimwiki parser emits: in
+/// the editor, colour comes from the `:Colorize` family writing literal
+/// `<span style="color:…">` markup, which round-trips through export via
+/// `valid_html_tags`. `ColorNode` exists so an alternate syntax (or a future
+/// `[color]` markup) can feed the renderer's `color_dic` / `color_tag_template`
+/// path directly; see `HtmlRenderer::render_color`.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ColorNode {
+    pub span: Span,
+    pub color: String,
+    pub children: Vec<InlineNode>,
+}

crates/nuwiki-core/src/ast/link.rs

@@ -0,0 +1,66 @@
+//! Link-related AST nodes and supporting types.
+//!
+//! All nodes here are inline (`InlineNode` variants);
+//! they live in their own module because the link model has enough surface
+//! area (kinds, anchors, interwiki indirection) to warrant separation.
+
+use std::collections::HashMap;
+
+use super::inline::InlineNode;
+use super::span::Span;
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum LinkKind {
+    Wiki,
+    Interwiki,
+    Diary,
+    File,
+    Local,
+    Raw,
+    AnchorOnly,
+}
+
+#[derive(Debug, Clone, Default, PartialEq, Eq, Hash)]
+pub struct LinkTarget {
+    pub kind: LinkKind,
+    pub path: Option<String>,
+    pub wiki_index: Option<usize>,
+    pub wiki_name: Option<String>,
+    pub anchor: Option<String>,
+    pub is_absolute: bool,
+    pub is_directory: bool,
+}
+
+impl Default for LinkKind {
+    fn default() -> Self {
+        Self::Wiki
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct WikiLinkNode {
+    pub span: Span,
+    pub target: LinkTarget,
+    pub description: Option<Vec<InlineNode>>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ExternalLinkNode {
+    pub span: Span,
+    pub url: String,
+    pub description: Option<Vec<InlineNode>>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct TransclusionNode {
+    pub span: Span,
+    pub url: String,
+    pub alt: Option<String>,
+    pub attrs: HashMap<String, String>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct RawUrlNode {
+    pub span: Span,
+    pub url: String,
+}

crates/nuwiki-core/src/ast/mod.rs

@@ -0,0 +1,54 @@
+//! AST types for nuwiki documents.
+//!
+//! Types here are syntax-agnostic — both vimwiki and (eventually) markdown
+//! produce the same node shapes.
+
+pub mod block;
+pub mod inline;
+pub mod link;
+pub mod span;
+pub mod visit;
+
+pub use block::{
+    BlockNode, BlockquoteNode, CheckboxState, CommentNode, DefinitionItemNode, DefinitionListNode,
+    ErrorNode, HeadingNode, HorizontalRuleNode, ListItemNode, ListNode, ListSymbol, MathBlockNode,
+    ParagraphNode, PreformattedNode, TableAlign, TableCellNode, TableNode, TableRowNode, TagNode,
+    TagScope,
+};
+pub use inline::{
+    inline_text, BoldItalicNode, BoldNode, CodeNode, ColorNode, InlineNode, ItalicNode, Keyword,
+    KeywordNode, MathInlineNode, SoftBreakNode, StrikethroughNode, SubscriptNode, SuperscriptNode,
+    TextNode,
+};
+pub use link::{
+    ExternalLinkNode, LinkKind, LinkTarget, RawUrlNode, TransclusionNode, WikiLinkNode,
+};
+pub use span::{Position, Span};
+pub use visit::{
+    walk_block, walk_definition_item, walk_document, walk_inline, walk_list, walk_list_item,
+    walk_table, walk_table_row, Visitor,
+};
+
+/// Root of an AST.
+#[derive(Debug, Clone, Default, PartialEq, Eq)]
+pub struct DocumentNode {
+    pub span: Span,
+    pub children: Vec<BlockNode>,
+    pub metadata: PageMetadata,
+}
+
+/// Document-level placeholders parsed from `%title` / `%nohtml` / `%template`
+/// / `%date` directives, plus the file-level tag accumulator.
+///
+/// New fields are added at the bottom; consumers should construct with
+/// `..Default::default()` to stay forward-compatible.
+#[derive(Debug, Clone, Default, PartialEq, Eq)]
+pub struct PageMetadata {
+    pub title: Option<String>,
+    pub nohtml: bool,
+    pub template: Option<String>,
+    pub date: Option<String>,
+    /// File-scope tags parsed by the tag lexer. Convenience
+    /// projection of the `BlockNode::Tag` nodes whose scope is `File`.
+    pub tags: Vec<String>,
+}

crates/nuwiki-core/src/ast/span.rs

@@ -0,0 +1,33 @@
+//! Source positions and spans carried on every AST node.
+//!
+//! Positions are 0-indexed; `column` is a byte offset
+//! within a line; `offset` is a byte offset from the start of the document.
+
+#[derive(Debug, Clone, Copy, Default, PartialEq, Eq, Hash, PartialOrd, Ord)]
+pub struct Position {
+    pub line: u32,
+    pub column: u32,
+    pub offset: usize,
+}
+
+impl Position {
+    pub const fn new(line: u32, column: u32, offset: usize) -> Self {
+        Self {
+            line,
+            column,
+            offset,
+        }
+    }
+}
+
+#[derive(Debug, Clone, Copy, Default, PartialEq, Eq, Hash)]
+pub struct Span {
+    pub start: Position,
+    pub end: Position,
+}
+
+impl Span {
+    pub const fn new(start: Position, end: Position) -> Self {
+        Self { start, end }
+    }
+}

crates/nuwiki-core/src/ast/visit.rs

@@ -0,0 +1,230 @@
+//! Open-recursion AST visitor.
+//!
+//! Each `visit_*` method has a default body that calls the corresponding
+//! `walk_*` free function, which in turn dispatches back into the trait.
+//! Override any method to short-circuit or augment traversal — call the
+//! matching `walk_*` from your override to keep descending.
+//!
+//! Pattern follows `rustc`'s and `syn`'s visitors.
+//!
+//! Read-only for now. A `VisitorMut` flavor can be added when transformations
+//! become relevant.
+
+use super::block::{
+    BlockNode, BlockquoteNode, CommentNode, DefinitionItemNode, DefinitionListNode, ErrorNode,
+    HeadingNode, HorizontalRuleNode, ListItemNode, ListNode, MathBlockNode, ParagraphNode,
+    PreformattedNode, TableCellNode, TableNode, TableRowNode, TagNode,
+};
+use super::inline::{
+    BoldItalicNode, BoldNode, CodeNode, ColorNode, InlineNode, ItalicNode, KeywordNode,
+    MathInlineNode, StrikethroughNode, SubscriptNode, SuperscriptNode, TextNode,
+};
+use super::link::{ExternalLinkNode, RawUrlNode, TransclusionNode, WikiLinkNode};
+use super::DocumentNode;
+
+pub trait Visitor {
+    // Top level
+    fn visit_document(&mut self, node: &DocumentNode) {
+        walk_document(self, node);
+    }
+
+    // Block dispatch
+    fn visit_block(&mut self, node: &BlockNode) {
+        walk_block(self, node);
+    }
+
+    // Inline dispatch
+    fn visit_inline(&mut self, node: &InlineNode) {
+        walk_inline(self, node);
+    }
+
+    // Block leaves and recursive block nodes
+    fn visit_heading(&mut self, node: &HeadingNode) {
+        for child in &node.children {
+            self.visit_inline(child);
+        }
+    }
+    fn visit_paragraph(&mut self, node: &ParagraphNode) {
+        for child in &node.children {
+            self.visit_inline(child);
+        }
+    }
+    fn visit_horizontal_rule(&mut self, _node: &HorizontalRuleNode) {}
+    fn visit_blockquote(&mut self, node: &BlockquoteNode) {
+        for child in &node.children {
+            self.visit_block(child);
+        }
+    }
+    fn visit_preformatted(&mut self, _node: &PreformattedNode) {}
+    fn visit_math_block(&mut self, _node: &MathBlockNode) {}
+    fn visit_list(&mut self, node: &ListNode) {
+        walk_list(self, node);
+    }
+    fn visit_list_item(&mut self, node: &ListItemNode) {
+        walk_list_item(self, node);
+    }
+    fn visit_definition_list(&mut self, node: &DefinitionListNode) {
+        for item in &node.items {
+            self.visit_definition_item(item);
+        }
+    }
+    fn visit_definition_item(&mut self, node: &DefinitionItemNode) {
+        walk_definition_item(self, node);
+    }
+    fn visit_table(&mut self, node: &TableNode) {
+        walk_table(self, node);
+    }
+    fn visit_table_row(&mut self, node: &TableRowNode) {
+        walk_table_row(self, node);
+    }
+    fn visit_table_cell(&mut self, node: &TableCellNode) {
+        for child in &node.children {
+            self.visit_inline(child);
+        }
+    }
+    fn visit_comment(&mut self, _node: &CommentNode) {}
+    fn visit_tag(&mut self, _node: &TagNode) {}
+    fn visit_error(&mut self, _node: &ErrorNode) {}
+
+    // Inline leaves and recursive inline nodes
+    fn visit_text(&mut self, _node: &TextNode) {}
+    fn visit_bold(&mut self, node: &BoldNode) {
+        for child in &node.children {
+            self.visit_inline(child);
+        }
+    }
+    fn visit_italic(&mut self, node: &ItalicNode) {
+        for child in &node.children {
+            self.visit_inline(child);
+        }
+    }
+    fn visit_bold_italic(&mut self, node: &BoldItalicNode) {
+        for child in &node.children {
+            self.visit_inline(child);
+        }
+    }
+    fn visit_strikethrough(&mut self, node: &StrikethroughNode) {
+        for child in &node.children {
+            self.visit_inline(child);
+        }
+    }
+    fn visit_code(&mut self, _node: &CodeNode) {}
+    fn visit_superscript(&mut self, node: &SuperscriptNode) {
+        for child in &node.children {
+            self.visit_inline(child);
+        }
+    }
+    fn visit_subscript(&mut self, node: &SubscriptNode) {
+        for child in &node.children {
+            self.visit_inline(child);
+        }
+    }
+    fn visit_math_inline(&mut self, _node: &MathInlineNode) {}
+    fn visit_keyword(&mut self, _node: &KeywordNode) {}
+    fn visit_color(&mut self, node: &ColorNode) {
+        for child in &node.children {
+            self.visit_inline(child);
+        }
+    }
+    fn visit_wiki_link(&mut self, node: &WikiLinkNode) {
+        if let Some(desc) = &node.description {
+            for child in desc {
+                self.visit_inline(child);
+            }
+        }
+    }
+    fn visit_external_link(&mut self, node: &ExternalLinkNode) {
+        if let Some(desc) = &node.description {
+            for child in desc {
+                self.visit_inline(child);
+            }
+        }
+    }
+    fn visit_transclusion(&mut self, _node: &TransclusionNode) {}
+    fn visit_raw_url(&mut self, _node: &RawUrlNode) {}
+}
+
+pub fn walk_document<V: Visitor + ?Sized>(v: &mut V, node: &DocumentNode) {
+    for child in &node.children {
+        v.visit_block(child);
+    }
+}
+
+pub fn walk_block<V: Visitor + ?Sized>(v: &mut V, node: &BlockNode) {
+    match node {
+        BlockNode::Heading(n) => v.visit_heading(n),
+        BlockNode::Paragraph(n) => v.visit_paragraph(n),
+        BlockNode::HorizontalRule(n) => v.visit_horizontal_rule(n),
+        BlockNode::Blockquote(n) => v.visit_blockquote(n),
+        BlockNode::Preformatted(n) => v.visit_preformatted(n),
+        BlockNode::MathBlock(n) => v.visit_math_block(n),
+        BlockNode::List(n) => v.visit_list(n),
+        BlockNode::DefinitionList(n) => v.visit_definition_list(n),
+        BlockNode::Table(n) => v.visit_table(n),
+        BlockNode::Comment(n) => v.visit_comment(n),
+        BlockNode::Tag(n) => v.visit_tag(n),
+        BlockNode::Error(n) => v.visit_error(n),
+    }
+}
+
+pub fn walk_inline<V: Visitor + ?Sized>(v: &mut V, node: &InlineNode) {
+    match node {
+        InlineNode::Text(n) => v.visit_text(n),
+        InlineNode::Bold(n) => v.visit_bold(n),
+        InlineNode::Italic(n) => v.visit_italic(n),
+        InlineNode::BoldItalic(n) => v.visit_bold_italic(n),
+        InlineNode::Strikethrough(n) => v.visit_strikethrough(n),
+        InlineNode::Code(n) => v.visit_code(n),
+        InlineNode::Superscript(n) => v.visit_superscript(n),
+        InlineNode::Subscript(n) => v.visit_subscript(n),
+        InlineNode::MathInline(n) => v.visit_math_inline(n),
+        InlineNode::Keyword(n) => v.visit_keyword(n),
+        InlineNode::Color(n) => v.visit_color(n),
+        InlineNode::WikiLink(n) => v.visit_wiki_link(n),
+        InlineNode::ExternalLink(n) => v.visit_external_link(n),
+        InlineNode::Transclusion(n) => v.visit_transclusion(n),
+        InlineNode::RawUrl(n) => v.visit_raw_url(n),
+        // A soft break carries no content/children — treated as whitespace.
+        InlineNode::SoftBreak(_) => {}
+    }
+}
+
+pub fn walk_list<V: Visitor + ?Sized>(v: &mut V, node: &ListNode) {
+    for item in &node.items {
+        v.visit_list_item(item);
+    }
+}
+
+pub fn walk_list_item<V: Visitor + ?Sized>(v: &mut V, node: &ListItemNode) {
+    for child in &node.children {
+        v.visit_inline(child);
+    }
+    if let Some(sublist) = &node.sublist {
+        v.visit_list(sublist);
+    }
+}
+
+pub fn walk_definition_item<V: Visitor + ?Sized>(v: &mut V, node: &DefinitionItemNode) {
+    if let Some(term) = &node.term {
+        for child in term {
+            v.visit_inline(child);
+        }
+    }
+    for definition in &node.definitions {
+        for child in definition {
+            v.visit_inline(child);
+        }
+    }
+}
+
+pub fn walk_table<V: Visitor + ?Sized>(v: &mut V, node: &TableNode) {
+    for row in &node.rows {
+        v.visit_table_row(row);
+    }
+}
+
+pub fn walk_table_row<V: Visitor + ?Sized>(v: &mut V, node: &TableRowNode) {
+    for cell in &node.cells {
+        v.visit_table_cell(cell);
+    }
+}

crates/nuwiki-core/src/date.rs

@@ -0,0 +1,541 @@
+//! Date primitives for the diary subsystem.
+//!
+//! This crate deliberately doesn't pull in `chrono` or `time`. The diary feature
+//! only needs `YYYY-MM-DD` parsing/formatting and ±1 day arithmetic, which
+//! is small enough that the dependency cost outweighs the convenience.
+//!
+//! Date math uses Howard Hinnant's days-from-civil algorithm
+//! (<https://howardhinnant.github.io/date_algorithms.html>) — proleptic
+//! Gregorian, valid for any year representable in `i32`.
+
+use std::cmp::Ordering;
+use std::fmt;
+use std::time::{SystemTime, UNIX_EPOCH};
+
+/// A calendar day, decoupled from any time zone or wall-clock context.
+///
+/// Validation: `parse` and `from_ymd` reject impossible dates (month 0,
+/// day 0, day > days-in-month, etc.). The struct fields are public so
+/// callers can read them, but constructing one with bogus values bypasses
+/// validation — use the constructors.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub struct DiaryDate {
+    pub year: i32,
+    pub month: u8,
+    pub day: u8,
+}
+
+impl DiaryDate {
+    /// Build a date from its components, validating the calendar.
+    pub fn from_ymd(year: i32, month: u8, day: u8) -> Option<Self> {
+        if !(1..=12).contains(&month) {
+            return None;
+        }
+        if day < 1 || day > days_in_month(year, month) {
+            return None;
+        }
+        Some(Self { year, month, day })
+    }
+
+    /// Parse a strict `YYYY-MM-DD` string. Rejects any deviation — extra
+    /// whitespace, missing zero-padding, alternate separators, etc.
+    pub fn parse(s: &str) -> Option<Self> {
+        let b = s.as_bytes();
+        if b.len() != 10 || b[4] != b'-' || b[7] != b'-' {
+            return None;
+        }
+        let y = parse_digits(&b[0..4])?;
+        let m = parse_digits(&b[5..7])? as u8;
+        let d = parse_digits(&b[8..10])? as u8;
+        Self::from_ymd(y as i32, m, d)
+    }
+
+    /// Format as `YYYY-MM-DD`. Years outside 0..=9999 still render
+    /// numerically; the diary use case never produces those.
+    pub fn format(&self) -> String {
+        format!("{:04}-{:02}-{:02}", self.year, self.month, self.day)
+    }
+
+    /// UTC "today" computed from `SystemTime::now()`. The diary
+    /// commands intentionally use UTC — local-time semantics depend on a
+    /// timezone DB we don't ship and would surprise users crossing DST
+    /// boundaries.
+    pub fn today_utc() -> Self {
+        let secs = SystemTime::now()
+            .duration_since(UNIX_EPOCH)
+            .map(|d| d.as_secs())
+            .unwrap_or(0);
+        let days = (secs / 86_400) as i64;
+        from_days(days)
+    }
+
+    pub fn next_day(&self) -> Self {
+        from_days(to_days(self.year, self.month, self.day) + 1)
+    }
+
+    pub fn prev_day(&self) -> Self {
+        from_days(to_days(self.year, self.month, self.day) - 1)
+    }
+
+    /// Internal — exposed for tests that want to verify ordering against
+    /// the days-from-epoch representation.
+    pub fn to_days_epoch(&self) -> i64 {
+        to_days(self.year, self.month, self.day)
+    }
+}
+
+impl PartialOrd for DiaryDate {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
+impl Ord for DiaryDate {
+    fn cmp(&self, other: &Self) -> Ordering {
+        self.to_days_epoch().cmp(&other.to_days_epoch())
+    }
+}
+
+impl fmt::Display for DiaryDate {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        write!(f, "{}", self.format())
+    }
+}
+
+fn parse_digits(b: &[u8]) -> Option<u32> {
+    if b.is_empty() {
+        return None;
+    }
+    let mut n: u32 = 0;
+    for ch in b {
+        if !ch.is_ascii_digit() {
+            return None;
+        }
+        n = n.checked_mul(10)?.checked_add((*ch - b'0') as u32)?;
+    }
+    Some(n)
+}
+
+fn is_leap(year: i32) -> bool {
+    (year % 4 == 0 && year % 100 != 0) || (year % 400 == 0)
+}
+
+fn days_in_month(year: i32, month: u8) -> u8 {
+    match month {
+        1 | 3 | 5 | 7 | 8 | 10 | 12 => 31,
+        4 | 6 | 9 | 11 => 30,
+        2 => {
+            if is_leap(year) {
+                29
+            } else {
+                28
+            }
+        }
+        _ => 0,
+    }
+}
+
+/// Howard Hinnant's `days_from_civil`: 1970-01-01 → 0.
+fn to_days(year: i32, month: u8, day: u8) -> i64 {
+    let y = year as i64 - if (month as i64) <= 2 { 1 } else { 0 };
+    let era = y.div_euclid(400);
+    let yoe = y.rem_euclid(400);
+    let m = month as i64;
+    let doy = (153 * (if m > 2 { m - 3 } else { m + 9 }) + 2) / 5 + day as i64 - 1;
+    let doe = yoe * 365 + yoe / 4 - yoe / 100 + doy;
+    era * 146_097 + doe - 719_468
+}
+
+// ============================================================
+// Diary frequency / period
+// ============================================================
+//
+// `DiaryFrequency` is the user-configured cadence (mirrors vimwiki's
+// `diary_frequency` setting). `DiaryPeriod` is the *concrete* diary
+// entry — a specific day, ISO week, calendar month, or year — that
+// the LSP commands compute and the renderer addresses.
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum DiaryFrequency {
+    Daily,
+    Weekly,
+    Monthly,
+    Yearly,
+}
+
+impl DiaryFrequency {
+    /// Parse the string form used in `WikiConfig.diary_frequency`.
+    /// Falls back to `Daily` for unrecognised values — vimwiki's
+    /// permissive behaviour for stale configs.
+    pub fn parse(s: &str) -> Self {
+        match s.trim().to_ascii_lowercase().as_str() {
+            "weekly" | "week" => Self::Weekly,
+            "monthly" | "month" => Self::Monthly,
+            "yearly" | "year" => Self::Yearly,
+            _ => Self::Daily,
+        }
+    }
+
+    pub fn as_str(self) -> &'static str {
+        match self {
+            Self::Daily => "daily",
+            Self::Weekly => "weekly",
+            Self::Monthly => "monthly",
+            Self::Yearly => "yearly",
+        }
+    }
+}
+
+/// A specific diary entry, identified by its calendar period. The four
+/// variants share the same `format` / `parse` / `next` / `prev` /
+/// `today_utc` surface so the LSP dispatcher can stay frequency-agnostic.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum DiaryPeriod {
+    Day(DiaryDate),
+    /// ISO 8601 week — year/week pair, week ∈ 1..=53.
+    Week {
+        iso_year: i32,
+        iso_week: u8,
+    },
+    Month {
+        year: i32,
+        month: u8,
+    },
+    Year {
+        year: i32,
+    },
+}
+
+impl DiaryPeriod {
+    /// "Now" for the given frequency, in UTC. See `DiaryDate::today_utc`
+    /// for the timezone rationale.
+    pub fn today_utc(freq: DiaryFrequency) -> Self {
+        let today = DiaryDate::today_utc();
+        match freq {
+            DiaryFrequency::Daily => Self::Day(today),
+            DiaryFrequency::Weekly => Self::week_containing(today),
+            DiaryFrequency::Monthly => Self::Month {
+                year: today.year,
+                month: today.month,
+            },
+            DiaryFrequency::Yearly => Self::Year { year: today.year },
+        }
+    }
+
+    /// File-stem format — also the format vimwiki uses for diary link
+    /// targets (`[[diary:2026-W19]]`, `[[diary:2026-05]]`, etc.).
+    ///
+    ///   Day(2026-05-12) → `2026-05-12`
+    ///   Week(2026, 19)  → `2026-W19`
+    ///   Month(2026, 5)  → `2026-05`
+    ///   Year(2026)      → `2026`
+    pub fn format(&self) -> String {
+        match self {
+            Self::Day(d) => d.format(),
+            Self::Week { iso_year, iso_week } => format!("{iso_year:04}-W{iso_week:02}"),
+            Self::Month { year, month } => format!("{year:04}-{month:02}"),
+            Self::Year { year } => format!("{year:04}"),
+        }
+    }
+
+    /// Inverse of `format` — recognise any of the four stems. Returns
+    /// `None` for inputs that don't match any flavour.
+    pub fn parse(s: &str) -> Option<Self> {
+        // Day: `YYYY-MM-DD`
+        if let Some(d) = DiaryDate::parse(s) {
+            return Some(Self::Day(d));
+        }
+        let b = s.as_bytes();
+        // Week: `YYYY-Www`
+        if b.len() == 8 && b[4] == b'-' && (b[5] == b'W' || b[5] == b'w') {
+            let y = parse_digits(&b[0..4])? as i32;
+            let w = parse_digits(&b[6..8])? as u8;
+            if (1..=53).contains(&w) {
+                return Some(Self::Week {
+                    iso_year: y,
+                    iso_week: w,
+                });
+            }
+        }
+        // Month: `YYYY-MM`
+        if b.len() == 7 && b[4] == b'-' {
+            let y = parse_digits(&b[0..4])? as i32;
+            let m = parse_digits(&b[5..7])? as u8;
+            if (1..=12).contains(&m) {
+                return Some(Self::Month { year: y, month: m });
+            }
+        }
+        // Year: `YYYY`
+        if b.len() == 4 {
+            let y = parse_digits(&b[0..4])? as i32;
+            return Some(Self::Year { year: y });
+        }
+        None
+    }
+
+    pub fn next(&self) -> Self {
+        match *self {
+            Self::Day(d) => Self::Day(d.next_day()),
+            Self::Week { iso_year, iso_week } => {
+                // Step forward one week: take the Monday of this ISO week,
+                // add 7 days, recompute the (year, week) it falls in.
+                let mon = monday_of_iso_week(iso_year, iso_week);
+                let next_mon = from_days(to_days(mon.year, mon.month, mon.day) + 7);
+                let (iso_year, iso_week) = iso_year_week(&next_mon);
+                Self::Week { iso_year, iso_week }
+            }
+            Self::Month { year, month } => {
+                if month == 12 {
+                    Self::Month {
+                        year: year + 1,
+                        month: 1,
+                    }
+                } else {
+                    Self::Month {
+                        year,
+                        month: month + 1,
+                    }
+                }
+            }
+            Self::Year { year } => Self::Year { year: year + 1 },
+        }
+    }
+
+    pub fn prev(&self) -> Self {
+        match *self {
+            Self::Day(d) => Self::Day(d.prev_day()),
+            Self::Week { iso_year, iso_week } => {
+                let mon = monday_of_iso_week(iso_year, iso_week);
+                let prev_mon = from_days(to_days(mon.year, mon.month, mon.day) - 7);
+                let (iso_year, iso_week) = iso_year_week(&prev_mon);
+                Self::Week { iso_year, iso_week }
+            }
+            Self::Month { year, month } => {
+                if month == 1 {
+                    Self::Month {
+                        year: year - 1,
+                        month: 12,
+                    }
+                } else {
+                    Self::Month {
+                        year,
+                        month: month - 1,
+                    }
+                }
+            }
+            Self::Year { year } => Self::Year { year: year - 1 },
+        }
+    }
+
+    /// Compute the ISO week containing the given calendar date.
+    pub fn week_containing(d: DiaryDate) -> Self {
+        let (iso_year, iso_week) = iso_year_week(&d);
+        Self::Week { iso_year, iso_week }
+    }
+
+    /// First calendar day of this period — useful for chronological
+    /// sorting across mixed-frequency entries.
+    ///
+    ///   Day(d)      → d
+    ///   Week(y,w)   → Monday of that ISO week
+    ///   Month(y,m)  → 1st of that month
+    ///   Year(y)     → Jan 1st of that year
+    pub fn first_day(&self) -> DiaryDate {
+        match *self {
+            Self::Day(d) => d,
+            Self::Week { iso_year, iso_week } => monday_of_iso_week(iso_year, iso_week),
+            Self::Month { year, month } => {
+                DiaryDate::from_ymd(year, month, 1).expect("validated month")
+            }
+            Self::Year { year } => DiaryDate::from_ymd(year, 1, 1).expect("Jan 1st always valid"),
+        }
+    }
+}
+
+impl fmt::Display for DiaryPeriod {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        write!(f, "{}", self.format())
+    }
+}
+
+/// How a *weekly* diary entry is named.
+///
+///   `Iso`  — ISO-week label `YYYY-Www` (nuwiki's original scheme; the week
+///            always begins on Monday, `week_start` is ignored).
+///   `Date` — the week-start day's calendar date `YYYY-MM-DD`, matching
+///            upstream vimwiki; honours `week_start`.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum WeeklyStyle {
+    Iso,
+    Date,
+}
+
+impl WeeklyStyle {
+    /// Parse the config string. `date`/`vimwiki` → `Date`; anything else
+    /// (including `iso`/`week`/empty) → `Iso`.
+    pub fn parse(s: &str) -> Self {
+        match s.trim().to_ascii_lowercase().as_str() {
+            "date" | "vimwiki" | "day" => Self::Date,
+            _ => Self::Iso,
+        }
+    }
+}
+
+/// Day a week begins on, in ISO numbering (Monday = 1 … Sunday = 7).
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct WeekStart(u8);
+
+impl WeekStart {
+    /// Parse a weekday name (`monday`..`sunday`); unknown → Monday.
+    pub fn parse(s: &str) -> Self {
+        let n = match s.trim().to_ascii_lowercase().as_str() {
+            "tuesday" => 2,
+            "wednesday" => 3,
+            "thursday" => 4,
+            "friday" => 5,
+            "saturday" => 6,
+            "sunday" => 7,
+            _ => 1, // monday (default)
+        };
+        Self(n)
+    }
+
+    pub fn monday() -> Self {
+        Self(1)
+    }
+
+    fn iso(self) -> i64 {
+        self.0 as i64
+    }
+}
+
+/// Diary navigation policy: frequency plus, for weekly diaries, the naming
+/// scheme and week-start day. Built from a wiki's `diary_*` config so the
+/// LSP commands can compute today / next / prev without re-deriving the
+/// rules. `Daily`/`Monthly`/`Yearly` ignore the weekly fields.
+#[derive(Debug, Clone, Copy)]
+pub struct DiaryCalendar {
+    pub freq: DiaryFrequency,
+    pub weekly_style: WeeklyStyle,
+    pub week_start: WeekStart,
+}
+
+impl DiaryCalendar {
+    pub fn new(freq: DiaryFrequency, weekly_style: WeeklyStyle, week_start: WeekStart) -> Self {
+        Self {
+            freq,
+            weekly_style,
+            week_start,
+        }
+    }
+
+    fn is_weekly_date(&self) -> bool {
+        self.freq == DiaryFrequency::Weekly && self.weekly_style == WeeklyStyle::Date
+    }
+
+    /// "Now" as a diary period for this calendar. Date-mode weekly snaps
+    /// today back to the most recent week-start day (a plain `Day`, so the
+    /// file is `YYYY-MM-DD` like upstream); everything else defers to
+    /// [`DiaryPeriod::today_utc`].
+    pub fn today(&self) -> DiaryPeriod {
+        if self.is_weekly_date() {
+            DiaryPeriod::Day(self.week_start_of(DiaryDate::today_utc()))
+        } else {
+            DiaryPeriod::today_utc(self.freq)
+        }
+    }
+
+    /// The period one cadence-step after `p`.
+    pub fn next(&self, p: DiaryPeriod) -> DiaryPeriod {
+        self.step(p, 1)
+    }
+
+    /// The period one cadence-step before `p`.
+    pub fn prev(&self, p: DiaryPeriod) -> DiaryPeriod {
+        self.step(p, -1)
+    }
+
+    fn step(&self, p: DiaryPeriod, dir: i64) -> DiaryPeriod {
+        // Date-mode weekly: the period is a `Day` on a week-start; a step is
+        // ±7 days (re-snapped so an off-week-start pivot still lands cleanly).
+        if self.is_weekly_date() {
+            if let DiaryPeriod::Day(d) = p {
+                let start = self.week_start_of(d);
+                return DiaryPeriod::Day(add_days(start, 7 * dir));
+            }
+        }
+        if dir >= 0 {
+            p.next()
+        } else {
+            p.prev()
+        }
+    }
+
+    /// The most recent `week_start` weekday on or before `d`.
+    fn week_start_of(&self, d: DiaryDate) -> DiaryDate {
+        let wd = iso_weekday(&d) as i64; // 1..=7
+        let back = (wd - self.week_start.iso()).rem_euclid(7);
+        add_days(d, -back)
+    }
+}
+
+/// Shift a calendar date by `delta` days (may be negative).
+fn add_days(d: DiaryDate, delta: i64) -> DiaryDate {
+    from_days(to_days(d.year, d.month, d.day) + delta)
+}
+
+/// Day-of-week with Monday=1 … Sunday=7 (ISO 8601).
+fn iso_weekday(d: &DiaryDate) -> u8 {
+    // Zeller-style via days-from-epoch: 1970-01-01 was a Thursday.
+    let days = to_days(d.year, d.month, d.day);
+    // (days + 3) mod 7 puts Monday at 0 … Sunday at 6; shift to 1..=7.
+    let r = (days + 3).rem_euclid(7);
+    (r as u8) + 1
+}
+
+/// ISO 8601 `(year, week)` for a calendar date. The ISO year may differ
+/// from the calendar year for early January / late December dates.
+fn iso_year_week(d: &DiaryDate) -> (i32, u8) {
+    // Standard algorithm: shift to the Thursday of the same ISO week,
+    // then ISO year = thursday.year, ISO week = ordinal_day(thursday) / 7 + 1.
+    let weekday = iso_weekday(d) as i64; // 1..=7
+    let days = to_days(d.year, d.month, d.day);
+    let thursday = from_days(days + 4 - weekday);
+    let jan1 = to_days(thursday.year, 1, 1);
+    let ordinal = to_days(thursday.year, thursday.month, thursday.day) - jan1 + 1;
+    let week = ((ordinal - 1) / 7 + 1) as u8;
+    (thursday.year, week)
+}
+
+/// Date of the Monday that opens the given ISO `(year, week)`.
+fn monday_of_iso_week(iso_year: i32, iso_week: u8) -> DiaryDate {
+    // Find Jan 4th of iso_year — always in ISO week 1 by definition.
+    let jan4 = DiaryDate {
+        year: iso_year,
+        month: 1,
+        day: 4,
+    };
+    let jan4_weekday = iso_weekday(&jan4) as i64; // 1..=7
+    let week1_monday_days = to_days(jan4.year, jan4.month, jan4.day) - (jan4_weekday - 1);
+    let target = week1_monday_days + (iso_week as i64 - 1) * 7;
+    from_days(target)
+}
+
+/// Inverse of [`to_days`].
+fn from_days(days: i64) -> DiaryDate {
+    let z = days + 719_468;
+    let era = if z >= 0 { z } else { z - 146_096 } / 146_097;
+    let doe = (z - era * 146_097) as u64;
+    let yoe = (doe - doe / 1460 + doe / 36_524 - doe / 146_096) / 365;
+    let y = yoe as i64 + era * 400;
+    let doy = doe - (365 * yoe + yoe / 4 - yoe / 100);
+    let mp = (5 * doy + 2) / 153;
+    let d = (doy - (153 * mp + 2) / 5 + 1) as u8;
+    let m_int = if mp < 10 { mp + 3 } else { mp - 9 };
+    let year = y + if m_int <= 2 { 1 } else { 0 };
+    DiaryDate {
+        year: year as i32,
+        month: m_int as u8,
+        day: d,
+    }
+}

crates/nuwiki-core/src/lib.rs

@@ -0,0 +1,10 @@
+//! Core parser, AST, and renderer for nuwiki.
+//!
+//! This crate is editor-independent and must never depend on `nuwiki-lsp` or
+//! `nuwiki-ls`.
+
+pub mod ast;
+pub mod date;
+pub mod listsyms;
+pub mod render;
+pub mod syntax;

crates/nuwiki-core/src/listsyms.rs

@@ -0,0 +1,198 @@
+//! 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;
+
+/// Default rejected/cancelled marker (vimwiki's `listsym_rejected`).
+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>,
+    rejected: char,
+}
+
+impl Default for ListSyms {
+    fn default() -> Self {
+        Self {
+            progression: " .oOX".chars().collect(),
+            rejected: REJECTED,
+        }
+    }
+}
+
+impl ListSyms {
+    /// Build a palette from a glyph string, with the default rejected marker.
+    /// 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 {
+        Self::new_with_rejected(s, REJECTED)
+    }
+
+    /// Like [`new`](Self::new) but with a custom rejected/cancelled glyph
+    /// (vimwiki's `listsym_rejected`).
+    pub fn new_with_rejected(s: &str, rejected: char) -> Self {
+        let progression: Vec<char> = s.chars().collect();
+        if progression.len() < 2 {
+            Self {
+                rejected,
+                ..Self::default()
+            }
+        } else {
+            Self {
+                progression,
+                rejected,
+            }
+        }
+    }
+
+    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 {
+        self.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 == self.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 self.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_rejected_glyph_is_honoured() {
+        // `listsym_rejected = '✗'`: that glyph lexes/snaps as Rejected and the
+        // default `-` no longer does.
+        let s = ListSyms::new_with_rejected(" .oOX", '✗');
+        assert_eq!(s.rejected_glyph(), '✗');
+        assert_eq!(s.state_of('✗'), Some(CheckboxState::Rejected));
+        assert_eq!(s.glyph_for_rate(-1), '✗');
+        assert_eq!(s.state_of('-'), None); // no longer the rejected marker
+                                           // A short palette still keeps the custom rejected glyph.
+        assert_eq!(ListSyms::new_with_rejected("x", '✗').rejected_glyph(), '✗');
+    }
+
+    #[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/render/mod.rs

@@ -0,0 +1,26 @@
+//! Document renderers.
+//!
+//! A `Renderer` is writer-based, takes a `DocumentNode`, and
+//! emits its representation into any `Write`. Errors propagate through
+//! `io::Result` — there's no separate per-renderer error type so the trait
+//! stays object-safe (`Box<dyn Renderer>` is useful for the LSP later).
+
+pub mod html;
+
+pub use html::HtmlRenderer;
+
+use std::io::{self, Write};
+
+use crate::ast::DocumentNode;
+
+pub trait Renderer {
+    fn render(&self, doc: &DocumentNode, w: &mut dyn Write) -> io::Result<()>;
+
+    /// Convenience: render into a `String`. Renderers that emit non-UTF-8
+    /// bytes should override or skip; the default assumes UTF-8 output.
+    fn render_to_string(&self, doc: &DocumentNode) -> io::Result<String> {
+        let mut buf = Vec::new();
+        self.render(doc, &mut buf)?;
+        String::from_utf8(buf).map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))
+    }
+}

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

@@ -0,0 +1,115 @@
+//! Syntax plugin interface.
+//!
+//! Each syntax (vimwiki today, markdown later) implements the same shape:
+//! a `Lexer` produces a `TokenStream`, a `Parser` consumes it and produces a
+//! `DocumentNode`. A `SyntaxPlugin` is the type-erased facade that the LSP
+//! layer holds in a `SyntaxRegistry`.
+
+pub mod registry;
+pub mod vimwiki;
+
+pub use registry::SyntaxRegistry;
+
+use crate::ast::DocumentNode;
+
+/// Eager token buffer produced by a `Lexer`. The newtype keeps the door
+/// open to a lazy/streaming variant later without breaking
+/// callers.
+#[derive(Debug, Clone, Default, PartialEq, Eq)]
+pub struct TokenStream<T> {
+    tokens: Vec<T>,
+}
+
+impl<T> TokenStream<T> {
+    pub fn new() -> Self {
+        Self { tokens: Vec::new() }
+    }
+
+    pub fn from_vec(tokens: Vec<T>) -> Self {
+        Self { tokens }
+    }
+
+    pub fn into_vec(self) -> Vec<T> {
+        self.tokens
+    }
+
+    pub fn as_slice(&self) -> &[T] {
+        &self.tokens
+    }
+
+    pub fn len(&self) -> usize {
+        self.tokens.len()
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.tokens.is_empty()
+    }
+
+    pub fn iter(&self) -> std::slice::Iter<'_, T> {
+        self.tokens.iter()
+    }
+
+    pub fn push(&mut self, token: T) {
+        self.tokens.push(token);
+    }
+}
+
+impl<T> From<Vec<T>> for TokenStream<T> {
+    fn from(tokens: Vec<T>) -> Self {
+        Self::from_vec(tokens)
+    }
+}
+
+impl<T> IntoIterator for TokenStream<T> {
+    type Item = T;
+    type IntoIter = std::vec::IntoIter<T>;
+
+    fn into_iter(self) -> Self::IntoIter {
+        self.tokens.into_iter()
+    }
+}
+
+impl<'a, T> IntoIterator for &'a TokenStream<T> {
+    type Item = &'a T;
+    type IntoIter = std::slice::Iter<'a, T>;
+
+    fn into_iter(self) -> Self::IntoIter {
+        self.tokens.iter()
+    }
+}
+
+/// Lexer half of a syntax plugin. Operates on raw source text and emits a
+/// `TokenStream` whose token type is syntax-specific (e.g. `VimwikiToken`).
+pub trait Lexer {
+    type Token;
+
+    fn lex(&self, text: &str) -> TokenStream<Self::Token>;
+}
+
+/// Parser half of a syntax plugin. Consumes a `TokenStream` and produces a
+/// `DocumentNode`. Parsing is resilient — malformed input
+/// becomes `BlockNode::Error(ErrorNode)`, never a hard failure.
+pub trait Parser {
+    type Token;
+
+    fn parse(&self, tokens: TokenStream<Self::Token>) -> DocumentNode;
+}
+
+/// Type-erased syntax plugin held by the `SyntaxRegistry`.
+///
+/// Implementations typically own a `Lexer` and a `Parser` and chain them
+/// inside `parse`. The token type is deliberately hidden so the registry can
+/// hold heterogeneous plugins behind a single trait object.
+///
+/// `Send + Sync` is required — the LSP server stores plugins behind an `Arc`
+/// and shares them across request handler tasks.
+pub trait SyntaxPlugin: Send + Sync {
+    fn id(&self) -> &str;
+    fn display_name(&self) -> &str;
+
+    /// File extensions associated with this syntax. Each entry includes the
+    /// leading dot, e.g. `[".wiki"]`.
+    fn file_extensions(&self) -> &[&str];
+
+    fn parse(&self, text: &str) -> DocumentNode;
+}

crates/nuwiki-core/src/syntax/registry.rs

@@ -0,0 +1,58 @@
+//! Registry of syntax plugins.
+//!
+//! Plugins are stored behind `Arc<dyn SyntaxPlugin>` so the LSP layer can
+//! hand a cheap clone to per-document tasks without re-locking the registry.
+
+use std::sync::Arc;
+
+use super::SyntaxPlugin;
+
+#[derive(Default, Clone)]
+pub struct SyntaxRegistry {
+    plugins: Vec<Arc<dyn SyntaxPlugin>>,
+}
+
+impl SyntaxRegistry {
+    pub fn new() -> Self {
+        Self::default()
+    }
+
+    /// Register a plugin owned directly. The most common shape — implementer
+    /// hands over a value, registry takes ownership.
+    pub fn register<P>(&mut self, plugin: P)
+    where
+        P: SyntaxPlugin + 'static,
+    {
+        self.plugins.push(Arc::new(plugin));
+    }
+
+    /// Look up a plugin by its `id`.
+    pub fn get(&self, id: &str) -> Option<&dyn SyntaxPlugin> {
+        self.plugins
+            .iter()
+            .find(|p| p.id() == id)
+            .map(|p| p.as_ref())
+    }
+
+    /// Look up a plugin by file extension. `ext` should include the leading
+    /// dot (e.g. `".wiki"`) — matches the plugin's `file_extensions()`.
+    pub fn detect_from_extension(&self, ext: &str) -> Option<&dyn SyntaxPlugin> {
+        self.plugins
+            .iter()
+            .find(|p| p.file_extensions().iter().any(|e| *e == ext))
+            .map(|p| p.as_ref())
+    }
+
+    /// Iterate over registered plugins in registration order.
+    pub fn iter(&self) -> impl Iterator<Item = &dyn SyntaxPlugin> + '_ {
+        self.plugins.iter().map(|p| p.as_ref())
+    }
+
+    pub fn len(&self) -> usize {
+        self.plugins.len()
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.plugins.is_empty()
+    }
+}

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

@@ -0,0 +1,50 @@
+//! Vimwiki syntax plugin.
+
+pub mod lexer;
+pub mod parser;
+
+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
+/// registry can hand out a single trait object.
+#[derive(Debug, Default, Clone)]
+pub struct VimwikiSyntax;
+
+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 {
+    fn id(&self) -> &str {
+        "vimwiki"
+    }
+
+    fn display_name(&self) -> &str {
+        "Vimwiki"
+    }
+
+    fn file_extensions(&self) -> &[&str] {
+        &[".wiki"]
+    }
+
+    fn parse(&self, text: &str) -> DocumentNode {
+        let tokens = VimwikiLexer::new().lex(text);
+        VimwikiParser::new().parse(tokens)
+    }
+}