Resolves the 18 findings from the parallel codebase review, tracked in
development/vimwiki-gap.md.
Correctness / perf:
- Wiki.config -> Arc<WikiConfig> so cloning a Wiki is a refcount bump (R5)
- WorkspaceIndex::remove is no longer O(n^2): a per-source contributions
map limits the scan to buckets the source actually wrote into (R6)
- render_color now expands color_tag_template (__STYLE__/__CONTENT__),
consuming the previously-dead field; ColorNode documented as an
extension point; 3 renderer tests added (R3/R4)
- wiki_root_for returns empty/nil on no-match instead of falling back to
the first wiki (R2); auto_header honours links_space_char (R7)
Cleanup / dedup:
- Remove dead #[allow(dead_code)] stubs + uncalled pub helpers, narrow
imports (R10/R11)
- Dedup span_of_inline x3 -> InlineNode::span() (R12)
- diary_step single read lock; page_captions single pass (R13)
- Lua auto_header loop -> wiki_list(); detect_current_symbol cleanup
(R14/R18)
Client / docs:
- :VimwikiNormalizeLink Vim cmds -> <q-args> (R17); ftplugin header fix
(R16); vars.vim multi-wiki limitation documented (R15)
- Document 19 config options in README.md + doc/nuwiki.txt; fix
list_margin/shiftwidth doc and stale comments (R1/R9)
- R8 investigated, confirmed not a real bug (documented)
Verified: Neovim harness 307, Vim harness 301/18/21, Rust suite 568, all
0 failed; clippy clean; fresh parallel-agent audit found no regressions.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Implements the deferred half of these options: a soft line break inside a
paragraph / list item can render as <br /> instead of a space.
- Parser: soft breaks now emit a dedicated InlineNode::SoftBreak node
(parse_inline_seq K::Newline + the list-continuation synthetic join) instead
of collapsing to Text(" "). Added SoftBreakNode + the variant.
- All inline consumers treat SoftBreak as whitespace: AST visitor (no-op),
text extraction (index.rs + lib.rs → space), nav/semantic_tokens span_of_inline,
semantic emit (skip like Text), parser span_of_inline.
- Renderer: render_inlines_break renders top-level SoftBreaks as the
context-specific string; render_paragraph uses text_ignore_newline,
render_list_item uses list_ignore_newline; default render_inline → space.
with_newline_handling builder; export.rs wires both from HtmlConfig.
Default (true/true = upstream default) behaviour is unchanged (space). Updated
3 core tests + 1 parser test that asserted the old Text(" ") shape to treat
SoftBreak as whitespace. New test render_page_html_ignore_newline_controls_soft_breaks.
Gap doc: P3 Config item closed — only the deferred markdown cluster remains.
Full rust suite + clippy clean; Neovim 307, Vim 301/18/21.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
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>
Adds STOPPED alongside TODO/DONE/STARTED/FIXME/FIXED/XXX across the stack:
lexer, AST Keyword enum, HTML renderer (class="stopped"), LSP keyword_str,
and the Vim syntax red group (nuwikiKeywordAttn). Grouped with the
attention/pending keywords (red) for both export and in-editor highlighting.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Lines indented strictly past a list item's marker now attach to that
item rather than becoming a sibling paragraph. Matches upstream
vimwiki + Markdown convention.
Before:
- first item
continuing here ← rendered as a separate <p>
- second
Now:
- first item continuing here ← one <li>, joined with a space
- second
The detection lives in `parse_list_item`: after the item's first
line, we loop over subsequent lines and treat each as continuation
when its leading whitespace exceeds the marker's column. Two
flavours need handling because the lexer emits different tokens
based on indent width:
* 1..3 leading spaces → `Text(s)` with the whitespace embedded;
we trim it off the first text token before re-running
`parse_inline_seq`.
* 4+ leading spaces → `BlockquoteIndent` + bare `Text`; we
advance past the indent token and use the text as-is.
Continuation also works inside nested lists (`level + 1` threshold
adapts to the marker's own indent) and is bounded by blank lines /
unindented content / sibling-or-shallower list markers.
Tests: 6 new in `crates/nuwiki-core/tests/list_continuation.rs`
covering single-line continuation, multi-line continuation, blank-
line termination, unindented termination, nested-list continuation,
and the HTML-renderer regression (single `<ul>` with two `<li>`s,
no orphan `<p>`).
Gates: 456 Rust / 39 Neovim / 12 Vim.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
`{{url|alt|key="quoted value"}}` syntax was already parsed end-to-end
through to `TransclusionNode::attrs`, but the value retained the
surrounding `"…"` (or `'…'`) characters verbatim. The HTML renderer
then escaped them into `"`, producing `style=""border: 1px""`
instead of clean `style="border: 1px"`.
Extracted a small `insert_attr` helper that splits on `=`, trims, and
strips matching single- or double-quote pairs from the value before
inserting. Unquoted values pass through unchanged.
Tests: 4 new in crates/nuwiki-core/tests/transclusion_attrs.rs
covering double-quote stripping, single-quote stripping, unquoted
pass-through, and the clean-HTML rendering shape (no `"`).
Note: this commit covers the "transclusion attrs" half of the
original Cluster 8 audit. The "multi-line list-item continuation
paragraphs" half requires lexer + parser changes (continuation lines
indented past the marker should attach to the prior list item
instead of becoming a sibling paragraph) and is deferred.
Gates: 429 Rust / 39 Neovim / 12 Vim.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The Markdown-style alignment anchors on a table's header-separator
row now propagate from the lexer through the parser into the AST,
and the HTML renderer emits a `style="text-align: …;"` attribute on
each affected `<th>` / `<td>`.
|:--| → left
|--:| → right
|:--:| → center
|---| → default (no style attr)
Encoding choice: rather than threading the source string into the
parser, the lexer now extracts per-cell alignment when it recognises
the separator row and carries the `Vec<TableAlign>` inline on the
`TableHeaderRow` token. The parser unpacks it into `TableNode`'s
new `alignments: Vec<TableAlign>` field. Cells past the end of the
vector render with the default alignment.
`TableAlign` lives in `nuwiki_core::ast::block` alongside the table
nodes and is re-exported via `nuwiki_core::ast`.
Tests: 4 new in crates/nuwiki-core/tests/vimwiki_table_alignment.rs
covering plain dashes, all three anchor flavours, the HTML output
shape, and the no-style-attr case. The existing
`vimwiki_lexer::table_header_separator_row` was updated to assert
the new payload shape.
Gates: 425 Rust / 39 Neovim / 12 Vim.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Vimwiki `:tag1:tag2:` syntax across the full stack. P11 resolved as
strict-vimwiki — no markdown `#tag` alias.
AST (nuwiki-core):
- `TagScope { File, Heading(usize), Standalone }` mirrors vimwiki's
three placement-determined meanings: lines 0-1 → File; one or two
lines below the most recent heading → Heading(idx); otherwise
Standalone.
- `TagNode { span, tags: Vec<String>, scope }` + `BlockNode::Tag`.
- `PageMetadata.tags: Vec<String>` accumulates the file-scope tag
names — convenience projection of the Tag blocks whose scope is
File. New field is additive (Default derive) so existing consumers
using `..Default::default()` keep working.
- `Visitor::visit_tag` lands with a default body and `walk_block`
dispatches the new arm.
Lexer (`syntax/vimwiki/lexer.rs`):
- `VimwikiTokenKind::Tag(Vec<String>)` block-level token.
- `try_lex_tag` runs in the normal-line dispatcher right after heading
recognition; demands the *whole trimmed line* match `:name:name…:`,
with non-empty whitespace-free names between adjacent colons. Empty
segments (`::tag::`), whitespace inside a name, or single-colon
lines are all rejected. Doesn't fight `Term::` (definition-list
marker) because that pattern requires text before the `::`.
- Indentation is tolerated; the span covers the trimmed run.
Parser (`syntax/vimwiki/parser.rs`):
- `ParseState` gains `headings_seen` + `last_heading_line`.
- `parse_heading` bumps both on every closed heading so the tag
parser can detect the heading window.
- `parse_tag` reads the Tag token, computes the scope via
`scope_for_tag(tag_line)`, eats the trailing newline, and emits the
TagNode.
- `parse_document` post-processes each block: when a Tag with
`scope == File` is emitted, its names are pushed (de-duped) into
`metadata.tags`.
Renderer (`render/html.rs`):
- `render_tag` emits `<div class="tags"><span class="tag"
id="tag-name">name</span>...</div>`. The `id` is what makes
`[[Page#tag-name]]` jump to the rendered tag.
Semantic tokens (`semantic_tokens.rs`):
- New `TOKEN_TAG = 19` / `vimwikiTag` so editors highlight tag lines
distinctly. `emit_block` adds the Tag arm.
Workspace index (`index.rs`):
- `IndexedPage.tags: Vec<TagInfo>` records each tag occurrence on a
page with its span + scope.
- `WorkspaceIndex.tags_by_name: HashMap<String, Vec<TagOccurrence>>`
is the reverse map for cross-page anchor resolution + the Phase 13
`nuwiki.tags.search` command.
- `upsert` populates both; `remove` scrubs the reverse map and drops
empty entries; `tags_for(name)` is the public lookup.
LSP handlers (`lib.rs`):
- `heading_range_in` falls back to tag matching when the anchor
isn't a heading slug — so `[[Page#some-tag]]` lands on the matching
TagNode. Heading anchors win over tag anchors when both match (the
vimwiki precedence).
- `workspace/symbol` includes tags alongside headings, named `:tag:`
and kinded `PROPERTY` so editor UIs can distinguish them.
Tests (20 new):
- Syntax (15 in `vimwiki_tags.rs`):
lexer recognition + rejection of `::tag::`, whitespace-in-name,
Term::-conflict; parser scope paths (file, heading-1-below,
heading-2-below, standalone, no-heading, line-1 priority);
metadata dedup; renderer HTML + escaping.
- LSP index (5 in `tags_index_and_lsp.rs`):
per-page tag population, cross-page `tags_by_name` aggregation,
re-upsert replaces stale tags, `remove` cleans the reverse map,
heading-scope tag indexed but not in file metadata.
All 191 v1.0 + Phase 11 tests still pass — backward compatible.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Hand-rolled recursive-descent parser over the token stream. Resilient
per SPEC §6.7: never aborts the document — anything the dispatcher
can't claim becomes BlockNode::Error and the cursor advances, and
unterminated wikilinks / transclusions / delimiters fall back to
literal text.
Block coverage: every variant from §6.4 — Heading (with inline
content between markers), Paragraph (with soft-break across single
newlines), HorizontalRule, Blockquote (both `>` and 4-space styles),
Preformatted, MathBlock, List (with checkbox + indent-driven
sublists), DefinitionList, Table (header separator promotes the
preceding row, plus colspan / rowspan cells), Comment (single +
multiline), and page-level placeholders threaded into PageMetadata.
Inline pairing: left-to-right scan with "find next matching delim",
recursing on the slice between. `_*x*_` becomes Italic(Bold(...)),
mirroring SPEC §6.7 precedence. URLs inside `[[ ]]` route to
ExternalLinkNode; everything else to WikiLinkNode with full
LinkTarget classification (Wiki / Interwiki numbered + named /
Diary / File / Local / AnchorOnly / directory / root-relative /
filesystem-absolute).
VimwikiSyntax registers as id="vimwiki", extensions=[".wiki"] and
chains lexer + parser inside SyntaxPlugin::parse.
SPEC §4 updated to record the hand-rolled choice with rationale —
span-bearing tokens fit awkwardly into winnow/nom combinator style;
resilience and recovery are explicit in code instead.
Tests (41 new parser cases) cover every AST construct, inline
precedence including nested bold/italic, every LinkKind, transclusion
alt + attrs, resilience on unterminated links, and end-to-end
SyntaxRegistry dispatch.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two-pass, hand-rolled lexer per SPEC.md §6.6. Block pass walks the
source line by line; recognised constructs emit structural tokens, and
text-bearing line content is fed to the inline pass. Multi-line fences
(`{{{ }}}`, `{{$ }}$`, `%%+ +%%`) flip a `BlockMode` so subsequent
lines are captured raw until the matching closer.
VimwikiToken covers the full §9 feature checklist:
- headings (1–6, centered), horizontal rule, blockquotes (`>` and
4-space), all list-marker variants, checkboxes, definition `::`,
tables (separators, header row, col/row span)
- preformatted fences (with language + key=val attrs), math blocks
(with `%env%`), single- and multi-line comments, all four
page placeholders (%title / %nohtml / %template / %date)
- inline: bold/italic/strike/super/sub delimiters, inline code, inline
math, all six keywords (word-bounded), wikilinks (with description
separator), transclusions (with attr separator), raw URLs across
http(s)/ftp/mailto/file schemes — trailing sentence punctuation
stripped from URL spans
Spans are byte-accurate per SPEC §6.5 (0-indexed line + byte column +
absolute byte offset). The lexer is permissive: every delimiter is
emitted, the parser will pair them and fall back to literal text on
mismatches.
Tests (41 cases) cover one example per construct in §9, plus
word-boundary keyword detection, list-vs-bold disambiguation,
indent-vs-list disambiguation, and a span-correctness check.
README now carries a per-phase status table; SPEC top-line status
advanced to Phase 3 complete.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add the Lexer/Parser/SyntaxPlugin traits and the SyntaxRegistry per
SPEC.md §6.3.
- `Lexer<Token>` and `Parser<Token>` are typed building blocks: each
syntax owns its own token alphabet (vimwiki today, markdown later).
- `TokenStream<T>` is a `Vec<T>` newtype so the eager-vs-lazy choice
(§6.6) can flip later without breaking callers.
- `SyntaxPlugin` is type-erased: it exposes id / display_name /
file_extensions / parse(text) so the registry can hold heterogeneous
plugins behind a single trait object. Implementations chain their
Lexer + Parser inside parse().
- `SyntaxPlugin: Send + Sync` so the LSP server can stash plugins in
an Arc and share them across handler tasks.
- `SyntaxRegistry` stores `Arc<dyn SyntaxPlugin>`, supports lookup by
id and by file extension, iteration in registration order.
Tests cover TokenStream round-trip, full lex→parse chain through a
mock plugin, registry lookup hits and misses, and a compile-time
assert_send_sync::<SyntaxRegistry>().
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
gffranco