feat: render richer blocks and inline nodes (task 0004) #3

alexion · 2026-07-29T07:35:37-04:00

alexion commented

2026-07-29 07:35:37 -04:00

Task: .claude/tasks/0004-richer-blocks-and-inline.md

Summary

Extends @kitchen-md/core and the CLI renderer to cover the rest of the everyday Markdown body, so kitchen view renders lists, blockquotes, code, thematic breaks, and inline emphasis faithfully.

Core AST gains ListBlock (ordered flag + ListItemBlock container items), BlockquoteBlock, CodeBlock (optional language + literal text), ThematicBreakBlock, and inline EmphasisNode, StrongNode, CodeSpanNode, LinkNode.
Raw fallback: any remark block or inline node core does not model falls through to RawBlock / RawInline, whose verbatim value is position-sliced from the original input — never re-stringified — so unmodelled constructs round-trip byte-for-byte.
remark-gfm is added to the pipeline; tables and strikethrough parse as their own nodes and therefore reach the raw fallback. The translation layer recurses into container children rather than collapsing them.
Renderer: bulleted/numbered lists (one item per line), literal code blocks, a thematic-break rule, blockquotes, and inline styling (emphasis italic, strong bold, code span inverse, links show inline content only).

Deviations

ListItemBlock is scoped to ListBlock.items, not the top-level Block union. A list item exists only inside a list, so it is dispatched through renderItemContent rather than renderBlock, keeping the type as tight as the spec's shape (raised by the Spec review below and addressed).
CodeBlock.lang is omitted rather than set to undefined when a fence has no language, so the parser output deep-equals { type: "code", value: … }.
"Content not annotation-parsed" holds structurally. Code blocks and code spans store their literal value and are never fed to translateInline; annotation parsing itself lands in task 0006.

No acceptance criteria were dropped.

Review

Risk

Overall: Low

Blast radius: Low — three source files in one CLI package plus tests; internal functions, no external callers affected.
Reversibility: Low — pure code additions/renames, no migrations or persisted schema; trivial git revert.
Test coverage: Low — each new node type, the raw fallbacks (byte-for-byte source), the list-item shape, and styling are covered by ~20 added tests alongside the diff.
Sensitive domain: Low — no auth, permissions, payments, concurrency, or data migration; the one input-slicing path guards undefined offsets by returning "".
Size & complexity: Low — moderate additions but straightforward switch-based dispatch, no tangled control flow or shared mutable state.
Runtime criticality: Low — dev/CLI Markdown rendering tool, not a production hot path.

Unaddressed findings

Standards: none. No documented-standard violations; the only baseline smell noted was inherent Shotgun Surgery (adding a node kind touches the union, the parse case, and the render case) — that is the intrinsic cost of an explicit AST with exhaustive dispatch, mitigated by the exhaustive switch arms, and not worth refactoring.

Spec: none outstanding. The one finding (the ListItemBlock widening) was fixed in this branch, so it is already in the diff.

Task: `.claude/tasks/0004-richer-blocks-and-inline.md` ## Summary Extends `@kitchen-md/core` and the CLI renderer to cover the rest of the everyday Markdown body, so `kitchen view` renders lists, blockquotes, code, thematic breaks, and inline emphasis faithfully. - **Core AST** gains `ListBlock` (ordered flag + `ListItemBlock` container items), `BlockquoteBlock`, `CodeBlock` (optional language + literal text), `ThematicBreakBlock`, and inline `EmphasisNode`, `StrongNode`, `CodeSpanNode`, `LinkNode`. - **Raw fallback**: any remark block or inline node core does not model falls through to `RawBlock` / `RawInline`, whose verbatim `value` is position-sliced from the original input — never re-stringified — so unmodelled constructs round-trip byte-for-byte. - **remark-gfm** is added to the pipeline; tables and strikethrough parse as their own nodes and therefore reach the raw fallback. The translation layer recurses into container children rather than collapsing them. - **Renderer**: bulleted/numbered lists (one item per line), literal code blocks, a thematic-break rule, blockquotes, and inline styling (emphasis italic, strong bold, code span inverse, links show inline content only). ### Deviations - **`ListItemBlock` is scoped to `ListBlock.items`, not the top-level `Block` union.** A list item exists only inside a list, so it is dispatched through `renderItemContent` rather than `renderBlock`, keeping the type as tight as the spec's shape (raised by the Spec review below and addressed). - **`CodeBlock.lang` is omitted rather than set to `undefined`** when a fence has no language, so the parser output deep-equals `{ type: "code", value: … }`. - **"Content not annotation-parsed" holds structurally.** Code blocks and code spans store their literal `value` and are never fed to `translateInline`; annotation parsing itself lands in task 0006. No acceptance criteria were dropped. ## Review ### Risk **Overall: Low** - Blast radius: Low — three source files in one CLI package plus tests; internal functions, no external callers affected. - Reversibility: Low — pure code additions/renames, no migrations or persisted schema; trivial git revert. - Test coverage: Low — each new node type, the raw fallbacks (byte-for-byte source), the list-item shape, and styling are covered by ~20 added tests alongside the diff. - Sensitive domain: Low — no auth, permissions, payments, concurrency, or data migration; the one input-slicing path guards undefined offsets by returning `""`. - Size & complexity: Low — moderate additions but straightforward switch-based dispatch, no tangled control flow or shared mutable state. - Runtime criticality: Low — dev/CLI Markdown rendering tool, not a production hot path. ### Unaddressed findings **Standards:** none. No documented-standard violations; the only baseline smell noted was inherent Shotgun Surgery (adding a node kind touches the union, the parse case, and the render case) — that is the intrinsic cost of an explicit AST with exhaustive dispatch, mitigated by the exhaustive `switch` arms, and not worth refactoring. **Spec:** none outstanding. The one finding (the `ListItemBlock` widening) was fixed in this branch, so it is already in the diff.

alexion added 1 commit 2026-07-29 07:35:37 -04:00

feat: render richer blocks and inline nodes (task 0004) 22befaaa71

Extend the core parser and the CLI renderer to cover the everyday Markdown
body beyond headings and paragraphs: lists, blockquotes, fenced code,
thematic breaks, and inline emphasis, strong, code spans, and links.

Add remark-gfm to the pipeline so tables and strikethrough parse as their
own nodes. Any block or inline node core does not model falls through to
RawBlock / RawInline, whose verbatim value is position-sliced from the
original input so unmodelled constructs round-trip byte-for-byte.

alexion merged commit 22befaaa71 into main

2026-07-29 19:47:11 -04:00

alexion deleted branch task-0004-richer-blocks-and-inline

2026-07-29 19:47:11 -04:00

This repo is archived. You cannot comment on pull requests.

1 Participants

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: alexion/kitchen-md#3