omena-parser

omena-parser is the green-field parser track for the future Omena CSS engine.

It lives next to the current engine-style-parser and does not replace product behavior until parity gates are met. The crate starts with the stable public parser surface, tokenizer, CST builder, recovery vocabulary, and dialect-extension seams that later full grammar work will fill in.

Current scope:

• ParseResult over a cstree green root.
• ParserCstEquivalenceSummaryV0 verifies that parser CST nodes and tokens use the shared omena-syntax SyntaxKind contract with source-text round-trip and typed-wrapper evidence.
• LexResult exposes token kind, original range, and token text for differential token checks.
• omena-parser-summary exposes the parser-owned parity-lite summary used by the public parser lane for selectors, CSS Modules values, keyframes, and structural CSS-family counts.
• omena-parser-css-modules-intermediate exposes the parser-owned CSS Modules intermediate index summary used by the parser index-bridge gate, including values, custom properties, Sass symbols, wrappers, keyframes, composes, and nested BEM selector metadata.
• omena-parser-canonical-candidate, omena-parser-evaluator-candidates, and omena-parser-canonical-producer expose the parser-owned wrapper summaries used by the parser public-product lane.
• Recursive-descent parser core coverage is explicit: stylesheet/rule/ declaration entry points, Selectors L4 CST slices, registered at-rule preludes, CSS nesting, SCSS/Sass/Less dialect statements, Bogus recovery, and style-fact extraction are ready. The complete external CSS-family spec mirror remains a separate conformance target.
• Typed CST wrapper accessors for the current stylesheet, rule, selector, declaration, value, component-value, component-value-list, simple-block, custom-property-value, and at-rule nodes.
• Typed Bogus-node wrapper accessors for recovery-aware consumers.
• CSS Syntax entry points for rule lists, component values, component-value lists, comma-separated component-value lists, and simple blocks.
• CSS custom property declarations parse values as arbitrary component-value lists.
• Pratt value parser core coverage is explicit: unary +/-, additive and multiplicative precedence, parenthesized expressions, function argument lists, specialized CSS value function families, and value-level recovery are ready. The full CSS property-value grammar registry remains a separate product-cutover target.
• Functional pseudo-classes that carry selector lists (:is, :where, :has, :not, :local, :global) now surface nested selector-list CST nodes and isolate malformed selector-list items as Bogus selectors.
• :nth-child(), :nth-last-child(), :nth-of-type(), and :nth-last-of-type() expose formula arguments, including of <selector-list>, as structured CST nodes.
• :has() arguments are represented as relative selector lists, preserving leading combinators for downstream selector semantics.
• :lang() and :dir() expose dedicated argument CST nodes for linguistic selector semantics.
• Attribute selectors expose dedicated name, matcher, value, and modifier nodes, including Selectors L4 case-sensitivity flags.
• Missing block-close recovery markers represented as BogusTrivia.
• Panic-free tokenizer for CSS-family source slices using char-boundary-safe cursor movement.
• Initial dialect classification for CSS, SCSS, Sass, and Less tokens.
• TokenSet recovery scaffolding and parser boundary summary.