nx_css_parse.nx

nx_css_parse.nx -- CSS parser for the Nishi browser. Phase 2 (CSS subset) of docs/NISHI_BROWSER_ROADMAP.md, second primitive after nx_css_tokenize.nx. Consumes the CssToken stream produced by nx_css_next_token and emits a CssRule AST suitable for cascade + apply by nx_css_apply (next primitive, queued). Subset: - one simple selector per rule (TAG / CLASS / ID) - one or more declarations per rule, separated by ';' - each declaration's value is ONE of: NUMBER + UNIT (e.g., "16px" -- emitted as NUMBER then IDENT by the tokenizer; parser pairs them) HEXCOLOR (e.g., "#ff0000") IDENT (e.g., "red" or "sans-serif") STRING (e.g., "Arial") - rules are separated by whitespace (no comma-separated selector lists in Phase 2 -- one selector per rule, repeated for groups) What it does NOT handle yet (named Phase 2b improvements): - selector lists with commas (group as multiple rules) - compound selectors (h1.foo#bar) (one type/class/id per rule) - descendant chains (header h1) (single simple selector only) - multi-token values (margin: 8px 16px;) - !important - shorthand expansion (border, font, margin) - @-rules (silently skipped to next rule) Subset is sufficient to drive nx_css_apply on andelinwest.com-class hand-written CSS. Per cardinal feedback-honest-perf-verdict: gap list is EXACT. genealogy_id: w3c_css_syntax_module_level_3_parser_section + substrate_browser_phase_2_second_primitive lineage_id: nishi_browser_css_parse_v1 nx_safety_envelope: intended_use: "CSS parser -- browser front-end style apply" sil_target: SIL1