@diagrammo/dgmo 0.47.0 → 0.48.0

package/src/completion.ts

@@ -1,2668 +0,0 @@
-/**
- * Diagram symbol extraction API + completion registry.
- *
- * Provides:
- * - DiagramSymbols interface + extractDiagramSymbols() dispatch
- * - COMPLETION_REGISTRY: chart-type → directives map (for editor autocomplete)
- * - CHART_TYPES: array of { name, description } for chart type completion
- * - METADATA_KEY_SET: derived set of all known directive keys
- *
- * Each diagram type registers its own extractor via registerExtractor().
- * All built-in extractors are registered at module init below.
- */
-
-import { extractSymbols as extractErSymbols } from './er/parser';
-import { extractSymbols as extractFlowchartSymbols } from './graph/flowchart-parser';
-import { extractSymbols as extractInfraSymbols } from './infra/parser';
-import { extractSymbols as extractClassSymbols } from './class/parser';
-import { extractPertSymbols } from './pert/parser';
-import {
-  parseFirstLine,
-  ALL_CHART_TYPES,
-  measureIndent,
-} from './utils/parsing';
-import { RECOGNIZED_COLOR_NAMES } from './colors';
-// Closed enum sets owned by their respective parsers — imported (never
-// hand-copied) so completion can't drift from the grammar (one-oracle rule).
-import { ALL_MARKERS } from './raci/variants';
-import { STATE_KEYWORDS as WIREFRAME_STATE_KEYWORDS } from './wireframe/parser';
-
-const RECOGNIZED_COLOR_SET: ReadonlySet<string> = new Set(
-  RECOGNIZED_COLOR_NAMES
-);
-// Read chart-type descriptions directly from the source-of-truth data
-// module instead of via dgmo-router.ts. dgmo-router imports every
-// parser, and the parsers (Class/ER/Infra/Pert/Flowchart) type-only
-// import DiagramSymbols back from this file — creating a hub of cycles
-// through completion ↔ dgmo-router. Going through chart-types.ts (a
-// leaf module with zero imports) breaks 7 of the 10 known cycles.
-import { chartTypes } from './chart-types';
-
-// ============================================================
-// Symbol extraction
-// ============================================================
-
-// Types live in ./completion-types so the chart-type parsers can
-// import them without taking a cycle through this file.
-import type { ChartType, DiagramSymbols, ExtractFn } from './completion-types';
-export type { ChartType, DiagramSymbols, ExtractFn };
-
-const extractorRegistry = new Map<ChartType, ExtractFn>();
-
-export function registerExtractor(kind: ChartType, fn: ExtractFn): void {
-  extractorRegistry.set(kind, fn);
-}
-
-/**
- * Extract diagram symbols from document text.
- * Returns null if the chart type is unknown or has no registered extractor.
- */
-export function extractDiagramSymbols(docText: string): DiagramSymbols | null {
-  // Parse chartType from first line — bare type name.
-  let chartType: string | null = null;
-  for (const line of docText.split('\n')) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-    const result = parseFirstLine(trimmed);
-    if (result) {
-      chartType = result.chartType;
-    }
-    break; // only check the first non-empty, non-comment line
-  }
-  if (!chartType) return null;
-  const fn = extractorRegistry.get(chartType);
-  if (!fn) return null;
-  const result = fn(docText);
-  // Populate `aliases` uniformly for every chart type so downstream
-  // editor surfaces don't need per-extractor branches.
-  const aliases = extractAliasDeclarations(docText);
-  return Object.keys(aliases).length > 0 ? { ...result, aliases } : result;
-}
-
-// ============================================================
-// Completion registry
-// ============================================================
-
-/** Specification for a single directive: description + optional enumerated values. */
-export interface DirectiveValueSpec {
-  description: string;
-  values?: string[];
-}
-
-/** Specification for a chart type's directives. */
-export interface DirectiveSpec {
-  directives: Record<string, DirectiveValueSpec>;
-}
-
-// Global directives applied to every chart type
-const GLOBAL_DIRECTIVES: Record<string, DirectiveValueSpec> = {
-  palette: {
-    description: 'Color palette name',
-    values: [
-      'nord',
-      'catppuccin',
-      'tokyo-night',
-      'atlas',
-      'blueprint',
-      'slate',
-      'tidewater',
-    ],
-  },
-  theme: {
-    description: 'Color theme',
-    values: ['light', 'dark', 'transparent'],
-  },
-  'no-title': {
-    description: 'Hide the diagram title',
-  },
-};
-
-function withGlobals(
-  directives: Record<string, DirectiveValueSpec> = {}
-): DirectiveSpec {
-  return { directives: { ...GLOBAL_DIRECTIVES, ...directives } };
-}
-
-/** Chart-type → directive specifications. Every chart type has at least palette + theme. */
-export const COMPLETION_REGISTRY = new Map<string, DirectiveSpec>([
-  // ── Data charts ──────────────────────────────────────────
-  [
-    'bar',
-    withGlobals({
-      stack: {
-        description:
-          'Multi-series block header → stacked bars (one bar/category)',
-      },
-      group: {
-        description:
-          'Multi-series block header → clustered (side-by-side) bars',
-      },
-      'x-label': { description: 'X-axis label' },
-      'y-label': { description: 'Y-axis label' },
-      'orientation-horizontal': { description: 'Switch to horizontal bars' },
-      'no-value': { description: 'Hide value labels atop each bar' },
-      color: { description: 'Bar color override' },
-    }),
-  ],
-  [
-    'line',
-    withGlobals({
-      series: { description: 'Series name(s)' },
-      fill: { description: 'Fill under the line (area chart)' },
-      'x-label': { description: 'X-axis label' },
-      'y-label': { description: 'Y-axis label (left axis)' },
-      'y-right-label': {
-        description: 'Right y-axis label (dual-axis; group series beneath it)',
-      },
-      'no-value': { description: 'Hide value labels at each point' },
-    }),
-  ],
-  [
-    'pie',
-    withGlobals({
-      hole: {
-        description:
-          'Doughnut ring (optional ratio 0–0.9); shows the total in center',
-      },
-      'no-center-total': {
-        description: 'Hide the center total on a doughnut/hole pie',
-      },
-      'no-name': { description: 'Hide name from segment labels' },
-      'no-value': { description: 'Hide value from segment labels' },
-      'no-percent': { description: 'Hide percent from segment labels' },
-    }),
-  ],
-  [
-    'polar-area',
-    withGlobals({
-      'no-name': { description: 'Hide name from segment labels' },
-      'no-value': { description: 'Hide value from segment labels' },
-      'no-percent': { description: 'Hide percent from segment labels' },
-    }),
-  ],
-  [
-    'radar',
-    withGlobals({
-      'no-value': { description: 'Hide value labels at each vertex' },
-    }),
-  ],
-
-  // ── Extended charts ──────────────────────────────────────
-  [
-    'scatter',
-    withGlobals({
-      'no-name': { description: 'Hide point labels' },
-      'x-label': { description: 'X-axis label' },
-      'y-label': { description: 'Y-axis label' },
-      'size-label': { description: 'Size axis label' },
-    }),
-  ],
-  [
-    'heatmap',
-    withGlobals({
-      columns: { description: 'Column labels (required)' },
-      'no-value': { description: 'Hide cell value text' },
-    }),
-  ],
-  ['sankey', withGlobals()],
-  [
-    'funnel',
-    withGlobals({
-      'no-name': { description: 'Hide left-side name labels' },
-      'no-value': { description: 'Hide right-side value labels' },
-    }),
-  ],
-  [
-    'function',
-    withGlobals({
-      x: { description: 'X-axis range (start to end)' },
-      'x-label': { description: 'X-axis label' },
-      'y-label': { description: 'Y-axis label' },
-      shade: { description: 'Fill area below curves with translucent color' },
-    }),
-  ],
-
-  // ── Visualizations ───────────────────────────────────────
-  ['slope', withGlobals()],
-  [
-    'wordcloud',
-    withGlobals({
-      rotate: {
-        description: 'Word rotation',
-        values: ['none', 'mixed', 'angled'],
-      },
-      max: { description: 'Maximum word count' },
-      size: { description: 'Font size range (min, max)' },
-    }),
-  ],
-  [
-    'arc',
-    withGlobals({
-      order: {
-        description: 'Node ordering',
-        values: ['appearance', 'name', 'group', 'degree'],
-      },
-    }),
-  ],
-  [
-    'timeline',
-    withGlobals({
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'event-line',
-    withGlobals({
-      'no-scale': { description: 'Space events evenly instead of by date' },
-      side: {
-        description: 'Card placement: side above | below (default alternate)',
-      },
-      'no-box': {
-        description: 'Card-less label/rule/description style (slides)',
-      },
-      'no-legend': { description: 'Hide the tag legend' },
-      TBD: {
-        description:
-          'Date prefix for a not-yet-scheduled future event (e.g. "TBD Console Port")',
-      },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  ['venn', withGlobals()],
-  [
-    'quadrant',
-    withGlobals({
-      'x-label': { description: 'X-axis labels (low, high)' },
-      'y-label': { description: 'Y-axis labels (low, high)' },
-    }),
-  ],
-
-  // ── Diagrams ─────────────────────────────────────────────
-  [
-    'sequence',
-    withGlobals({
-      activations: {
-        description: 'Show activation bars',
-        values: ['on', 'off'],
-      },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'flowchart',
-    // Spec §5 §4.6: direction-lr, orientation-vertical, solid-fill, no-notes
-    withGlobals({
-      'direction-lr': { description: 'Switch to left-to-right layout' },
-      'orientation-vertical': {
-        description: 'Use vertical orientation for ranks',
-      },
-      'no-notes': { description: 'Suppress all node note boxes' },
-    }),
-  ],
-  ['class', withGlobals({})],
-  [
-    'er',
-    // Spec §9 §8.5: notation (chen/crow), active-tag.
-    withGlobals({
-      notation: {
-        description: 'ER notation style',
-        values: ['chen', 'crow'],
-      },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'org',
-    // Spec §7 §6.5: direction-tb, sub-node-label, show-sub-node-count,
-    // hide, active-tag. solid-fill via SOLID_FILL_CAPABLE.
-    withGlobals({
-      'direction-tb': { description: 'Switch to top-to-bottom layout' },
-      'sub-node-label': { description: 'Label for sub-nodes' },
-      'show-sub-node-count': { description: 'Show sub-node counts' },
-      hide: { description: 'Hide tag:value pairs' },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'kanban',
-    // Spec §11 §10.4: hide, active-tag.
-    withGlobals({
-      hide: { description: 'Hide tag:value pairs' },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  // RACI / RASCI / DACI — one chart type (`raci`); the variant (RACI, RASCI,
-  // or DACI) is inferred from the markers used (D → DACI, S → RASCI).
-  [
-    'raci',
-    withGlobals({
-      roles: {
-        description:
-          'Declare role column order (inline `roles A, B, C` or indented block with per-role pipe metadata)',
-      },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'c4',
-    // Spec §8 §7.7: direction-tb, active-tag.
-    withGlobals({
-      'direction-tb': { description: 'Switch to top-to-bottom layout' },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'state',
-    // Spec §6 §5.6: direction-tb, solid-fill, no-notes.
-    withGlobals({
-      'direction-tb': { description: 'Switch to top-to-bottom layout' },
-      'no-notes': { description: 'Suppress all state note boxes' },
-    }),
-  ],
-  [
-    'sitemap',
-    withGlobals({
-      'direction-tb': { description: 'Switch to top-to-bottom layout' },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'infra',
-    withGlobals({
-      'direction-tb': { description: 'Switch to top-to-bottom layout' },
-      animate: { description: 'Enable traffic animation' },
-      'default-latency-ms': { description: 'Default latency for all nodes' },
-      'default-uptime': { description: 'Default uptime for all nodes' },
-      'default-rps': { description: 'Default RPS capacity for all nodes' },
-      'slo-availability': { description: 'SLO availability target (0-1)' },
-      'slo-p90-latency-ms': { description: 'SLO p90 latency target in ms' },
-      'slo-warning-margin': { description: 'SLO warning margin percentage' },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'pert',
-    withGlobals({
-      'time-unit': {
-        description: 'Time unit for activity durations',
-        values: ['min', 'h', 'd', 'bd', 'w', 'm', 'q', 'y'],
-      },
-      confidence: {
-        description: 'Confidence factor for M-only durations',
-        values: ['high', 'medium', 'low'],
-      },
-      direction: { description: 'Layout direction', values: ['LR', 'TB'] },
-      'node-detail': {
-        description: 'Node visual density',
-        values: ['compact', 'full'],
-      },
-      trials: {
-        description:
-          'Monte Carlo trial count (auto-derived from activity count)',
-      },
-      seed: { description: 'Monte Carlo PRNG seed (auto-derived from title)' },
-      'scrubber-trials': {
-        description: 'Fast-MC trials for the duration scrubber (default 300)',
-      },
-      'start-date': { description: 'Project start date (YYYY-MM-DD or now)' },
-      'end-date': { description: 'Project end date (YYYY-MM-DD)' },
-      'sprint-length': { description: 'Sprint duration (e.g. 2w)' },
-      'sprint-number': { description: 'Starting sprint number' },
-      'sprint-start': { description: 'Sprint start date (YYYY-MM-DD)' },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'gantt',
-    // Spec §13 §12.2 Options.
-    withGlobals({
-      start: { description: 'Project start date (YYYY-MM-DD)' },
-      'today-marker': {
-        description: 'Today marker (bare = on, or YYYY-MM-DD date)',
-      },
-      sort: { description: 'Sort order', values: ['time', 'group', 'tag'] },
-      'critical-path': { description: 'Show critical path' },
-      'no-dependencies': { description: 'Hide dependency arrows' },
-      'sprint-length': { description: 'Sprint duration (e.g. 2w)' },
-      'sprint-number': { description: 'Starting sprint number' },
-      'sprint-start': { description: 'Sprint start date (YYYY-MM-DD)' },
-      'active-tag': { description: 'Active tag group name' },
-      // Legacy positive form `dependencies` — kept for back-compat. Use
-      // `no-dependencies` to suppress dependency arrows in new code.
-      dependencies: { description: 'Show dependencies (legacy form)' },
-    }),
-  ],
-  [
-    'boxes-and-lines',
-    withGlobals({
-      direction: { description: 'Layout direction', values: ['LR', 'TB'] },
-      'active-tag': { description: 'Active tag group name' },
-      hide: { description: 'Hide tag:value pairs' },
-      heat: {
-        description:
-          'Label for the value→colour ramp, with an optional trailing [low] [high] color pair (pairs with the `heat:` key)',
-      },
-      'show-values': { description: 'Print box values as text' },
-    }),
-  ],
-  [
-    'swimlane',
-    withGlobals({
-      direction: { description: 'Layout direction', values: ['LR', 'TB'] },
-      lane: { description: 'Declare a lane (row) with an optional color' },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'version-control',
-    withGlobals({
-      direction: {
-        description: 'Layout direction',
-        values: ['LR', 'TB', 'BT'],
-      },
-      merge: { description: 'Merge a branch into the active branch' },
-      'cherry-pick': { description: 'Copy a commit onto the active branch' },
-      rebase: { description: 'Replay a branch onto another (rebase X onto Y)' },
-      reset: { description: 'Move a branch pointer back (reset X to commit)' },
-      revert: { description: 'Add an inverse commit undoing a commit' },
-      ref: {
-        description:
-          'Pointer at a commit (ref origin/main at commit) — remotes / HEAD',
-      },
-      note: { description: 'Numbered step annotation on the current commit' },
-      'no-labels': { description: 'Hide commit messages' },
-      'no-lanes': { description: 'Hide branch lanes' },
-      'no-head': { description: 'Hide the HEAD marker' },
-    }),
-  ],
-  [
-    'mindmap',
-    withGlobals({
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'wireframe',
-    withGlobals({
-      mobile: { description: 'Use mobile (narrow vertical) layout' },
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'tech-radar',
-    // Spec §20 documents one directive: `show-blip-legend`. `rings` is a
-    // structural block keyword; quadrant/ring/trend/color are pipe metadata
-    // that live in PIPE_METADATA.
-    withGlobals({
-      'show-blip-legend': {
-        description: 'Render the four-column blip listing alongside the radar',
-      },
-    }),
-  ],
-  [
-    'cycle',
-    withGlobals({
-      'direction-counterclockwise': {
-        description: 'Reverse cycle direction to counterclockwise',
-      },
-      'circle-nodes': {
-        description: 'Render nodes as circles instead of rectangles',
-      },
-    }),
-  ],
-  [
-    'journey-map',
-    // Spec §22 directives: `active-tag`. `persona` is a
-    // structural keyword (like `tag` / `roles`), not a directive.
-    // `solid-fill` is added via SOLID_FILL_CAPABLE below.
-    withGlobals({
-      'active-tag': { description: 'Active tag group name' },
-    }),
-  ],
-  [
-    'pyramid',
-    // Spec §23.5 documents `inverted`; `solid-fill` is added via
-    // SOLID_FILL_CAPABLE below (working but not yet in spec §23.5).
-    // `color`/`description` are layer pipe-metadata, not directives.
-    withGlobals({
-      inverted: { description: 'Flip apex to the bottom (funnel orientation)' },
-    }),
-  ],
-  [
-    'ring',
-    // Per spec §24.5 the only chart-specific directive is `solid-fill`,
-    // applied via SOLID_FILL_CAPABLE below. `color`/`description` are
-    // layer pipe-metadata, not directives — they live in PIPE_METADATA.
-    withGlobals({}),
-  ],
-  [
-    'treemap',
-    // Hierarchy sized by a bare trailing number; color modes + opt-outs.
-    withGlobals({
-      heat: {
-        description:
-          'Name the color-by-value ramp (`heat <Label> [low] [high]`); pairs with the per-node `heat:` key',
-      },
-      depth: {
-        description:
-          'Render N levels; deeper subtrees collapse to a drillable block',
-      },
-      'no-values': { description: 'Hide value labels' },
-      'no-percent': { description: 'Hide percentage labels' },
-      'no-headers': { description: 'Hide parent header bars' },
-      'no-legend': { description: 'Hide the legend' },
-      radial: {
-        description:
-          'Render as a sunburst / hierarchical pie (concentric rings)',
-      },
-    }),
-  ],
-  [
-    'block',
-    // Author-controlled grid; columns are inferred from placement, so the only
-    // directives are an explicit `columns` override and `no-legend`.
-    withGlobals({
-      columns: {
-        description:
-          'Grid width (columns); inferred from the widest row if omitted',
-      },
-      'no-legend': { description: 'Hide the tag legend' },
-    }),
-  ],
-  [
-    'map',
-    // Geographic map directives (§24B.2/.7). Cosmetics are ON by default — the
-    // only switches are bare `no-*` opt-outs, surfaced proactively so a
-    // zero-config map still hints at what can be turned off. `poi`/`route` are
-    // content keywords, not directives; metadata keys (heat/size/width/label/style)
-    // live in the reserved-key registry.
-    withGlobals({
-      'region-heat': {
-        description:
-          'Label for the region value→colour ramp, with an optional trailing [low] [high] color pair (pairs with the `heat:` key)',
-      },
-      'poi-size': {
-        description:
-          'Label for the POI value→marker-size channel (pairs with the `size:` key)',
-      },
-      'flow-width': {
-        description:
-          'Label for the edge/leg value→line-thickness channel (pairs with the `width:` key)',
-      },
-      locale: {
-        description:
-          'Default country/state for bare place names, e.g. locale US-GA',
-      },
-      'active-tag': {
-        description: 'Which tag group leads when several are present',
-      },
-      caption: { description: 'Caption line (data-source attribution)' },
-      'no-title': { description: 'Suppress the title banner' },
-      'no-legend': { description: 'Suppress the legend' },
-      'no-coastline': {
-        description: 'Turn off coastal water-lines (on by default)',
-      },
-      'no-relief': {
-        description: 'Turn off mountain-range relief shading (on by default)',
-      },
-      'no-context-labels': {
-        description: 'Turn off orientation labels for water + nearby countries',
-      },
-      'no-region-labels': {
-        description: 'Turn off subdivision name labels (on by default)',
-      },
-      'no-region-heat-value': {
-        description:
-          'Turn off the heat value shown under each region (on by default)',
-      },
-      'no-poi-labels': { description: 'Turn off POI labels (on by default)' },
-      'no-colorize': {
-        description:
-          'Force plain green-land reference dress (regions are auto-coloured by default)',
-      },
-      'no-cities': {
-        description:
-          'Turn off the subtle city dots scattered across the basemap (on by default)',
-      },
-      'no-cluster-pois': {
-        description:
-          'Always fan out coincident POI markers instead of collapsing them into a count badge',
-      },
-    }),
-  ],
-]);
-
-// ── Cross-chart-type bare-keyword option: `solid-fill` ──────────
-// Adds the directive to every chart type whose renderer actually responds to
-// it (renders shapes at full intent saturation when set). Chart types where
-// the keyword is a no-op (gantt/infra/tech-radar/event-line deliberately opt
-// out; quadrant/scatter-dot/wordcloud/line/arc/slope have no tinted shape
-// fill) intentionally don't list it — keeps the completion popup honest.
-const SOLID_FILL_CAPABLE = new Set([
-  'flowchart',
-  'state',
-  'sequence',
-  'c4',
-  'org',
-  'kanban',
-  'journey-map',
-  'mindmap',
-  'cycle',
-  'pyramid',
-  'ring',
-  'funnel',
-  'class',
-  'er',
-  'sitemap',
-  'boxes-and-lines',
-  'wireframe',
-  'bar',
-  'pie',
-  'polar-area',
-  'radar',
-  'scatter',
-  'pert',
-  'block',
-  'swimlane',
-  'treemap',
-  'heatmap',
-  'venn',
-  'timeline',
-]);
-for (const [type, spec] of COMPLETION_REGISTRY) {
-  if (SOLID_FILL_CAPABLE.has(type)) {
-    spec.directives['solid-fill'] = {
-      description:
-        'Render shapes with full intent color instead of the default 25% tint',
-    };
-  }
-}
-
-// ============================================================
-// Chart types array (for chart type completion popup)
-// ============================================================
-
-/** All chart types with descriptions, for chart type autocomplete. */
-const CHART_TYPE_DESCRIPTIONS: Record<string, string> = Object.fromEntries(
-  chartTypes.map((c) => [c.id, c.description])
-);
-
-export const CHART_TYPES: ReadonlyArray<{ name: string; description: string }> =
-  [...ALL_CHART_TYPES].map((name) => ({
-    name,
-    description: CHART_TYPE_DESCRIPTIONS[name] ?? name,
-  }));
-
-// ============================================================
-// Entity types for `is a` declarations
-// ============================================================
-
-/**
- * Entity types for `Name is a <type>` declarations, keyed by chart type.
- * Values are sourced from parser constants (VALID_PARTICIPANT_TYPES,
- * C4_IS_A_RE).
- */
-export const ENTITY_TYPES = new Map<string, string[]>([
-  ['sequence', ['actor', 'database', 'queue', 'cache']],
-  [
-    'c4',
-    ['person', 'system', 'container', 'component', 'external', 'database'],
-  ],
-]);
-
-// ============================================================
-// Structural keywords for line-leading completion
-// ============================================================
-
-/**
- * Chart-type-specific structural keywords offered on an empty/start-of-line in
- * the data zone (block openers like `loop`, section headers like `containers`,
- * the `tag` block declaration, etc.). This is the single source of truth for
- * the editor's structural-keyword popup — every entry MUST be a token the
- * corresponding parser actually recognizes (validated by the
- * completion-conformance suite). Do NOT add removed/diagnostic-only tokens
- * (e.g. cycle's `no-descriptions`) or tokens the parser ignores.
- *
- * Chart types not listed here have no structural keywords (most data charts).
- */
-export const STRUCTURAL_KEYWORDS = new Map<string, string[]>([
-  ['sequence', ['if', 'else', 'loop', 'parallel', 'note', 'tag']],
-  ['gantt', ['era', 'marker', 'holiday', 'workweek', 'parallel', 'tag']],
-  ['c4', ['containers', 'components', 'deployment', 'tag']],
-  ['timeline', ['era', 'marker', 'tag']],
-  ['org', ['tag']],
-  ['kanban', ['tag']],
-  ['sitemap', ['tag']],
-  ['infra', ['tag']],
-  ['pert', ['tag']],
-  ['mindmap', ['tag']],
-  ['treemap', ['tag']],
-  ['block', ['tag']],
-  ['boxes-and-lines', ['tag']],
-  ['swimlane', ['lane', 'tag']],
-  [
-    'version-control',
-    ['merge', 'cherry-pick', 'rebase', 'reset', 'revert', 'ref', 'note'],
-  ],
-  ['er', ['tag']],
-  ['cycle', ['direction-counterclockwise', 'circle-nodes']],
-  ['journey-map', ['persona', 'tag']],
-  ['raci', ['roles']],
-  ['tech-radar', ['rings']],
-  [
-    'wireframe',
-    [
-      'nav',
-      'tabs',
-      'table',
-      'image',
-      'modal',
-      'skeleton',
-      'alert',
-      'progress',
-      'chart',
-      'mobile',
-      'tag',
-    ],
-  ],
-  ['class', ['abstract', 'interface', 'enum', 'extends', 'implements']],
-]);
-
-/**
- * Chart types that support `tag` block declarations (and thus the
- * `alias`/`default` sub-keywords inside a tag block). Derived from
- * STRUCTURAL_KEYWORDS so the two can never drift — a chart supports tag blocks
- * iff it offers the `tag` keyword.
- */
-export const TAG_SUPPORTING_TYPES: ReadonlySet<string> = new Set(
-  [...STRUCTURAL_KEYWORDS]
-    .filter(([, kws]) => kws.includes('tag'))
-    .map(([type]) => type)
-);
-
-// ============================================================
-// Reference grammar descriptor (slot 5 — entity reference)
-// ============================================================
-
-/** Whether a chart type has a reference position (arrow/operator → a prior
- *  declaration) and which operators open it. The single oracle that drives
- *  the library extractor audit, the app's entity-trigger map, and the
- *  conformance drift-guard — so the three can never disagree. v1 models
- *  reference-after-operator only (NOT metadata-target references). */
-export interface ReferenceGrammar {
-  hasReferenceGrammar: boolean;
-  /** Literal operator tokens that introduce a reference (e.g. `->`, `<->`,
-   *  `~>`, `+`). The app derives its trigger regex from these (allowing the
-   *  labeled `-label->` / `~label~>` variants where applicable). */
-  referenceOperators: string[];
-}
-
-/**
- * Per-type reference grammar. Grounded in each parser's arrow/operator
- * grammar — NOT hand-typed blind. Types with no reference position (org,
- * kanban, mindmap, all data charts, the radial/visualization types) declare
- * `hasReferenceGrammar: false`. `map` references (`route A ~> B`) are owned by
- * the app's bespoke geo-completion path, so it stays `false` here to avoid
- * double-handling.
- */
-export const REFERENCE_GRAMMAR = new Map<string, ReferenceGrammar>([
-  // Diagrams with a genuine reference position
-  ['sequence', { hasReferenceGrammar: true, referenceOperators: ['->', '~>'] }],
-  ['flowchart', { hasReferenceGrammar: true, referenceOperators: ['->'] }],
-  ['state', { hasReferenceGrammar: true, referenceOperators: ['->'] }],
-  ['sitemap', { hasReferenceGrammar: true, referenceOperators: ['->'] }],
-  [
-    'c4',
-    {
-      hasReferenceGrammar: true,
-      referenceOperators: ['->', '<->', '~>', '<~>'],
-    },
-  ],
-  // ER/class/infra reference grammars are irregular (cardinality / typed
-  // relationship operators / indented arrows); the app keeps a special-cased
-  // trigger for those, but they're still reference-grammar types here so the
-  // drift-guard covers their extractors.
-  ['er', { hasReferenceGrammar: true, referenceOperators: ['--'] }],
-  [
-    'class',
-    {
-      hasReferenceGrammar: true,
-      referenceOperators: ['-->', '--|>', '..|>', '*--', 'o--', '..>'],
-    },
-  ],
-  ['infra', { hasReferenceGrammar: true, referenceOperators: ['->'] }],
-  ['gantt', { hasReferenceGrammar: true, referenceOperators: ['->'] }],
-  ['pert', { hasReferenceGrammar: true, referenceOperators: ['->'] }],
-  ['arc', { hasReferenceGrammar: true, referenceOperators: ['->'] }],
-  ['sankey', { hasReferenceGrammar: true, referenceOperators: ['->', '--'] }],
-  [
-    'boxes-and-lines',
-    { hasReferenceGrammar: true, referenceOperators: ['->', '<->'] },
-  ],
-  // Venn references prior sets via the `+` intersection operator (not an arrow).
-  ['venn', { hasReferenceGrammar: true, referenceOperators: ['+'] }],
-]);
-
-// Every other registered chart type has no reference position.
-for (const type of ALL_CHART_TYPES) {
-  if (!REFERENCE_GRAMMAR.has(type)) {
-    REFERENCE_GRAMMAR.set(type, {
-      hasReferenceGrammar: false,
-      referenceOperators: [],
-    });
-  }
-}
-
-// ============================================================
-// Closed value enums (slot 7) sourced from parser constants
-// ============================================================
-
-/**
- * RACI marker alphabet (slot-7 enum in the role-assignment value position).
- * Consumed by editor completion. The chart type is always `raci`; the variant
- * (RACI / RASCI / DACI) is inferred from the markers used, so completion offers
- * the full union (R / A / S / C / I / D) sourced from `raci/variants.ts`'s
- * `ALL_MARKERS` — the marker set can never drift from the parser.
- */
-export const RACI_MARKER_ALPHABETS: ReadonlyMap<string, readonly string[]> =
-  new Map([['raci', [...ALL_MARKERS]]]);
-
-/**
- * Closed set of wireframe element state flags (slot-7 trailing enum, e.g.
- * `(Submit) primary destructive`). Sourced from the wireframe parser's
- * `STATE_KEYWORDS` — exported there and consumed here, never re-typed.
- */
-export const WIREFRAME_FLAGS: readonly string[] = [...WIREFRAME_STATE_KEYWORDS];
-
-// ============================================================
-// Pipe metadata for inline `| key value` on data lines
-// ============================================================
-
-/** Specification for a single pipe metadata key. */
-export interface PipeKeySpec {
-  description: string;
-  values?: string[];
-}
-
-/**
- * Pipe metadata keys for inline `| key value` on data lines.
- * Keyed by chart type → { context-name: keys }.
- *
- * Contexts are open-ended. The two universal ones are:
- *   - `node` — the default for any non-arrow line
- *   - `edge` — lines containing an arrow (`->`, `--`)
- *
- * Charts with richer line types declare additional contexts:
- *   - raci: `role`, `phase`, `assignment`
- *   - ring / pyramid: `layer`
- *   - tech-radar: `quadrant`, `blip`
- *   - journey-map: `step`
- *
- * IMPORTANT: NEVER add 'sequence' here. The `|` character in sequence
- * diagrams separates display names from identifiers and tag metadata.
- * Adding sequence would trigger false pipe-metadata completions on every `|`.
- */
-export type PipeContextMap = Record<string, Record<string, PipeKeySpec>>;
-
-export const PIPE_METADATA = new Map<string, PipeContextMap>([
-  [
-    'infra',
-    {
-      node: {
-        description: { description: 'Node description text' },
-        instances: {
-          description: 'Instance count or auto-scaling range (N-M)',
-        },
-        'latency-ms': { description: 'Per-request latency in milliseconds' },
-        'max-rps': { description: 'Max requests per second per instance' },
-        'cache-hit': { description: 'Cache hit percentage (0-100)' },
-        'firewall-block': { description: 'Traffic blocked percentage' },
-        'ratelimit-rps': { description: 'Max RPS allowed through' },
-        'cb-error-threshold': {
-          description: 'Circuit breaker error threshold %',
-        },
-        'cb-latency-threshold-ms': {
-          description: 'Circuit breaker latency threshold',
-        },
-        uptime: { description: 'Component availability (0-1)' },
-        concurrency: { description: 'Concurrent request limit' },
-        'duration-ms': { description: 'Processing duration' },
-        'cold-start-ms': { description: 'Function cold-start time' },
-        buffer: { description: 'Queue/buffer capacity' },
-        'drain-rate': { description: 'Queue drain rate' },
-        'retention-hours': { description: 'Data retention period' },
-        partitions: { description: 'Queue/stream partition count' },
-        'slo-availability': { description: 'Node availability target (0-1)' },
-        'slo-p90-latency-ms': { description: 'Node p90 latency target' },
-        'slo-warning-margin': { description: 'Node SLO warning margin' },
-      },
-      edge: {
-        split: { description: 'Traffic split percentage (e.g., 60%)' },
-        fanout: { description: 'Fanout multiplier (integer >= 1)' },
-      },
-    },
-  ],
-  [
-    'c4',
-    {
-      node: {
-        description: { description: 'Element description' },
-        tech: { description: 'Technology stack' },
-        technology: { description: 'Technology stack (alias for tech)' },
-      },
-      edge: {},
-    },
-  ],
-  [
-    'gantt',
-    {
-      node: {
-        duration: { description: 'Task duration (e.g., 30bd, 5d, 1.5w)' },
-        start: { description: 'Explicit start date (e.g., 2024-01-15)' },
-        progress: { description: 'Task progress (0–100)' },
-        offset: { description: 'Task start offset (e.g., 8bd, -3bd)' },
-      },
-      edge: {
-        // Gantt "edge" = dependency arrow (TaskA -> TaskB | offset 2bd)
-        offset: { description: 'Dependency offset (e.g., 2bd, -1w)' },
-      },
-    },
-  ],
-  [
-    'boxes-and-lines',
-    {
-      node: {
-        description: { description: 'Node description text' },
-        heat: { description: 'Numeric value for the heat (colour) ramp' },
-      },
-      edge: {},
-    },
-  ],
-  [
-    'mindmap',
-    {
-      node: {
-        description: { description: 'Node description text' },
-        collapsed: { description: 'Collapse node subtree by default' },
-      },
-      edge: {},
-    },
-  ],
-  [
-    // Tech-radar pipe metadata (spec §20). Two contexts:
-    //   - quadrant: top-level quadrant headers (`Tools | quadrant: top-left, color: blue`)
-    //   - blip: indented blip lines (`  Vite | ring: Adopt, trend: up`)
-    'tech-radar',
-    {
-      quadrant: {
-        quadrant: {
-          description: 'Quadrant position',
-          values: ['top-left', 'top-right', 'bottom-left', 'bottom-right'],
-        },
-        color: { description: 'Override quadrant color' },
-      },
-      blip: {
-        ring: { description: 'Ring assignment (must match a declared ring)' },
-        trend: {
-          description: 'Blip trend indicator',
-          values: ['new', 'up', 'down', 'stable'],
-        },
-      },
-    },
-  ],
-  [
-    'cycle',
-    {
-      node: {
-        color: { description: 'Node fill color (palette name)' },
-        span: { description: 'Relative arc distance to next node' },
-        description: { description: 'Node description text' },
-      },
-      edge: {
-        color: { description: 'Edge stroke color (palette name)' },
-        width: { description: 'Edge stroke width in pixels' },
-      },
-    },
-  ],
-  [
-    // RACI pipe metadata (spec §24A). Two contexts:
-    //   - role: declarations inside the `roles` block (`Cap | color: red`)
-    //   - phase: bracketed phase headers (`[Departure] | color: teal`)
-    'raci',
-    {
-      role: {
-        color: { description: 'Role column tint (palette name)' },
-      },
-      phase: {
-        color: { description: 'Phase bar tint (palette name)' },
-      },
-    },
-  ],
-  [
-    // Ring layer pipe metadata (spec §24.4). One context: `layer`.
-    'ring',
-    {
-      layer: {
-        color: { description: 'Ring color (palette name)' },
-        description: { description: 'Layer description (one-liner shorthand)' },
-      },
-    },
-  ],
-  [
-    // Pyramid layer pipe metadata (spec §23.4). Identical surface to ring.
-    'pyramid',
-    {
-      layer: {
-        color: { description: 'Layer color (palette name)' },
-        description: { description: 'Layer description (one-liner shorthand)' },
-      },
-    },
-  ],
-  [
-    // Journey-map step pipe metadata (spec §22).
-    // Tag aliases (e.g. `ch: Web`) are user-defined via the `tag` block
-    // and resolved dynamically — not part of the static map.
-    'journey-map',
-    {
-      step: {
-        score: { description: 'Step score (1–5 integer; high = good)' },
-        emotion: { description: 'Emotional state at this step' },
-        description: { description: 'Step description text' },
-        pain: { description: 'Pain point at this step' },
-        opportunity: { description: 'Opportunity at this step' },
-        thought: { description: 'User thought at this step' },
-      },
-    },
-  ],
-  [
-    'org',
-    {
-      node: {
-        description: { description: 'Person/team description' },
-        role: { description: 'Role or job title' },
-        location: { description: 'Office location' },
-        email: { description: 'Email address' },
-        phone: { description: 'Phone number' },
-      },
-    },
-  ],
-  [
-    'er',
-    {
-      node: {
-        description: { description: 'Entity description' },
-        domain: { description: 'Domain grouping' },
-      },
-    },
-  ],
-  [
-    'class',
-    {
-      node: {
-        description: { description: 'Class description' },
-      },
-    },
-  ],
-  [
-    'kanban',
-    {
-      node: {
-        description: { description: 'Card description' },
-        assignee: { description: 'Assigned person' },
-        due: { description: 'Due date (YYYY-MM-DD)' },
-      },
-    },
-  ],
-  [
-    'sitemap',
-    {
-      node: {
-        description: { description: 'Page description' },
-        status: { description: 'Page status' },
-      },
-    },
-  ],
-  [
-    'pert',
-    {
-      node: {
-        description: { description: 'Activity description' },
-        confidence: {
-          description: 'Confidence factor',
-          values: ['high', 'medium', 'low'],
-        },
-        collapsed: { description: 'Collapse activity detail' },
-      },
-    },
-  ],
-  [
-    'timeline',
-    {
-      node: {
-        description: { description: 'Event description' },
-        duration: { description: 'Event duration (e.g., 30d, 1.5y)' },
-      },
-    },
-  ],
-]);
-
-// ============================================================
-// Derived metadata key set
-// ============================================================
-
-/** All known directive keys, derived from COMPLETION_REGISTRY. Includes implicit keys. */
-export const METADATA_KEY_SET: ReadonlySet<string> = new Set([
-  'chart',
-  'title', // implicit directives recognized as metadata
-  ...[...COMPLETION_REGISTRY.values()].flatMap((spec) =>
-    Object.keys(spec.directives)
-  ),
-]);
-
-// ============================================================
-// Sequence extractor
-// ============================================================
-
-// Universal Name Handling: source/target accept multi-word + "quoted" names.
-// `[^|]+?` captures greedy-to-pipe-or-arrow; the arrow alternation acts as
-// the boundary. Caller strips quotes via stripQuotes().
-const SEQ_ARROW_RE =
-  /^(?:"([^"]+)"|([^|"]+?))\s+(->|-.*->|~>|~.*~>)\s+(?:"([^"]+)"|([^|"]+?))(?:\s|\|.*)?$/;
-const SEQ_IS_A_RE = /^(?:"([^"]+)"|([^|":]+?))\s+is\s+an?\s+/i;
-const SEQ_SECTION_RE = /^==/;
-const SEQ_STRUCTURAL_RE = /^(if|else|loop|parallel|end)\b/i;
-
-function extractSequenceSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    // Skip first line (chart type)
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    // Skip metadata lines
-    // split(/\s+/) on non-empty `trimmed` always yields at least one element.
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    // Skip sections, structural keywords
-    if (SEQ_SECTION_RE.test(trimmed)) continue;
-    if (SEQ_STRUCTURAL_RE.test(trimmed)) continue;
-
-    // Arrow lines: A -> B, A -label-> B, A ~> B
-    const arrowMatch = trimmed.match(SEQ_ARROW_RE);
-    if (arrowMatch) {
-      const src = (arrowMatch[1] ?? arrowMatch[2] ?? '').trim();
-      const dst = (arrowMatch[4] ?? arrowMatch[5] ?? '').trim();
-      if (src && !entities.includes(src)) entities.push(src);
-      if (dst && !entities.includes(dst)) entities.push(dst);
-      continue;
-    }
-
-    // Type declarations: A is a person, A is an actor
-    const isAMatch = trimmed.match(SEQ_IS_A_RE);
-    if (isAMatch) {
-      const name = (isAMatch[1] ?? isAMatch[2] ?? '').trim();
-      if (name && !entities.includes(name)) entities.push(name);
-      continue;
-    }
-  }
-
-  return {
-    kind: 'sequence',
-    entities,
-  };
-}
-
-// ============================================================
-// State extractor
-// ============================================================
-
-const STATE_ARROW_RE = /^(\S+)\s+->\s+(\S+)/;
-
-function extractStateSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    // Skip metadata lines
-    // split(/\s+/) on non-empty `trimmed` always yields at least one element.
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    const arrowMatch = trimmed.match(STATE_ARROW_RE);
-    if (arrowMatch) {
-      // Regex captured groups 1 and 2 by successful match; split('|')[0] always defined.
-      const src = arrowMatch[1]!.split('|')[0]!.trim();
-      const dst = arrowMatch[2]!.split('|')[0]!.trim();
-      if (src && !entities.includes(src)) entities.push(src);
-      if (dst && !entities.includes(dst)) entities.push(dst);
-    }
-  }
-
-  return { kind: 'state', entities };
-}
-
-// ============================================================
-// Tag declaration extraction
-// ============================================================
-
-// Matches tag declarations in both forms:
-// - `tag Name alias x` (explicit alias keyword)
-// - `tag Name x` (shorthand: 1-4 lowercase chars = alias, matching parser's isAliasToken)
-const TAG_DECL_EXPLICIT_RE = /^tag\s+(\S+)\s+alias\s+(\S+)/i;
-const TAG_DECL_SHORT_RE = /^tag\s+(\S+)\s+([a-z]{1,4})(?:\s|$)/;
-
-/**
- * Extract tag declarations from document text.
- * Returns a map of alias (or full name) → array of tag values.
- * Keys preserve original case for display; use case-insensitive lookup.
- */
-export function extractTagDeclarations(docText: string): Map<string, string[]> {
-  const result = new Map<string, string[]>();
-  const lines = docText.split('\n');
-  let currentAlias: string | null = null;
-  let currentValues: string[] = [];
-
-  for (let i = 0; i < lines.length; i++) {
-    // In-bounds by loop guard.
-    const raw = lines[i]!;
-    const trimmed = raw.trim();
-
-    // Check for tag declaration — try explicit `alias` keyword first, then shorthand
-    const tagMatch =
-      trimmed.match(TAG_DECL_EXPLICIT_RE) ?? trimmed.match(TAG_DECL_SHORT_RE);
-    if (tagMatch) {
-      // Save previous tag group
-      if (currentAlias !== null) {
-        result.set(currentAlias, currentValues);
-      }
-      // Both regexes capture groups 1 (and 2 for explicit) on successful match.
-      const name = tagMatch[1]!;
-      const alias = tagMatch[2] ?? name;
-      currentAlias = alias;
-      currentValues = [];
-      continue;
-    }
-    // Also match bare `tag Name` (no alias) — fall through with name as key
-    if (/^tag\s+(\S+)\s*$/i.test(trimmed)) {
-      if (currentAlias !== null) {
-        result.set(currentAlias, currentValues);
-      }
-      // Regex captured group 1 by successful re-match (test passed above).
-      currentAlias = trimmed.match(/^tag\s+(\S+)/i)![1]!;
-      currentValues = [];
-      continue;
-    }
-
-    // Collect indented tag values
-    if (
-      currentAlias !== null &&
-      raw.length > 0 &&
-      (raw[0] === ' ' || raw[0] === '\t')
-    ) {
-      if (trimmed && !trimmed.startsWith('//')) {
-        // Strip trailing-token color (§1.5): `Frontend blue` → `Frontend`.
-        // Whitespace-split; if the last token is a recognized color word,
-        // drop it; otherwise the whole trimmed string is the value.
-        const lastSpaceIdx = trimmed.lastIndexOf(' ');
-        const value =
-          lastSpaceIdx > 0 &&
-          RECOGNIZED_COLOR_SET.has(trimmed.substring(lastSpaceIdx + 1))
-            ? trimmed.substring(0, lastSpaceIdx).trim()
-            : trimmed;
-        if (value) currentValues.push(value);
-      }
-      continue;
-    }
-
-    // Non-indented non-tag line ends the current tag block
-    if (currentAlias !== null && trimmed) {
-      result.set(currentAlias, currentValues);
-      currentAlias = null;
-      currentValues = [];
-    }
-  }
-
-  // Save last tag group
-  if (currentAlias !== null) {
-    result.set(currentAlias, currentValues);
-  }
-
-  return result;
-}
-
-// ============================================================
-// Universal alias extractor (`Name as <alias>` postfix)
-// ============================================================
-
-// Postfix-alias form on any name-slot line. Caller-agnostic — runs
-// over the full document so every chart-type extractor can populate
-// `DiagramSymbols.aliases` consistently.
-const ALIAS_POSTFIX_DECL_RE =
-  /(?:^|[^|/])\s*(.+?)\s+as\s+([A-Za-z][A-Za-z0-9_]{0,11})\s*(?:\|.*)?$/;
-
-/**
- * Scan document text for `Name as <alias>` declarations.
- *
- * Returns `Record<alias, canonical-fragment>`. The canonical may
- * still carry color/type modifiers (the per-parser logic peels
- * those off at parse time); for autocomplete display purposes the
- * raw fragment is good enough.
- *
- * Pure helper — does NOT enforce strict-ordering, collisions, or
- * other semantic rules. Those are parser-side checks.
- */
-export function extractAliasDeclarations(
-  docText: string
-): Record<string, string> {
-  const aliases: Record<string, string> = {};
-  for (const raw of docText.split('\n')) {
-    const trimmed = raw.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-    const match = trimmed.match(ALIAS_POSTFIX_DECL_RE);
-    if (!match) continue;
-    // Regex captured groups 1 and 2 by successful match.
-    const canonical = match[1]!.trim();
-    const alias = match[2]!;
-    // Skip if canonical itself looks structural (arrow / pipe-only / brackets-only)
-    if (!canonical || canonical === '[' || canonical === ']') continue;
-    if (!(alias in aliases)) {
-      aliases[alias] = canonical;
-    }
-  }
-  return aliases;
-}
-
-// ============================================================
-// Sitemap extractor
-// ============================================================
-
-const SITEMAP_CONTAINER_RE = /^\[([^\]]+)\]/;
-const SITEMAP_ARROW_RE = /^-.*->\s*(.+)$/;
-const SITEMAP_BARE_ARROW_RE = /^->\s*(.+)$/;
-const SITEMAP_METADATA_RE = /^([^:]+):\s*(.+)$/;
-
-function extractSitemapSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-  let inTagBlock = false;
-  let lastNodeIndent = -1;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    // Skip metadata lines
-    // split(/\s+/) on non-empty `trimmed` always yields at least one element.
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    // Track tag blocks
-    if (/^tag\s+/i.test(trimmed)) {
-      inTagBlock = true;
-      continue;
-    }
-    const indent = line.search(/\S/);
-    if (inTagBlock) {
-      if (indent > 0) continue;
-      inTagBlock = false;
-    }
-
-    // Containers: [GroupName]
-    const containerMatch = trimmed.match(SITEMAP_CONTAINER_RE);
-    if (containerMatch) {
-      // Regex captured group 1 by successful match; split('|')[0] always defined.
-      const name = containerMatch[1]!.split('|')[0]!.trim();
-      if (name && !entities.includes(name)) entities.push(name);
-      lastNodeIndent = indent;
-      continue;
-    }
-
-    // Arrows: -> Target or -label-> Target
-    const bareArrow = trimmed.match(SITEMAP_BARE_ARROW_RE);
-    const labeledArrow = !bareArrow ? trimmed.match(SITEMAP_ARROW_RE) : null;
-    if (bareArrow || labeledArrow) {
-      // split('|')[0] always defined on any string.
-      const target = (bareArrow?.[1] ?? labeledArrow?.[1] ?? '')
-        .split('|')[0]!
-        .trim();
-      if (target && !entities.includes(target)) entities.push(target);
-      continue;
-    }
-
-    // Indented metadata under a node (key: value) — skip
-    if (
-      indent > 0 &&
-      lastNodeIndent >= 0 &&
-      indent > lastNodeIndent &&
-      SITEMAP_METADATA_RE.test(trimmed)
-    ) {
-      continue;
-    }
-
-    // Page label (anything else that's not special)
-    // split('|')[0] always defined on any string.
-    const label = trimmed.split('|')[0]!.trim();
-    if (label) {
-      if (!entities.includes(label)) entities.push(label);
-      lastNodeIndent = indent;
-    }
-  }
-
-  return { kind: 'sitemap', entities };
-}
-
-// ============================================================
-// C4 extractor
-// ============================================================
-
-const C4_ELEMENT_RE = /^(person|system|container|component)\s+(.+)$/i;
-const C4_IS_A_RE =
-  /^(.+?)\s+is\s+an?\s+(person|system|container|component|external|database)\b/i;
-const C4_ARROW_RE =
-  /^(\S+)\s+(?:->|-.*->|~>|~.*~>|<->|<-.*->|<~>|<~.*~>)\s+(\S+)/;
-const C4_SECTION_RE = /^(containers|components|deployment)\s*$/i;
-
-function extractC4Symbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-  let inTagBlock = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    // split(/\s+/) on non-empty `trimmed` always yields at least one element.
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    if (/^tag\s+/i.test(trimmed)) {
-      inTagBlock = true;
-      continue;
-    }
-    const indent = line.search(/\S/);
-    if (inTagBlock) {
-      if (indent > 0) continue;
-      inTagBlock = false;
-    }
-
-    // Skip section headers
-    if (C4_SECTION_RE.test(trimmed)) continue;
-
-    // Element declaration: person Name, system Name, etc.
-    const elemMatch = trimmed.match(C4_ELEMENT_RE);
-    if (elemMatch) {
-      // Regex captured group 2 by successful match; split('|')[0] always defined.
-      const name = elemMatch[2]!.split('|')[0]!.trim();
-      if (name && !entities.includes(name)) entities.push(name);
-      continue;
-    }
-
-    // Is-a declaration: Name is a person
-    const isAMatch = trimmed.match(C4_IS_A_RE);
-    if (isAMatch) {
-      // Regex captured group 1 by successful match; split('|')[0] always defined.
-      const name = isAMatch[1]!.split('|')[0]!.trim();
-      if (name && !entities.includes(name)) entities.push(name);
-      continue;
-    }
-
-    // Arrow lines: Source -> Target, Source ~> Target, etc.
-    const arrowMatch = trimmed.match(C4_ARROW_RE);
-    if (arrowMatch) {
-      // Regex captured groups 1 and 2 by successful match; split('|')[0] always defined.
-      const src = arrowMatch[1]!.split('|')[0]!.trim();
-      const dst = arrowMatch[2]!.split('|')[0]!.trim();
-      if (src && !entities.includes(src)) entities.push(src);
-      if (dst && !entities.includes(dst)) entities.push(dst);
-      continue;
-    }
-  }
-
-  return {
-    kind: 'c4',
-    entities,
-  };
-}
-
-// ============================================================
-// Gantt extractor
-// ============================================================
-
-const GANTT_LEGACY_DURATION_RE =
-  /^(\d+(?:\.\d+)?)(min|bd|d|w|m|q|y|h|s)\??\s+(.+)$/;
-const GANTT_LEGACY_DATE_RE = /^(\d{4}-\d{2}-\d{2}(?:\s\d{2}:\d{2})?)\s+(.+)$/;
-const GANTT_GROUP_RE = /^\[(.+?)\]/;
-const GANTT_STRUCTURAL_RE = /^(era|marker|holiday|workweek|parallel)\b/i;
-const GANTT_META_KEY_RE = /\b(?:duration|start):\s/;
-
-function extractGanttSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-  let inTagBlock = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    // split(/\s+/) on non-empty `trimmed` always yields at least one element.
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    if (/^tag\s+/i.test(trimmed)) {
-      inTagBlock = true;
-      continue;
-    }
-    const indent = line.search(/\S/);
-    if (inTagBlock) {
-      if (indent > 0) continue;
-      inTagBlock = false;
-    }
-
-    // Groups: [GroupName]
-    const groupMatch = trimmed.match(GANTT_GROUP_RE);
-    if (groupMatch) {
-      const name = groupMatch[1]!.trim();
-      if (name && !entities.includes(name)) entities.push(name);
-      continue;
-    }
-
-    // New syntax: Task Name duration: 5d or Task Name start: 2024-01-15
-    // (checked before structural keyword skip so "Era of Innovation duration: 5d" isn't skipped)
-    if (GANTT_META_KEY_RE.test(trimmed)) {
-      const cutIdx = trimmed.search(
-        /\b(?:duration|start|progress|offset|color|description):\s/
-      );
-      if (cutIdx > 0) {
-        let taskName = trimmed.substring(0, cutIdx).trim();
-        const arrowIdx = taskName.indexOf('->');
-        if (arrowIdx > 0)
-          taskName = taskName
-            .substring(0, arrowIdx)
-            .replace(/-[^>]*$/, '')
-            .trim();
-        if (taskName && !entities.includes(taskName)) entities.push(taskName);
-        continue;
-      }
-    }
-
-    // Skip structural keywords (after new-syntax check so "Era of Innovation duration: 5d" isn't skipped)
-    if (GANTT_STRUCTURAL_RE.test(trimmed)) continue;
-
-    // Legacy: Tasks by duration: 30d Task Name
-    const durMatch = trimmed.match(GANTT_LEGACY_DURATION_RE);
-    if (durMatch) {
-      let taskName = durMatch[3]!.split('|')[0]!.trim();
-      const arrowIdx = taskName.indexOf('->');
-      if (arrowIdx > 0)
-        taskName = taskName
-          .substring(0, arrowIdx)
-          .replace(/-[^>]*$/, '')
-          .trim();
-      if (taskName && !entities.includes(taskName)) entities.push(taskName);
-      continue;
-    }
-
-    // Legacy: Tasks by date: 2024-01-15 Task Name
-    const dateMatch = trimmed.match(GANTT_LEGACY_DATE_RE);
-    if (dateMatch) {
-      let taskName = dateMatch[2]!.split('|')[0]!.trim();
-      const arrowIdx = taskName.indexOf('->');
-      if (arrowIdx > 0)
-        taskName = taskName
-          .substring(0, arrowIdx)
-          .replace(/-[^>]*$/, '')
-          .trim();
-      if (taskName && !entities.includes(taskName)) entities.push(taskName);
-      continue;
-    }
-  }
-
-  return { kind: 'gantt', entities };
-}
-
-// ============================================================
-// Boxes-and-lines extractor
-// ============================================================
-
-const BL_ARROW_RE = /^(\S+)\s+(?:-.*)?(?:->|<->)\s+(\S+)/;
-
-function extractBoxesAndLinesSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-  let inTagBlock = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    // split(/\s+/) on non-empty `trimmed` always yields at least one element.
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    if (/^tag\s+/i.test(trimmed)) {
-      inTagBlock = true;
-      continue;
-    }
-    const indent = line.search(/\S/);
-    if (inTagBlock) {
-      if (indent > 0) continue;
-      inTagBlock = false;
-    }
-
-    // Skip groups
-    if (/^\[.+?\]/.test(trimmed)) continue;
-
-    // Edge lines
-    const arrowMatch = trimmed.match(BL_ARROW_RE);
-    if (arrowMatch) {
-      // Regex captured groups 1 and 2 by successful match; split('|')[0] always defined.
-      const src = arrowMatch[1]!.split('|')[0]!.trim();
-      const dst = arrowMatch[2]!.split('|')[0]!.trim();
-      if (src && !entities.includes(src)) entities.push(src);
-      if (dst && !entities.includes(dst)) entities.push(dst);
-      continue;
-    }
-
-    // Node lines
-    // split('|')[0] and chained split('[')[0] always defined on any string.
-    const label = trimmed.split('|')[0]!.split('[')[0]!.trim();
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return { kind: 'boxes-and-lines', entities };
-}
-
-// ============================================================
-// Org extractor
-// ============================================================
-
-const ORG_GROUP_RE = /^\[(.+?)\]/;
-
-function extractOrgSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-  let inTagBlock = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    if (/^tag\s+/i.test(trimmed)) {
-      inTagBlock = true;
-      continue;
-    }
-    const indent = line.search(/\S/);
-    if (inTagBlock) {
-      if (indent > 0) continue;
-      inTagBlock = false;
-    }
-
-    // Team/group headers: [Team Name]
-    const groupMatch = trimmed.match(ORG_GROUP_RE);
-    if (groupMatch) {
-      const name = groupMatch[1]!.split('|')[0]!.trim();
-      if (name && !entities.includes(name)) entities.push(name);
-      continue;
-    }
-
-    // Skip indented metadata lines (key: value)
-    if (indent > 0 && /^[a-z]+\s*:/.test(trimmed)) continue;
-
-    // Person name (indent 0 or direct child)
-    const label = trimmed.split('|')[0]!.trim();
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return { kind: 'org', entities };
-}
-
-// ============================================================
-// Kanban extractor
-// ============================================================
-
-const KANBAN_COLUMN_RE = /^\[(.+?)\]/;
-
-function extractKanbanSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-  let inTagBlock = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    if (/^tag\s+/i.test(trimmed)) {
-      inTagBlock = true;
-      continue;
-    }
-    const indent = line.search(/\S/);
-    if (inTagBlock) {
-      if (indent > 0) continue;
-      inTagBlock = false;
-    }
-
-    // Column headers: [Column Name]
-    const colMatch = trimmed.match(KANBAN_COLUMN_RE);
-    if (colMatch) {
-      const name = colMatch[1]!.split('|')[0]!.trim();
-      if (name && !entities.includes(name)) entities.push(name);
-      continue;
-    }
-
-    // Card names (indented under columns)
-    if (indent > 0) {
-      const label = trimmed.split('|')[0]!.trim();
-      if (label && !entities.includes(label)) entities.push(label);
-    }
-  }
-
-  return { kind: 'kanban', entities };
-}
-
-// ============================================================
-// Mindmap extractor
-// ============================================================
-
-function extractMindmapSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-  let inTagBlock = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    if (/^tag\s+/i.test(trimmed)) {
-      inTagBlock = true;
-      continue;
-    }
-    const indent = line.search(/\S/);
-    if (inTagBlock) {
-      if (indent > 0) continue;
-      inTagBlock = false;
-    }
-
-    // Skip indented metadata (description:, collapsed:)
-    if (/^(description|collapsed)\s*:/i.test(trimmed)) continue;
-
-    // Node name (at any indent level)
-    const label = trimmed.split('|')[0]!.trim();
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return { kind: 'mindmap', entities };
-}
-
-// ============================================================
-// Treemap extractor
-// ============================================================
-
-function extractTreemapSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-  let inTagBlock = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    // Directives and tag blocks are not node entities.
-    if (
-      /^(depth|heat|no-[a-z]+)\s/i.test(trimmed) ||
-      /^(no-[a-z]+|radial)$/i.test(trimmed)
-    )
-      continue;
-    if (/^tag\s+/i.test(trimmed)) {
-      inTagBlock = true;
-      continue;
-    }
-    const indent = line.search(/\S/);
-    if (inTagBlock) {
-      if (indent > 0) continue;
-      inTagBlock = false;
-    }
-
-    // Node name: strip same-line metadata + the bare trailing value number.
-    let label = trimmed.split(/\s+\w+:/)[0]!.trim();
-    label = label.replace(/\s+-?\d[\d_,.]*$/, '').trim();
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return { kind: 'treemap', entities };
-}
-
-/** Block-diagram symbols: every `[Label]` on a grid row (skips tag blocks and
- *  the `columns` / `no-*` directives). */
-function extractBlockSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-  let inTagBlock = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-    if (/^(columns|no-[a-z]+)\b/i.test(trimmed)) continue;
-    if (/^tag\s+/i.test(trimmed)) {
-      inTagBlock = true;
-      continue;
-    }
-    const indent = line.search(/\S/);
-    if (inTagBlock) {
-      if (indent > 0) continue;
-      inTagBlock = false;
-    }
-    const re = /\[([^\]]*)\]/g;
-    let m: RegExpExecArray | null;
-    while ((m = re.exec(trimmed)) !== null) {
-      const label = m[1]!.trim();
-      if (label && !entities.includes(label)) entities.push(label);
-    }
-  }
-
-  return { kind: 'block', entities };
-}
-
-// ============================================================
-// Pyramid extractor
-// ============================================================
-
-function extractPyramidSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-    if (firstToken === 'inverted' || firstToken === 'solid-fill') continue;
-
-    // Skip indented description lines
-    if (line[0] === ' ' || line[0] === '\t') continue;
-
-    // Layer name (strip pipe metadata)
-    const label = trimmed.split('|')[0]!.trim();
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return { kind: 'pyramid', entities };
-}
-
-// ============================================================
-// Ring extractor
-// ============================================================
-
-function extractRingSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-    if (firstToken === 'solid-fill') continue;
-
-    // Skip indented description lines
-    if (line[0] === ' ' || line[0] === '\t') continue;
-
-    // Layer name (strip pipe metadata)
-    const label = trimmed.split('|')[0]!.trim();
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return { kind: 'ring', entities };
-}
-
-// ============================================================
-// Arc extractor
-// ============================================================
-
-const ARC_ARROW_RE = /^(\S+)\s+(?:->|-[^>]*->)\s+(\S+)/;
-
-function extractArcSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    const arrowMatch = trimmed.match(ARC_ARROW_RE);
-    if (arrowMatch) {
-      const src = arrowMatch[1]!.split('|')[0]!.trim();
-      const dst = arrowMatch[2]!.split('|')[0]!.trim();
-      if (src && !entities.includes(src)) entities.push(src);
-      if (dst && !entities.includes(dst)) entities.push(dst);
-    }
-  }
-
-  return { kind: 'arc', entities };
-}
-
-// ============================================================
-// Sankey extractor
-// ============================================================
-
-// Sankey links accept both directed `->` and undirected `--` (echarts.ts §1.5).
-const SANKEY_ARROW_RE = /^(.+?)\s+(?:->|--)\s+(.+?)\s+(\d[\d,_.]*)/;
-
-function extractSankeySymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    const arrowMatch = trimmed.match(SANKEY_ARROW_RE);
-    if (arrowMatch) {
-      const src = arrowMatch[1]!.split('|')[0]!.trim();
-      const dst = arrowMatch[2]!.split('|')[0]!.trim();
-      if (src && !entities.includes(src)) entities.push(src);
-      if (dst && !entities.includes(dst)) entities.push(dst);
-    } else {
-      // Standalone node declaration (just a name, possibly with color)
-      const label = trimmed.split('|')[0]!.trim();
-      if (label && !entities.includes(label)) entities.push(label);
-    }
-  }
-
-  return { kind: 'sankey', entities };
-}
-
-// ============================================================
-// Timeline extractor
-// ============================================================
-
-const TIMELINE_ERA_RE = /^era\s+/i;
-const TIMELINE_MARKER_RE = /^marker\s+/i;
-
-const TIMELINE_SCHEDULING_RE = /\b(?:start|end|duration)\s*:/;
-
-function extractTimelineSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-  let inTagBlock = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-    if (TIMELINE_ERA_RE.test(trimmed) || TIMELINE_MARKER_RE.test(trimmed))
-      continue;
-
-    if (/^tag\s+/i.test(trimmed)) {
-      inTagBlock = true;
-      continue;
-    }
-    const indent = line.search(/\S/);
-    if (inTagBlock) {
-      if (indent > 0) continue;
-      inTagBlock = false;
-    }
-
-    let label: string;
-    if (TIMELINE_SCHEDULING_RE.test(trimmed)) {
-      label = trimmed
-        .replace(/\b(?:start|end|duration|color|description)\s*:.*$/, '')
-        .split('|')[0]!
-        .trim();
-    } else {
-      label = trimmed
-        .replace(
-          /^\d{4}(?:-\d{2}(?:-\d{2})?)?\s*(?:->\s*\d{4}(?:-\d{2}(?:-\d{2})?)?)?\s*/,
-          ''
-        )
-        .split('|')[0]!
-        .trim();
-    }
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return { kind: 'timeline', entities };
-}
-
-// ============================================================
-// Venn extractor
-// ============================================================
-
-function extractVennSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    // Skip indented intersection lines
-    if (line[0] === ' ' || line[0] === '\t') continue;
-
-    // Skip intersection rows — those are references to prior sets, not
-    // declarations. Set NAMES + aliases come from the declaration lines below
-    // (the `+`-completion offers those). Match the parser's detection (d3.ts:
-    // any `+` on the line = intersection), not just the spaced form.
-    if (trimmed.includes('+')) continue;
-
-    // Set declaration: `Name [as <alias>] [color]`. Emit both the clean name
-    // and the alias so `Set + ` reference completion can offer either token.
-    const work = trimmed.split('|')[0]!.trim();
-    const asMatch = work.match(/^(.+?)\s+as\s+([A-Za-z][\w-]*)\b/i);
-    if (asMatch) {
-      const name = asMatch[1]!.trim();
-      const alias = asMatch[2]!;
-      if (name && !entities.includes(name)) entities.push(name);
-      if (alias && !entities.includes(alias)) entities.push(alias);
-      continue;
-    }
-    // No alias — strip a trailing color token if present.
-    const colorMatch = work.match(/^(.+?)\s+(\S+)$/);
-    const label =
-      colorMatch && RECOGNIZED_COLOR_SET.has(colorMatch[2]!.toLowerCase())
-        ? colorMatch[1]!.trim()
-        : work;
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return { kind: 'venn', entities };
-}
-
-// ============================================================
-// Quadrant extractor
-// ============================================================
-
-const QUADRANT_POSITION_RE =
-  /^(top-right|top-left|bottom-right|bottom-left)\s+/i;
-
-function extractQuadrantSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-    if (QUADRANT_POSITION_RE.test(trimmed)) continue;
-
-    // Point name (may have coordinates: Name x,y)
-    const parts = trimmed.split(/\s+\d/);
-    const label = (parts[0] ?? '').split('|')[0]!.trim();
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return { kind: 'quadrant', entities };
-}
-
-// ============================================================
-// Slope extractor
-// ============================================================
-
-function extractSlopeSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-    if (firstToken === 'period') continue;
-
-    // Data row: Label value1 value2 [color]
-    // Extract just the label (everything before first number)
-    const numIdx = trimmed.search(/\s\d/);
-    const label =
-      numIdx > 0
-        ? trimmed.slice(0, numIdx).trim()
-        : trimmed.split('|')[0]!.trim();
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return { kind: 'slope', entities };
-}
-
-// ============================================================
-// Generic data chart extractor
-// ============================================================
-
-const SERIES_RE = /^series\s+(.+)$/i;
-// "A -> B value" / "A -- B value" flow-link rows (chord/sankey via the shared
-// extractor). Both endpoints captured; trailing weight optional.
-const DATA_EDGE_RE = /^(.+?)\s+(?:->|--)\s+(.+?)(?:\s+-?\d[\d,_.]*)?\s*$/;
-
-function extractDataChartSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let chartType = 'bar';
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-      if (firstToken) chartType = firstToken;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    // Series declarations: "series Revenue, Expenses"
-    const seriesMatch = trimmed.match(SERIES_RE);
-    if (seriesMatch) {
-      for (const s of seriesMatch[1]!.split(',')) {
-        const name = s.trim().split(/\s+/)[0]!;
-        if (name && !entities.includes(name)) entities.push(name);
-      }
-      continue;
-    }
-
-    // Edge rows (chord / flow links): "A -> B value" / "A -- B value" — recover
-    // BOTH endpoints. Without this the trailing-number scan below mangles the
-    // whole "A -> B" into a single junk entity.
-    const edgeMatch = trimmed.match(DATA_EDGE_RE);
-    if (edgeMatch) {
-      const src = edgeMatch[1]!.trim();
-      const dst = edgeMatch[2]!.trim();
-      if (src && !entities.includes(src)) entities.push(src);
-      if (dst && !entities.includes(dst)) entities.push(dst);
-      continue;
-    }
-
-    // Label-expression rows (function curves §15.4): "Name: expression". ONLY
-    // for `function` — other data charts have no colon-form and would otherwise
-    // truncate legitimate colon-bearing labels (`12:00 5` → `12`). Column-0
-    // only, so indented `key: value` metadata is never misread as an entity.
-    if (chartType === 'function') {
-      const indent = line.length - line.trimStart().length;
-      const colonIdx = trimmed.indexOf(':');
-      if (indent === 0 && colonIdx > 0) {
-        const beforeColon = trimmed.slice(0, colonIdx).trim();
-        if (beforeColon && !METADATA_KEY_SET.has(beforeColon.toLowerCase())) {
-          if (!entities.includes(beforeColon)) entities.push(beforeColon);
-          continue;
-        }
-      }
-    }
-
-    // Data rows: "Label value [value...] [color]"
-    const numIdx = trimmed.search(/\s-?\d/);
-    if (numIdx > 0) {
-      const label = trimmed.slice(0, numIdx).trim();
-      if (label && !entities.includes(label)) entities.push(label);
-    }
-  }
-
-  return { kind: chartType, entities };
-}
-
-// ============================================================
-// Wireframe extractor
-// ============================================================
-
-// Wireframe element grammar is not a numeric-row variant, so it earns its own
-// extractor: `[label]` fields, `(button)`, `{a | b}` dropdowns/selects, with
-// possibly several elements per line. Returns the element labels as entities.
-const WF_FIELD_RE = /\[([^\]]*)\]/g;
-const WF_BUTTON_RE = /\(([^)]+)\)/g;
-const WF_DROPDOWN_RE = /\{([^}]+)\}/g;
-
-function extractWireframeSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  const push = (s: string): void => {
-    // Strip any legacy trailing `| meta` inside a bracket; the `{a|b}` pipe is
-    // handled separately by the dropdown split below.
-    const t = s.split('|')[0]!.trim();
-    if (t && !entities.includes(t)) entities.push(t);
-  };
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-
-    // `[label]` input/group fields.
-    for (const m of trimmed.matchAll(WF_FIELD_RE)) {
-      const inner = m[1]!.trim();
-      if (inner) push(inner);
-    }
-    // `(button)` — skip radio markers `(*)` / `( )`.
-    for (const m of trimmed.matchAll(WF_BUTTON_RE)) {
-      const inner = m[1]!.trim();
-      if (inner === '*' || inner === '') continue;
-      push(inner);
-    }
-    // `{opt1 | opt2}` dropdown/select — each option is an entity (carve-out pipe).
-    for (const m of trimmed.matchAll(WF_DROPDOWN_RE)) {
-      for (const opt of m[1]!.split('|')) {
-        const t = opt.trim();
-        if (t && !entities.includes(t)) entities.push(t);
-      }
-    }
-  }
-
-  return { kind: 'wireframe', entities };
-}
-
-// ============================================================
-// Register built-in extractors
-// ============================================================
-
-registerExtractor('er', extractErSymbols);
-registerExtractor('flowchart', extractFlowchartSymbols);
-registerExtractor('infra', extractInfraSymbols);
-registerExtractor('class', extractClassSymbols);
-registerExtractor('sequence', extractSequenceSymbols);
-registerExtractor('state', extractStateSymbols);
-registerExtractor('sitemap', extractSitemapSymbols);
-registerExtractor('c4', extractC4Symbols);
-registerExtractor('gantt', extractGanttSymbols);
-registerExtractor('pert', extractPertSymbols);
-registerExtractor('boxes-and-lines', extractBoxesAndLinesSymbols);
-registerExtractor('tech-radar', extractTechRadarSymbols);
-registerExtractor('cycle', extractCycleSymbols);
-registerExtractor('journey-map', extractJourneyMapSymbols);
-registerExtractor('raci', extractRaciSymbols);
-registerExtractor('org', extractOrgSymbols);
-registerExtractor('kanban', extractKanbanSymbols);
-registerExtractor('mindmap', extractMindmapSymbols);
-registerExtractor('treemap', extractTreemapSymbols);
-registerExtractor('block', extractBlockSymbols);
-registerExtractor('pyramid', extractPyramidSymbols);
-registerExtractor('ring', extractRingSymbols);
-registerExtractor('arc', extractArcSymbols);
-registerExtractor('sankey', extractSankeySymbols);
-registerExtractor('timeline', extractTimelineSymbols);
-registerExtractor('venn', extractVennSymbols);
-registerExtractor('quadrant', extractQuadrantSymbols);
-registerExtractor('slope', extractSlopeSymbols);
-registerExtractor('bar', extractDataChartSymbols);
-registerExtractor('line', extractDataChartSymbols);
-registerExtractor('pie', extractDataChartSymbols);
-registerExtractor('polar-area', extractDataChartSymbols);
-registerExtractor('radar', extractDataChartSymbols);
-registerExtractor('scatter', extractDataChartSymbols);
-registerExtractor('heatmap', extractDataChartSymbols);
-registerExtractor('funnel', extractDataChartSymbols);
-// `function` (`Name: expr`) and `wordcloud` (`Word weight`) had NO extractor
-// registered — extractDiagramSymbols returned null for them. The generalized
-// shared extractor now handles the colon-label form, so register both.
-registerExtractor('function', extractDataChartSymbols);
-registerExtractor('wordcloud', extractDataChartSymbols);
-registerExtractor('wireframe', extractWireframeSymbols);
-
-function extractTechRadarSymbols(docText: string): DiagramSymbols {
-  const entities: string[] = [];
-
-  // Extract ring names and aliases from the rings block
-  const lines = docText.split('\n');
-  let inRings = false;
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (trimmed.toLowerCase() === 'rings') {
-      inRings = true;
-      continue;
-    }
-    if (inRings) {
-      if (!trimmed || (line[0] !== ' ' && line[0] !== '\t')) {
-        inRings = false;
-        continue;
-      }
-      // Parse ring name (and alias)
-      const aliasMatch = trimmed.match(/^(.+?)\s+(?:alias|aka)\s+(\S+)\s*$/i);
-      if (aliasMatch) {
-        // Regex captured groups 1 and 2 by successful match.
-        entities.push(aliasMatch[1]!.trim());
-        entities.push(aliasMatch[2]!.trim());
-      } else {
-        entities.push(trimmed);
-      }
-    }
-  }
-
-  return { kind: 'tech-radar', entities };
-}
-
-// ============================================================
-// Cycle extractor
-// ============================================================
-
-function extractCycleSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    // Skip directives/metadata
-    // split(/\s+/) on non-empty `trimmed` always yields at least one element.
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-    if (
-      firstToken === 'direction-counterclockwise' ||
-      firstToken === 'circle-nodes'
-    )
-      continue;
-
-    // Skip indented lines (descriptions, edges)
-    if (line[0] === ' ' || line[0] === '\t') continue;
-
-    // Node label (strip pipe metadata)
-    // split('|')[0] always defined on any string.
-    const label = trimmed.split('|')[0]!.trim();
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return {
-    kind: 'cycle',
-    entities,
-  };
-}
-
-// ============================================================
-// RACI / RASCI / DACI extractor
-// ============================================================
-//
-// Extract role names, task names, and phase labels for editor
-// autocomplete. Mirrors the lightweight per-line scan pattern used
-// by other extractors (cycle / journey-map) — does NOT rebuild the
-// full AST.
-
-const RACI_PHASE_RE = /^\[(.+)\]\s*$/;
-const RACI_ROLES_DIRECTIVE_RE = /^roles\s+(.+)$/i;
-const RACI_VARIANT_DIRECTIVE_RE = /^variant\s+(.+)$/i;
-const RACI_ROLE_ASSIGNMENT_RE = /^([^:]+):\s*(.*)$/;
-
-function extractRaciSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let chartType = 'raci';
-  let pastFirstLine = false;
-  let underTask = false;
-
-  const push = (s: string): void => {
-    const trimmed = s.trim();
-    if (trimmed && !entities.includes(trimmed)) entities.push(trimmed);
-  };
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      // split(/\s+/) on non-empty `trimmed` always yields at least one element.
-      const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-      if (firstToken === 'raci') {
-        chartType = firstToken;
-      }
-      continue;
-    }
-
-    const indent = measureIndent(line);
-
-    // Header directives
-    if (indent === 0) {
-      const rolesMatch = trimmed.match(RACI_ROLES_DIRECTIVE_RE);
-      if (rolesMatch) {
-        // Regex captured group 1 by successful match.
-        for (const r of rolesMatch[1]!.split(',')) push(r);
-        continue;
-      }
-      if (RACI_VARIANT_DIRECTIVE_RE.test(trimmed)) continue;
-      // split(/\s+/) on non-empty `trimmed` always yields at least one element.
-      if (METADATA_KEY_SET.has(trimmed.split(/\s+/)[0]!.toLowerCase()))
-        continue;
-      if (
-        trimmed.toLowerCase() === 'draft' ||
-        trimmed.toLowerCase() === 'solid-fill'
-      )
-        continue;
-    }
-
-    // [Phase Label]
-    const phaseMatch = trimmed.match(RACI_PHASE_RE);
-    if (phaseMatch && indent === 0) {
-      // Regex captured group 1 by successful match.
-      push(phaseMatch[1]!);
-      underTask = false;
-      continue;
-    }
-
-    // Role assignment (Role: markers) — only valid under a task
-    const roleMatch = trimmed.match(RACI_ROLE_ASSIGNMENT_RE);
-    if (underTask && roleMatch) {
-      // Strip a possible trailing `# annotation`
-      // Regex captured group 1 by successful match.
-      const rolePart = roleMatch[1]!.trim();
-      push(rolePart);
-      continue;
-    }
-
-    // Otherwise: treat the line as a task name. `#` is NOT a comment
-    // character in DGMO (`//` is) — task names are used verbatim.
-    push(trimmed);
-    underTask = true;
-  }
-
-  return {
-    kind: chartType,
-    entities,
-  };
-}
-
-function extractJourneyMapSymbols(docText: string): DiagramSymbols {
-  const lines = docText.split('\n');
-  const entities: string[] = [];
-  let pastFirstLine = false;
-
-  for (const line of lines) {
-    const trimmed = line.trim();
-    if (!trimmed || trimmed.startsWith('//')) continue;
-
-    if (!pastFirstLine) {
-      pastFirstLine = true;
-      continue;
-    }
-
-    // Skip directives/metadata at indent 0
-    // split(/\s+/) on non-empty `trimmed` always yields at least one element.
-    const firstToken = trimmed.split(/\s+/)[0]!.toLowerCase();
-    if (METADATA_KEY_SET.has(firstToken)) continue;
-    if (firstToken === 'persona' || firstToken === 'tag') continue;
-
-    const isIndented = line[0] === ' ' || line[0] === '\t';
-
-    // Skip deep-indented lines (annotations, descriptions under steps)
-    // but keep singly-indented lines (steps within phases)
-    if (isIndented) {
-      // Annotation/description keywords — skip
-      if (/^(pain|opportunity|thought|description)\s*:/i.test(trimmed))
-        continue;
-      // Tag group entries — skip
-      if (/^\S+\([^)]+\)/.test(trimmed)) continue;
-    }
-
-    // Phase header
-    const phaseMatch = trimmed.match(/^\[(.+?)\]$/);
-    if (phaseMatch) {
-      // Regex captured group 1 by successful match.
-      entities.push(phaseMatch[1]!.trim());
-      continue;
-    }
-
-    // Step label (strip pipe metadata) — works for both indent 0 and indented steps
-    // split('|')[0] always defined on any string.
-    const label = trimmed.split('|')[0]!.trim();
-    if (label && !entities.includes(label)) entities.push(label);
-  }
-
-  return {
-    kind: 'journey-map',
-    entities,
-  };
-}