@opendata-ai/openchart-engine 8.0.0-rc.12 → 8.0.0-rc.3

package/src/compiler/validate.ts

@@ -0,0 +1,1871 @@
+/**
+ * Runtime spec validation.
+ *
+ * TypeScript catches compile-time errors for specs written in code.
+ * This module catches runtime errors for specs coming from JSON, APIs,
+ * or Claude-generated output where the TypeScript compiler can't help.
+ *
+ * Every error includes a machine-readable code and an actionable suggestion
+ * so consumers (and LLMs) can fix issues programmatically.
+ */
+
+import {
+  type FieldType,
+  inferFieldType,
+  MARK_ENCODING_RULES,
+  MARK_TYPES,
+  type MarkType,
+  resolveSchemeName,
+  SUPPORTED_SCHEME_NAMES,
+  type VizSpec,
+} from '@opendata-ai/openchart-core';
+
+import type { ValidationError, ValidationResult } from './types';
+
+// ---------------------------------------------------------------------------
+// Helpers
+// ---------------------------------------------------------------------------
+
+const VALID_FIELD_TYPES = new Set<string>(['quantitative', 'temporal', 'nominal', 'ordinal']);
+
+const VALID_DARK_MODES = new Set<string>(['auto', 'force', 'off']);
+
+/** Check if a value looks like a parseable date. */
+function isParseableDate(value: unknown): boolean {
+  if (value instanceof Date) return !Number.isNaN(value.getTime());
+  if (typeof value === 'string') {
+    const d = new Date(value);
+    return !Number.isNaN(d.getTime());
+  }
+  if (typeof value === 'number') return true;
+  return false;
+}
+
+/** Check if a value is numeric. */
+function isNumeric(value: unknown): boolean {
+  if (typeof value === 'number') return Number.isFinite(value);
+  if (typeof value === 'string') {
+    const n = Number(value);
+    return !Number.isNaN(n) && Number.isFinite(n);
+  }
+  return false;
+}
+
+/**
+ * Levenshtein edit distance between two strings (case-insensitive). Used to
+ * turn a misspelled field reference into a "did you mean" repair hint.
+ */
+function editDistance(a: string, b: string): number {
+  const s = a.toLowerCase();
+  const t = b.toLowerCase();
+  const m = s.length;
+  const n = t.length;
+  if (m === 0) return n;
+  if (n === 0) return m;
+  // Single-row DP: prev[j] holds distance for the previous source char.
+  let prev = Array.from({ length: n + 1 }, (_, j) => j);
+  let curr = new Array<number>(n + 1);
+  for (let i = 1; i <= m; i++) {
+    curr[0] = i;
+    for (let j = 1; j <= n; j++) {
+      const cost = s[i - 1] === t[j - 1] ? 0 : 1;
+      curr[j] = Math.min(prev[j] + 1, curr[j - 1] + 1, prev[j - 1] + cost);
+    }
+    [prev, curr] = [curr, prev];
+  }
+  return prev[n];
+}
+
+/**
+ * Given a misspelled field name and the actual data columns, return the nearest
+ * column when it is close enough to be a plausible typo, else undefined.
+ *
+ * The threshold scales with word length (at most ~40% of the target length, and
+ * never more than 3 edits) so short columns don't match unrelated names.
+ */
+function nearestColumn(field: string, columns: string[]): string | undefined {
+  let best: string | undefined;
+  let bestDistance = Number.POSITIVE_INFINITY;
+  for (const column of columns) {
+    const distance = editDistance(field, column);
+    if (distance < bestDistance) {
+      bestDistance = distance;
+      best = column;
+    }
+  }
+  if (best === undefined) return undefined;
+  const threshold = Math.min(3, Math.max(1, Math.floor(best.length * 0.4)));
+  return bestDistance <= threshold ? best : undefined;
+}
+
+/**
+ * Build the "did you mean" clause appended to DATA_FIELD_MISSING suggestions.
+ * Returns an empty string when no column is a plausible near-match.
+ */
+function didYouMean(field: string, columns: string[]): string {
+  const match = nearestColumn(field, columns);
+  return match ? ` Did you mean "${match}"?` : '';
+}
+
+// ---------------------------------------------------------------------------
+// Range mark validation
+// ---------------------------------------------------------------------------
+
+const VALID_RANGE_STYLES = new Set(['dumbbell', 'arrow', 'bar']);
+
+/**
+ * Range marks need an orientation-dependent encoding pair: x + x2 when the
+ * category axis is y (horizontal, the common editorial form), or y + y2 when
+ * the category axis is x (vertical). The static MARK_ENCODING_RULES table
+ * can't express "x2 required only when y is categorical", so this check
+ * enforces it with errors naming exactly what to add.
+ */
+function validateRangeSpec(spec: Record<string, unknown>, errors: ValidationError[]): void {
+  const encoding = spec.encoding as Record<string, Record<string, unknown> | undefined>;
+  const data = spec.data as Record<string, unknown>[];
+  const x = encoding.x;
+  const y = encoding.y;
+  // Missing x/y already produced MISSING_FIELD errors from the generic loop.
+  if (!x?.field || typeof x.field !== 'string' || !y?.field || typeof y.field !== 'string') {
+    return;
+  }
+
+  const xType = (x.type as string | undefined) ?? inferFieldType(data, x.field);
+  const yType = (y.type as string | undefined) ?? inferFieldType(data, y.field);
+  const xIsCategory = xType === 'nominal' || xType === 'ordinal';
+  const yIsCategory = yType === 'nominal' || yType === 'ordinal';
+
+  if (xIsCategory && yIsCategory) {
+    errors.push({
+      message:
+        'Spec error: range chart requires a quantitative value axis, but both encoding.x and encoding.y are categorical',
+      path: 'encoding',
+      code: 'ENCODING_MISMATCH',
+      suggestion:
+        'Map the range start to a quantitative channel: x (with x2 for the end) for horizontal ranges, or y (with y2) for vertical ranges',
+    });
+    return;
+  }
+  if (!xIsCategory && !yIsCategory) {
+    errors.push({
+      message:
+        'Spec error: range chart requires a nominal or ordinal category axis, but neither encoding.x nor encoding.y is categorical',
+      path: 'encoding',
+      code: 'ENCODING_MISMATCH',
+      suggestion:
+        'Set encoding.y to a nominal field for horizontal ranges (the common form), or encoding.x for vertical ranges',
+    });
+    return;
+  }
+
+  const availableColumns = data.length > 0 ? Object.keys(data[0]).join(', ') : '';
+  if (yIsCategory) {
+    // Horizontal: x is the range start, x2 the end.
+    if (!encoding.x2) {
+      errors.push({
+        message:
+          'Spec error: range chart requires encoding.x2 (the range end) when encoding.y is the category axis, but none was provided',
+        path: 'encoding.x2',
+        code: 'MISSING_FIELD',
+        suggestion: `Add encoding.x2 with the field holding the range's end value, e.g. x2: { field: "endValue" }. Available columns: ${availableColumns}`,
+      });
+    }
+    if (encoding.y2) {
+      errors.push({
+        message:
+          'Spec error: encoding.y2 is not used on a horizontal range chart (category axis on y); the range end belongs on x2',
+        path: 'encoding.y2',
+        code: 'INVALID_VALUE',
+        suggestion:
+          'Remove encoding.y2, or move the category to encoding.x and the values to y/y2 for a vertical range chart',
+      });
+    }
+  } else {
+    // Vertical: y is the range start, y2 the end.
+    if (!encoding.y2) {
+      errors.push({
+        message:
+          'Spec error: range chart requires encoding.y2 (the range end) when encoding.x is the category axis, but none was provided',
+        path: 'encoding.y2',
+        code: 'MISSING_FIELD',
+        suggestion: `Add encoding.y2 with the field holding the range's end value, e.g. y2: { field: "endValue" }. Available columns: ${availableColumns}`,
+      });
+    }
+    if (encoding.x2) {
+      errors.push({
+        message:
+          'Spec error: encoding.x2 is not used on a vertical range chart (category axis on x); the range end belongs on y2',
+        path: 'encoding.x2',
+        code: 'INVALID_VALUE',
+        suggestion:
+          'Remove encoding.x2, or move the category to encoding.y and the values to x/x2 for a horizontal range chart',
+      });
+    }
+  }
+
+  // markDef options: style must be one of the known variants.
+  const mark = spec.mark;
+  if (mark && typeof mark === 'object' && !Array.isArray(mark)) {
+    const style = (mark as Record<string, unknown>).style;
+    if (style !== undefined && !VALID_RANGE_STYLES.has(style as string)) {
+      errors.push({
+        message: `Spec error: mark.style "${style}" is not valid for range marks. Must be one of: dumbbell, arrow, bar`,
+        path: 'mark.style',
+        code: 'INVALID_VALUE',
+        suggestion:
+          'Use mark: { type: "range", style: "dumbbell" } (default), "arrow" for directional change, or "bar" for a plain floating range bar',
+      });
+    }
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Calendar mark validation
+// ---------------------------------------------------------------------------
+
+const VALID_WEEK_STARTS = new Set(['monday', 'sunday']);
+
+/**
+ * Calendar heatmaps take daily dates on x and a quantitative value on color;
+ * the weeks-x-weekdays geometry is computed internally, so there is no y
+ * channel. This check enforces the parts the static MARK_ENCODING_RULES
+ * table can't express: no y, a temporal x even when the type is inferred,
+ * every date parseable, one row per day (daily granularity), and valid
+ * mark-level weekStart/cellRadius options.
+ */
+function validateCalendarSpec(spec: Record<string, unknown>, errors: ValidationError[]): void {
+  const encoding = spec.encoding as Record<string, Record<string, unknown> | undefined>;
+  const data = spec.data as Record<string, unknown>[];
+
+  if (encoding.y) {
+    errors.push({
+      message:
+        'Spec error: calendar chart does not accept encoding.y (the weeks x weekdays layout is computed from the dates)',
+      path: 'encoding.y',
+      code: 'ENCODING_MISMATCH',
+      suggestion:
+        'Remove encoding.y. A calendar needs only x (temporal, daily dates) and color (quantitative).',
+    });
+  }
+
+  // Mark-level options (weekStart / cellRadius) on the MarkDef object form.
+  if (spec.mark && typeof spec.mark === 'object') {
+    const markDef = spec.mark as Record<string, unknown>;
+    if (markDef.weekStart !== undefined && !VALID_WEEK_STARTS.has(markDef.weekStart as string)) {
+      errors.push({
+        message: `Spec error: mark.weekStart "${markDef.weekStart}" is not a valid week start`,
+        path: 'mark.weekStart',
+        code: 'INVALID_VALUE',
+        suggestion:
+          'Use mark: { type: "calendar", weekStart: "monday" } (default, ISO weeks) or "sunday"',
+      });
+    }
+    if (
+      markDef.cellRadius !== undefined &&
+      (typeof markDef.cellRadius !== 'number' ||
+        !Number.isFinite(markDef.cellRadius) ||
+        markDef.cellRadius < 0)
+    ) {
+      errors.push({
+        message: `Spec error: mark.cellRadius must be a non-negative number, got ${JSON.stringify(markDef.cellRadius)}`,
+        path: 'mark.cellRadius',
+        code: 'INVALID_VALUE',
+        suggestion:
+          'Use a pixel radius like cellRadius: 1 (default) or cellRadius: 0 for square corners',
+      });
+    }
+  }
+
+  const x = encoding.x;
+  // Missing x already produced a MISSING_FIELD error from the generic loop.
+  if (!x?.field || typeof x.field !== 'string') return;
+
+  // The generic loop rejects a declared non-temporal type against the
+  // encoding rules table; this catches the inferred case (type omitted).
+  const xType = (x.type as string | undefined) ?? inferFieldType(data, x.field);
+  if (xType !== 'temporal') {
+    if (!x.type) {
+      errors.push({
+        message: `Spec error: calendar chart requires temporal encoding.x, but "${x.field}" was inferred as ${xType}`,
+        path: 'encoding.x',
+        code: 'ENCODING_MISMATCH',
+        suggestion: `Provide daily dates in "${x.field}" (ISO 8601 strings like "2024-01-15" or Date objects) and declare x: { field: "${x.field}", type: "temporal" }`,
+      });
+    }
+    return;
+  }
+
+  // Transforms (e.g. a timeUnit aggregation) reshape the data before the
+  // calendar sees it, so per-row date checks only apply to untransformed specs.
+  if (Array.isArray(spec.transform) && spec.transform.length > 0) return;
+
+  // Every date must parse; report the first offender by row index.
+  for (let i = 0; i < data.length; i++) {
+    const value = data[i][x.field];
+    if (value != null && !isParseableDate(value)) {
+      errors.push({
+        message: `Spec error: encoding.x.field "${x.field}" contains an unparseable date at data[${i}]: ${JSON.stringify(value)}`,
+        path: `data[${i}].${x.field}`,
+        code: 'INVALID_VALUE',
+        suggestion: `Fix data[${i}].${x.field} to a parseable date (ISO 8601 strings like "2024-01-15" or Date objects)`,
+      });
+      return;
+    }
+  }
+
+  // Daily granularity: more than one row on the same UTC day means the data
+  // is sub-daily (or duplicated) and cells would silently overwrite.
+  const seenDays = new Map<string, number>();
+  for (let i = 0; i < data.length; i++) {
+    const value = data[i][x.field];
+    if (value == null) continue;
+    const date = value instanceof Date ? value : new Date(value as string | number);
+    if (Number.isNaN(date.getTime())) continue;
+    const dayKey = `${date.getUTCFullYear()}-${date.getUTCMonth()}-${date.getUTCDate()}`;
+    const firstIndex = seenDays.get(dayKey);
+    if (firstIndex !== undefined) {
+      errors.push({
+        message: `Spec error: calendar chart requires daily granularity, but data[${firstIndex}] and data[${i}] fall on the same day (${JSON.stringify(value)})`,
+        path: `data[${i}].${x.field}`,
+        code: 'ENCODING_MISMATCH',
+        suggestion: `Aggregate to one row per day first, e.g. transform: [{ timeUnit: "yearmonthdate", field: "${x.field}", as: "day" }] plus an aggregate over the value field, or pre-aggregate the data.`,
+      });
+      return;
+    }
+    seenDays.set(dayKey, i);
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Chart validation
+// ---------------------------------------------------------------------------
+
+function validateChartSpec(spec: Record<string, unknown>, errors: ValidationError[]): void {
+  const markType =
+    typeof spec.mark === 'string' ? spec.mark : (spec.mark as Record<string, unknown>)?.type;
+
+  // Check data
+  if (!Array.isArray(spec.data)) {
+    // Near-miss: the VL `data: { url }` form. Fetching is out of scope for a
+    // rendering library, so reject it with a pointed message. The `{ values }`
+    // form is unwrapped by the pre-validation sugar expansion and never
+    // reaches this check on the compile path.
+    if (spec.data && typeof spec.data === 'object' && 'url' in (spec.data as object)) {
+      errors.push({
+        message: 'Spec error: data.url is not supported (openchart does not fetch remote data)',
+        path: 'data.url',
+        code: 'INVALID_VALUE',
+        suggestion:
+          'Fetch the data in your application and provide the rows inline: data: [...] or data: { values: [...] }',
+      });
+      return;
+    }
+    errors.push({
+      message: 'Spec error: "data" must be an array',
+      path: 'data',
+      code: 'INVALID_TYPE',
+      suggestion: 'Provide data as an array of objects, e.g. data: [{ x: 1, y: 2 }]',
+    });
+    return; // Can't validate further without data
+  }
+
+  if (spec.data.length === 0) {
+    errors.push({
+      message: 'Spec error: "data" must be a non-empty array',
+      path: 'data',
+      code: 'EMPTY_DATA',
+      suggestion: 'Add at least one data row, e.g. data: [{ x: 1, y: 2 }]',
+    });
+    return;
+  }
+
+  // Validate data entries are objects
+  const firstRow = spec.data[0] as unknown;
+  if (typeof firstRow !== 'object' || firstRow === null || Array.isArray(firstRow)) {
+    errors.push({
+      message: 'Spec error: each item in "data" must be a plain object',
+      path: 'data[0]',
+      code: 'INVALID_TYPE',
+      suggestion:
+        'Each data item should be an object with key-value pairs, e.g. { name: "Alice", value: 10 }',
+    });
+    return;
+  }
+
+  // Check encoding exists
+  if (!spec.encoding || typeof spec.encoding !== 'object') {
+    const rules = MARK_ENCODING_RULES[markType as MarkType];
+    const requiredChannels = Object.entries(rules)
+      .filter(([, rule]) => rule.required)
+      .map(([ch]) => ch);
+    errors.push({
+      message: `Spec error: ${markType} chart requires an "encoding" object`,
+      path: 'encoding',
+      code: 'MISSING_FIELD',
+      suggestion: `Add an encoding object with required channels: ${requiredChannels.join(', ')}. Example: encoding: { ${requiredChannels.map((ch) => `${ch}: { field: "...", type: "..." }`).join(', ')} }`,
+    });
+    return;
+  }
+
+  const rules = MARK_ENCODING_RULES[markType as MarkType];
+  const encoding = spec.encoding as Record<string, unknown>;
+  const dataColumns = new Set(Object.keys(firstRow as Record<string, unknown>));
+  const columnList = [...dataColumns];
+  const availableColumns = columnList.join(', ');
+
+  // Validate required channels. Arc's canonical channel is theta; waffle and
+  // parliament still key on y with theta as an alias. For backward compat,
+  // validateSpec (a public entry point that callers may invoke on un-expanded
+  // specs) accepts y as satisfying a theta requirement and vice versa on
+  // theta-capable marks.
+  const THETA_MARKS = new Set(['arc', 'waffle', 'parliament']);
+  const isThetaMark = THETA_MARKS.has(markType as string);
+  for (const [channel, rule] of Object.entries(rules)) {
+    if (!rule.required) continue;
+    const satisfiedByAlias =
+      isThetaMark && ((channel === 'theta' && encoding.y) || (channel === 'y' && encoding.theta));
+    if (!encoding[channel] && !satisfiedByAlias) {
+      const allowedTypes = rule.allowedTypes.join(' or ');
+      const suggestedChannel = channel === 'y' && isThetaMark ? 'theta' : channel;
+      errors.push({
+        message: `Spec error: ${markType} chart requires encoding.${suggestedChannel} but none was provided`,
+        path: `encoding.${suggestedChannel}`,
+        code: 'MISSING_FIELD',
+        suggestion: `Add encoding.${suggestedChannel} with a field from your data (${availableColumns}) and type (${allowedTypes}). Example: ${suggestedChannel}: { field: "${[...dataColumns][0] ?? 'myField'}", type: "${rule.allowedTypes[0]}" }`,
+      });
+    }
+  }
+
+  // Beeswarm axis combination: the channel rules mark both axes optional, so
+  // enforce the shape here. Exactly one positional channel is the quantitative
+  // value axis; the other, when present, is the nominal/ordinal lane channel.
+  // Only declared types are checked; omitted types are inferred later.
+  if (markType === 'beeswarm') {
+    const xChannel = encoding.x as Record<string, unknown> | undefined;
+    const yChannel = encoding.y as Record<string, unknown> | undefined;
+    const xType = xChannel?.type as string | undefined;
+    const yType = yChannel?.type as string | undefined;
+    if (!xChannel && !yChannel) {
+      errors.push({
+        message:
+          'Spec error: beeswarm chart requires a quantitative encoding.x or encoding.y (the value axis)',
+        path: 'encoding',
+        code: 'MISSING_FIELD',
+        suggestion: `Add the value axis, e.g. x: { field: "${[...dataColumns][0] ?? 'myField'}", type: "quantitative" }. Optionally add a nominal channel on the other axis for grouped swarms.`,
+      });
+    } else if (xType === 'quantitative' && yType === 'quantitative') {
+      errors.push({
+        message:
+          'Spec error: beeswarm chart accepts only one quantitative axis (the value axis), but both encoding.x and encoding.y are quantitative',
+        path: 'encoding',
+        code: 'ENCODING_MISMATCH',
+        suggestion:
+          'Keep the value axis quantitative and change the other channel to nominal/ordinal (grouped swarm lanes), or remove it for a single swarm. For two quantitative axes use mark: "point".',
+      });
+    } else if (
+      xType !== 'quantitative' &&
+      yType !== 'quantitative' &&
+      (!xChannel || !!xType) &&
+      (!yChannel || !!yType)
+    ) {
+      errors.push({
+        message:
+          'Spec error: beeswarm chart requires one quantitative axis (the value axis), but neither encoding.x nor encoding.y is quantitative',
+        path: 'encoding',
+        code: 'ENCODING_MISMATCH',
+        suggestion:
+          'Set the value axis to type: "quantitative" (x for a horizontal swarm, y for a vertical one).',
+      });
+    }
+  }
+
+  // Range marks: orientation-dependent x2/y2 requirement + mark.style options
+  if (markType === 'range') {
+    validateRangeSpec(spec, errors);
+  }
+
+  // Calendar marks: temporal daily x, no y, parseable dates, mark options
+  if (markType === 'calendar') {
+    validateCalendarSpec(spec, errors);
+  }
+
+  // Near-miss: VL's string expression form of calculate. A restricted string
+  // grammar is deliberately not supported (decision: structured form only);
+  // point authors at the structured equivalent instead.
+  if (Array.isArray(spec.transform)) {
+    const transforms = spec.transform as Record<string, unknown>[];
+    for (let i = 0; i < transforms.length; i++) {
+      const t = transforms[i];
+      if (t && typeof t === 'object' && typeof t.calculate === 'string') {
+        errors.push({
+          message: `Spec error: transform[${i}].calculate must be a structured expression object, not a string`,
+          path: `transform[${i}].calculate`,
+          code: 'INVALID_TYPE',
+          suggestion:
+            'Use the structured form for single operations, e.g. calculate: { op: "/", field: "a", field2: "b" } with as: "ratio". For running or grouped computations use the window or aggregate transforms.',
+        });
+      }
+    }
+  }
+
+  // Collect fields that transforms will create, so we don't reject them
+  const transformFields = new Set<string>();
+  if (Array.isArray(spec.transform)) {
+    for (const t of spec.transform as Record<string, unknown>[]) {
+      // filter/bin/calculate/timeUnit/fold write their output(s) to `as`.
+      if (typeof t.as === 'string') transformFields.add(t.as);
+      if (Array.isArray(t.as)) {
+        for (const f of t.as) {
+          if (typeof f === 'string') transformFields.add(f);
+        }
+      }
+      // aggregate/window carry their output field names inside their op arrays,
+      // and aggregate preserves its groupby fields on the output rows.
+      const opList = (t.aggregate ?? t.window) as Record<string, unknown>[] | undefined;
+      if (Array.isArray(opList)) {
+        for (const op of opList) {
+          if (op && typeof op.as === 'string') transformFields.add(op.as);
+        }
+      }
+      if (Array.isArray(t.groupby)) {
+        for (const g of t.groupby) {
+          if (typeof g === 'string') transformFields.add(g);
+        }
+      }
+    }
+  }
+
+  // Validate provided channels
+  for (const [channel, channelSpec] of Object.entries(encoding)) {
+    if (!channelSpec || typeof channelSpec !== 'object') continue;
+
+    // Tooltip can be an array of encoding channels
+    if (channel === 'tooltip' && Array.isArray(channelSpec)) {
+      for (let i = 0; i < channelSpec.length; i++) {
+        const elem = channelSpec[i] as Record<string, unknown> | null;
+        if (!elem || typeof elem !== 'object') continue;
+        if (!elem.field || typeof elem.field !== 'string') {
+          errors.push({
+            message: `Spec error: encoding.tooltip[${i}] must have a "field" string`,
+            path: `encoding.tooltip[${i}].field`,
+            code: 'MISSING_FIELD',
+            suggestion: `Add a field name from your data columns: ${availableColumns}`,
+          });
+          continue;
+        }
+        if (!dataColumns.has(elem.field) && !transformFields.has(elem.field)) {
+          errors.push({
+            message: `Spec error: encoding.tooltip[${i}].field "${elem.field}" does not exist in data. Available columns: ${availableColumns}`,
+            path: `encoding.tooltip[${i}].field`,
+            code: 'DATA_FIELD_MISSING',
+            suggestion: `Use one of the available data columns: ${availableColumns}.${didYouMean(elem.field, columnList)}`,
+          });
+        }
+        if (elem.type && !VALID_FIELD_TYPES.has(elem.type as string)) {
+          errors.push({
+            message: `Spec error: encoding.tooltip[${i}].type "${elem.type}" is not valid. Must be one of: ${[...VALID_FIELD_TYPES].join(', ')}`,
+            path: `encoding.tooltip[${i}].type`,
+            code: 'INVALID_VALUE',
+            suggestion: `Use one of: ${[...VALID_FIELD_TYPES].join(', ')}`,
+          });
+        }
+      }
+      continue;
+    }
+
+    const channelObj = channelSpec as Record<string, unknown>;
+    const channelRule = rules[channel as keyof typeof rules];
+
+    // Skip ConditionalValueDef channels (they have 'condition' instead of 'field')
+    if ('condition' in channelObj) continue;
+
+    // Check field exists
+    if (!channelObj.field || typeof channelObj.field !== 'string') {
+      errors.push({
+        message: `Spec error: encoding.${channel} must have a "field" string`,
+        path: `encoding.${channel}.field`,
+        code: 'MISSING_FIELD',
+        suggestion: `For constant colors, use mark.fill (e.g., mark: { type: "bar", fill: "#1b7fa3" }) instead of encoding.${channel}. Encoding channels require a data field: ${availableColumns}`,
+      });
+      continue;
+    }
+
+    // Check field references a column in data (or will be created by a transform)
+    if (!dataColumns.has(channelObj.field) && !transformFields.has(channelObj.field)) {
+      errors.push({
+        message: `Spec error: encoding.${channel}.field "${channelObj.field}" does not exist in data. Available columns: ${availableColumns}`,
+        path: `encoding.${channel}.field`,
+        code: 'DATA_FIELD_MISSING',
+        suggestion: `Use one of the available data columns: ${availableColumns}.${didYouMean(channelObj.field, columnList)}`,
+      });
+    }
+
+    // Check field type is valid
+    if (channelObj.type && !VALID_FIELD_TYPES.has(channelObj.type as string)) {
+      errors.push({
+        message: `Spec error: encoding.${channel}.type "${channelObj.type}" is not valid. Must be one of: ${[...VALID_FIELD_TYPES].join(', ')}`,
+        path: `encoding.${channel}.type`,
+        code: 'INVALID_VALUE',
+        suggestion: `Use one of: ${[...VALID_FIELD_TYPES].join(', ')}`,
+      });
+    }
+
+    // Check scale.scheme names. Known names (including VL aliases) are
+    // resolved to palette ranges by the pre-validation sugar expansion, so a
+    // scheme surviving to this point on the compile path is an unknown name.
+    const channelScale = channelObj.scale as Record<string, unknown> | undefined;
+    if (
+      channelScale &&
+      typeof channelScale.scheme === 'string' &&
+      !resolveSchemeName(channelScale.scheme)
+    ) {
+      errors.push({
+        message: `Spec error: encoding.${channel}.scale.scheme "${channelScale.scheme}" is not a supported scheme name`,
+        path: `encoding.${channel}.scale.scheme`,
+        code: 'INVALID_VALUE',
+        suggestion: `Use one of the supported scheme names: ${[...SUPPORTED_SCHEME_NAMES].join(', ')}. Or provide explicit colors via scale.range.`,
+      });
+    }
+
+    // Check field type is allowed for this channel
+    if (channelRule && channelObj.type && channelRule.allowedTypes.length > 0) {
+      if (!channelRule.allowedTypes.includes(channelObj.type as FieldType)) {
+        errors.push({
+          message: `Spec error: encoding.${channel} for ${markType} chart does not accept type "${channelObj.type}". Allowed types: ${channelRule.allowedTypes.join(', ')}`,
+          path: `encoding.${channel}.type`,
+          code: 'ENCODING_MISMATCH',
+          suggestion: `Change encoding.${channel}.type to one of: ${channelRule.allowedTypes.join(', ')}`,
+        });
+      }
+    }
+
+    // Check field values match declared type
+    if (channelObj.type && channelObj.field && dataColumns.has(channelObj.field as string)) {
+      const data = spec.data as Record<string, unknown>[];
+      const fieldName = channelObj.field as string;
+      const fieldType = channelObj.type as string;
+      // Sample up to 5 values for type checking
+      const sampleSize = Math.min(5, data.length);
+
+      if (fieldType === 'temporal') {
+        let nonDateCount = 0;
+        for (let i = 0; i < sampleSize; i++) {
+          const val = data[i][fieldName];
+          if (val != null && !isParseableDate(val)) {
+            nonDateCount++;
+          }
+        }
+        if (nonDateCount > 0) {
+          errors.push({
+            message: `Spec error: encoding.${channel}.field "${fieldName}" is declared as temporal but contains non-date values`,
+            path: `encoding.${channel}`,
+            code: 'ENCODING_MISMATCH',
+            suggestion: `Either change the type to "nominal" or ensure "${fieldName}" values are parseable dates (ISO 8601 strings like "2024-01-15" or Date objects)`,
+          });
+        }
+      }
+
+      if (fieldType === 'quantitative') {
+        let nonNumericCount = 0;
+        for (let i = 0; i < sampleSize; i++) {
+          const val = data[i][fieldName];
+          if (val != null && !isNumeric(val)) {
+            nonNumericCount++;
+          }
+        }
+        if (nonNumericCount > 0) {
+          errors.push({
+            message: `Spec error: encoding.${channel}.field "${fieldName}" is declared as quantitative but contains non-numeric values`,
+            path: `encoding.${channel}`,
+            code: 'ENCODING_MISMATCH',
+            suggestion: `Either change the type to "nominal" or ensure "${fieldName}" values are numbers`,
+          });
+        }
+      }
+    }
+  }
+
+  // Validate highlight on the color channel
+  if (encoding) {
+    const colorCh = (encoding as Record<string, Record<string, unknown> | undefined>).color;
+    if (colorCh && colorCh.highlight != null) {
+      const hlVal = colorCh.highlight;
+      const isValidType =
+        typeof hlVal === 'string' ||
+        (Array.isArray(hlVal) && hlVal.every((v: unknown) => typeof v === 'string'));
+      if (!isValidType) {
+        errors.push({
+          message: 'Spec error: encoding.color.highlight must be a string or array of strings',
+          path: 'encoding.color.highlight',
+          code: 'INVALID_TYPE',
+          suggestion: 'Provide a series name (string) or an array of series names.',
+        });
+      } else {
+        const colorType =
+          (colorCh.type as string | undefined) ??
+          (colorCh.field && Array.isArray(spec.data) && spec.data.length > 0
+            ? inferFieldType(spec.data as Record<string, unknown>[], colorCh.field as string)
+            : undefined);
+        if (colorType === 'quantitative') {
+          errors.push({
+            message:
+              'Spec error: encoding.color.highlight is not supported on quantitative color channels',
+            path: 'encoding.color.highlight',
+            code: 'INVALID_VALUE',
+            suggestion:
+              'Highlight works with nominal/ordinal color (categorical series). Remove highlight or change encoding.color.type to nominal.',
+          });
+        }
+      }
+    }
+  }
+
+  // Validate facet channel if provided
+  if (encoding.facet && typeof encoding.facet === 'object') {
+    const facet = encoding.facet as Record<string, unknown>;
+    if (!facet.field || typeof facet.field !== 'string') {
+      errors.push({
+        message: 'Spec error: encoding.facet.field is required and must be a non-empty string',
+        path: 'encoding.facet.field',
+        code: 'MISSING_FIELD',
+        suggestion:
+          'Add a field property to encoding.facet (e.g. { field: "category", type: "nominal" })',
+      });
+    } else if (
+      !dataColumns.has(facet.field as string) &&
+      !transformFields.has(facet.field as string)
+    ) {
+      errors.push({
+        message: `Spec error: encoding.facet.field "${facet.field}" does not exist in data. Available columns: ${availableColumns}`,
+        path: 'encoding.facet.field',
+        code: 'DATA_FIELD_MISSING',
+        suggestion: `Use one of the available data columns: ${availableColumns}.${didYouMean(facet.field as string, columnList)}`,
+      });
+    }
+    if (!facet.type || (facet.type !== 'nominal' && facet.type !== 'ordinal')) {
+      errors.push({
+        message: `Spec error: encoding.facet.type must be "nominal" or "ordinal"${facet.type ? `, got "${facet.type}"` : ''}`,
+        path: 'encoding.facet.type',
+        code: facet.type ? 'INVALID_VALUE' : 'MISSING_FIELD',
+        suggestion: 'Set encoding.facet.type to "nominal" or "ordinal"',
+      });
+    }
+    if (facet.columns !== undefined) {
+      const cols = Number(facet.columns);
+      if (!Number.isInteger(cols) || cols < 1) {
+        errors.push({
+          message: 'Spec error: encoding.facet.columns must be a positive integer',
+          path: 'encoding.facet.columns',
+          code: 'INVALID_VALUE',
+          suggestion: 'Set columns to a positive integer (e.g. 3)',
+        });
+      }
+    }
+  }
+
+  // Validate darkMode if provided
+  if (spec.darkMode !== undefined && !VALID_DARK_MODES.has(spec.darkMode as string)) {
+    errors.push({
+      message: `Spec error: darkMode must be "auto", "force", or "off"`,
+      path: 'darkMode',
+      code: 'INVALID_VALUE',
+      suggestion:
+        'Use one of: "auto" (system preference), "force" (always dark), or "off" (always light)',
+    });
+  }
+
+  // Validate youDrawIt: line marks only, single-series only, "from" required.
+  if (spec.youDrawIt !== undefined && spec.youDrawIt !== false) {
+    if (markType !== 'line') {
+      errors.push({
+        message: `Spec error: youDrawIt is only supported on line charts, but mark is "${markType}"`,
+        path: 'youDrawIt',
+        code: 'ENCODING_MISMATCH',
+        suggestion: 'Remove youDrawIt or change mark to "line".',
+      });
+    } else {
+      const config = spec.youDrawIt as Record<string, unknown>;
+      if (
+        typeof config !== 'object' ||
+        Array.isArray(config) ||
+        (config.from !== 0 && !config.from)
+      ) {
+        errors.push({
+          message: 'Spec error: youDrawIt.from is required (the x value where drawing starts)',
+          path: 'youDrawIt.from',
+          code: 'MISSING_FIELD',
+          suggestion:
+            'Set youDrawIt.from to an x value from your data, e.g. youDrawIt: { from: "2010" }',
+        });
+      } else if (typeof config.from !== 'string' && typeof config.from !== 'number') {
+        errors.push({
+          message: 'Spec error: youDrawIt.from must be a string or number',
+          path: 'youDrawIt.from',
+          code: 'INVALID_TYPE',
+          suggestion: 'Provide the x value as a string or number matching your data.',
+        });
+      }
+
+      const data = Array.isArray(spec.data) ? (spec.data as Record<string, unknown>[]) : [];
+      const colorCh = encoding?.color as Record<string, unknown> | undefined;
+      if (colorCh && typeof colorCh === 'object' && typeof colorCh.field === 'string') {
+        const colorType =
+          (colorCh.type as string | undefined) ?? inferFieldType(data, colorCh.field);
+        if (colorType !== 'quantitative') {
+          const distinct = new Set(data.map((row) => row[colorCh.field as string]));
+          if (distinct.size > 1) {
+            errors.push({
+              message:
+                'Spec error: youDrawIt only supports single-series line charts, but encoding.color has multiple distinct values',
+              path: 'youDrawIt',
+              code: 'ENCODING_MISMATCH',
+              suggestion:
+                'Remove encoding.color (or filter the data to one series) to use youDrawIt.',
+            });
+          }
+        }
+      }
+    }
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Table validation
+// ---------------------------------------------------------------------------
+
+function validateTableSpec(spec: Record<string, unknown>, errors: ValidationError[]): void {
+  if (!Array.isArray(spec.data)) {
+    errors.push({
+      message: 'Spec error: "data" must be an array',
+      path: 'data',
+      code: 'INVALID_TYPE',
+      suggestion: 'Provide data as an array of objects, e.g. data: [{ name: "Alice", age: 30 }]',
+    });
+    return;
+  }
+
+  if (spec.data.length === 0) {
+    errors.push({
+      message: 'Spec error: "data" must be a non-empty array',
+      path: 'data',
+      code: 'EMPTY_DATA',
+      suggestion: 'Add at least one data row to the data array',
+    });
+    return;
+  }
+
+  if (!Array.isArray(spec.columns)) {
+    errors.push({
+      message: 'Spec error: table spec requires a "columns" array',
+      path: 'columns',
+      code: 'MISSING_FIELD',
+      suggestion:
+        'Add a columns array defining which data fields to display, e.g. columns: [{ key: "name" }, { key: "age" }]',
+    });
+    return;
+  }
+
+  const data = spec.data as Record<string, unknown>[];
+  const firstRow = data[0];
+  if (typeof firstRow !== 'object' || firstRow === null || Array.isArray(firstRow)) {
+    errors.push({
+      message: 'Spec error: each item in "data" must be a plain object',
+      path: 'data[0]',
+      code: 'INVALID_TYPE',
+      suggestion:
+        'Each data item should be an object with key-value pairs, e.g. { name: "Alice", age: 30 }',
+    });
+    return;
+  }
+
+  const dataColumns = new Set(Object.keys(firstRow as Record<string, unknown>));
+  const columnList = [...dataColumns];
+  const availableColumns = columnList.join(', ');
+  const columns = spec.columns as Record<string, unknown>[];
+
+  for (let i = 0; i < columns.length; i++) {
+    const col = columns[i];
+    if (!col || typeof col !== 'object') {
+      errors.push({
+        message: `Spec error: columns[${i}] must be an object`,
+        path: `columns[${i}]`,
+        code: 'INVALID_TYPE',
+        suggestion: 'Each column entry should be an object, e.g. { key: "fieldName" }',
+      });
+      continue;
+    }
+
+    // Check key exists
+    if (!col.key || typeof col.key !== 'string') {
+      errors.push({
+        message: `Spec error: columns[${i}] must have a "key" string`,
+        path: `columns[${i}].key`,
+        code: 'MISSING_FIELD',
+        suggestion: `Add a key referencing a data field. Available columns: ${availableColumns}`,
+      });
+      continue;
+    }
+
+    // Check key references a field in data
+    if (!dataColumns.has(col.key as string)) {
+      errors.push({
+        message: `Spec error: columns[${i}].key "${col.key}" does not exist in data. Available columns: ${availableColumns}`,
+        path: `columns[${i}].key`,
+        code: 'DATA_FIELD_MISSING',
+        suggestion: `Use one of the available data columns: ${availableColumns}.${didYouMean(col.key as string, columnList)}`,
+      });
+    }
+
+    // Check at most one visual enhancement
+    const visuals = ['heatmap', 'bar', 'sparkline', 'image', 'flag', 'categoryColors'].filter(
+      (v) => col[v] != null && col[v] !== false,
+    );
+    if (visuals.length > 1) {
+      errors.push({
+        message: `Spec error: columns[${i}] has multiple visual features (${visuals.join(', ')}). Only one is allowed per column.`,
+        path: `columns[${i}]`,
+        code: 'INVALID_VALUE',
+        suggestion: `Keep only one visual feature per column. Remove all but one of: ${visuals.join(', ')}`,
+      });
+    }
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Graph validation
+// ---------------------------------------------------------------------------
+
+function validateGraphSpec(spec: Record<string, unknown>, errors: ValidationError[]): void {
+  // Validate nodes array exists and is non-empty
+  if (!Array.isArray(spec.nodes)) {
+    errors.push({
+      message: 'Spec error: graph spec requires a "nodes" array',
+      path: 'nodes',
+      code: 'MISSING_FIELD',
+      suggestion:
+        'Add a nodes array with objects that have an "id" field, e.g. nodes: [{ id: "a" }, { id: "b" }]',
+    });
+    return; // Can't validate further without nodes
+  }
+
+  if (spec.nodes.length === 0) {
+    errors.push({
+      message: 'Spec error: "nodes" must be a non-empty array',
+      path: 'nodes',
+      code: 'EMPTY_DATA',
+      suggestion: 'Add at least one node, e.g. nodes: [{ id: "a" }]',
+    });
+    return;
+  }
+
+  // Validate each node has a string id
+  const nodeIds = new Set<string>();
+  const nodes = spec.nodes as Record<string, unknown>[];
+  for (let i = 0; i < nodes.length; i++) {
+    const node = nodes[i];
+    if (!node || typeof node !== 'object') {
+      errors.push({
+        message: `Spec error: nodes[${i}] must be an object`,
+        path: `nodes[${i}]`,
+        code: 'INVALID_TYPE',
+        suggestion: 'Each node must be an object with at least an "id" field, e.g. { id: "a" }',
+      });
+      continue;
+    }
+    if (typeof node.id !== 'string' || node.id === '') {
+      errors.push({
+        message: `Spec error: nodes[${i}] must have a non-empty string "id" field`,
+        path: `nodes[${i}].id`,
+        code: 'MISSING_FIELD',
+        suggestion: 'Add a string id to the node, e.g. { id: "node1" }',
+      });
+    } else {
+      nodeIds.add(node.id);
+    }
+  }
+
+  // Validate edges array exists
+  if (!Array.isArray(spec.edges)) {
+    errors.push({
+      message: 'Spec error: graph spec requires an "edges" array',
+      path: 'edges',
+      code: 'MISSING_FIELD',
+      suggestion: 'Add an edges array (can be empty), e.g. edges: [{ source: "a", target: "b" }]',
+    });
+    return;
+  }
+
+  // Validate each edge has string source and target that reference existing nodes
+  const edges = spec.edges as Record<string, unknown>[];
+  for (let i = 0; i < edges.length; i++) {
+    const edge = edges[i];
+    if (!edge || typeof edge !== 'object') {
+      errors.push({
+        message: `Spec error: edges[${i}] must be an object`,
+        path: `edges[${i}]`,
+        code: 'INVALID_TYPE',
+        suggestion:
+          'Each edge must be an object with "source" and "target" fields, e.g. { source: "a", target: "b" }',
+      });
+      continue;
+    }
+
+    if (typeof edge.source !== 'string' || edge.source === '') {
+      errors.push({
+        message: `Spec error: edges[${i}] must have a non-empty string "source" field`,
+        path: `edges[${i}].source`,
+        code: 'MISSING_FIELD',
+        suggestion: 'Add a source node id, e.g. { source: "a", target: "b" }',
+      });
+    } else if (nodeIds.size > 0 && !nodeIds.has(edge.source)) {
+      errors.push({
+        message: `Spec error: edges[${i}].source "${edge.source}" does not reference an existing node id`,
+        path: `edges[${i}].source`,
+        code: 'DATA_FIELD_MISSING',
+        suggestion: `Use one of the existing node ids: ${[...nodeIds].slice(0, 5).join(', ')}${nodeIds.size > 5 ? '...' : ''}`,
+      });
+    }
+
+    if (typeof edge.target !== 'string' || edge.target === '') {
+      errors.push({
+        message: `Spec error: edges[${i}] must have a non-empty string "target" field`,
+        path: `edges[${i}].target`,
+        code: 'MISSING_FIELD',
+        suggestion: 'Add a target node id, e.g. { source: "a", target: "b" }',
+      });
+    } else if (nodeIds.size > 0 && !nodeIds.has(edge.target)) {
+      errors.push({
+        message: `Spec error: edges[${i}].target "${edge.target}" does not reference an existing node id`,
+        path: `edges[${i}].target`,
+        code: 'DATA_FIELD_MISSING',
+        suggestion: `Use one of the existing node ids: ${[...nodeIds].slice(0, 5).join(', ')}${nodeIds.size > 5 ? '...' : ''}`,
+      });
+    }
+  }
+
+  // Validate encoding fields exist on at least the first node/edge
+  if (spec.encoding && typeof spec.encoding === 'object') {
+    const encoding = spec.encoding as Record<string, unknown>;
+    const firstNode = nodes[0] as Record<string, unknown>;
+    const firstEdge = edges.length > 0 ? (edges[0] as Record<string, unknown>) : null;
+    const nodeFields = firstNode ? new Set(Object.keys(firstNode)) : new Set<string>();
+    const edgeFields = firstEdge ? new Set(Object.keys(firstEdge)) : new Set<string>();
+
+    const nodeChannels = ['nodeColor', 'nodeSize', 'nodeLabel'] as const;
+    for (const channel of nodeChannels) {
+      const ch = encoding[channel] as Record<string, unknown> | undefined;
+      if (ch?.field && typeof ch.field === 'string' && !nodeFields.has(ch.field)) {
+        errors.push({
+          message: `Spec error: encoding.${channel}.field "${ch.field}" does not exist on nodes. Available fields: ${[...nodeFields].join(', ')}`,
+          path: `encoding.${channel}.field`,
+          code: 'DATA_FIELD_MISSING',
+          suggestion: `Use one of the node fields: ${[...nodeFields].join(', ')}`,
+        });
+      }
+    }
+
+    const edgeChannels = ['edgeColor', 'edgeWidth'] as const;
+    for (const channel of edgeChannels) {
+      const ch = encoding[channel] as Record<string, unknown> | undefined;
+      if (ch?.field && typeof ch.field === 'string' && firstEdge && !edgeFields.has(ch.field)) {
+        errors.push({
+          message: `Spec error: encoding.${channel}.field "${ch.field}" does not exist on edges. Available fields: ${[...edgeFields].join(', ')}`,
+          path: `encoding.${channel}.field`,
+          code: 'DATA_FIELD_MISSING',
+          suggestion: `Use one of the edge fields: ${[...edgeFields].join(', ')}`,
+        });
+      }
+    }
+  }
+
+  // Validate layout type if specified
+  if (spec.layout && typeof spec.layout === 'object') {
+    const layout = spec.layout as Record<string, unknown>;
+    if (layout.type && layout.type !== 'force') {
+      errors.push({
+        message: `Spec error: layout.type "${layout.type}" is not supported. Only "force" is currently supported`,
+        path: 'layout.type',
+        code: 'INVALID_VALUE',
+        suggestion:
+          'Use layout.type: "force" or omit the layout field to use the default force layout',
+      });
+    }
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Sankey validation
+// ---------------------------------------------------------------------------
+
+function validateSankeySpec(spec: Record<string, unknown>, errors: ValidationError[]): void {
+  // Validate data
+  if (!Array.isArray(spec.data)) {
+    errors.push({
+      message: 'Spec error: sankey spec requires a "data" array',
+      path: 'data',
+      code: 'INVALID_TYPE',
+      suggestion:
+        'Provide data as an array of objects, e.g. data: [{ source: "A", target: "B", value: 10 }]',
+    });
+    return;
+  }
+
+  if (spec.data.length === 0) {
+    errors.push({
+      message: 'Spec error: "data" must be a non-empty array',
+      path: 'data',
+      code: 'EMPTY_DATA',
+      suggestion: 'Add at least one data row, e.g. data: [{ source: "A", target: "B", value: 10 }]',
+    });
+    return;
+  }
+
+  const firstRow = spec.data[0] as unknown;
+  if (typeof firstRow !== 'object' || firstRow === null || Array.isArray(firstRow)) {
+    errors.push({
+      message: 'Spec error: each item in "data" must be a plain object',
+      path: 'data[0]',
+      code: 'INVALID_TYPE',
+      suggestion:
+        'Each data item should be an object, e.g. { source: "A", target: "B", value: 10 }',
+    });
+    return;
+  }
+
+  // Validate encoding
+  if (!spec.encoding || typeof spec.encoding !== 'object') {
+    errors.push({
+      message:
+        'Spec error: sankey spec requires an "encoding" object with source, target, and value channels',
+      path: 'encoding',
+      code: 'MISSING_FIELD',
+      suggestion:
+        'Add an encoding object, e.g. encoding: { source: { field: "source", type: "nominal" }, target: { field: "target", type: "nominal" }, value: { field: "value", type: "quantitative" } }',
+    });
+    return;
+  }
+
+  const encoding = spec.encoding as Record<string, unknown>;
+  const dataColumns = new Set(Object.keys(firstRow as Record<string, unknown>));
+  const availableColumns = [...dataColumns].join(', ');
+
+  // Required channels
+  for (const channel of ['source', 'target', 'value'] as const) {
+    const ch = encoding[channel] as Record<string, unknown> | undefined;
+    if (!ch || typeof ch !== 'object') {
+      errors.push({
+        message: `Spec error: sankey encoding requires "${channel}" channel`,
+        path: `encoding.${channel}`,
+        code: 'MISSING_FIELD',
+        suggestion: `Add encoding.${channel} with a field from your data (${availableColumns}). Example: ${channel}: { field: "${[...dataColumns][0] ?? 'myField'}", type: "${channel === 'value' ? 'quantitative' : 'nominal'}" }`,
+      });
+      continue;
+    }
+
+    if (!ch.field || typeof ch.field !== 'string') {
+      errors.push({
+        message: `Spec error: encoding.${channel} must have a "field" string`,
+        path: `encoding.${channel}.field`,
+        code: 'MISSING_FIELD',
+        suggestion: `Add a field name from your data columns: ${availableColumns}`,
+      });
+      continue;
+    }
+
+    if (!dataColumns.has(ch.field as string)) {
+      errors.push({
+        message: `Spec error: encoding.${channel}.field "${ch.field}" does not exist in data. Available columns: ${availableColumns}`,
+        path: `encoding.${channel}.field`,
+        code: 'DATA_FIELD_MISSING',
+        suggestion: `Use one of the available data columns: ${availableColumns}`,
+      });
+    }
+  }
+
+  // Validate darkMode if provided
+  if (spec.darkMode !== undefined && !VALID_DARK_MODES.has(spec.darkMode as string)) {
+    errors.push({
+      message: 'Spec error: darkMode must be "auto", "force", or "off"',
+      path: 'darkMode',
+      code: 'INVALID_VALUE',
+      suggestion:
+        'Use one of: "auto" (system preference), "force" (always dark), or "off" (always light)',
+    });
+  }
+}
+
+// ---------------------------------------------------------------------------
+// TileMap validation
+// ---------------------------------------------------------------------------
+
+function validateTileMapSpec(spec: Record<string, unknown>, errors: ValidationError[]): void {
+  // Validate data (can be record or array)
+  if (!spec.data || typeof spec.data !== 'object') {
+    errors.push({
+      message: 'Spec error: tilemap spec requires a "data" field (record or array)',
+      path: 'data',
+      code: 'INVALID_TYPE',
+      suggestion:
+        'Provide data as either a record mapping state codes to values (e.g. { "CA": 12000, "TX": 8500 }) or an array of objects with state and value fields',
+    });
+    return;
+  }
+
+  // If data is an object (record), validate it has at least one entry
+  if (!Array.isArray(spec.data) && Object.keys(spec.data as Record<string, unknown>).length === 0) {
+    errors.push({
+      message: 'Spec error: "data" must have at least one entry',
+      path: 'data',
+      code: 'EMPTY_DATA',
+      suggestion: 'Add at least one state-value pair, e.g. { "CA": 12000 }',
+    });
+    return;
+  }
+
+  // If data is an array, validate it's non-empty
+  if (Array.isArray(spec.data)) {
+    if (spec.data.length === 0) {
+      errors.push({
+        message: 'Spec error: "data" array must be non-empty',
+        path: 'data',
+        code: 'EMPTY_DATA',
+        suggestion: 'Add at least one data row',
+      });
+      return;
+    }
+
+    const firstRow = spec.data[0] as unknown;
+    if (typeof firstRow !== 'object' || firstRow === null || Array.isArray(firstRow)) {
+      errors.push({
+        message: 'Spec error: each item in "data" must be a plain object',
+        path: 'data[0]',
+        code: 'INVALID_TYPE',
+        suggestion: 'Each data item should be an object, e.g. { state: "CA", value: 12000 }',
+      });
+      return;
+    }
+
+    // If data is array, encoding is required
+    if (!spec.encoding || typeof spec.encoding !== 'object') {
+      errors.push({
+        message:
+          'Spec error: tilemap spec with array data requires an "encoding" object with state and value channels',
+        path: 'encoding',
+        code: 'MISSING_FIELD',
+        suggestion:
+          'Add an encoding object, e.g. encoding: { state: { field: "state", type: "nominal" }, value: { field: "value", type: "quantitative" } }',
+      });
+      return;
+    }
+
+    const encoding = spec.encoding as Record<string, unknown>;
+    const dataColumns = new Set(Object.keys(firstRow as Record<string, unknown>));
+    const availableColumns = [...dataColumns].join(', ');
+
+    // Required channels
+    for (const channel of ['state', 'value'] as const) {
+      const ch = encoding[channel] as Record<string, unknown> | undefined;
+      if (!ch || typeof ch !== 'object') {
+        errors.push({
+          message: `Spec error: tilemap encoding requires "${channel}" channel`,
+          path: `encoding.${channel}`,
+          code: 'MISSING_FIELD',
+          suggestion: `Add encoding.${channel} with a field from your data (${availableColumns}). Example: ${channel}: { field: "${[...dataColumns][0] ?? 'myField'}", type: "${channel === 'value' ? 'quantitative' : 'nominal'}" }`,
+        });
+        continue;
+      }
+
+      if (!ch.field || typeof ch.field !== 'string') {
+        errors.push({
+          message: `Spec error: encoding.${channel} must have a "field" string`,
+          path: `encoding.${channel}.field`,
+          code: 'MISSING_FIELD',
+          suggestion: `Add a field name from your data columns: ${availableColumns}`,
+        });
+        continue;
+      }
+
+      if (!dataColumns.has(ch.field as string)) {
+        errors.push({
+          message: `Spec error: encoding.${channel}.field "${ch.field}" does not exist in data. Available columns: ${availableColumns}`,
+          path: `encoding.${channel}.field`,
+          code: 'DATA_FIELD_MISSING',
+          suggestion: `Use one of the available data columns: ${availableColumns}`,
+        });
+      }
+    }
+  }
+
+  // Validate darkMode if provided
+  if (spec.darkMode !== undefined && !VALID_DARK_MODES.has(spec.darkMode as string)) {
+    errors.push({
+      message: 'Spec error: darkMode must be "auto", "force", or "off"',
+      path: 'darkMode',
+      code: 'INVALID_VALUE',
+      suggestion:
+        'Use one of: "auto" (system preference), "force" (always dark), or "off" (always light)',
+    });
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Map validation
+// ---------------------------------------------------------------------------
+
+function validateMapSpec(spec: Record<string, unknown>, errors: ValidationError[]): void {
+  if (!spec.geo || typeof spec.geo !== 'object') {
+    errors.push({
+      message: 'Spec error: map spec requires a "geo" object with a TopoJSON "features" field',
+      path: 'geo',
+      code: 'MISSING_FIELD',
+      suggestion: 'Add a geo object, e.g. geo: { features: topoJsonData, projection: "identity" }',
+    });
+    return;
+  }
+
+  const geo = spec.geo as Record<string, unknown>;
+  if (!geo.features) {
+    errors.push({
+      message: 'Spec error: geo requires a "features" field containing TopoJSON topology',
+      path: 'geo.features',
+      code: 'MISSING_FIELD',
+      suggestion:
+        'Import a TopoJSON file and pass it as geo.features, e.g. import topo from "us-atlas/states-albers-10m.json"',
+    });
+    return;
+  }
+
+  if (!spec.encoding || typeof spec.encoding !== 'object') {
+    errors.push({
+      message: 'Spec error: map spec requires an "encoding" object with key and color channels',
+      path: 'encoding',
+      code: 'MISSING_FIELD',
+      suggestion:
+        'Add an encoding object, e.g. encoding: { key: { field: "id", type: "nominal" }, color: { field: "value", type: "quantitative" } }',
+    });
+    return;
+  }
+
+  // Validate geo.focus if present
+  if ('focus' in geo && geo.focus !== undefined && geo.focus !== null) {
+    const focus = geo.focus;
+    if (typeof focus === 'string' || typeof focus === 'number') {
+      // valid: single id
+    } else if (Array.isArray(focus)) {
+      for (let i = 0; i < focus.length; i++) {
+        if (typeof focus[i] !== 'string' && typeof focus[i] !== 'number') {
+          errors.push({
+            message: `Spec error: geo.focus array element [${i}] must be a string or number`,
+            path: `geo.focus[${i}]`,
+            code: 'INVALID_TYPE',
+            suggestion: 'Each element in the focus array should be a feature id (string or number)',
+          });
+        }
+      }
+    } else if (typeof focus === 'object') {
+      const focusObj = focus as Record<string, unknown>;
+      if (!('features' in focusObj) || focusObj.features == null) {
+        errors.push({
+          message: 'Spec error: geo.focus object must have a "features" property',
+          path: 'geo.focus.features',
+          code: 'MISSING_FIELD',
+          suggestion:
+            'Provide feature id(s) to focus on, e.g. geo: { focus: { features: "36", padding: 32 } }',
+        });
+      } else {
+        const feats = focusObj.features;
+        if (typeof feats !== 'string' && typeof feats !== 'number' && !Array.isArray(feats)) {
+          errors.push({
+            message:
+              'Spec error: geo.focus.features must be a string, number, or array of strings/numbers',
+            path: 'geo.focus.features',
+            code: 'INVALID_TYPE',
+            suggestion: 'Use a feature id or array of ids, e.g. "36" or ["36", "34"]',
+          });
+        } else if (Array.isArray(feats)) {
+          for (let i = 0; i < feats.length; i++) {
+            if (typeof feats[i] !== 'string' && typeof feats[i] !== 'number') {
+              errors.push({
+                message: `Spec error: geo.focus.features[${i}] must be a string or number`,
+                path: `geo.focus.features[${i}]`,
+                code: 'INVALID_TYPE',
+                suggestion: 'Each element should be a feature id (string or number)',
+              });
+            }
+          }
+        }
+        if (focusObj.padding !== undefined && typeof focusObj.padding !== 'number') {
+          errors.push({
+            message: 'Spec error: geo.focus.padding must be a number',
+            path: 'geo.focus.padding',
+            code: 'INVALID_TYPE',
+            suggestion: 'Provide padding as a number in map-local units, e.g. padding: 32',
+          });
+        }
+      }
+    } else {
+      errors.push({
+        message:
+          'Spec error: geo.focus must be a string, number, array of ids, or object with { features, padding? }',
+        path: 'geo.focus',
+        code: 'INVALID_TYPE',
+        suggestion:
+          'Use a feature id, array of ids, or { features: "36", padding: 16 }. Pass null to clear focus.',
+      });
+    }
+  }
+
+  const encoding = spec.encoding as Record<string, unknown>;
+  const hasPoints = spec.points && typeof spec.points === 'object' && !Array.isArray(spec.points);
+
+  // key is always required
+  const keyCh = encoding.key as Record<string, unknown> | undefined;
+  if (!keyCh || typeof keyCh !== 'object') {
+    errors.push({
+      message: 'Spec error: map encoding requires "key" channel',
+      path: 'encoding.key',
+      code: 'MISSING_FIELD',
+      suggestion: 'Add encoding.key, e.g. key: { field: "id", type: "nominal" }',
+    });
+  } else if (!keyCh.field || typeof keyCh.field !== 'string') {
+    errors.push({
+      message: 'Spec error: encoding.key must have a "field" string',
+      path: 'encoding.key.field',
+      code: 'MISSING_FIELD',
+      suggestion: 'Add a field name, e.g. key: { field: "id" }',
+    });
+  }
+
+  // color is required UNLESS points is present (basemap-only mode)
+  const colorCh = encoding.color as Record<string, unknown> | undefined;
+  if (!hasPoints) {
+    if (!colorCh || typeof colorCh !== 'object') {
+      errors.push({
+        message:
+          'Spec error: map encoding requires "color" channel (or add a "points" layer for basemap-only mode)',
+        path: 'encoding.color',
+        code: 'MISSING_FIELD',
+        suggestion: 'Add encoding.color, e.g. color: { field: "value", type: "quantitative" }',
+      });
+    } else if (!colorCh.field || typeof colorCh.field !== 'string') {
+      errors.push({
+        message: 'Spec error: encoding.color must have a "field" string',
+        path: 'encoding.color.field',
+        code: 'MISSING_FIELD',
+        suggestion: 'Add a field name, e.g. color: { field: "value" }',
+      });
+    }
+  } else if (
+    colorCh &&
+    typeof colorCh === 'object' &&
+    (!colorCh.field || typeof colorCh.field !== 'string')
+  ) {
+    errors.push({
+      message: 'Spec error: encoding.color must have a "field" string when provided',
+      path: 'encoding.color.field',
+      code: 'MISSING_FIELD',
+      suggestion: 'Add a field name, e.g. color: { field: "value" }',
+    });
+  }
+
+  // Validate points layer if present
+  if (hasPoints) {
+    const points = spec.points as Record<string, unknown>;
+    if (!Array.isArray(points.data)) {
+      errors.push({
+        message: 'Spec error: points.data must be an array of data rows',
+        path: 'points.data',
+        code: 'INVALID_TYPE',
+        suggestion: 'Provide data as an array of objects with lat/lng fields',
+      });
+    }
+    for (const ch of ['longitude', 'latitude'] as const) {
+      const enc = points[ch] as Record<string, unknown> | undefined;
+      if (!enc || typeof enc !== 'object') {
+        errors.push({
+          message: `Spec error: points requires "${ch}" encoding channel`,
+          path: `points.${ch}`,
+          code: 'MISSING_FIELD',
+          suggestion: `Add points.${ch}, e.g. ${ch}: { field: "${ch === 'longitude' ? 'lon' : 'lat'}", type: "quantitative" }`,
+        });
+      } else if (!enc.field || typeof enc.field !== 'string') {
+        errors.push({
+          message: `Spec error: points.${ch} must have a "field" string`,
+          path: `points.${ch}.field`,
+          code: 'MISSING_FIELD',
+          suggestion: `Add a field name, e.g. ${ch}: { field: "${ch === 'longitude' ? 'lon' : 'lat'}" }`,
+        });
+      }
+    }
+  }
+}
+
+// ---------------------------------------------------------------------------
+// BarList validation
+// ---------------------------------------------------------------------------
+
+function validateBarListSpec(spec: Record<string, unknown>, errors: ValidationError[]): void {
+  if (!Array.isArray(spec.data)) {
+    errors.push({
+      message: 'Spec error: barlist spec requires a "data" array',
+      path: 'data',
+      code: 'INVALID_TYPE',
+      suggestion:
+        'Provide data as an array of objects, e.g. data: [{ label: "Category A", value: 42 }]',
+    });
+    return;
+  }
+
+  if (spec.data.length === 0) {
+    errors.push({
+      message: 'Spec error: "data" array must be non-empty',
+      path: 'data',
+      code: 'EMPTY_DATA',
+      suggestion: 'Add at least one data row',
+    });
+    return;
+  }
+
+  const firstRow = spec.data[0] as unknown;
+  if (typeof firstRow !== 'object' || firstRow === null || Array.isArray(firstRow)) {
+    errors.push({
+      message: 'Spec error: each item in "data" must be a plain object',
+      path: 'data[0]',
+      code: 'INVALID_TYPE',
+      suggestion: 'Each data item should be an object, e.g. { label: "Category A", value: 42 }',
+    });
+    return;
+  }
+
+  if (!spec.encoding || typeof spec.encoding !== 'object') {
+    errors.push({
+      message:
+        'Spec error: barlist spec requires an "encoding" object with label and value channels',
+      path: 'encoding',
+      code: 'MISSING_FIELD',
+      suggestion:
+        'Add an encoding object, e.g. encoding: { label: { field: "name", type: "nominal" }, value: { field: "count", type: "quantitative" } }',
+    });
+    return;
+  }
+
+  const encoding = spec.encoding as Record<string, unknown>;
+  const dataColumns = new Set(Object.keys(firstRow as Record<string, unknown>));
+  const availableColumns = [...dataColumns].join(', ');
+
+  for (const channel of ['label', 'value'] as const) {
+    const ch = encoding[channel] as Record<string, unknown> | undefined;
+    if (!ch || typeof ch !== 'object') {
+      errors.push({
+        message: `Spec error: barlist encoding requires "${channel}" channel`,
+        path: `encoding.${channel}`,
+        code: 'MISSING_FIELD',
+        suggestion: `Add encoding.${channel} with a field from your data (${availableColumns}). Example: ${channel}: { field: "${[...dataColumns][0] ?? 'myField'}", type: "${channel === 'value' ? 'quantitative' : 'nominal'}" }`,
+      });
+      continue;
+    }
+
+    if (!ch.field || typeof ch.field !== 'string') {
+      errors.push({
+        message: `Spec error: encoding.${channel} must have a "field" string`,
+        path: `encoding.${channel}.field`,
+        code: 'MISSING_FIELD',
+        suggestion: `Add a field name from your data columns: ${availableColumns}`,
+      });
+      continue;
+    }
+
+    if (!dataColumns.has(ch.field as string)) {
+      errors.push({
+        message: `Spec error: encoding.${channel}.field "${ch.field}" does not exist in data. Available columns: ${availableColumns}`,
+        path: `encoding.${channel}.field`,
+        code: 'DATA_FIELD_MISSING',
+        suggestion: `Use one of the available data columns: ${availableColumns}`,
+      });
+    }
+  }
+
+  // Validate optional encoding channels
+  for (const channel of ['subtitle', 'color', 'tooltip'] as const) {
+    const ch = encoding[channel] as Record<string, unknown> | undefined;
+    if (!ch) continue;
+    const field = ch.field;
+    if (field && typeof field === 'string' && !dataColumns.has(field)) {
+      errors.push({
+        message: `Spec error: encoding.${channel}.field "${field}" does not exist in data. Available columns: ${availableColumns}`,
+        path: `encoding.${channel}.field`,
+        code: 'DATA_FIELD_MISSING',
+        suggestion: `Use one of the available data columns: ${availableColumns}`,
+      });
+    }
+  }
+
+  if (spec.darkMode !== undefined && !VALID_DARK_MODES.has(spec.darkMode as string)) {
+    errors.push({
+      message: 'Spec error: darkMode must be "auto", "force", or "off"',
+      path: 'darkMode',
+      code: 'INVALID_VALUE',
+      suggestion:
+        'Use one of: "auto" (system preference), "force" (always dark), or "off" (always light)',
+    });
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Layer validation
+// ---------------------------------------------------------------------------
+
+function validateLayerSpec(spec: Record<string, unknown>, errors: ValidationError[]): void {
+  const layer = spec.layer as unknown[];
+
+  if (layer.length === 0) {
+    errors.push({
+      message: 'Spec error: "layer" must be a non-empty array',
+      path: 'layer',
+      code: 'EMPTY_DATA',
+      suggestion: 'Add at least one layer with a mark and encoding',
+    });
+    return;
+  }
+
+  for (let i = 0; i < layer.length; i++) {
+    const child = layer[i];
+    if (!child || typeof child !== 'object' || Array.isArray(child)) {
+      errors.push({
+        message: `Spec error: layer[${i}] must be an object`,
+        path: `layer[${i}]`,
+        code: 'INVALID_TYPE',
+        suggestion:
+          'Each layer must be a chart spec (with mark) or a nested layer spec (with layer)',
+      });
+      continue;
+    }
+
+    const childObj = child as Record<string, unknown>;
+    const isNestedLayer = 'layer' in childObj && Array.isArray(childObj.layer);
+    const isChildChart = 'mark' in childObj;
+
+    if (!isNestedLayer && !isChildChart) {
+      errors.push({
+        message: `Spec error: layer[${i}] must have a "mark" field or a "layer" array`,
+        path: `layer[${i}]`,
+        code: 'MISSING_FIELD',
+        suggestion:
+          'Each layer must be a chart spec (with mark + encoding) or a nested layer spec (with layer array)',
+      });
+      continue;
+    }
+
+    if (isNestedLayer) {
+      validateLayerSpec(childObj, errors);
+    } else if (isChildChart) {
+      // Validate mark type
+      const mark = childObj.mark;
+      let markValue: string | undefined;
+      if (typeof mark === 'string') {
+        markValue = mark;
+      } else if (mark && typeof mark === 'object' && !Array.isArray(mark)) {
+        markValue = (mark as Record<string, unknown>).type as string | undefined;
+      }
+
+      if (!markValue || !MARK_TYPES.has(markValue)) {
+        errors.push({
+          message: `Spec error: layer[${i}].mark "${markValue ?? String(mark)}" is not a valid mark type`,
+          path: `layer[${i}].mark`,
+          code: 'INVALID_VALUE',
+          suggestion: `Change mark to one of: ${[...MARK_TYPES].join(', ')}`,
+        });
+        continue;
+      }
+
+      // Child layers can inherit data and encoding from parent, so only validate
+      // if the child has its own data (or the parent provides shared data).
+      const hasOwnData = Array.isArray(childObj.data) && (childObj.data as unknown[]).length > 0;
+      const parentHasData = Array.isArray(spec.data) && (spec.data as unknown[]).length > 0;
+
+      if (hasOwnData || parentHasData) {
+        // Build a merged spec for validation purposes
+        const mergedForValidation = { ...childObj };
+        if (!hasOwnData && parentHasData) {
+          mergedForValidation.data = spec.data;
+        }
+        // Merge encoding: parent fields are inherited unless child overrides
+        if (spec.encoding && typeof spec.encoding === 'object') {
+          mergedForValidation.encoding = {
+            ...(spec.encoding as Record<string, unknown>),
+            ...((childObj.encoding as Record<string, unknown>) ?? {}),
+          };
+        }
+        if (mergedForValidation.data && mergedForValidation.encoding) {
+          validateChartSpec(mergedForValidation, errors);
+        }
+      }
+    }
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Public API
+// ---------------------------------------------------------------------------
+
+/**
+ * Validate a spec at runtime.
+ *
+ * Checks structure, required fields, encoding rules, data shape, and
+ * field type compatibility. Returns structured errors with machine-readable
+ * codes and actionable suggestions for each problem found.
+ */
+export function validateSpec(spec: unknown): ValidationResult {
+  const errors: ValidationError[] = [];
+
+  // Basic shape check
+  if (!spec || typeof spec !== 'object' || Array.isArray(spec)) {
+    return {
+      valid: false,
+      errors: [
+        {
+          message: 'Spec error: spec must be a non-null object',
+          code: 'INVALID_TYPE',
+          suggestion:
+            'Pass a spec object with at least a "mark" field for charts, e.g. { mark: "line", data: [...], encoding: {...} }',
+        },
+      ],
+      normalized: null,
+    };
+  }
+
+  const obj = spec as Record<string, unknown>;
+
+  // Determine spec type via structural discrimination:
+  // - Layer specs have a 'layer' array
+  // - Chart specs have a 'mark' field (string or object with type property)
+  // - Table specs have type: 'table'
+  // - Graph specs have type: 'graph'
+  // - Sankey specs have type: 'sankey'
+  // - TileMap specs have type: 'tilemap'
+  const hasLayer = 'layer' in obj && Array.isArray(obj.layer);
+  const hasMark = 'mark' in obj;
+  const isTable = obj.type === 'table';
+  const isGraph = obj.type === 'graph';
+  const isSankey = obj.type === 'sankey';
+  const isTileMap = obj.type === 'tilemap';
+  const isMap = obj.type === 'map';
+  const isBarList = obj.type === 'barlist';
+  const isLayer =
+    hasLayer && !isTable && !isGraph && !isSankey && !isTileMap && !isMap && !isBarList;
+  const isChart =
+    hasMark && !hasLayer && !isTable && !isGraph && !isSankey && !isTileMap && !isMap && !isBarList;
+
+  if (
+    !isChart &&
+    !isTable &&
+    !isGraph &&
+    !isSankey &&
+    !isTileMap &&
+    !isMap &&
+    !isBarList &&
+    !isLayer
+  ) {
+    // Near-misses for VL composition operators that are unsupported by decision
+    if ('hconcat' in obj || 'vconcat' in obj) {
+      const key = 'hconcat' in obj ? 'hconcat' : 'vconcat';
+      return {
+        valid: false,
+        errors: [
+          {
+            message: `Spec error: "${key}" composition is not supported`,
+            path: key,
+            code: 'INVALID_VALUE',
+            suggestion:
+              'Render each chart in its own container instead. For small multiples of the same chart, use the facet encoding channel: encoding.facet = { field, type }',
+          },
+        ],
+        normalized: null,
+      };
+    }
+    if ('facet' in obj && 'spec' in obj) {
+      return {
+        valid: false,
+        errors: [
+          {
+            message: 'Spec error: the top-level "facet" operator is not supported',
+            path: 'facet',
+            code: 'INVALID_VALUE',
+            suggestion:
+              'Use the facet encoding channel on a regular chart spec instead: encoding.facet = { field: "category", type: "nominal" }',
+          },
+        ],
+        normalized: null,
+      };
+    }
+    return {
+      valid: false,
+      errors: [
+        {
+          message:
+            'Spec error: spec must have a "mark" field for charts, a "layer" array for layered charts, or a "type" field for tables/graphs/sankey/tilemap/map/barlist',
+          path: 'mark',
+          code: 'MISSING_FIELD',
+          suggestion: `Add a "mark" field for charts (e.g. mark: "bar"), a "layer" array for layered charts, or a "type" field (type: "table", type: "graph", type: "sankey", type: "tilemap", type: "map", or type: "barlist"). Valid mark types: ${[...MARK_TYPES].join(', ')}`,
+        },
+      ],
+      normalized: null,
+    };
+  }
+
+  // For layer specs, validate each child layer recursively
+  if (isLayer) {
+    validateLayerSpec(obj, errors);
+  }
+
+  // For chart specs, validate the mark field
+  if (isChart) {
+    const mark = obj.mark;
+    let markValue: string | undefined;
+
+    if (typeof mark === 'string') {
+      markValue = mark;
+    } else if (mark && typeof mark === 'object' && !Array.isArray(mark)) {
+      markValue = (mark as Record<string, unknown>).type as string | undefined;
+    }
+
+    if (!markValue || !MARK_TYPES.has(markValue)) {
+      return {
+        valid: false,
+        errors: [
+          {
+            message: `Spec error: "${markValue ?? String(mark)}" is not a valid mark type. Valid mark types: ${[...MARK_TYPES].join(', ')}`,
+            path: 'mark',
+            code: 'INVALID_VALUE',
+            suggestion: `Change mark to one of: ${[...MARK_TYPES].join(', ')}`,
+          },
+        ],
+        normalized: null,
+      };
+    }
+
+    validateChartSpec(obj, errors);
+  } else if (isTable) {
+    validateTableSpec(obj, errors);
+  } else if (isGraph) {
+    validateGraphSpec(obj, errors);
+  } else if (isSankey) {
+    validateSankeySpec(obj, errors);
+  } else if (isTileMap) {
+    validateTileMapSpec(obj, errors);
+  } else if (isMap) {
+    validateMapSpec(obj, errors);
+  } else if (isBarList) {
+    validateBarListSpec(obj, errors);
+  }
+
+  if (errors.length > 0) {
+    return { valid: false, errors, normalized: null };
+  }
+
+  return {
+    valid: true,
+    errors: [],
+    normalized: spec as VizSpec,
+  };
+}