feat(formula): align CEL stdlib catalog with runtime (#1928)

The authoring catalog (introspectScope / CEL_STDLIB_FUNCTIONS) advertised 25
functions but only 8 were registered; 14 faulted at runtime — authors (incl.
AI) were told to call functions that don't exist (daysBetween, abs, round, min,
max, upper, lower, len, isEmpty, contains, startsWith, endsWith, matches,
date/datetime).

Register the genuinely-useful set in registerStdLib (dyn-lenient sigs + internal
coercion so a Field.date string arg still works) and reconcile the catalog so
every advertised entry resolves. Adds a drift-guard test evaluating every
CEL_STDLIB_FUNCTIONS entry. Pure additions — no change to existing expressions.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: os-zhuang

.changeset/cel-stdlib-catalog-align.md

@@ -0,0 +1,17 @@
+---
+"@objectstack/formula": minor
+---
+
+feat(formula): register the CEL functions the authoring catalog advertises (daysBetween, abs, round, min, max, upper, lower, contains, startsWith, endsWith, matches, len, isEmpty, date, datetime)
+
+`introspectScope` / `CEL_STDLIB_FUNCTIONS` advertised 25 functions to authors
+(incl. AI), but only 8 were registered — 14 faulted at runtime (`daysBetween`,
+`abs`, `round`, `min`, `max`, `upper`, `lower`, `len`, `isEmpty`, `contains`,
+`startsWith`, `endsWith`, `matches`, plus `date`/`datetime`). Authors were told
+to call functions that don't exist (e.g. `daysBetween` for "days remaining").
+
+Register the genuinely-useful set in `registerStdLib` with dyn-lenient signatures
+(so a `Field.date` arriving as a string still works) and internal coercion, and
+reconcile the catalog so every advertised entry resolves — guarded by a test that
+evaluates every `CEL_STDLIB_FUNCTIONS` entry. Pure additions; no behavior change
+to existing expressions.

packages/formula/src/cel-engine.test.ts

@@ -1,6 +1,7 @@
 import { describe, expect, it } from 'vitest';
 
 import { celEngine } from './cel-engine';
+import { CEL_STDLIB_FUNCTIONS } from './validate';
 import type { Expression } from '@objectstack/spec';
 
 const cel = (source: string): Expression => ({ dialect: 'cel', source });
@@ -296,4 +297,77 @@ describe('celEngine', () => {
       expect(r).toEqual({ ok: true, value: true });
     });
   });
+
+  // #1928 follow-up — the authoring catalog (`introspectScope`) advertised 25
+  // functions but only 8 were registered; 14 faulted at runtime. These cover the
+  // newly-registered stdlib, plus a drift-guard that every advertised function resolves.
+  describe('stdlib catalog (registered functions)', () => {
+    const now = new Date('2026-06-16T00:00:00Z');
+
+    it('daysBetween counts whole days (negative when reversed)', () => {
+      expect(celEngine.evaluate(cel('daysBetween(today(), daysFromNow(7))'), { now }))
+        .toEqual({ ok: true, value: 7 });
+      expect(celEngine.evaluate(cel('daysBetween(today(), daysAgo(3))'), { now }))
+        .toEqual({ ok: true, value: -3 });
+    });
+
+    it('daysBetween coerces a date-string field arg (no manual hydration)', () => {
+      const r = celEngine.evaluate(cel('daysBetween(today(), record.due)'), {
+        now, record: { due: '2026-06-26' },
+      });
+      expect(r).toEqual({ ok: true, value: 10 });
+    });
+
+    it('date / datetime parse ISO strings to a timestamp', () => {
+      const r = celEngine.evaluate(cel('date("2026-03-15") < datetime("2026-03-16T08:00:00Z")'), {});
+      expect(r).toEqual({ ok: true, value: true });
+    });
+
+    it('abs / round / min / max', () => {
+      expect(celEngine.evaluate(cel('abs(record.x)'), { record: { x: -3.5 } })).toEqual({ ok: true, value: 3.5 });
+      expect(celEngine.evaluate(cel('round(2.6)'), {})).toEqual({ ok: true, value: 3 });
+      expect(celEngine.evaluate(cel('min(record.a, record.b)'), { record: { a: 3, b: 7 } })).toEqual({ ok: true, value: 3 });
+      expect(celEngine.evaluate(cel('max(record.a, record.b)'), { record: { a: 3, b: 7 } })).toEqual({ ok: true, value: 7 });
+    });
+
+    it('string ops: upper / lower / contains / startsWith / endsWith / matches', () => {
+      expect(celEngine.evaluate(cel('upper(record.s)'), { record: { s: 'hi' } })).toEqual({ ok: true, value: 'HI' });
+      expect(celEngine.evaluate(cel('lower("HI")'), {})).toEqual({ ok: true, value: 'hi' });
+      expect(celEngine.evaluate(cel('contains(record.s, "ell")'), { record: { s: 'hello' } })).toEqual({ ok: true, value: true });
+      expect(celEngine.evaluate(cel('startsWith("hello", "he")'), {})).toEqual({ ok: true, value: true });
+      expect(celEngine.evaluate(cel('endsWith("hello", "lo")'), {})).toEqual({ ok: true, value: true });
+      expect(celEngine.evaluate(cel('matches("a1", "a.")'), {})).toEqual({ ok: true, value: true });
+    });
+
+    it('len / isEmpty over strings and lists', () => {
+      expect(celEngine.evaluate(cel('len(record.items)'), { record: { items: [1, 2, 3] } })).toEqual({ ok: true, value: 3 });
+      expect(celEngine.evaluate(cel('isEmpty("")'), {})).toEqual({ ok: true, value: true });
+      expect(celEngine.evaluate(cel('isEmpty(record.items)'), { record: { items: [] } })).toEqual({ ok: true, value: true });
+      expect(celEngine.evaluate(cel('isEmpty(record.items)'), { record: { items: [1] } })).toEqual({ ok: true, value: false });
+    });
+
+    // Drift guard: introspectScope promises these to authors; every one must resolve.
+    it('every advertised CEL_STDLIB_FUNCTIONS entry resolves at runtime', () => {
+      const call: Record<string, string> = {
+        now: 'now()', today: 'today()', daysFromNow: 'daysFromNow(30)', daysAgo: 'daysAgo(7)',
+        daysBetween: 'daysBetween(today(), daysFromNow(7))', date: 'date("2026-03-15")',
+        datetime: 'datetime("2026-03-15T08:00:00Z")', abs: 'abs(-3.5)', round: 'round(2.6)',
+        min: 'min(1, 2)', max: 'max(1, 2)', upper: 'upper("hi")', lower: 'lower("HI")',
+        trim: 'trim(" x ")', contains: 'contains("hello", "ell")', startsWith: 'startsWith("hi", "h")',
+        endsWith: 'endsWith("hi", "i")', matches: 'matches("a1", "a.")', joinNonEmpty: 'joinNonEmpty(["a", "b"], "-")',
+        isBlank: 'isBlank("")', isEmpty: 'isEmpty([])', coalesce: 'coalesce(null, "x")', len: 'len("ab")',
+        size: 'size([1, 2])', has: 'has(record.a)', int: 'int("3")', string: 'string(3)',
+        bool: 'bool("true")', double: 'double("3.5")', timestamp: 'timestamp("2026-01-01T00:00:00Z")',
+        duration: 'duration("3600s")',
+      };
+      const unresolved: string[] = [];
+      for (const fn of CEL_STDLIB_FUNCTIONS) {
+        const src = call[fn];
+        expect(src, `no probe call defined for advertised function \`${fn}\``).toBeTruthy();
+        const r = celEngine.evaluate(cel(src), { now, record: { a: 1 } });
+        if (!r.ok) unresolved.push(`${fn}: ${r.error.message.split('\n')[0]}`);
+      }
+      expect(unresolved, `advertised functions that fault at runtime:\n${unresolved.join('\n')}`).toEqual([]);
+    });
+  });
 });

packages/formula/src/stdlib.ts

@@ -22,6 +22,16 @@ function startOfDayUtc(d: Date): Date {
   return out;
 }
 
+/** Coerce a CEL value (Date | ISO string | epoch number) to a Date. */
+function toDate(v: unknown): Date {
+  if (v instanceof Date) return v;
+  if (typeof v === 'number' || typeof v === 'bigint') return new Date(Number(v));
+  return new Date(String(v));
+}
+
+/** One UTC day in milliseconds. */
+const MS_PER_DAY = 86_400_000;
+
 /** Add `n` days to a Date in UTC; returns a new Date. */
 function addDaysUtc(d: Date, n: number): Date {
   const out = new Date(d.getTime());
@@ -101,9 +111,58 @@ export function registerStdLib(
         }
         return parts.join(separator);
       },
+    )
+    // ── Dates ────────────────────────────────────────────────────────────
+    // Whole days from `a` to `b` (negative if `b` is before `a`). The common
+    // shape is `daysBetween(today(), record.due)` for "days remaining". Args are
+    // coerced (Date | ISO string | epoch) so a `Field.date` that arrives as a
+    // string still works without the caller hydrating it.
+    .registerFunction(
+      'daysBetween(dyn, dyn): int',
+      (a: unknown, b: unknown) =>
+        BigInt(Math.round((toDate(b).getTime() - toDate(a).getTime()) / MS_PER_DAY)),
+    )
+    // Parse an ISO date / date-time string to a Timestamp. `date` and `datetime`
+    // are aliases — both accept either form (the field's own type decides the
+    // intent); kept distinct because authors reach for whichever reads clearer.
+    .registerFunction('date(dyn): google.protobuf.Timestamp', (s: unknown) => toDate(s))
+    .registerFunction('datetime(dyn): google.protobuf.Timestamp', (s: unknown) => toDate(s))
+    // ── Numbers ──────────────────────────────────────────────────────────
+    .registerFunction('abs(dyn): double', (x: unknown) => Math.abs(Number(x)))
+    .registerFunction('round(dyn): int', (x: unknown) => BigInt(Math.round(Number(x))))
+    // min/max return the smaller/larger operand verbatim (type preserved) rather
+    // than a coerced copy, so `min(record.a, record.b)` keeps int-ness when both
+    // are ints. Comparison is numeric.
+    .registerFunction('min(dyn, dyn): dyn', (a: unknown, b: unknown) => (Number(a) <= Number(b) ? a : b))
+    .registerFunction('max(dyn, dyn): dyn', (a: unknown, b: unknown) => (Number(a) >= Number(b) ? a : b))
+    // ── Strings ──────────────────────────────────────────────────────────
+    // Free-function forms of the common string ops. CEL also exposes some as
+    // receiver methods (`s.contains(x)`), but the authoring catalog advertises
+    // the bare-call form, so register it to match what authors are told to use.
+    .registerFunction('upper(dyn): string', (s: unknown) => String(s ?? '').toUpperCase())
+    .registerFunction('lower(dyn): string', (s: unknown) => String(s ?? '').toLowerCase())
+    .registerFunction('contains(dyn, dyn): bool', (s: unknown, sub: unknown) => String(s ?? '').includes(String(sub ?? '')))
+    .registerFunction('startsWith(dyn, dyn): bool', (s: unknown, p: unknown) => String(s ?? '').startsWith(String(p ?? '')))
+    .registerFunction('endsWith(dyn, dyn): bool', (s: unknown, p: unknown) => String(s ?? '').endsWith(String(p ?? '')))
+    .registerFunction('matches(dyn, dyn): bool', (s: unknown, re: unknown) => new RegExp(String(re ?? '')).test(String(s ?? '')))
+    // ── Collections ──────────────────────────────────────────────────────
+    // `len` mirrors CEL's built-in `size()` for strings/lists/maps; `isEmpty` is
+    // the inverse-of-non-empty companion to `isBlank` (true for null, '', []).
+    .registerFunction('len(dyn): int', (v: unknown) => BigInt(lengthOf(v)))
+    .registerFunction(
+      'isEmpty(dyn): bool',
+      (v: unknown) => v === null || v === undefined || lengthOf(v) === 0,
     );
 }
 
+/** Length of a string / list / map (0 for scalars and null). */
+function lengthOf(v: unknown): number {
+  if (v === null || v === undefined) return 0;
+  if (typeof v === 'string' || Array.isArray(v)) return v.length;
+  if (typeof v === 'object') return Object.keys(v as Record<string, unknown>).length;
+  return 0;
+}
+
 /**
  * Register mixed `double <op> int` / `int <op> double` arithmetic overloads.
  *

packages/formula/src/validate.ts

@@ -256,10 +256,21 @@ export function introspectScope(role: FieldRole, schema?: ExprSchemaHint): {
   };
 }
 
-/** Public catalog of CEL stdlib functions available in expressions. */
+/**
+ * Public catalog of CEL functions available in expressions — what `introspectScope`
+ * advertises to authors (incl. AI). Every entry MUST actually resolve at runtime:
+ * either registered in `registerStdLib` or a verified cel-js built-in. Drifting this
+ * list ahead of the runtime tells the author to call functions that fault (#1928).
+ */
 export const CEL_STDLIB_FUNCTIONS: string[] = [
-  'now', 'today', 'daysFromNow', 'daysBetween', 'date', 'datetime', 'timestamp',
-  'isBlank', 'isEmpty', 'coalesce', 'len', 'size', 'int', 'float', 'string', 'bool',
-  'upper', 'lower', 'trim', 'contains', 'startsWith', 'endsWith', 'matches',
-  'has', 'min', 'max', 'abs', 'round',
+  // Dates (registered stdlib)
+  'now', 'today', 'daysFromNow', 'daysAgo', 'daysBetween', 'date', 'datetime',
+  // Numbers (registered stdlib)
+  'abs', 'round', 'min', 'max',
+  // Strings (registered stdlib)
+  'upper', 'lower', 'trim', 'contains', 'startsWith', 'endsWith', 'matches', 'joinNonEmpty',
+  // Collections / null-ish (registered stdlib)
+  'isBlank', 'isEmpty', 'coalesce', 'len',
+  // cel-js built-ins (verified to resolve)
+  'size', 'has', 'int', 'string', 'bool', 'double', 'timestamp', 'duration',
 ];