fix(objectql): the value-shape detail prescribes the key rename, not the missing-pair type error

.changeset/value-shape-detail-prefers-unrecognized-keys.md

@@ -0,0 +1,13 @@
+---
+"@objectstack/objectql": patch
+---
+
+`os migrate value-shapes` now prescribes the key rename on a legacy `{latitude, longitude}` location, instead of reporting the missing-pair type error.
+
+A value-shape rejection was read positionally — `parse.error.issues[0]` — at both places the value-shape detail is produced: the write path's warn-first / strict branch, and the exported `valueShapeViolation` the scan imports. zod reports per-member issues before the object-level `unrecognized_keys` one, so on a value whose keys were **renamed** the actionable message sorts last and was discarded. A `location` stored as `{latitude, longitude}` — the exact legacy shape the scan's own header names as one it exists to find — reported `Invalid input: expected number, received undefined`, leaving an operator to derive a rename that edit distance cannot reach (`latitude` -> `lat`), while `LocationValueSchema` had built the prescription and thrown it away.
+
+Both readers now prefer the undeclared-key issue when the rejection carries one, through a single shared helper — two readings of the same rejection drifting by one clause is how one path prescribes the rename and the other does not. The affected strings are the `os migrate value-shapes` finding `detail`, the warn-first `[value-shape]` log line, and the `invalid_value_shape` error's `detail` under strict enforcement.
+
+⛔ No verdict moves. The same values are flagged, the same writes are rejected or admitted, and the deployment gate opens on exactly the same evidence — only the operator-facing text changes.
+
+Scoped by measurement rather than by assumption: of the sixteen types these readers cover, only `location` and `address` are backed by a key-closed object schema, so only they can emit `unrecognized_keys` at all — for the other fourteen the preference cannot change a single character. Both classes it does reach curate the alias map that makes the undeclared key the more actionable half. The defect reaches `address` as well as `location`: every address member being optional rules out a *missing*-member type error, but not a *wrong-typed* declared one, which still sorts ahead of the undeclared-key issue.

packages/objectql/src/validation/record-validator.ts

@@ -487,6 +487,47 @@ export function coerceBooleanFields<T extends Record<string, unknown>>(
   return (copy ?? row) as T;
 }
 
+/**
+ * The one parse issue an author can act on, out of everything zod reported for
+ * a value-shape rejection.
+ *
+ * NOT `issues[0]`. zod reports per-member issues before the object-level
+ * `unrecognized_keys` one, so on a value whose keys were RENAMED the actionable
+ * message sorts LAST. A `location` stored as `{latitude, longitude}` — the
+ * exact legacy shape `scan-value-shapes` names in its own header as one the
+ * scan exists to find — reports:
+ *
+ *     [0] invalid_type       lat   Invalid input: expected number, received undefined
+ *     [1] invalid_type       lng   Invalid input: expected number, received undefined
+ *     [2] unrecognized_keys        ... Did you mean `latitude` -> `lat`, `longitude` -> `lng`? ...
+ *
+ * The rename IS the prescription, and edit distance cannot reach it
+ * (`latitude` -> `lat`), which is exactly why `LocationValueSchema` curates an
+ * `aliases` map. Reading positionally builds that hint and then discards it,
+ * leaving an operator running `os migrate value-shapes` to derive the rename
+ * themselves while the identically-shaped `address` case is handed it.
+ *
+ * The preference is a NO-OP for every other class rather than merely harmless
+ * to it, which is what makes it safe as a blanket rule: of the sixteen types
+ * these readers cover, only `location` and `address` are backed by a
+ * `strictObject`, so only they can emit `unrecognized_keys` at all. The
+ * reference and file-reference classes are strings, `composite` / `record` /
+ * `repeater` / `vector` are open records and arrays, `json` is `z.unknown()`,
+ * and the one deliberately loose object shape (`FileValueSchema`) never refuses
+ * a key. For the other fourteen this cannot change a single character; both
+ * classes it does reach curate the alias map that makes the undeclared key the
+ * more actionable half of the rejection.
+ *
+ * Shared by both readers for the reason `valueShapeViolation` is exported
+ * rather than re-derived one layer up: two readings of the same rejection
+ * drifting by one clause is how one path prescribes the rename and the other
+ * hands out `expected number, received undefined`.
+ */
+function valueShapeDetail(error: { issues: ReadonlyArray<{ code: string; message: string }> }): string {
+  const { issues } = error;
+  return (issues.find((i) => i.code === 'unrecognized_keys') ?? issues[0])?.message ?? 'invalid value shape';
+}
+
 function validateOne(
   name: string,
   def: FieldDef,
@@ -792,7 +833,7 @@ function validateOne(
   if (REFERENCE_VALUE_TYPES.has(t) || FILE_REFERENCE_TYPES.has(t) || STRUCTURED_JSON_TYPES.has(t)) {
     const parsed = shapeSchemaFor(def).safeParse(value);
     if (!parsed.success) {
-      const detail = parsed.error.issues[0]?.message ?? 'invalid value shape';
+      const detail = valueShapeDetail(parsed.error);
       const isMedia = FILE_REFERENCE_TYPES.has(t);
       if (isMedia ? mediaStrictEffective(mediaStrict) : valueShapeStrictEffective(valueStrict)) {
         return fail('invalid_type', { type: t, detail }, 'invalid_value_shape');
@@ -931,7 +972,9 @@ export function mediaPostureSetByEnv(): boolean {
  * not recognise, which is precisely the borrowed-evidence failure the ADR's
  * addendum forbids one layer up.
  *
- * Returns the first parse issue's message, or `null` when the value conforms.
+ * Returns the actionable parse issue's message (see `valueShapeDetail` — the
+ * undeclared-key one when present, which is the half carrying the rename),
+ * or `null` when the value conforms.
  * A value that is missing (per `isMissing`) is never a violation: absence is
  * the `required` check's business, not the shape contract's.
  */
@@ -943,7 +986,7 @@ export function valueShapeViolation(def: FieldDef, value: unknown): string | nul
   }
   const parsed = shapeSchemaFor(def).safeParse(value);
   if (parsed.success) return null;
-  return parsed.error.issues[0]?.message ?? 'invalid value shape';
+  return valueShapeDetail(parsed.error);
 }
 
 /**
@@ -1009,7 +1052,11 @@ export interface AdmittedValueShapeViolation {
   field: string;
   /** The declared field type, so a report can group by what went wrong. */
   type: string;
-  /** The first parse issue — the prescription an author acts on. */
+  /**
+   * The prescription an author acts on — `valueShapeDetail`'s reading of the
+   * rejection, the same string the warn-first log line and the
+   * `os migrate value-shapes` finding carry. Not positional.
+   */
   detail: string;
 }
 

packages/objectql/src/validation/scan-value-shapes.test.ts

@@ -176,6 +176,27 @@ describe('scanValueShapes (ADR-0104 D1 / #3438)', () => {
     ).not.toThrow();
   });
 
+  it('#15490: an undeclared key outranks a same-value member error — on address too, not just location', async () => {
+    // The sweep this fix owes measured every class the two readers cover, and
+    // found only `location` and `address` able to emit `unrecognized_keys` at
+    // all (the rest are strings, open records, arrays, or `z.unknown()`). It
+    // also refuted the reason `address` was believed immune: every member being
+    // OPTIONAL rules out a MISSING-member `invalid_type`, but says nothing about
+    // a WRONG-TYPED declared one — which still sorts ahead of the object-level
+    // issue. So the defect reaches this class too; the contrast case that made
+    // it look location-only just happened to carry no type error.
+    const engine = makeEngine({
+      contact: [{ id: 'c1', addr: { street: 5, postal_code: '98101' } }],
+    });
+    const report = await scanValueShapes(engine, silent);
+
+    expect(report.blocking).toBe(1);
+    const addr = report.findings.find((f) => f.field === 'addr')!;
+    // Positionally this rejection reads `[invalid_type(street), unrecognized_keys]`.
+    expect(addr.detail).toContain('`postal_code` \u2192 `postalCode`');
+    expect(addr.detail).not.toContain('expected string, received number');
+  });
+
   it('the scan counts exactly what strict mode rejects — one predicate, not two', async () => {
     // The anti-drift property: every value the scan flags must also be a write
     // rejection under the strict gate, and every value it passes must write.
@@ -186,6 +207,25 @@ describe('scanValueShapes (ADR-0104 D1 / #3438)', () => {
     const report = await scanValueShapes(engine, silent);
     expect(report.blocking).toBe(1);
 
+    // #15490: the fixture above was already the renamed pair, but the pin
+    // asserted only the COUNT — so the operator-facing prescription was free to
+    // be the wrong half of the rejection and every gate stayed green. zod sorts
+    // the two missing-member `invalid_type` issues (`lat`, `lng`) ahead of the
+    // object-level `unrecognized_keys` one, so a positional `issues[0]` read
+    // reported `expected number, received undefined` and threw away the rename
+    // the schema curates an `aliases` map to produce. Assert the PRESCRIPTION,
+    // not just that something was flagged: `detail` is documented as "the
+    // prescription an author acts on".
+    const geo = report.findings.find((f) => f.field === 'geo')!;
+    expect(geo.detail).toContain('`latitude` \u2192 `lat`');
+    expect(geo.detail).toContain('`longitude` \u2192 `lng`');
+    expect(geo.detail).toContain('this location value');
+    // The half that was being shown instead — naming it keeps the pin able to
+    // see a regression back to the positional read.
+    expect(geo.detail).not.toContain('expected number, received undefined');
+    // Customer-facing refusal text carries no internal issue id.
+    expect(geo.detail).not.toMatch(/#\d+/);
+
     expect(() =>
       validateRecord(OBJECTS.contact, { ...flagged }, 'update', { valueShapeStrict: true }),
     ).toThrow(ValidationError);

packages/objectql/src/validation/scan-value-shapes.ts

@@ -48,7 +48,14 @@ export interface ValueShapeFinding {
   count: number;
   /** A few offending record ids, so an operator can go and look. */
   sampleRecordIds: string[];
-  /** The first parse issue seen — the prescription an author acts on. */
+  /**
+   * The prescription an author acts on: the undeclared-key issue when the
+   * rejection carries one (it names the key AND the rename its schema
+   * curates), else the first issue seen. Deliberately not positional —
+   * zod sorts per-member issues ahead of the object-level one, so on the
+   * renamed `{latitude, longitude}` location this header names, the
+   * actionable message is last.
+   */
   detail: string;
 }