@enricai/barnacle 1.12.19 → 1.12.21

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -29,6 +29,7 @@ exports.harvestPersonaBindings = harvestPersonaBindings;
29
29
  exports.resolveCompositePersonaFields = resolveCompositePersonaFields;
30
30
  exports.collectConditionalGraphQLFieldNames = collectConditionalGraphQLFieldNames;
31
31
  exports.inferZodSchemaFromSamples = inferZodSchemaFromSamples;
32
+ exports.detectAggregateUnitBasisFindings = detectAggregateUnitBasisFindings;
32
33
  exports.selectPayloadAction = selectPayloadAction;
33
34
  exports.selectReturnAction = selectReturnAction;
34
35
  exports.selectEffectiveResponseBody = selectEffectiveResponseBody;
@@ -51,6 +52,9 @@ exports.compileActionSteps = compileActionSteps;
51
52
  exports.collectHeaderBindings = collectHeaderBindings;
52
53
  exports.deriveProducerBoundaryBindings = deriveProducerBoundaryBindings;
53
54
  exports.emitMultiStepExecuteHttp = emitMultiStepExecuteHttp;
55
+ exports.detectDrillDownFoldPlan = detectDrillDownFoldPlan;
56
+ exports.parseFoldReturnSpec = parseFoldReturnSpec;
57
+ exports.resolveFoldPlan = resolveFoldPlan;
54
58
  exports.buildContractChecklist = buildContractChecklist;
55
59
  exports.emitContractTs = emitContractTs;
56
60
  exports.assertRequiredUrlFieldsReferenced = assertRequiredUrlFieldsReferenced;
@@ -62,6 +66,7 @@ const node_path_1 = require("node:path");
62
66
  const ats_field_vocabulary_1 = require("../lib/ats-field-vocabulary");
63
67
  const errors_1 = require("../lib/errors");
64
68
  const logging_1 = require("../lib/logging");
69
+ const merge_folded_primary_bodies_1 = require("../lib/merge-folded-primary-bodies");
65
70
  const plugin_api_version_1 = require("../plugins/plugin-api-version");
66
71
  const plugin_manifest_envelope_1 = require("../plugins/plugin-manifest-envelope");
67
72
  const capture_filters_1 = require("../recon/capture-filters");
@@ -509,7 +514,7 @@ function collectConditionalGraphQLFieldNames(query) {
509
514
  * what the endpoint actually returns rather than what one capture happened to
510
515
  * show.
511
516
  */
512
- function inferZodSchemaFromSamples(samples, depth = 0, indent = "", opts = {}) {
517
+ function inferZodSchemaFromSamples(samples, depth = 0, indent = "", opts = {}, path = []) {
513
518
  const maxDepth = opts.maxDepth ?? DEFAULT_MAX_INFER_DEPTH;
514
519
  if (depth > maxDepth)
515
520
  return "z.unknown()";
@@ -547,7 +552,7 @@ function inferZodSchemaFromSamples(samples, depth = 0, indent = "", opts = {}) {
547
552
  const items = nonNull.flat();
548
553
  if (items.length === 0)
549
554
  return wrap("z.array(z.unknown())");
550
- return wrap(`z.array(${inferZodSchemaFromSamples(items, depth + 1, indent, opts)})`);
555
+ return wrap(`z.array(${inferZodSchemaFromSamples(items, depth + 1, indent, opts, path)})`);
551
556
  }
552
557
  if (kind === "object") {
553
558
  const objects = nonNull;
@@ -559,15 +564,22 @@ function inferZodSchemaFromSamples(samples, depth = 0, indent = "", opts = {}) {
559
564
  // the generated file on first lint:fix.
560
565
  const fields = keys
561
566
  .map((k) => {
567
+ const fieldPath = [...path, k];
562
568
  const valuesForKey = objects.filter((o) => k in o).map((o) => o[k]);
563
- const expr = inferZodSchemaFromSamples(valuesForKey, depth + 1, inner, opts);
569
+ const expr = inferZodSchemaFromSamples(valuesForKey, depth + 1, inner, opts, fieldPath);
564
570
  // Seen on some samples but not others (the endpoint omits it sometimes)
565
571
  // OR the query marks it @include/@skip-conditional (the server can
566
572
  // legally omit it regardless of what this sample happened to show):
567
573
  // either signal alone is enough to require callers to guard the field.
568
574
  const isOptional = valuesForKey.length < objects.length || (opts.conditionalFieldNames?.has(k) ?? false);
569
575
  const optional = isOptional ? `${expr}.optional()` : expr;
570
- return `${inner}${isValidJsIdentifier(k) ? k : JSON.stringify(k)}: ${optional}`;
576
+ const findings = opts.aggregateUnitBasisFindingsByPath?.get(fieldPath.join("."));
577
+ const described = findings?.length
578
+ ? `${optional}.describe(${JSON.stringify(`Derived: ${findings
579
+ .map((finding) => `equals the sum of "${finding.unitFieldName}" across every entry of "${finding.breakdownPath.join(".")}"`)
580
+ .join("; also ")}.`)})`
581
+ : optional;
582
+ return `${inner}${isValidJsIdentifier(k) ? k : JSON.stringify(k)}: ${described}`;
571
583
  })
572
584
  .join(",\n");
573
585
  const objectExpr = `z.object({\n${fields},\n${indent}})${opts.looseServerResponse ? ".loose()" : ""}`;
@@ -581,6 +593,198 @@ function inferZodSchemaFromSamples(samples, depth = 0, indent = "", opts = {}) {
581
593
  function inferZodSchema(value, depth = 0, indent = "", opts = {}) {
582
594
  return inferZodSchemaFromSamples([value], depth, indent, opts);
583
595
  }
596
+ /** Bound on how far {@link detectAggregateUnitBasisFindings} descends into a
597
+ * response tree, matching {@link DEFAULT_MAX_INFER_DEPTH} so detection never
598
+ * outruns the schema inference it feeds. */
599
+ const MAX_AGGREGATE_UNIT_BASIS_DEPTH = DEFAULT_MAX_INFER_DEPTH;
600
+ /** Relative tolerance for aggregate-vs-sum comparison, wide enough to absorb
601
+ * float rounding in tax/currency math but tight enough that two merely
602
+ * similar-magnitude fields won't pass by coincidence. */
603
+ const AGGREGATE_UNIT_BASIS_RELATIVE_EPSILON = 1e-6;
604
+ const AGGREGATE_UNIT_BASIS_ABSOLUTE_EPSILON = 1e-6;
605
+ function isPlainObject(value) {
606
+ return typeof value === "object" && value !== null && !Array.isArray(value);
607
+ }
608
+ /** A "map of objects" is a dynamically-keyed collection (guest index, room id,
609
+ * line-item id, ...) where every value is itself a record — the per-unit
610
+ * breakdown shape, as opposed to a plain nested object or a scalar map. */
611
+ function isMapOfObjects(value) {
612
+ if (!isPlainObject(value))
613
+ return false;
614
+ const values = Object.values(value);
615
+ if (values.length === 0)
616
+ return false;
617
+ return values.every(isPlainObject);
618
+ }
619
+ /**
620
+ * Collects, for one merged object node, every numeric field that could be an
621
+ * aggregate: fields directly on the node, and fields one level deeper inside
622
+ * a plain (non-map) nested object — the shape the report evidences
623
+ * (`price.summary.total`, where `summary` is a single object, not a
624
+ * dynamically-keyed map).
625
+ */
626
+ function collectAggregateCandidates(instances, path, keys) {
627
+ const candidates = [];
628
+ for (const key of keys) {
629
+ if (instances.some((instance) => typeof instance[key] === "number")) {
630
+ candidates.push({
631
+ fieldName: key,
632
+ path: [...path, key],
633
+ getValue: (instance) => typeof instance[key] === "number" ? instance[key] : undefined,
634
+ });
635
+ }
636
+ const nestedObjects = instances.map((instance) => instance[key]).filter(isPlainObject);
637
+ if (nestedObjects.length === 0 || nestedObjects.some(isMapOfObjects))
638
+ continue;
639
+ const innerKeys = new Set(nestedObjects.flatMap((o) => Object.keys(o)));
640
+ for (const innerKey of innerKeys) {
641
+ if (!nestedObjects.some((o) => typeof o[innerKey] === "number"))
642
+ continue;
643
+ candidates.push({
644
+ fieldName: innerKey,
645
+ path: [...path, key, innerKey],
646
+ getValue: (instance) => {
647
+ const nested = instance[key];
648
+ return isPlainObject(nested) && typeof nested[innerKey] === "number"
649
+ ? nested[innerKey]
650
+ : undefined;
651
+ },
652
+ });
653
+ }
654
+ }
655
+ return candidates;
656
+ }
657
+ function collectBreakdownCandidates(instances, keys) {
658
+ const candidates = [];
659
+ for (const key of keys) {
660
+ if (!instances.some((instance) => isMapOfObjects(instance[key])))
661
+ continue;
662
+ candidates.push({
663
+ key,
664
+ getEntries: (instance) => {
665
+ const value = instance[key];
666
+ return isMapOfObjects(value) ? Object.values(value) : undefined;
667
+ },
668
+ });
669
+ }
670
+ return candidates;
671
+ }
672
+ /**
673
+ * Checks one aggregate/breakdown candidate pair against every instance at
674
+ * this node and records a finding only when the sum relation holds in ALL
675
+ * evidence. A single sample that violates the relation, or a breakdown entry
676
+ * missing the shared field, disqualifies the pair entirely -- proving the
677
+ * relation from a subset of samples is exactly the false-positive pattern
678
+ * this detector exists to avoid (a field that merely happens to share
679
+ * magnitude with a sum on some responses).
680
+ */
681
+ function evaluateAggregateUnitBasisPair(instances, breakdownPath, aggregate, breakdown, findings) {
682
+ let confirmedSamples = 0;
683
+ let maxEntries = 0;
684
+ let sawEvidence = false;
685
+ for (const instance of instances) {
686
+ const aggregateValue = aggregate.getValue(instance);
687
+ const entries = breakdown.getEntries(instance);
688
+ if (aggregateValue === undefined || entries === undefined || entries.length === 0)
689
+ continue;
690
+ sawEvidence = true;
691
+ maxEntries = Math.max(maxEntries, entries.length);
692
+ const unitValues = entries.map((entry) => entry[aggregate.fieldName]);
693
+ if (!unitValues.every((v) => typeof v === "number"))
694
+ return;
695
+ const sum = unitValues.reduce((total, v) => total + v, 0);
696
+ const tolerance = Math.max(Math.abs(aggregateValue) * AGGREGATE_UNIT_BASIS_RELATIVE_EPSILON, AGGREGATE_UNIT_BASIS_ABSOLUTE_EPSILON);
697
+ if (Math.abs(sum - aggregateValue) > tolerance)
698
+ return;
699
+ confirmedSamples++;
700
+ }
701
+ if (!sawEvidence || confirmedSamples === 0)
702
+ return;
703
+ // A single-entry breakdown makes aggregate == unit trivially true and
704
+ // proves nothing about summation -- at least one sample must carry a real
705
+ // multi-entry breakdown for the relation to be evidence of aggregation.
706
+ if (maxEntries < 2)
707
+ return;
708
+ findings.push({
709
+ aggregatePath: aggregate.path,
710
+ breakdownPath,
711
+ unitFieldName: aggregate.fieldName,
712
+ sampleCount: confirmedSamples,
713
+ maxBreakdownEntries: maxEntries,
714
+ });
715
+ }
716
+ function detectAggregateUnitBasisAtNode(instances, path, findings) {
717
+ const keys = new Set(instances.flatMap((instance) => Object.keys(instance)));
718
+ const aggregateCandidates = collectAggregateCandidates(instances, path, keys);
719
+ const breakdownCandidates = collectBreakdownCandidates(instances, keys);
720
+ for (const aggregate of aggregateCandidates) {
721
+ // The immediate child key the aggregate was found under -- either the
722
+ // aggregate's own key (direct case) or the plain-object container it was
723
+ // found nested inside (one-level-deeper case).
724
+ const aggregateContainerKey = aggregate.path[path.length];
725
+ for (const breakdown of breakdownCandidates) {
726
+ if (aggregateContainerKey === breakdown.key)
727
+ continue;
728
+ evaluateAggregateUnitBasisPair(instances, [...path, breakdown.key], aggregate, breakdown, findings);
729
+ }
730
+ }
731
+ }
732
+ function walkAggregateUnitBasis(values, path, depth, findings) {
733
+ if (depth > MAX_AGGREGATE_UNIT_BASIS_DEPTH)
734
+ return;
735
+ const instances = values.filter(isPlainObject);
736
+ if (instances.length > 0) {
737
+ detectAggregateUnitBasisAtNode(instances, path, findings);
738
+ const keys = new Set(instances.flatMap((instance) => Object.keys(instance)));
739
+ for (const key of keys) {
740
+ const childValues = instances
741
+ .map((instance) => instance[key])
742
+ .filter((v) => v !== undefined && v !== null);
743
+ if (childValues.length === 0)
744
+ continue;
745
+ walkAggregateUnitBasis(childValues, [...path, key], depth + 1, findings);
746
+ }
747
+ }
748
+ const arrayItems = values.filter((v) => Array.isArray(v)).flat();
749
+ if (arrayItems.length > 0) {
750
+ walkAggregateUnitBasis(arrayItems, path, depth + 1, findings);
751
+ }
752
+ }
753
+ /**
754
+ * Finds every object path where a numeric field's value observably equals
755
+ * the sum of a same-named numeric field carried by every entry of a sibling
756
+ * map-of-objects field -- the party-total/per-guest-breakdown shape
757
+ * generalized to any aggregate/per-unit pair, regardless of field or site
758
+ * naming, so recon-generate can annotate the price basis instead of leaving
759
+ * every plugin author and consumer to re-derive it independently and
760
+ * diverge. See docs/architecture.md, "Why the generator annotates
761
+ * aggregate/per-unit basis instead of deriving it".
762
+ *
763
+ * Pure and read-only: walks the merged sample set the same way
764
+ * {@link inferZodSchemaFromSamples} does, but only ever compares numbers
765
+ * already present in the samples -- it emits no schema and mutates nothing.
766
+ */
767
+ function detectAggregateUnitBasisFindings(samples) {
768
+ const findings = [];
769
+ walkAggregateUnitBasis(samples.filter((s) => s !== undefined && s !== null), [], 0, findings);
770
+ return findings;
771
+ }
772
+ /** Groups {@link detectAggregateUnitBasisFindings} output by aggregate path so
773
+ * {@link inferZodSchemaFromSamples} can look findings up per-field in O(1). */
774
+ function groupAggregateUnitBasisFindingsByPath(samples) {
775
+ const byPath = new Map();
776
+ for (const finding of detectAggregateUnitBasisFindings(samples)) {
777
+ const key = finding.aggregatePath.join(".");
778
+ const existing = byPath.get(key);
779
+ if (existing) {
780
+ existing.push(finding);
781
+ }
782
+ else {
783
+ byPath.set(key, [finding]);
784
+ }
785
+ }
786
+ return byPath;
787
+ }
584
788
  function deriveMinTime(rateLimits) {
585
789
  const first = rateLimits.find((f) => f.safeRps !== null);
586
790
  return first?.safeRps ? Math.floor(1000 / first.safeRps) : 200;
@@ -683,11 +887,50 @@ function selectReturnAction(steps) {
683
887
  * submission flow's `executeHttp` and its inferred type/schema have to agree
684
888
  * on which call they describe, or the emitted type disagrees with the value
685
889
  * actually returned. Falls back to the replay body for single-endpoint sites.
890
+ *
891
+ * A resolved drill-down fold plan (see {@link resolveFoldPlan}) bypasses
892
+ * `selectReturnAction` entirely, exactly as `emitMultiStepExecuteHttp` does
893
+ * when it emits the per-item loop-and-merge — so the shape this infers has to
894
+ * be the FOLDED primary body, not the plain one, or the schema would omit
895
+ * every field the fold adds. Both resolve through `resolveFoldPlan` with the
896
+ * same `foldReturnSpec` so they can never disagree on whether a fold applies.
897
+ * When multiple plans resolve, each plan's own folded body is merged via
898
+ * `mergeFoldedPrimaryBodies` into one combined object — the same merge
899
+ * `emitMultiStepExecuteHttp` performs for `return { data }` — so the inferred
900
+ * shape and the runtime return value always describe the same call, and two
901
+ * plans that share a top-level array key (e.g. the same paginated primary
902
+ * endpoint drilled into twice) both contribute their items instead of one
903
+ * clobbering the other. A plan whose folded body isn't a plain object can't
904
+ * be merged meaningfully, so in that case this falls back to the LAST plan's
905
+ * folded body alone, exactly as before this merge was introduced.
686
906
  */
687
- function selectEffectiveResponseBody(isSubmissionFlow, actionSteps, replayResponseBody) {
907
+ function selectEffectiveResponseBody(isSubmissionFlow, actionSteps, replayResponseBody, foldReturnSpec = null) {
688
908
  if (!isSubmissionFlow)
689
909
  return replayResponseBody;
690
- return selectReturnAction(actionSteps)?.capture.responseBody ?? replayResponseBody;
910
+ const foldPlans = resolveFoldPlan(actionSteps, foldReturnSpec);
911
+ const lastFoldPlan = foldPlans[foldPlans.length - 1] ?? null;
912
+ if (foldPlans.length <= 1) {
913
+ if (lastFoldPlan)
914
+ return foldResponseBodyForShapeInference(actionSteps, lastFoldPlan);
915
+ return selectReturnAction(actionSteps)?.capture.responseBody ?? replayResponseBody;
916
+ }
917
+ // Plans sharing a primaryStepIndex (independent arrays on one primary
918
+ // response) MUST fold onto the SAME accumulating body, not independent
919
+ // copies of the original — folding each in isolation and then merging via
920
+ // mergeFoldedPrimaryBodies would array-concat one plan's folded items with
921
+ // the other's still-unfolded ones, duplicating every item.
922
+ const plansByPrimaryStep = new Map();
923
+ for (const plan of foldPlans) {
924
+ plansByPrimaryStep.set(plan.primaryStepIndex, [
925
+ ...(plansByPrimaryStep.get(plan.primaryStepIndex) ?? []),
926
+ plan,
927
+ ]);
928
+ }
929
+ const foldedBodies = [...plansByPrimaryStep.values()].map((plans) => plans.reduce((body, plan) => foldResponseBodyForShapeInference(actionSteps, plan, body), actionSteps[plans[0].primaryStepIndex].capture.responseBody));
930
+ if (foldedBodies.every(isPlainObject)) {
931
+ return (0, merge_folded_primary_bodies_1.mergeFoldedPrimaryBodies)(...foldedBodies);
932
+ }
933
+ return foldResponseBodyForShapeInference(actionSteps, lastFoldPlan);
691
934
  }
692
935
  /**
693
936
  * A capture's own URL is not guaranteed parseable (see the try/catch in
@@ -1171,16 +1414,24 @@ function resolveManifestActionSequence(runRoot, captures) {
1171
1414
  * JWT refresh, reference-lookup) that a browser fires incidentally. Absent
1172
1415
  * patterns preserve the noise heuristic exactly.
1173
1416
  *
1417
+ * When the flow declares a `foldReturnSpec`, a GET whose URL matches its
1418
+ * `endpointPattern` is admitted despite the GET drop above — the same scoped
1419
+ * rule `buildFoldPlanFromSpec` later uses to resolve the fold plan, so a
1420
+ * spec-declared GET drill-down survives to reach it instead of being
1421
+ * dropped before the fold pipeline ever sees it. Every other GET is still
1422
+ * dropped.
1423
+ *
1174
1424
  * Exported for tests: this predicate decides what a generated plugin will POST
1175
1425
  * at a live site, and it is the only gate between a browser's incidental
1176
1426
  * chatter and the emitted hot path.
1177
1427
  */
1178
- function extractActionSequence(captures, submitPatterns = null) {
1428
+ function extractActionSequence(captures, submitPatterns = null, foldReturnSpec = null) {
1179
1429
  const matchesSubmit = compileSubmitMatcher(submitPatterns);
1430
+ const matchesFoldReturn = compileFoldReturnEndpointMatcher(foldReturnSpec);
1180
1431
  return captures
1181
1432
  .map((capture, index) => ({ capture, index }))
1182
1433
  .filter(({ capture }) => {
1183
- if (capture.method === "GET")
1434
+ if (capture.method === "GET" && !matchesFoldReturn(capture))
1184
1435
  return false;
1185
1436
  if (capture.status < 200 || capture.status >= 300)
1186
1437
  return false;
@@ -1202,11 +1453,23 @@ function extractActionSequence(captures, submitPatterns = null) {
1202
1453
  * what let a chronologically-first fallback pick an unrelated query. Host is
1203
1454
  * NOT a filter criterion, matching {@link extractActionSequence}.
1204
1455
  *
1456
+ * When the flow declares a `foldReturnSpec`, a non-mutation capture whose
1457
+ * URL matches its `endpointPattern` is admitted despite the query drop
1458
+ * above, mirroring {@link extractActionSequence}'s GET admission. So is a
1459
+ * non-mutation capture whose response resolves the spec's own `resultsPath`
1460
+ * — the GraphQL-primary read op the drill-down folds onto, which
1461
+ * `endpointPattern` (naming the drill, not the primary) never matches on its
1462
+ * own; without this a declared spec would admit the drill-down but leave
1463
+ * `resolveFoldPlan` with no primary capture to resolve `resultsPath`
1464
+ * against. Every other non-mutation capture is still dropped.
1465
+ *
1205
1466
  * Exported for tests: this predicate decides what a generated GraphQL plugin
1206
1467
  * will send at a live site.
1207
1468
  */
1208
- function extractGraphQLActionSequence(captures, submitPatterns = null) {
1469
+ function extractGraphQLActionSequence(captures, submitPatterns = null, foldReturnSpec = null) {
1209
1470
  const matchesSubmit = compileSubmitMatcher(submitPatterns);
1471
+ const matchesFoldReturn = compileFoldReturnEndpointMatcher(foldReturnSpec);
1472
+ const matchesFoldReturnResults = compileFoldReturnResultsMatcher(foldReturnSpec);
1210
1473
  return captures
1211
1474
  .map((capture, index) => ({ capture, index }))
1212
1475
  .filter(({ capture }) => {
@@ -1216,7 +1479,9 @@ function extractGraphQLActionSequence(captures, submitPatterns = null) {
1216
1479
  return false;
1217
1480
  if (!matchesSubmit(capture))
1218
1481
  return false;
1219
- return capture.query !== null && /^\s*mutation\b/.test(capture.query);
1482
+ if (capture.query !== null && /^\s*mutation\b/.test(capture.query))
1483
+ return true;
1484
+ return matchesFoldReturn(capture) || matchesFoldReturnResults(capture);
1220
1485
  });
1221
1486
  }
1222
1487
  /**
@@ -1287,8 +1552,10 @@ function jsonBodyLeafValues(requestPostData) {
1287
1552
  if (parsed === undefined)
1288
1553
  return null;
1289
1554
  const values = [];
1290
- for (const { value } of walkStringLeaves(parsed))
1291
- values.push(value);
1555
+ for (const { value } of walkAllPrimitiveLeaves(parsed)) {
1556
+ if (value !== null)
1557
+ values.push(String(value));
1558
+ }
1292
1559
  return values;
1293
1560
  }
1294
1561
  /**
@@ -2222,9 +2489,25 @@ function locateFormEnvelopePath(parsedBody) {
2222
2489
  * whole blob is caller-supplied; the generator can't reach inside a value it
2223
2490
  * has delegated wholesale.
2224
2491
  *
2492
+ * A candidate whose leaves include a value already threaded from elsewhere in
2493
+ * the fold — either a PRIOR step's response (e.g. a per-item join token
2494
+ * wrapped in a bulk-lookup array like `{"tokens":["<prior-response-value>"]}`)
2495
+ * OR the fold's own PRIMARY ITEM field feeding the immediate drill step's
2496
+ * request (e.g. `{"orderIds":["<primary-item-value>"]}`) — is never swallowed
2497
+ * here, no matter how array/object-shaped it looks: both are threaded
2498
+ * dependent-drill-down coordinates, not caller-supplied history data. The
2499
+ * prior-step-response case must stay reachable for `interpolateStateValues`
2500
+ * (Pass 1, which runs AFTER this pass); the primary-item case must stay
2501
+ * reachable as literal text for the fold-loop's own `parameterize` pass
2502
+ * (which runs even later, once rendering enters the per-item loop) to find
2503
+ * and swap for `${item.<field>}`. Freezing either into an opaque
2504
+ * `${JSON.stringify(payload.tokens)}` blob here would silently drop the
2505
+ * value the request depends on, breaking the fold at exactly the case an
2506
+ * ARRAY/OBJECT-wrapped join field represents.
2507
+ *
2225
2508
  * Site-agnostic: operates only on the recon body's own shape.
2226
2509
  */
2227
- function applyStructuredValuePayloadSubstitutions(template, parsedBody, outStructuredKeys) {
2510
+ function applyStructuredValuePayloadSubstitutions(template, parsedBody, outStructuredKeys, priorStepStateValues = new Set()) {
2228
2511
  if (parsedBody === null || typeof parsedBody !== "object" || Array.isArray(parsedBody)) {
2229
2512
  return template;
2230
2513
  }
@@ -2248,6 +2531,11 @@ function applyStructuredValuePayloadSubstitutions(template, parsedBody, outStruc
2248
2531
  Object.keys(value).length > 0;
2249
2532
  if (!isNonEmptyArray && !isNestedObject)
2250
2533
  continue;
2534
+ if (priorStepStateValues.size > 0) {
2535
+ const carriesThreadedValue = [...walkAllPrimitiveLeaves(value)].some(({ value: leaf }) => leaf !== null && priorStepStateValues.has(String(leaf)));
2536
+ if (carriesThreadedValue)
2537
+ continue;
2538
+ }
2251
2539
  const keyMarker = `"${key}":`;
2252
2540
  const markerIdx = result.indexOf(keyMarker);
2253
2541
  if (markerIdx === -1)
@@ -2358,11 +2646,29 @@ function* walkSetCookiePairs(rawSetCookie) {
2358
2646
  * Exception: values in `PLACEHOLDER_STATE_VALUES` are skipped entirely so
2359
2647
  * the LATER non-placeholder occurrence at the same JSON path becomes the
2360
2648
  * canonical binding instead.
2649
+ *
2650
+ * `forceIncludeValues` (see {@link collectDependentDrillDownChainValues})
2651
+ * bypasses `MIN_STATE_VALUE_LENGTH` for the specific values it names — a
2652
+ * value already confirmed, by the fold-chain detector itself, to be threaded
2653
+ * from one dependent-drill-down chain hop's response into the next hop's
2654
+ * request is exactly as legitimate a produced state value as a long one; a
2655
+ * length floor exists to keep an UNRELATED short value (an enum code, a page
2656
+ * number) from being mistaken for reused state by blind substring/value
2657
+ * matching, and a value the chain detector already confirmed is threaded
2658
+ * carries no such ambiguity. Every other filter (MAX length, placeholder,
2659
+ * shielded UUID, GET-non-UUID) still applies.
2361
2660
  */
2362
2661
  /** Exported for unit testing — lets tests exercise the produces[] walk (body
2363
2662
  * AND header/cookie origins) directly against synthetic Capture sequences. */
2364
- function indexStateValues(captures, shieldedUuids = new Set(), actionCaptureIndices = new Set()) {
2663
+ function indexStateValues(captures, shieldedUuids = new Set(), actionCaptureIndices = new Set(), forceIncludeValues = new Set()) {
2365
2664
  const index = new Map();
2665
+ // Computed structurally off the SAME captures being indexed (no
2666
+ // foldReturnSpec available at this layer) — a spec-declared fold's own
2667
+ // chain values reach here via the caller-supplied `forceIncludeValues`
2668
+ // (see recon-generate's top-level `collectDependentDrillDownChainValues`
2669
+ // call), so this indexes a chain-produced value regardless of whether the
2670
+ // fold plan that confirmed it is structural or spec-declared.
2671
+ const chainForceIncludeValues = collectDependentDrillDownChainValues(captures.map((capture) => ({ capture })), null);
2366
2672
  // First pass: identify the earliest origin among ACTION captures for each
2367
2673
  // value. Action-only earliest-origin tracking is what compileActionSteps'
2368
2674
  // produces[] check needs — it ignores non-action captures (telemetry GETs,
@@ -2381,7 +2687,13 @@ function indexStateValues(captures, shieldedUuids = new Set(), actionCaptureIndi
2381
2687
  const rawSetCookie = Object.entries(c.responseHeaders).find(([k]) => k.toLowerCase() === "set-cookie")?.[1];
2382
2688
  if (rawSetCookie !== undefined) {
2383
2689
  for (const { name, value } of walkSetCookiePairs(rawSetCookie)) {
2384
- if (value.length < MIN_STATE_VALUE_LENGTH)
2690
+ // Same chain/force exemption as the body-value MIN_STATE_VALUE_LENGTH
2691
+ // floor below: a cookie-sourced value the fold-chain detector already
2692
+ // confirmed is threaded into a later hop's request is exactly as
2693
+ // legitimate as a long one, so it must not be dropped for being short.
2694
+ if (value.length < MIN_STATE_VALUE_LENGTH &&
2695
+ !chainForceIncludeValues.has(value) &&
2696
+ !forceIncludeValues.has(value))
2385
2697
  continue;
2386
2698
  if (value.length > MAX_COOKIE_STATE_VALUE_LENGTH)
2387
2699
  continue;
@@ -2397,6 +2709,31 @@ function indexStateValues(captures, shieldedUuids = new Set(), actionCaptureIndi
2397
2709
  }
2398
2710
  }
2399
2711
  }
2712
+ // Non-cookie response headers are only indexed for values the fold-chain
2713
+ // detector already confirmed are threaded from this hop's response into a
2714
+ // later hop's request (`chainForceIncludeValues`) — unlike Set-Cookie,
2715
+ // which is always a plausible token mint, an arbitrary header (e.g.
2716
+ // `X-Conversation-Id`) is indexed as producible state only when chain
2717
+ // detection itself has already established that reuse, so this never
2718
+ // sweeps every header value as noise.
2719
+ for (const [headerName, headerValue] of Object.entries(c.responseHeaders)) {
2720
+ if (headerName.toLowerCase() === "set-cookie")
2721
+ continue;
2722
+ if (!chainForceIncludeValues.has(headerValue))
2723
+ continue;
2724
+ if (headerValue.length > MAX_COOKIE_STATE_VALUE_LENGTH)
2725
+ continue;
2726
+ if (PLACEHOLDER_STATE_VALUES.has(headerValue))
2727
+ continue;
2728
+ if (!index.has(headerValue)) {
2729
+ index.set(headerValue, {
2730
+ value: headerValue,
2731
+ originIndex: i,
2732
+ path: [],
2733
+ headerOrigin: { sourceHeader: headerName },
2734
+ });
2735
+ }
2736
+ }
2400
2737
  if (c.responseBody === undefined || c.responseBody === null)
2401
2738
  continue;
2402
2739
  // For GET captures, only index UUID-shaped strings. GET captures (today,
@@ -2407,8 +2744,13 @@ function indexStateValues(captures, shieldedUuids = new Set(), actionCaptureIndi
2407
2744
  // e.g. "candidate" as a state value gets substituted INSIDE an already-
2408
2745
  // emitted ${candidateId} interpolation, producing ${${entityTypeCode}Id}.
2409
2746
  const isGet = c.method === "GET";
2410
- for (const { value, path } of walkStringLeaves(c.responseBody)) {
2411
- if (value.length < MIN_STATE_VALUE_LENGTH)
2747
+ for (const { value: rawValue, path } of walkAllPrimitiveLeaves(c.responseBody)) {
2748
+ if (rawValue === null)
2749
+ continue;
2750
+ const value = String(rawValue);
2751
+ if (value.length < MIN_STATE_VALUE_LENGTH &&
2752
+ !chainForceIncludeValues.has(value) &&
2753
+ !forceIncludeValues.has(value))
2412
2754
  continue;
2413
2755
  if (value.length > MAX_STATE_VALUE_LENGTH)
2414
2756
  continue;
@@ -2421,7 +2763,16 @@ function indexStateValues(captures, shieldedUuids = new Set(), actionCaptureIndi
2421
2763
  // corrupt T2/T3's already-substituted Values.
2422
2764
  if (shieldedUuids.has(value))
2423
2765
  continue;
2424
- if (isGet && !UUID_REGEX.test(value))
2766
+ // Same chain/force exemption as the MIN_STATE_VALUE_LENGTH floor above:
2767
+ // a value the fold-chain detector already confirmed is threaded from
2768
+ // this GET hop's response into a later hop's request is exactly as
2769
+ // legitimate as a UUID anchor, so it must not be dropped just because
2770
+ // this hop happens to be a GET rather than every existing fixture's
2771
+ // POST.
2772
+ if (isGet &&
2773
+ !UUID_REGEX.test(value) &&
2774
+ !chainForceIncludeValues.has(value) &&
2775
+ !forceIncludeValues.has(value))
2425
2776
  continue;
2426
2777
  if (!index.has(value)) {
2427
2778
  index.set(value, { value, originIndex: i, path });
@@ -2485,6 +2836,44 @@ function pathToAssertionType(path) {
2485
2836
  const key = isValidJsIdentifier(segment) ? segment : JSON.stringify(segment);
2486
2837
  return `{ ${key}: ${pathToAssertionType(path.slice(1))} }`;
2487
2838
  }
2839
+ /**
2840
+ * Same nesting as {@link pathToAssertionType} but the leaf types as
2841
+ * `Record<string, unknown>[]` instead of `string` — used to cast a step's
2842
+ * response down to the object-array field a {@link FoldPlan} located
2843
+ * (the primary results array, or the drill-down's per-item match array). An
2844
+ * {@link ARRAY_WILDCARD_SEGMENT} segment types as an array of whatever the
2845
+ * rest of the path resolves to, matching the `.flatMap` accessor
2846
+ * {@link pathToFoldAccessorExpr} emits for the same segment.
2847
+ */
2848
+ function foldArrayAssertionType(path) {
2849
+ if (path.length === 0)
2850
+ return "Record<string, unknown>[]";
2851
+ const segment = path[0];
2852
+ if (segment === ARRAY_WILDCARD_SEGMENT) {
2853
+ return `(${foldArrayAssertionType(path.slice(1))})[]`;
2854
+ }
2855
+ const key = isValidJsIdentifier(segment) ? segment : JSON.stringify(segment);
2856
+ return `{ ${key}: ${foldArrayAssertionType(path.slice(1))} }`;
2857
+ }
2858
+ /**
2859
+ * Builds a JS access expression reading `path` off of `expr`, generalizing
2860
+ * across every {@link ARRAY_WILDCARD_SEGMENT} in `path` via `.flatMap` so the
2861
+ * emitted accessor visits every element of that outer array instead of
2862
+ * freezing the single index that happened to contain the matched item during
2863
+ * detection (see {@link ARRAY_WILDCARD_SEGMENT}'s docstring). A path with no
2864
+ * wildcard segment degrades to the plain {@link pathToAccessor} chain.
2865
+ */
2866
+ function pathToFoldAccessorExpr(expr, path, depth = 0) {
2867
+ const wildcardIndex = path.indexOf(ARRAY_WILDCARD_SEGMENT);
2868
+ if (wildcardIndex === -1) {
2869
+ return `${expr}${pathToAccessor(path, { assertNonNull: false })}`;
2870
+ }
2871
+ const before = path.slice(0, wildcardIndex);
2872
+ const after = path.slice(wildcardIndex + 1);
2873
+ const groupVar = `g${depth}`;
2874
+ const outerExpr = `${expr}${pathToAccessor(before, { assertNonNull: false })}`;
2875
+ return `${outerExpr}.flatMap((${groupVar}) => ${pathToFoldAccessorExpr(groupVar, after, depth + 1)})`;
2876
+ }
2488
2877
  /** Suggests a JS-camelCase variable name for a state value path. Falls back
2489
2878
  * up the path if the tail is numeric or not a valid JS identifier. */
2490
2879
  function pathToVarName(path) {
@@ -2589,8 +2978,48 @@ function compileActionSteps(actions, stateIndex) {
2589
2978
  });
2590
2979
  }
2591
2980
  }
2981
+ // Non-cookie response-header-origin produces — mirrors the Set-Cookie
2982
+ // block above but for a plain header (e.g. `X-Price-Token`) whose value
2983
+ // `indexStateValues` indexed with `headerOrigin.sourceHeader` set to the
2984
+ // real header name. Only emitted when the value is actually consumed as
2985
+ // a REQUEST HEADER downstream (`usedValueTargetHeader`) — `createHttpClient`'s
2986
+ // `bind` option (see http-client.ts) is the only mechanism that can
2987
+ // thread a header-origin value forward, since the emitted response
2988
+ // variable never exposes response headers to the rest of the generated
2989
+ // code the way it exposes the parsed body.
2990
+ for (const [headerName, headerValue] of Object.entries(capture.responseHeaders)) {
2991
+ if (headerName.toLowerCase() === "set-cookie")
2992
+ continue;
2993
+ if (!usedValues.has(headerValue))
2994
+ continue;
2995
+ const sv = stateIndex.get(headerValue);
2996
+ if (!sv || sv.originIndex !== index || !sv.headerOrigin)
2997
+ continue;
2998
+ const targetHeader = usedValueTargetHeader.get(headerValue);
2999
+ if (!targetHeader)
3000
+ continue;
3001
+ let name = `${headerName.replace(/[^A-Za-z0-9]/g, "")}Header`;
3002
+ if (!/^[A-Za-z_$]/.test(name))
3003
+ name = `_${name}`;
3004
+ let suffix = 1;
3005
+ while (seenNames.has(name)) {
3006
+ suffix++;
3007
+ name = `${headerName.replace(/[^A-Za-z0-9]/g, "")}Header${suffix}`;
3008
+ }
3009
+ seenNames.add(name);
3010
+ produces.push({
3011
+ kind: "header",
3012
+ name,
3013
+ sourceHeader: sv.headerOrigin.sourceHeader,
3014
+ cookieName: sv.headerOrigin.cookieName,
3015
+ targetHeader,
3016
+ });
3017
+ }
2592
3018
  if (capture.responseBody !== undefined && capture.responseBody !== null) {
2593
- for (const { value, path } of walkStringLeaves(capture.responseBody)) {
3019
+ for (const { value: rawValue, path } of walkAllPrimitiveLeaves(capture.responseBody)) {
3020
+ if (rawValue === null)
3021
+ continue;
3022
+ const value = String(rawValue);
2594
3023
  if (!usedValues.has(value))
2595
3024
  continue;
2596
3025
  const sv = stateIndex.get(value);
@@ -2675,7 +3104,9 @@ function resolveResponsePathValue(responseBody, path) {
2675
3104
  return null;
2676
3105
  }
2677
3106
  }
2678
- return typeof cursor === "string" ? cursor : null;
3107
+ return typeof cursor === "string" || typeof cursor === "number" || typeof cursor === "boolean"
3108
+ ? String(cursor)
3109
+ : null;
2679
3110
  }
2680
3111
  /**
2681
3112
  * Replaces occurrences of state values in `template` with `${varName}`
@@ -3224,7 +3655,7 @@ function emitErrorSignalGuards(varName, urlPath, signals) {
3224
3655
  }
3225
3656
  /** Exported for unit testing — lets tests drive the multipart-upload code path directly
3226
3657
  * without going through the full emitContractTs pipeline. */
3227
- function emitMultiStepExecuteHttp(actions, inputBody, errorSignals, fieldNameMap, outDiscoveredFields, fieldOptionsMap, outDiscoveredOptionFields, outDiscoveredRawOptionFields, outDiscoveredAdditionalBodyKeys, baseUrl, baseUrlDerivedHeaders, tenantSubdomainHeaders, formSchema = null, personaBindings = new Map(), entryUrlParams = new Map(), shieldedUuids = new Set(), selectResolutions = [], outStructuredKeys = new Map(), rawCodeFields = new Map()) {
3658
+ function emitMultiStepExecuteHttp(actions, inputBody, errorSignals, fieldNameMap, outDiscoveredFields, fieldOptionsMap, outDiscoveredOptionFields, outDiscoveredRawOptionFields, outDiscoveredAdditionalBodyKeys, baseUrl, baseUrlDerivedHeaders, tenantSubdomainHeaders, formSchema = null, personaBindings = new Map(), entryUrlParams = new Map(), shieldedUuids = new Set(), selectResolutions = [], outStructuredKeys = new Map(), rawCodeFields = new Map(), foldReturnSpec = null) {
3228
3659
  // Walk the first action's request body to map each leaf string value to its
3229
3660
  // `payload.<accessor>` expression. The emit's second interpolation pass uses
3230
3661
  // this to substitute literal occurrences (e.g. "Reginald") with their
@@ -3397,6 +3828,66 @@ function emitMultiStepExecuteHttp(actions, inputBody, errorSignals, fieldNameMap
3397
3828
  const boundValues = new Set(payloadAccessorByValue.keys());
3398
3829
  // Captured literals that survived every pass — surfaced as a review TODO.
3399
3830
  const unboundLiteralKeys = new Set();
3831
+ // A fold target's join value can be threaded through a request HEADER
3832
+ // rather than the URL/body (see collectRequestValuesIncludingHeaders) —
3833
+ // the structural heuristic can't see those, but a flow-declared foldReturn
3834
+ // spec resolves the target anyway, and the drill step's per-item re-issue
3835
+ // below must still carry that header or every iteration replays the SAME
3836
+ // captured header value instead of re-keying it. Resolved once, up front
3837
+ // (fold plans depend only on `actions`/`foldReturnSpec`, not on Pass 1's
3838
+ // render), so Pass 1's header collection below knows which non-auth header
3839
+ // names are actually load-bearing for a resolved fold.
3840
+ const earlyFoldPlans = resolveFoldPlan(actions, foldReturnSpec);
3841
+ const joinCarryingHeaderNamesByStep = new Map();
3842
+ // A fold target's join value is the PRIMARY ITEM's own field, threaded into
3843
+ // the drill step's request text-literally by the fold-loop's own
3844
+ // `parameterize` pass (later, once rendering enters the per-item loop) — it
3845
+ // is never a prior STEP's produced state var, so `deriveStateVarByValue`
3846
+ // (built from `produces[]`) never sees it. Mechanism B
3847
+ // (`applyStructuredValuePayloadSubstitutions`) runs BEFORE that loop-aware
3848
+ // pass and only knows to spare a candidate carrying a prior step's state
3849
+ // value; without this, it freezes an array/object-wrapped join field (e.g.
3850
+ // `{"orderIds":["<item.orderId>"]}`) into an opaque
3851
+ // `${JSON.stringify(payload.orderIds)}` blob, destroying the literal text
3852
+ // `parameterize` needs to find and swap for `${item.orderId}` — every
3853
+ // iteration then replays one caller-supplied value instead of the item's
3854
+ // own. Collected once here (fold plans depend only on `actions`/
3855
+ // `foldReturnSpec`, not on Pass 1's render) so Pass 1 knows which values to
3856
+ // spare per step, mirroring the header-name collection above.
3857
+ const joinFieldValuesByStep = new Map();
3858
+ for (const plan of earlyFoldPlans) {
3859
+ const primaryItems = objectItemsAtPath(actions[plan.primaryStepIndex].capture.responseBody, plan.primaryArrayPath);
3860
+ for (const target of plan.targets) {
3861
+ const firstItem = primaryItems?.[target.primaryMatchedItemIndex];
3862
+ if (!firstItem)
3863
+ continue;
3864
+ for (const stepIndex of [target.drillStepIndex, ...target.chain]) {
3865
+ const headerNames = joinCarryingHeaderNamesByStep.get(stepIndex) ?? new Set();
3866
+ for (const [headerName, headerValue] of Object.entries(actions[stepIndex].capture.requestHeaders)) {
3867
+ const matchesJoinField = target.joinFields.some((field) => {
3868
+ const value = readValueAtPath(firstItem, field.split("."));
3869
+ return ((typeof value === "string" && value.length > 0 && value === headerValue) ||
3870
+ ((typeof value === "number" || typeof value === "boolean") &&
3871
+ String(value) === headerValue));
3872
+ });
3873
+ if (matchesJoinField)
3874
+ headerNames.add(headerName);
3875
+ }
3876
+ if (headerNames.size > 0)
3877
+ joinCarryingHeaderNamesByStep.set(stepIndex, headerNames);
3878
+ const joinValues = joinFieldValuesByStep.get(stepIndex) ?? new Set();
3879
+ for (const field of target.joinFields) {
3880
+ const value = readValueAtPath(firstItem, field.split("."));
3881
+ if (typeof value === "string" && value.length > 0)
3882
+ joinValues.add(value);
3883
+ else if (typeof value === "number" || typeof value === "boolean")
3884
+ joinValues.add(String(value));
3885
+ }
3886
+ if (joinValues.size > 0)
3887
+ joinFieldValuesByStep.set(stepIndex, joinValues);
3888
+ }
3889
+ }
3890
+ }
3400
3891
  // Pass 1: render every step's emitted strings; collect referenced var names.
3401
3892
  const rendered = [];
3402
3893
  for (let i = 0; i < actions.length; i++) {
@@ -3432,9 +3923,16 @@ function emitMultiStepExecuteHttp(actions, inputBody, errorSignals, fieldNameMap
3432
3923
  // (experienceData/educationData history, opaque eventData) BEFORE value
3433
3924
  // substitution reaches inside them: swallowing the entire array/object first
3434
3925
  // keeps interpolateStateValues from binding a code buried in the history
3435
- // sample (e.g. a work entry's state code) to an unrelated field.
3926
+ // sample (e.g. a work entry's state code) to an unrelated field. Excludes
3927
+ // any candidate that itself carries a PRIOR step's produced value (see
3928
+ // applyStructuredValuePayloadSubstitutions' docstring) — an array-wrapped
3929
+ // dependent-drill-down join field (e.g. a bulk `{"tokens":[...]}` lookup)
3930
+ // must stay reachable for state threading, not get frozen as caller data.
3436
3931
  const rawBodyWithStructuredSubs = parsedBody !== null
3437
- ? applyStructuredValuePayloadSubstitutions(rawBodyWithFormSubs, parsedBody, outStructuredKeys)
3932
+ ? applyStructuredValuePayloadSubstitutions(rawBodyWithFormSubs, parsedBody, outStructuredKeys, new Set([
3933
+ ...deriveStateVarByValue(prior).keys(),
3934
+ ...(joinFieldValuesByStep.get(i) ?? []),
3935
+ ]))
3438
3936
  : rawBodyWithFormSubs;
3439
3937
  // Whole-value caller coordinates bind here — after structured subs, BEFORE
3440
3938
  // state threading — so a composite coordinate (a jobLocation or jobSeqNo
@@ -3490,10 +3988,11 @@ function emitMultiStepExecuteHttp(actions, inputBody, errorSignals, fieldNameMap
3490
3988
  unboundLiteralKeys.add(key);
3491
3989
  }
3492
3990
  }
3991
+ const joinCarryingHeaderNames = joinCarryingHeaderNamesByStep.get(i);
3493
3992
  const perCallHeaders = {};
3494
3993
  for (const [k, v] of Object.entries(cap.requestHeaders)) {
3495
3994
  const lower = k.toLowerCase();
3496
- if (lower === "api-token" || lower === "authorization") {
3995
+ if (lower === "api-token" || lower === "authorization" || joinCarryingHeaderNames?.has(k)) {
3497
3996
  perCallHeaders[k] = interpolateStateValues(v, prior, payloadAccessorByValue);
3498
3997
  }
3499
3998
  }
@@ -3527,7 +4026,10 @@ function emitMultiStepExecuteHttp(actions, inputBody, errorSignals, fieldNameMap
3527
4026
  // validates any individual call. Without this override, HttpRequestInit.schema
3528
4027
  // would default to the client's z.unknown() and narrowing the caller-facing
3529
4028
  // contract would enforce that narrowed shape on every call in the chain.
3530
- const schemaExpr = inferZodSchema(cap.responseBody, 0, "", { looseServerResponse: true });
4029
+ const schemaExpr = inferZodSchema(cap.responseBody, 0, "", {
4030
+ looseServerResponse: true,
4031
+ aggregateUnitBasisFindingsByPath: groupAggregateUnitBasisFindingsByPath([cap.responseBody]),
4032
+ });
3531
4033
  rendered.push({ url, method: cap.method, headersExpr, bodyArg, schemaExpr });
3532
4034
  }
3533
4035
  // Identifier scan against the rendered text — captures `${foo}`, `${foo.bar}`,
@@ -3551,11 +4053,17 @@ function emitMultiStepExecuteHttp(actions, inputBody, errorSignals, fieldNameMap
3551
4053
  }
3552
4054
  }
3553
4055
  }
3554
- // The relevance-selected step's var is also referenced by the closing
3555
- // `return { data }` see selectReturnAction.
3556
- const returnAction = selectReturnAction(actions);
4056
+ // Every resolved drill-down fold plan bypasses selectReturnAction entirely:
4057
+ // each plan's primary step's array is folded in place (see the loop-and-merge
4058
+ // emitted below, one such loop per plan), so a primary step's var — not
4059
+ // whichever call selectReturnAction would otherwise pick — is what
4060
+ // `return { data }` must reference when any plan resolves.
4061
+ const foldPlans = earlyFoldPlans;
4062
+ const returnAction = foldPlans.length > 0 ? null : selectReturnAction(actions);
3557
4063
  if (returnAction)
3558
4064
  referencedNames.add(returnAction.varName);
4065
+ for (const plan of foldPlans)
4066
+ referencedNames.add(actions[plan.primaryStepIndex].varName);
3559
4067
  // Pass 2: emit. Skip response bindings that aren't referenced; skip
3560
4068
  // produces[] entries whose name isn't referenced. A step's response var
3561
4069
  // is still needed when at least one of its produces[] entries IS
@@ -3575,10 +4083,195 @@ function emitMultiStepExecuteHttp(actions, inputBody, errorSignals, fieldNameMap
3575
4083
  lines.push(` // TODO: unbound captured literal(s) — verify these carry caller data, not the recon capture's: ${[...unboundLiteralKeys].join(", ")}`);
3576
4084
  }
3577
4085
  const declaredNames = new Set();
4086
+ // Every non-terminal chain step past the drill step is folded into the SAME
4087
+ // per-item loop the drill step itself emits (see below) — it must not also
4088
+ // get the normal single-call treatment this pass gives every other step, or
4089
+ // its request would be issued a second time, unconditionally, outside the
4090
+ // loop.
4091
+ // Every step in a fold target's chain (the drill step and every further
4092
+ // step transitively dependent on it — see FoldTarget.chain) is emitted
4093
+ // together as a single per-item loop below, not by this pass's normal
4094
+ // per-step produce/response declarations: their responses and produces are
4095
+ // block-scoped to that loop and never escape to the rest of the function,
4096
+ // so none of them may run through the outer `declaredNames`/produceLines
4097
+ // bookkeeping below (that bookkeeping assumes function-scope declarations).
4098
+ const foldChainIndices = new Set(foldPlans.flatMap((plan) => plan.targets.flatMap((target) => target.chain)));
3578
4099
  for (let i = 0; i < actions.length; i++) {
3579
4100
  const step = actions[i];
3580
4101
  const cap = step.capture;
3581
4102
  const r = rendered[i];
4103
+ // Every independent fold target within a given plan — regardless of which
4104
+ // drill step starts it — is folded into ONE shared per-item loop over
4105
+ // that plan's primary array, emitted once, at the plan's FIRST target's
4106
+ // drillStepIndex: each target's chain calls and Object.assign both run
4107
+ // inside that same `for (const item of foldItems)` body, so a primary
4108
+ // item ends up with fields folded in from every independent dependent
4109
+ // drill-down of that plan, not just the first. When more than one
4110
+ // independent plan resolves (distinct primary arrays), each plan gets its
4111
+ // OWN loop block, anchored at that plan's own first target's
4112
+ // drillStepIndex, so each primary array is folded with only its own
4113
+ // drill-downs' matched fields.
4114
+ // Each chain step re-issues the SAME url/headers/bodyArg/schemaExpr this
4115
+ // pass already rendered for it, only with the captured join value(s)
4116
+ // swapped for the loop item's own field accessor (the drill step) or its
4117
+ // own produced response values (later chain steps, which already render
4118
+ // with `${producedName}` templates — see deriveStateVarByValue).
4119
+ const matchingPlanIndex = foldPlans.findIndex((plan) => plan.targets.length > 0 && plan.targets[0].drillStepIndex === i);
4120
+ if (matchingPlanIndex !== -1) {
4121
+ const foldPlan = foldPlans[matchingPlanIndex];
4122
+ const primaryStep = actions[foldPlan.primaryStepIndex];
4123
+ // Read at the plan's OWN path rather than re-running the DFS: a
4124
+ // flow-declared `resultsPath` (see FoldReturnSpec) can name a different
4125
+ // array than findObjectArrayField's first match.
4126
+ const primaryItems = objectItemsAtPath(primaryStep.capture.responseBody, foldPlan.primaryArrayPath);
4127
+ const primaryArrType = foldArrayAssertionType(foldPlan.primaryArrayPath);
4128
+ // Plan-level suffix mirrors the target-level suffix below: multiple
4129
+ // loop blocks now sharing the same function scope can't declare
4130
+ // unsuffixed `foldItems`/`item` locals without colliding. The
4131
+ // overwhelmingly common single-plan case keeps the original
4132
+ // unsuffixed names.
4133
+ const planSuffix = foldPlans.length > 1 ? String(matchingPlanIndex) : "";
4134
+ const foldItemsVar = `foldItems${planSuffix}`;
4135
+ const foldItemsExpr = pathToFoldAccessorExpr(`(${primaryStep.varName} as ${primaryArrType})`, foldPlan.primaryArrayPath);
4136
+ const itemVar = `item${planSuffix}`;
4137
+ lines.push(` const ${foldItemsVar} = ${foldItemsExpr};`, ` for (const ${itemVar} of ${foldItemsVar}) {`);
4138
+ for (const [targetIndex, target] of foldPlan.targets.entries()) {
4139
+ // `firstItem` decides which captured literal `parameterize` rewrites
4140
+ // — it must be the item at `primaryMatchedItemIndex`, the one THIS
4141
+ // target's drill request was actually built from, not always index
4142
+ // 0, and can differ per target even though every target now shares
4143
+ // the same runtime loop item.
4144
+ const firstItem = primaryItems?.[target.primaryMatchedItemIndex];
4145
+ if (!firstItem) {
4146
+ throw new Error(`emitMultiStepExecuteHttp: fold plan primary step ${primaryStep.varName} no longer resolves an object array at ${foldPlan.primaryArrayPath.join(".")} — the fold plan and this emitter have drifted out of sync`);
4147
+ }
4148
+ // Each target's chain variables and merge result get their own
4149
+ // suffixed local names so multiple independent targets sharing the
4150
+ // same loop body can each declare their own locals without
4151
+ // colliding on `foldMatches`/`foldMatch`. The overwhelmingly common
4152
+ // single-target case keeps the original unsuffixed names.
4153
+ const suffix = foldPlan.targets.length > 1 ? `${planSuffix}${targetIndex}` : planSuffix;
4154
+ const joinAccessor = (field) => `${itemVar}${pathToAccessor(field.split("."), { assertNonNull: false })}`;
4155
+ // A join field can reach the render either as the raw captured literal
4156
+ // (URL query params) or as an already-generic `${payload.<field>}`
4157
+ // reference (top-level JSON body keys — see
4158
+ // applyPayloadKeyValueSubstitutions, which payload-ifies every scalar
4159
+ // body key regardless of length, running BEFORE this fold branch ever
4160
+ // sees the value). Both must resolve to the loop item's own field, not
4161
+ // a caller-supplied payload value shared across every iteration.
4162
+ // Word-boundary anchored: a plain `.split(value).join(...)` would also
4163
+ // rewrite unrelated substrings that happen to contain the join value
4164
+ // (e.g. a "p1" product id colliding with a "/v1/" path segment or a
4165
+ // "p10" sibling id), corrupting parts of the request the join field
4166
+ // never touched.
4167
+ const replaceWholeValue = (haystack, value, replacement) => haystack.replace(new RegExp(`\\b${value.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}\\b`, "g"), replacement);
4168
+ const parameterize = (text) => target.joinFields.reduce((acc, field) => {
4169
+ const replacement = `\${${joinAccessor(field)}}`;
4170
+ // applyPayloadKeyValueSubstitutions only ever names a payload
4171
+ // accessor after the DRILL REQUEST's own top-level JSON key
4172
+ // (`${payload.sku}`), never after `field`'s dot path into the
4173
+ // PRIMARY ITEM — those are unrelated structures that only
4174
+ // happen to share a leaf name for a top-level join field. A
4175
+ // nested join field (e.g. `identifiers.sku`) must therefore
4176
+ // also match on its bare last segment, or the accessor swap
4177
+ // silently no-ops and leaves an undefined `payload.sku`
4178
+ // reference behind once the literal value itself has already
4179
+ // been replaced by the payload-key-value pass.
4180
+ const lastSegment = field.split(".").pop();
4181
+ const withAccessorSwapped = acc
4182
+ .split(`\${payload.${field}}`)
4183
+ .join(replacement)
4184
+ .split(`\${payload.${lastSegment}}`)
4185
+ .join(replacement);
4186
+ const value = readValueAtPath(firstItem, field.split("."));
4187
+ const stringValue = typeof value === "string" && value.length > 0
4188
+ ? value
4189
+ : typeof value === "number" || typeof value === "boolean"
4190
+ ? String(value)
4191
+ : null;
4192
+ return stringValue !== null
4193
+ ? replaceWholeValue(withAccessorSwapped, stringValue, replacement)
4194
+ : withAccessorSwapped;
4195
+ }, text);
4196
+ // Every chain step's response and produces are block-scoped to this
4197
+ // `for` — they never escape to the rest of the function. That is
4198
+ // exactly the constraint the previous (now-removed) throw enforced by
4199
+ // refusing to run at all: instead of failing, each chain step's
4200
+ // produces are re-declared here as loop-scoped locals, so a later
4201
+ // chain step's own request (already rendered with `${producedName}`
4202
+ // templates by the pass above, same as it would be for any two
4203
+ // sequential non-fold steps) resolves them from this narrower scope.
4204
+ const chainDeclared = new Set();
4205
+ for (const chainIndex of target.chain) {
4206
+ const chainStep = actions[chainIndex];
4207
+ const chainRendered = rendered[chainIndex];
4208
+ lines.push(` const ${chainStep.varName} = (await httpClient(\`${parameterize(chainRendered.url)}\`, {`, ` method: ${JSON.stringify(chainRendered.method)},`);
4209
+ const joined = [
4210
+ parameterize(chainRendered.headersExpr),
4211
+ parameterize(chainRendered.bodyArg),
4212
+ ]
4213
+ .filter((s) => s !== "")
4214
+ .join(" ");
4215
+ if (joined !== "")
4216
+ lines.push(` ${joined}`);
4217
+ lines.push(` schema: ${chainRendered.schemaExpr},`, ` })) as Record<string, unknown>;`);
4218
+ for (const p of chainStep.produces) {
4219
+ if (p.kind === "header")
4220
+ continue;
4221
+ if (chainDeclared.has(p.name))
4222
+ continue;
4223
+ if (!referencedNames.has(p.name))
4224
+ continue;
4225
+ chainDeclared.add(p.name);
4226
+ const assertion = pathToAssertionType(p.path);
4227
+ lines.push(` const ${p.name} = (${chainStep.varName} as ${assertion})${pathToAccessor(p.path, { assertNonNull: false })};`);
4228
+ }
4229
+ }
4230
+ const terminalStep = actions[target.chainTerminalIndex];
4231
+ // An empty chainArrayPath means the terminal step's response IS the
4232
+ // implicit one-item collection (see findAllObjectArrayFieldsOrWholeObject
4233
+ // / objectItemsAtPath's flat-object branch): the response is a flat
4234
+ // object at runtime, not an array. There is exactly one candidate, so
4235
+ // no join-field match is needed (or even possible against an array
4236
+ // API) — emit a direct object reference instead of the array
4237
+ // `.find()` machinery the multi-item branch below needs.
4238
+ if (target.chainArrayPath.length === 0) {
4239
+ lines.push(` const foldMatch${suffix} = ${terminalStep.varName} as Record<string, unknown>;`, ` Object.assign(${itemVar}, foldMatch${suffix} ?? {});`);
4240
+ }
4241
+ else {
4242
+ const foldMatchesExpr = pathToFoldAccessorExpr(`(${terminalStep.varName} as ${foldArrayAssertionType(target.chainArrayPath)})`, target.chainArrayPath);
4243
+ lines.push(` const foldMatches${suffix} = ${foldMatchesExpr};`, ` const foldMatch${suffix} = foldMatches${suffix}.find((m) => ${target.joinFields
4244
+ .map((f) => {
4245
+ const segments = f.split(".");
4246
+ // The drill-down response is a DIFFERENT payload than the
4247
+ // primary item, so it has no obligation to mirror the
4248
+ // primary item's own nesting for the join key (e.g. a
4249
+ // primary item's `identifiers.sku` is typically echoed
4250
+ // back flat, as `sku`, on the drill response). Try the
4251
+ // full nested path first (optional-chained, since an
4252
+ // intermediate segment may not exist on a flat response),
4253
+ // then fall back to the bare last segment.
4254
+ const lastSegment = segments[segments.length - 1];
4255
+ const optionalBracketAccessor = segments
4256
+ .map((segment) => `?.[${JSON.stringify(segment)}]`)
4257
+ .join("");
4258
+ const matchAccessor = segments.length > 1
4259
+ ? `(m${optionalBracketAccessor} ?? m[${JSON.stringify(lastSegment)}])`
4260
+ : `m[${JSON.stringify(lastSegment)}]`;
4261
+ return `String(${matchAccessor}) === String(${joinAccessor(f)})`;
4262
+ })
4263
+ .join(" && ")}) ?? foldMatches${suffix}[0];`, ` Object.assign(${itemVar}, foldMatch${suffix} ?? {});`);
4264
+ }
4265
+ }
4266
+ lines.push(` }`, "");
4267
+ continue;
4268
+ }
4269
+ // Every other chain step (already fully emitted, inline, by the fold
4270
+ // block above) must not also get the normal single-call treatment this
4271
+ // pass gives every step, or its request would be issued a second time,
4272
+ // unconditionally, outside the loop.
4273
+ if (foldChainIndices.has(i))
4274
+ continue;
3582
4275
  // Build the produce-extraction lines FIRST so the binding decision reflects
3583
4276
  // what is actually emitted, not a pre-scan predicate. A produce whose name
3584
4277
  // was already declared by an earlier step is de-dup-skipped here — and must
@@ -3690,8 +4383,38 @@ function emitMultiStepExecuteHttp(actions, inputBody, errorSignals, fieldNameMap
3690
4383
  lines.push(line);
3691
4384
  lines.push("");
3692
4385
  }
3693
- const returnVar = returnAction ? returnAction.varName : "undefined";
3694
- lines.push(` return { data: ${returnVar} };`);
4386
+ // When multiple plans resolve, every plan's own primary var is deep-merged
4387
+ // together into one combined result via mergeFoldedPrimaryBodies —
4388
+ // mirroring selectEffectiveResponseBody's merge — so the runtime return
4389
+ // value and the inferred shape always describe the same call. A plain
4390
+ // object-spread would silently drop one plan's array whenever two plans'
4391
+ // primary bodies share a top-level array key (e.g. the same paginated
4392
+ // primary endpoint drilled into twice). A primary whose OWN top-level body
4393
+ // isn't a plain object can't be merged meaningfully, so in that case this
4394
+ // falls back to the LAST plan's primary var alone, same as before this
4395
+ // merge was introduced.
4396
+ //
4397
+ // Deduped by var name (not one entry per plan): two plans anchored on the
4398
+ // SAME primary step (e.g. a structural plan and a spec-only plan each
4399
+ // resolving a different array on one shared response — see
4400
+ // mergeSpecPlanOntoSamePrimary) both mutate that ONE response object's own
4401
+ // arrays in place per their own loop above. Passing that same var into
4402
+ // mergeFoldedPrimaryBodies once per plan would concatenate every one of its
4403
+ // arrays with itself, duplicating every already-folded item.
4404
+ const lastFoldPlan = foldPlans[foldPlans.length - 1] ?? null;
4405
+ const uniquePrimaryVarNames = [
4406
+ ...new Set(foldPlans.map((plan) => actions[plan.primaryStepIndex].varName)),
4407
+ ];
4408
+ const everyPrimaryIsPlainObject = foldPlans.every((plan) => isPlainObject(actions[plan.primaryStepIndex].capture.responseBody));
4409
+ if (uniquePrimaryVarNames.length > 1 && everyPrimaryIsPlainObject) {
4410
+ lines.push(` return { data: mergeFoldedPrimaryBodies(${uniquePrimaryVarNames.join(", ")}) };`);
4411
+ }
4412
+ else {
4413
+ const returnVar = lastFoldPlan
4414
+ ? actions[lastFoldPlan.primaryStepIndex].varName
4415
+ : (returnAction?.varName ?? "undefined");
4416
+ lines.push(` return { data: ${returnVar} };`);
4417
+ }
3695
4418
  return lines.join("\n");
3696
4419
  }
3697
4420
  function summariseResponseShape(value) {
@@ -3843,23 +4566,986 @@ function findNumericFieldByName(value, pattern, path = []) {
3843
4566
  }
3844
4567
  return null;
3845
4568
  }
3846
- /** Depth-first search for the first array whose elements are (non-array)
4569
+ /** Filter predicate isolating an array's (non-array) object elements the
4570
+ * shape both {@link findAllObjectArrayFields} and {@link objectItemsAtPath}
4571
+ * treat as a "results array" candidate. */
4572
+ function isObjectArrayItem(v) {
4573
+ return v !== null && typeof v === "object" && !Array.isArray(v);
4574
+ }
4575
+ /** Sentinel path segment marking "every element of the array reached so
4576
+ * far", emitted by {@link findAllObjectArrayFields} in place of a literal
4577
+ * numeric index whenever it descends through an array to keep searching —
4578
+ * the array itself is a container of candidate groups, not a single fixed
4579
+ * one. Freezing the literal index of whichever group happened to contain
4580
+ * the matched item (the bug this sentinel fixes) meant a multi-element
4581
+ * outer array — e.g. a paginated/grouped response wrapping several
4582
+ * sub-collections — only ever resolved/iterated the ONE group seen during
4583
+ * detection. {@link objectItemsAtPath} and the fold-emission accessor
4584
+ * builders below both flatten across every element at this position instead
4585
+ * of indexing into one. */
4586
+ const ARRAY_WILDCARD_SEGMENT = "*";
4587
+ /** Depth-first search for every array whose elements are (non-array)
3847
4588
  * objects — the same "per-item response array" shape schema inference
3848
- * already resolves to when it emits `z.array(z.object({...}))`. */
3849
- function findObjectArrayField(value, path = []) {
4589
+ * already resolves to when it emits `z.array(z.object({...}))`. Ordered by
4590
+ * DFS/key order, so `[0]` is {@link findObjectArrayField}'s first match.
4591
+ * A path segment for an array index the search descended through (to keep
4592
+ * looking for a nested candidate array) is the {@link ARRAY_WILDCARD_SEGMENT}
4593
+ * sentinel, never a literal index — see its docstring. */
4594
+ function findAllObjectArrayFields(value, path = []) {
3850
4595
  if (value === null || typeof value !== "object")
3851
- return null;
4596
+ return [];
3852
4597
  if (Array.isArray(value)) {
3853
- const objectItems = value.filter((v) => v !== null && typeof v === "object" && !Array.isArray(v));
3854
- return objectItems.length > 0 ? { path, items: objectItems } : null;
4598
+ const objectItems = value.filter(isObjectArrayItem);
4599
+ const nestedCandidates = objectItems.flatMap((item) => findAllObjectArrayFields(item, [...path, ARRAY_WILDCARD_SEGMENT]));
4600
+ return objectItems.length > 0
4601
+ ? [{ path, items: objectItems }, ...nestedCandidates]
4602
+ : nestedCandidates;
4603
+ }
4604
+ return Object.entries(value).flatMap(([key, v]) => findAllObjectArrayFields(v, [...path, key]));
4605
+ }
4606
+ /** The first object-array field by DFS/key order — see
4607
+ * {@link findAllObjectArrayFields}. Every call site that must disambiguate
4608
+ * between several candidate arrays (e.g. a decoy array positioned earlier in
4609
+ * key order than the real one) uses {@link findAllObjectArrayFields}
4610
+ * directly instead of this first-match shortcut. */
4611
+ function findObjectArrayField(value, path = []) {
4612
+ return findAllObjectArrayFields(value, path)[0] ?? null;
4613
+ }
4614
+ /** {@link findAllObjectArrayFields}, widened to ALSO offer the flat
4615
+ * (non-array) whole object itself as a candidate — a detail-by-id
4616
+ * drill/chain response (e.g. `GET /widgets/{id}` returning the widget
4617
+ * object directly, not `{ widget: {...} }`) is exactly as foldable as a
4618
+ * one-element array would be. The flat entry's path is `[]` (the whole
4619
+ * body); {@link objectItemsAtPath} recognizes a flat object at a resolved
4620
+ * path the same way, so a {@link FoldTarget} built from this fallback
4621
+ * resolves correctly end to end. The flat candidate is appended AFTER every
4622
+ * real object-array candidate (never in place of one) so a caller that
4623
+ * needs to compare candidates by richness — see {@link
4624
+ * chainTerminalItemRichness} — can prefer the flat shape when it carries
4625
+ * more genuine per-item data than a small real nested object-array; a
4626
+ * caller that just wants the first real array (the common case) is
4627
+ * unaffected since it still comes first. */
4628
+ function findAllObjectArrayFieldsOrWholeObject(value, path = []) {
4629
+ const found = findAllObjectArrayFields(value, path);
4630
+ return isObjectArrayItem(value) ? [...found, { path, items: [value] }] : found;
4631
+ }
4632
+ /** The first candidate from {@link findAllObjectArrayFieldsOrWholeObject} —
4633
+ * the flat-object-aware counterpart of {@link findObjectArrayField}. */
4634
+ function findObjectArrayFieldOrWholeObject(value, path = []) {
4635
+ return findAllObjectArrayFieldsOrWholeObject(value, path)[0] ?? null;
4636
+ }
4637
+ /** Every string and numeric value present in a capture's outbound request —
4638
+ * its URL path segments and query parameters (always strings) and its JSON
4639
+ * body's string and numeric leaves (numeric leaves stringified) — the set a
4640
+ * drill-down request's threaded join value must appear in. Path segments are
4641
+ * included because REST-style APIs commonly thread a primary item's id as a
4642
+ * path segment (e.g. `/orders/{id}`) rather than a query param or body
4643
+ * field. Numeric leaves are included because a join key is just as often a
4644
+ * numeric id (threaded as a query param string or a JSON body number
4645
+ * literal) as a string one. */
4646
+ function collectRequestStringValues(capture) {
4647
+ const values = new Set();
4648
+ try {
4649
+ const url = new URL(capture.url);
4650
+ for (const v of url.searchParams.values())
4651
+ values.add(v);
4652
+ for (const segment of url.pathname.split("/").filter(Boolean))
4653
+ values.add(segment);
3855
4654
  }
3856
- for (const [key, v] of Object.entries(value)) {
3857
- const found = findObjectArrayField(v, [...path, key]);
3858
- if (found)
3859
- return found;
4655
+ catch {
4656
+ // Relative or malformed URL — no query params or path segments to contribute.
4657
+ }
4658
+ for (const v of jsonBodyLeafValues(capture.requestPostData) ?? [])
4659
+ values.add(v);
4660
+ const parsedBody = (() => {
4661
+ try {
4662
+ return typeof capture.requestPostData === "string" && capture.requestPostData.length > 0
4663
+ ? JSON.parse(capture.requestPostData)
4664
+ : undefined;
4665
+ }
4666
+ catch {
4667
+ return undefined;
4668
+ }
4669
+ })();
4670
+ if (parsedBody !== undefined) {
4671
+ for (const { value } of walkAllPrimitiveLeaves(parsedBody)) {
4672
+ if (typeof value === "number" || typeof value === "boolean")
4673
+ values.add(String(value));
4674
+ }
4675
+ }
4676
+ return values;
4677
+ }
4678
+ /**
4679
+ * Yields every string/numeric/boolean leaf reachable from `item` by walking nested
4680
+ * plain objects only (not arrays — a join key is a scalar field of the item
4681
+ * or one of its nested objects, never an element drawn from a nested array),
4682
+ * paired with its dot-separated path from `item`'s root. A bare top-level
4683
+ * field yields a single-segment path (e.g. `["sku"]`), matching every
4684
+ * existing joinFields entry's shape unchanged; a field nested inside an
4685
+ * object (e.g. `{ identifiers: { sku } }`) yields `["identifiers", "sku"]`.
4686
+ */
4687
+ function* walkItemFieldPaths(item, path = []) {
4688
+ for (const [k, v] of Object.entries(item)) {
4689
+ const childPath = [...path, k];
4690
+ if (v !== null && typeof v === "object" && !Array.isArray(v)) {
4691
+ yield* walkItemFieldPaths(v, childPath);
4692
+ continue;
4693
+ }
4694
+ yield { path: childPath, value: v };
4695
+ }
4696
+ }
4697
+ /**
4698
+ * Finds the ordered list of an array item's string/numeric/boolean field
4699
+ * paths whose values are threaded into `drillCapture`'s outbound request — the join key a
4700
+ * dependent drill-down call was built from. Each entry is a dot-separated
4701
+ * path (see {@link readValueAtPath} / {@link pathToAccessor}), so a bare
4702
+ * top-level field stays a single segment (e.g. `"sku"`) and a field nested
4703
+ * inside an object becomes e.g. `"identifiers.sku"`. Field order follows the
4704
+ * item's own key order (nested objects walked depth-first as encountered),
4705
+ * so a composite join (e.g. `accountId` + `region`) comes out in the same
4706
+ * order the primary response declares them, not sorted. Returns `[]` when no
4707
+ * field of the item threads into the request at all.
4708
+ */
4709
+ function findThreadedJoinFields(item, drillCapture) {
4710
+ const requestValues = collectRequestStringValues(drillCapture);
4711
+ if (requestValues.size === 0)
4712
+ return [];
4713
+ return [...walkItemFieldPaths(item)]
4714
+ .filter(({ value: v }) => (typeof v === "string" && v.length > 0 && requestValues.has(v)) ||
4715
+ (typeof v === "number" && requestValues.has(String(v))) ||
4716
+ (typeof v === "boolean" && requestValues.has(String(v))))
4717
+ .map(({ path }) => path.join("."));
4718
+ }
4719
+ /** Every string, numeric, and boolean leaf value present anywhere in a response —
4720
+ * the set a chained drill-down step's request must overlap with for that
4721
+ * step to count as depending on this response. Deliberately walks the WHOLE
4722
+ * body (not just object-array items, unlike {@link findThreadedJoinFields})
4723
+ * since a chained step can thread any response value, not only a per-item
4724
+ * join field. Also walks every response HEADER value, mirroring
4725
+ * {@link collectRequestValuesIncludingHeaders} on the request side, since a
4726
+ * chain hop can just as easily mint its join token in a response header
4727
+ * (e.g. a `Location` or custom correlation header) as in the body. */
4728
+ function collectResponseLeafValues(capture) {
4729
+ const values = new Set();
4730
+ for (const { value } of walkAllPrimitiveLeaves(capture.responseBody)) {
4731
+ if (typeof value === "string" && value.length > 0)
4732
+ values.add(value);
4733
+ if (typeof value === "number")
4734
+ values.add(String(value));
4735
+ if (typeof value === "boolean")
4736
+ values.add(String(value));
4737
+ }
4738
+ for (const v of Object.values(capture.responseHeaders))
4739
+ values.add(v);
4740
+ return values;
4741
+ }
4742
+ /**
4743
+ * Walks forward from `drillStepIndex`, following the transitive per-item
4744
+ * dependency chain: each subsequent step whose request threads a value out
4745
+ * of ANY step already in the chain is itself added to the chain, since its
4746
+ * own request — and therefore its response — only makes sense once per
4747
+ * matched primary item. Reuses {@link collectRequestValuesIncludingHeaders}
4748
+ * (URL, body, AND headers) so a chained step threading its join value
4749
+ * through a header, exactly like {@link buildFoldPlanFromSpec}'s own
4750
+ * threading, is still picked up. Returns the ordered step indices (starting
4751
+ * at `drillStepIndex`) plus the object-array path AND owning step index found
4752
+ * on the LAST chain step whose response actually has an object-array field,
4753
+ * OR — failing that — the last chain step whose response is a flat object
4754
+ * carrying MORE of its own per-item data than the chain's best terminal
4755
+ * found so far (see {@link chainTerminalItemRichness}). Either way this is
4756
+ * NOT necessarily `chain`'s last entry, since a step can be chained purely
4757
+ * because it threads a value onward (e.g. a status-check response with no
4758
+ * array of its own and no richer data than what came before) without itself
4759
+ * holding the foldable data. Falls back to `drillArrayPath`/`drillStepIndex`
4760
+ * (the immediate drill step's own array or flat object) when nothing
4761
+ * threads further, so `chain` degrades to `[drillStepIndex]` for the common
4762
+ * single-step case.
4763
+ */
4764
+ /** Disambiguates among every object-array (or flat-object) candidate on
4765
+ * `responseBody`, preferring the one whose items thread `capture`'s own
4766
+ * request/header values (see {@link findThreadedJoinFields}), falling back
4767
+ * to the richest candidate by {@link chainTerminalItemRichness} when none
4768
+ * thread. Both the immediate drill step ({@link scanPrimaryCandidateGroups})
4769
+ * and every later chained step ({@link computeFoldChain}) must disambiguate
4770
+ * identically — a decoy candidate positioned earlier in key order than the
4771
+ * real per-item array is exactly as invalid a pick on a chain hop as it is
4772
+ * on the immediate drill call — so this is the single implementation both
4773
+ * reuse instead of each doing its own first-DFS-match shortcut. */
4774
+ function selectDisambiguatedCandidate(responseBody, capture) {
4775
+ const candidates = findAllObjectArrayFieldsOrWholeObject(responseBody);
4776
+ if (candidates.length === 0)
4777
+ return null;
4778
+ const requestValues = collectRequestValuesIncludingHeaders(capture);
4779
+ const threaded = candidates.find((candidate) => candidate.items.some((item) => findThreadedJoinFields(item, capture).length > 0));
4780
+ if (threaded)
4781
+ return threaded;
4782
+ return candidates.reduce((richest, candidate) => {
4783
+ const candidateRichness = chainTerminalItemRichness(responseBody, candidate.path, requestValues);
4784
+ const richestSoFar = chainTerminalItemRichness(responseBody, richest.path, requestValues);
4785
+ return candidateRichness > richestSoFar ? candidate : richest;
4786
+ });
4787
+ }
4788
+ function computeFoldChain(actions, drillStepIndex, drillArrayPath) {
4789
+ const chain = [drillStepIndex];
4790
+ let chainArrayPath = drillArrayPath;
4791
+ let chainTerminalIndex = drillStepIndex;
4792
+ let chainTerminalRichness = chainTerminalItemRichness(actions[drillStepIndex].capture.responseBody, drillArrayPath, collectRequestValuesIncludingHeaders(actions[drillStepIndex].capture));
4793
+ for (let i = drillStepIndex + 1; i < actions.length; i++) {
4794
+ const candidate = actions[i];
4795
+ const requestValues = collectRequestValuesIncludingHeaders(candidate.capture);
4796
+ const dependsOnChain = chain.some((chainIndex) => {
4797
+ const chainStepCapture = actions[chainIndex].capture;
4798
+ const responseValues = collectResponseLeafValues(chainStepCapture);
4799
+ const echoedValues = collectRequestValuesIncludingHeaders(chainStepCapture);
4800
+ return [...responseValues].some((v) => !echoedValues.has(v) && requestValues.has(v));
4801
+ });
4802
+ if (!dependsOnChain)
4803
+ continue;
4804
+ chain.push(i);
4805
+ // Disambiguated identically to the immediate drill step (see
4806
+ // selectDisambiguatedCandidate): a decoy object-array field positioned
4807
+ // earlier in key order than the real per-item array must not win just
4808
+ // for being found first on THIS chain step's own response, any more
4809
+ // than it would on the immediate drill call. The winning candidate
4810
+ // still only displaces the terminal when it's STRICTLY richer than the
4811
+ // chain's best terminal so far, OR ties it with a genuine object-ARRAY
4812
+ // candidate (isGenuineArrayCandidate: candidateArray.path is non-empty).
4813
+ // A flat single-object candidate never wins a tie — otherwise a step
4814
+ // chained purely for threading a value onward (e.g. a `{ held: true }`
4815
+ // confirmation, never richer than the real per-item shape it merely
4816
+ // threads from) would always qualify as an implicit one-item collection
4817
+ // and collapse this into "always advance the terminal to the newest
4818
+ // chain member" regardless of whether that member actually holds
4819
+ // foldable data. But a later hop that DOES resolve to a real per-item
4820
+ // array, tied only because a same-shaped flat confirmation hop sits
4821
+ // earlier in the chain, is the genuine terminal and must still win —
4822
+ // see buildMulticallSingleShotSearchDrillDownRichnessTiedConfirmationHopChainedDependentActionSteps.
4823
+ const candidateArray = selectDisambiguatedCandidate(candidate.capture.responseBody, candidate.capture);
4824
+ if (!candidateArray)
4825
+ continue;
4826
+ const candidateRichness = chainTerminalItemRichness(candidate.capture.responseBody, candidateArray.path, requestValues);
4827
+ const isGenuineArrayCandidate = candidateArray.path.length > 0;
4828
+ const advancesOnTie = candidateRichness === chainTerminalRichness &&
4829
+ candidateRichness > 0 &&
4830
+ isGenuineArrayCandidate;
4831
+ if (candidateRichness > chainTerminalRichness || advancesOnTie) {
4832
+ chainArrayPath = candidateArray.path;
4833
+ chainTerminalIndex = i;
4834
+ chainTerminalRichness = candidateRichness;
4835
+ }
4836
+ }
4837
+ return { chain, chainArrayPath, chainTerminalIndex };
4838
+ }
4839
+ /** The per-item primitive-field richness of a chain terminal candidate at
4840
+ * `path` — {@link directPrimitiveChildCountExcludingEchoed} of the first item
4841
+ * {@link objectItemsAtPath} resolves there (an object-array item when `path`
4842
+ * names a real array, or the whole flat object when `path` is `[]`), or 0
4843
+ * when `path` resolves to nothing. `requestValues` excludes fields the
4844
+ * candidate's own response merely echoes back from its request (join keys,
4845
+ * threaded ids), so an echo can never be mistaken for genuine per-item data.
4846
+ * The single metric every {@link computeFoldChain} comparison — baseline,
4847
+ * array-branch, and flat-branch alike — uses, so a later step only ever
4848
+ * displaces the terminal by actually contributing more of its own data. */
4849
+ function chainTerminalItemRichness(responseBody, path, requestValues) {
4850
+ const items = objectItemsAtPath(responseBody, path);
4851
+ return items && items.length > 0
4852
+ ? directPrimitiveChildCountExcludingEchoed(items[0], requestValues)
4853
+ : 0;
4854
+ }
4855
+ /** Like {@link directPrimitiveChildCount}, but skips a field whose value was
4856
+ * itself already threaded INTO this response's own request (present in
4857
+ * `requestValues`) — a confirmation step routinely echoes the id/token it
4858
+ * was called with alongside a status flag, and that echo must not count as
4859
+ * genuine per-item data or a side-effect-only response (e.g. `{ token:
4860
+ * "t1", held: true }` echoing a threaded `token`) would out-rank the real
4861
+ * terminal on field count alone. */
4862
+ function directPrimitiveChildCountExcludingEchoed(obj, requestValues) {
4863
+ let n = 0;
4864
+ for (const v of Object.values(obj)) {
4865
+ if (v === null || (typeof v !== "object" && typeof v !== "function")) {
4866
+ if (typeof v === "string" || typeof v === "number" || typeof v === "boolean") {
4867
+ if (requestValues.has(String(v)))
4868
+ continue;
4869
+ }
4870
+ n++;
4871
+ }
4872
+ }
4873
+ return n;
4874
+ }
4875
+ /**
4876
+ * Scans every candidate object-array field of `actions[primaryIndex]`'s
4877
+ * response independently, each yielding its own {@link PrimaryScanGroup} of
4878
+ * targets when at least one later step threads a join field out of it — a
4879
+ * single primary response holding two genuinely unrelated per-item
4880
+ * collections (e.g. a search response with both a `results[]` and a
4881
+ * `facets[]` that are each independently drilled by their own later call)
4882
+ * must fold BOTH, not just whichever array the first qualifying drill-down
4883
+ * happens to thread from. `consumedIndices` is shared across every
4884
+ * candidate's scan (not reset per candidate) so a step already folded into
4885
+ * one array's target chain is never re-claimed as a fresh target thread of
4886
+ * a different, independent array on the same primary response.
4887
+ */
4888
+ function scanPrimaryCandidateGroups(actions, primaryIndex, globallyConsumedIndices) {
4889
+ const primary = actions[primaryIndex];
4890
+ const primaryCandidates = findAllObjectArrayFields(primary.capture.responseBody);
4891
+ if (primaryCandidates.length === 0)
4892
+ return [];
4893
+ const groups = [];
4894
+ // A step already consumed as a later member of an earlier target's
4895
+ // chain — from THIS primary array or an independent one on the same
4896
+ // primary response — threads its request from that target's own
4897
+ // response, not straight off the primary array, so it must not also be
4898
+ // picked up as a second, independent target of any candidate array.
4899
+ const consumedIndices = new Set();
4900
+ for (const primaryArray of primaryCandidates) {
4901
+ const targets = [];
4902
+ for (let drillIndex = primaryIndex + 1; drillIndex < actions.length; drillIndex++) {
4903
+ const drill = actions[drillIndex];
4904
+ if (drill === primary)
4905
+ continue;
4906
+ if (consumedIndices.has(drillIndex))
4907
+ continue;
4908
+ if (globallyConsumedIndices.has(drillIndex))
4909
+ continue;
4910
+ // Every item in THIS candidate array is searched, not just items[0]
4911
+ // — a flow that only ever drilled into a later item (never the
4912
+ // first) must still resolve.
4913
+ const primaryMatchedItemIndex = primaryArray.items.findIndex((item) => findThreadedJoinFields(item, drill.capture).length > 0);
4914
+ if (primaryMatchedItemIndex === -1)
4915
+ continue;
4916
+ const joinFields = findThreadedJoinFields(primaryArray.items[primaryMatchedItemIndex], drill.capture);
4917
+ // Widened to a flat (non-array) object response when the drill step has
4918
+ // no object-array field of its own — see
4919
+ // findAllObjectArrayFieldsOrWholeObject. A detail-by-id response (e.g.
4920
+ // `GET /widgets/{id}` returning the widget directly) is just as valid a
4921
+ // fold target as a one-element array would be; only skipping it, rather
4922
+ // than treating it as an implicit one-item collection, was the actual
4923
+ // root cause of dependent drill-downs never folding onto primary
4924
+ // results. Disambiguated via selectDisambiguatedCandidate — a per-item
4925
+ // drill array commonly echoes the join value(s) it was looked up by
4926
+ // (e.g. a `sku` search parameter mirrored back on each result), which
4927
+ // distinguishes it from a decoy array (e.g. an `errors[]` collection)
4928
+ // that never does. Not every real drill array echoes the join value,
4929
+ // though (e.g. a `productId`-keyed lookup returning `units[]` with no
4930
+ // `productId` field of its own), so when no candidate threads, the
4931
+ // richest candidate by per-item primitive-field count wins — the same
4932
+ // selection computeFoldChain applies to every later chained step, so a
4933
+ // decoy is never disambiguated differently on the immediate drill call
4934
+ // than it is one hop further down the chain.
4935
+ // An immediate hop with no object-array/flat-object candidate at all
4936
+ // (a bare token/id array, a scalar, an empty body) must not abandon
4937
+ // the whole candidate here — a LATER hop in the chain may still
4938
+ // thread forward to the real per-item data (e.g. an id array whose
4939
+ // values are looked up individually on a following step). Feed
4940
+ // computeFoldChain the empty-path baseline FoldTarget already
4941
+ // documents as its fallback contract, and only bail once the
4942
+ // resolved chain terminal itself has no real items to fold onto.
4943
+ const drillArray = selectDisambiguatedCandidate(drill.capture.responseBody, drill.capture);
4944
+ const drillArrayPath = drillArray?.path ?? [];
4945
+ const { chain, chainArrayPath, chainTerminalIndex } = computeFoldChain(actions, drillIndex, drillArrayPath);
4946
+ const chainTerminalItems = objectItemsAtPath(actions[chainTerminalIndex].capture.responseBody, chainArrayPath);
4947
+ if (!chainTerminalItems || chainTerminalItems.length === 0)
4948
+ continue;
4949
+ // `primaryArray.path` can carry an ARRAY_WILDCARD_SEGMENT (a matched
4950
+ // item nested inside a multi-element outer array — e.g. a
4951
+ // paginated/grouped response wrapping several sub-collections), in
4952
+ // which case `primaryMatchedItemIndex` above is only the LOCAL index
4953
+ // within the one group `primaryArray.items` happens to be. Re-resolve
4954
+ // it against the FLATTENED items every group at that path contributes,
4955
+ // by object identity (findAllObjectArrayFields and objectItemsAtPath
4956
+ // both read the same references, never cloning), so downstream
4957
+ // consumers reading through `objectItemsAtPath` — the emitter's
4958
+ // `firstItem` lookup and the shape-inference fold — land on the exact
4959
+ // same item regardless of which group it came from.
4960
+ const flattenedPrimaryItems = objectItemsAtPath(primary.capture.responseBody, primaryArray.path) ?? [];
4961
+ const globalMatchedItemIndex = flattenedPrimaryItems.indexOf(primaryArray.items[primaryMatchedItemIndex]);
4962
+ targets.push({
4963
+ joinFields,
4964
+ drillStepIndex: drillIndex,
4965
+ drillArrayPath,
4966
+ primaryMatchedItemIndex: globalMatchedItemIndex === -1 ? primaryMatchedItemIndex : globalMatchedItemIndex,
4967
+ chain,
4968
+ chainArrayPath,
4969
+ chainTerminalIndex,
4970
+ });
4971
+ // Everything past drillIndex already folded into this target's own
4972
+ // chain is per-item dependent on THIS drill-down, not independently
4973
+ // threaded off the primary array, so it must be skipped rather than
4974
+ // re-considered as a new target — by this candidate array's own
4975
+ // further scanning or any other candidate array's scan.
4976
+ for (const chainIndex of chain)
4977
+ consumedIndices.add(chainIndex);
4978
+ }
4979
+ if (targets.length > 0)
4980
+ groups.push({ primaryArrayPath: primaryArray.path, targets });
4981
+ }
4982
+ return groups;
4983
+ }
4984
+ function detectDrillDownFoldPlan(actions) {
4985
+ const plans = [];
4986
+ // Spans every primary candidate, not just the current one's own drill
4987
+ // scan: a step already folded into an earlier plan's chain — as either
4988
+ // its drill step or a later chained step — depended on that earlier
4989
+ // primary's response, not on this later primary's array, so it must
4990
+ // never be re-claimed as a fresh drill target for a subsequent primary.
4991
+ const globallyConsumedIndices = new Set();
4992
+ for (let primaryIndex = 0; primaryIndex < actions.length; primaryIndex++) {
4993
+ if (globallyConsumedIndices.has(primaryIndex))
4994
+ continue;
4995
+ const primary = actions[primaryIndex];
4996
+ const groups = scanPrimaryCandidateGroups(actions, primaryIndex, globallyConsumedIndices);
4997
+ if (groups.length === 0)
4998
+ continue;
4999
+ const primaryEndpointKey = endpointKey(primary.capture.url);
5000
+ // Each independent array group on this primary is resolved (and
5001
+ // freshest-wins-checked) on its own — a re-queried primary can have one
5002
+ // array whose freshest occurrence is a later re-query while a second,
5003
+ // unrelated array on the SAME step is only ever threaded from the
5004
+ // first occurrence, so the two groups must not be forced to share one
5005
+ // anchor index.
5006
+ for (const group of groups) {
5007
+ // A re-queried primary (same endpoint hit more than once, per
5008
+ // findRequeriedActions) can have MULTIPLE occurrences that each
5009
+ // independently thread a join key into the SAME later drill-down —
5010
+ // e.g. two "available-products" calls that both happen to contain the
5011
+ // item the drill-down looks up. selectReturnAction/selectPayloadAction
5012
+ // already establish freshest-wins for this exact re-queried-primary
5013
+ // case, so the plan must anchor on the LAST such occurrence, not the
5014
+ // first one the forward scan happens to reach.
5015
+ let freshestIndex = primaryIndex;
5016
+ let freshestGroup = group;
5017
+ for (let laterIndex = primaryIndex + 1; laterIndex < actions.length; laterIndex++) {
5018
+ if (globallyConsumedIndices.has(laterIndex))
5019
+ continue;
5020
+ const laterAction = actions[laterIndex];
5021
+ if (endpointKey(laterAction.capture.url) !== primaryEndpointKey)
5022
+ continue;
5023
+ const laterGroups = scanPrimaryCandidateGroups(actions, laterIndex, globallyConsumedIndices);
5024
+ const laterGroup = laterGroups.find((g) => JSON.stringify(g.primaryArrayPath) === JSON.stringify(freshestGroup.primaryArrayPath));
5025
+ if (laterGroup === undefined)
5026
+ continue;
5027
+ const threadsSameDrill = laterGroup.targets.some((laterTarget) => freshestGroup.targets.some((currentTarget) => currentTarget.drillStepIndex === laterTarget.drillStepIndex));
5028
+ if (!threadsSameDrill)
5029
+ continue;
5030
+ // A primary occurrence's targets all read from THAT occurrence's own
5031
+ // response array, so switching the anchor to a later occurrence can
5032
+ // only be done wholesale, not merged field-by-field. Doing so is only
5033
+ // safe when the later occurrence re-threads EVERY drill-down the
5034
+ // current anchor already covers — otherwise a drill-down target
5035
+ // unique to the earlier occurrence (one it threads independently of
5036
+ // the re-queried join key) would be silently dropped instead of
5037
+ // folded at all.
5038
+ const laterCoversEveryCurrentTarget = freshestGroup.targets.every((currentTarget) => laterGroup.targets.some((laterTarget) => laterTarget.drillStepIndex === currentTarget.drillStepIndex));
5039
+ if (!laterCoversEveryCurrentTarget)
5040
+ continue;
5041
+ freshestIndex = laterIndex;
5042
+ freshestGroup = laterGroup;
5043
+ }
5044
+ // Defer to the later occurrence: it will be picked up on its own turn
5045
+ // through the outer loop, once it is reached as `primaryIndex`.
5046
+ if (freshestIndex !== primaryIndex)
5047
+ continue;
5048
+ plans.push({
5049
+ primaryStepIndex: freshestIndex,
5050
+ primaryArrayPath: freshestGroup.primaryArrayPath,
5051
+ targets: freshestGroup.targets,
5052
+ });
5053
+ // A step already folded into this plan's chains — the drill step(s)
5054
+ // and everything threaded onward from them — was already merged
5055
+ // into this primary's own array; it must not be re-picked up as a
5056
+ // fresh PRIMARY (its response was already consumed here) nor as a
5057
+ // drill target for a later, independent primary. The primary index
5058
+ // itself is marked consumed once per group pushed (idempotent via
5059
+ // Set.add), since the step itself is only visited once regardless of
5060
+ // how many independent array groups it yields.
5061
+ globallyConsumedIndices.add(freshestIndex);
5062
+ for (const target of freshestGroup.targets) {
5063
+ for (const chainIndex of target.chain)
5064
+ globallyConsumedIndices.add(chainIndex);
5065
+ }
5066
+ }
5067
+ }
5068
+ return plans;
5069
+ }
5070
+ /**
5071
+ * Parses an object-form recon-flow.json's optional `foldReturn` declaration
5072
+ * into a typed {@link FoldReturnSpec}, or `null` when the flow is array-form,
5073
+ * doesn't declare `foldReturn`, or declares it with a non-string field —
5074
+ * mirroring the same null-safe pattern the flow loader already applies to
5075
+ * `submitEndpointPattern`/`submitBodyPattern`.
5076
+ */
5077
+ function parseFoldReturnSpec(flowFileContents) {
5078
+ try {
5079
+ const raw = JSON.parse(flowFileContents);
5080
+ if (Array.isArray(raw))
5081
+ return null;
5082
+ if (raw === null ||
5083
+ typeof raw !== "object" ||
5084
+ !("steps" in raw) ||
5085
+ !Array.isArray(raw.steps)) {
5086
+ return null;
5087
+ }
5088
+ const foldReturn = raw.foldReturn;
5089
+ if (foldReturn === undefined || foldReturn === null || typeof foldReturn !== "object") {
5090
+ return null;
5091
+ }
5092
+ const { endpointPattern, resultsPath, drillResultsPath, joinFields } = foldReturn;
5093
+ if (typeof endpointPattern !== "string" ||
5094
+ typeof resultsPath !== "string" ||
5095
+ (drillResultsPath !== undefined &&
5096
+ (typeof drillResultsPath !== "string" || drillResultsPath.length === 0)) ||
5097
+ !Array.isArray(joinFields) ||
5098
+ joinFields.length === 0 ||
5099
+ !joinFields.every((f) => typeof f === "string" && f.length > 0)) {
5100
+ return null;
5101
+ }
5102
+ return {
5103
+ endpointPattern,
5104
+ resultsPath,
5105
+ ...(drillResultsPath !== undefined ? { drillResultsPath } : {}),
5106
+ joinFields,
5107
+ };
5108
+ }
5109
+ catch {
5110
+ return null;
5111
+ }
5112
+ }
5113
+ /** Reads the value at an exact JSON path out of a response body, or
5114
+ * `undefined` when any segment doesn't resolve. The exactness is the point:
5115
+ * {@link findObjectArrayField} is a DFS FIRST-match, so it would silently
5116
+ * override a flow-declared `resultsPath` that names a later array. */
5117
+ function readValueAtPath(body, path) {
5118
+ return path.reduce((node, segment) => {
5119
+ if (node === null || typeof node !== "object")
5120
+ return undefined;
5121
+ return node[segment];
5122
+ }, body);
5123
+ }
5124
+ /** The object items of the array at `path` — the same subset
5125
+ * {@link findObjectArrayField} exposes as `items`, but anchored to a
5126
+ * caller-supplied path instead of discovered by DFS. `null` when `path`
5127
+ * doesn't resolve to an array holding at least one non-array object, UNLESS
5128
+ * `path` resolves to a flat (non-array) object itself, in which case that
5129
+ * object is treated as an implicit one-item collection — see {@link
5130
+ * findAllObjectArrayFieldsOrWholeObject}, which resolves a {@link
5131
+ * FoldTarget}'s `chainArrayPath` to exactly this whole-body shape for a
5132
+ * detail-by-id drill/chain response. An {@link ARRAY_WILDCARD_SEGMENT}
5133
+ * segment in `path` flattens across every element of the array reached at
5134
+ * that point — in DFS/outer-array order — instead of indexing into one,
5135
+ * mirroring the `.flatMap` accessor {@link pathToFoldAccessorExpr} emits for
5136
+ * the same segment, so plan resolution and codegen always agree on which
5137
+ * items a fold covers. */
5138
+ function objectItemsAtPath(body, path) {
5139
+ const wildcardIndex = path.indexOf(ARRAY_WILDCARD_SEGMENT);
5140
+ if (wildcardIndex === -1) {
5141
+ const value = readValueAtPath(body, path);
5142
+ if (Array.isArray(value)) {
5143
+ const items = value.filter(isObjectArrayItem);
5144
+ return items.length > 0 ? items : null;
5145
+ }
5146
+ return isObjectArrayItem(value) ? [value] : null;
5147
+ }
5148
+ const outer = readValueAtPath(body, path.slice(0, wildcardIndex));
5149
+ if (!Array.isArray(outer))
5150
+ return null;
5151
+ const after = path.slice(wildcardIndex + 1);
5152
+ const items = outer.flatMap((element) => objectItemsAtPath(element, after) ?? []);
5153
+ return items.length > 0 ? items : null;
5154
+ }
5155
+ /**
5156
+ * Builds a {@link FoldPlan} from a flow-declared {@link FoldReturnSpec}, so a
5157
+ * site author can express a fold the structural heuristic misses.
5158
+ *
5159
+ * Returns `null` — the same null-safe contract as
5160
+ * {@link detectDrillDownFoldPlan} — when `endpointPattern` is not a valid
5161
+ * regex, when `resultsPath` resolves to no object array on any action, when
5162
+ * no strictly-later action's URL matches `endpointPattern`, or when neither
5163
+ * the matched drill-down's own response nor any later chained hop off of it
5164
+ * holds an object array or flat object (the emitter folds `foldMatches[0]`
5165
+ * out of that collection, so a plan without one has nothing to merge).
5166
+ */
5167
+ /** Same value set as {@link collectRequestStringValues}, plus every request
5168
+ * header value — a flow-declared `foldReturn` exists specifically to cover
5169
+ * joins the structural heuristic can't see (most notably a value threaded
5170
+ * through a request HEADER), so matching a spec's `joinFields` against the
5171
+ * drill capture must search headers even though the structural heuristic
5172
+ * deliberately doesn't (see {@link collectRequestStringValues}'s docstring). */
5173
+ function collectRequestValuesIncludingHeaders(capture) {
5174
+ const values = collectRequestStringValues(capture);
5175
+ for (const v of Object.values(capture.requestHeaders))
5176
+ values.add(v);
5177
+ return values;
5178
+ }
5179
+ /** Finds which of `primaryItems` the drill call actually captured, by
5180
+ * checking every field named in `joinFields` against the drill request's
5181
+ * full value set (URL, body, and headers) — the item matches only when ALL
5182
+ * join fields resolve, since a composite join (e.g. `accountId` + `region`)
5183
+ * is only a real match when every field lines up together. Returns `null`
5184
+ * (never a guessed index) when no item fully matches. */
5185
+ function resolveSpecMatchedPrimaryItemIndex(primaryItems, joinFields, drillCapture) {
5186
+ const requestValues = collectRequestValuesIncludingHeaders(drillCapture);
5187
+ if (requestValues.size === 0)
5188
+ return null;
5189
+ const matchedIndex = primaryItems.findIndex((item) => joinFields.every((field) => {
5190
+ const value = readValueAtPath(item, field.split("."));
5191
+ return ((typeof value === "string" && value.length > 0 && requestValues.has(value)) ||
5192
+ (typeof value === "number" && requestValues.has(String(value))) ||
5193
+ (typeof value === "boolean" && requestValues.has(String(value))));
5194
+ }));
5195
+ return matchedIndex === -1 ? null : matchedIndex;
5196
+ }
5197
+ /** Like {@link resolveSpecMatchedPrimaryItemIndex}, but also looks upstream of
5198
+ * `drillStepIndex` for the join key when `drillStepIndex`'s own request
5199
+ * doesn't carry it — a `foldReturn` spec's `endpointPattern` naturally names
5200
+ * the chain TERMINAL (the response actually holding the data an author wants
5201
+ * folded), not the opaque entry hop that carries the join key onward (e.g.
5202
+ * as a header-threaded token). Walks every earlier step back to
5203
+ * `primaryStepIndex`, in reverse so the closest (least ambiguous) candidate
5204
+ * wins first, and accepts one only once {@link computeFoldChain} confirms it
5205
+ * actually chains FORWARD to `drillStepIndex` — otherwise an unrelated
5206
+ * earlier step matching the join key by coincidence could hijack the fold.
5207
+ * Returns the resolved `entryIndex` alongside the matched item index so the
5208
+ * caller can build the fold's `chain` starting from the step that ACTUALLY
5209
+ * carries the join key, not from `drillStepIndex` — a chain built from
5210
+ * `drillStepIndex` alone would never include this upstream entry hop, so it
5211
+ * would never be re-executed (header-parameterized) per primary item at
5212
+ * runtime. */
5213
+ function resolveSpecMatchedPrimaryItemIndexAlongChain(actions, primaryItems, joinFields, primaryStepIndex, drillStepIndex) {
5214
+ for (let entryIndex = drillStepIndex; entryIndex > primaryStepIndex; entryIndex--) {
5215
+ const matched = resolveSpecMatchedPrimaryItemIndex(primaryItems, joinFields, actions[entryIndex].capture);
5216
+ if (matched === null)
5217
+ continue;
5218
+ if (entryIndex === drillStepIndex)
5219
+ return { entryIndex, primaryMatchedItemIndex: matched };
5220
+ const { chain } = computeFoldChain(actions, entryIndex, []);
5221
+ if (chain.includes(drillStepIndex))
5222
+ return { entryIndex, primaryMatchedItemIndex: matched };
3860
5223
  }
3861
5224
  return null;
3862
5225
  }
5226
+ /**
5227
+ * Compiles a flow-declared {@link FoldReturnSpec.endpointPattern} into a
5228
+ * capture predicate, or a predicate that always returns `false` when `spec`
5229
+ * is `null` or its pattern isn't a valid regex — the same null-safe
5230
+ * try/catch shape {@link buildFoldPlanFromSpec} already applied inline,
5231
+ * shared here so the action-sequence extractors can admit a spec-matched
5232
+ * drill-down capture under the identical rule that later resolves its fold
5233
+ * plan, instead of dropping it before the fold pipeline ever sees it.
5234
+ */
5235
+ function compileFoldReturnEndpointMatcher(spec) {
5236
+ if (spec === null)
5237
+ return () => false;
5238
+ const endpointRx = (() => {
5239
+ try {
5240
+ return new RegExp(spec.endpointPattern);
5241
+ }
5242
+ catch {
5243
+ return null;
5244
+ }
5245
+ })();
5246
+ if (endpointRx === null)
5247
+ return () => false;
5248
+ return (capture) => endpointRx.test(capture.url);
5249
+ }
5250
+ /**
5251
+ * A capture predicate matching whichever capture actually holds
5252
+ * {@link FoldReturnSpec.resultsPath}'s object array — the flow's own PRIMARY
5253
+ * results source, as opposed to {@link compileFoldReturnEndpointMatcher}'s
5254
+ * `endpointPattern` match (the drill-down). REST's `extractActionSequence`
5255
+ * never needs this: a REST primary is a POST/non-GET and is admitted
5256
+ * unconditionally regardless of `foldReturnSpec`. GraphQL's primary is
5257
+ * always a `query`, and `extractGraphQLActionSequence` drops every
5258
+ * non-mutation capture by default, so without this predicate a declared
5259
+ * `foldReturnSpec` would admit only the drill-down capture and never the
5260
+ * read op whose response the drill-down folds onto — leaving
5261
+ * `buildFoldPlanFromSpec` with no `primaryStepIndex` to resolve against.
5262
+ */
5263
+ function compileFoldReturnResultsMatcher(spec) {
5264
+ if (spec === null)
5265
+ return () => false;
5266
+ const resultsPath = spec.resultsPath.split(".");
5267
+ return (capture) => objectItemsAtPath(capture.responseBody, resultsPath) !== null;
5268
+ }
5269
+ function buildFoldPlanFromSpec(actions, spec) {
5270
+ const primaryArrayPath = spec.resultsPath.split(".");
5271
+ const matchesFoldReturnEndpoint = compileFoldReturnEndpointMatcher(spec);
5272
+ let freshestPlan = null;
5273
+ for (let primaryStepIndex = 0; primaryStepIndex < actions.length; primaryStepIndex++) {
5274
+ const primaryItems = objectItemsAtPath(actions[primaryStepIndex].capture.responseBody, primaryArrayPath);
5275
+ if (!primaryItems)
5276
+ continue;
5277
+ for (let drillStepIndex = primaryStepIndex + 1; drillStepIndex < actions.length; drillStepIndex++) {
5278
+ const drill = actions[drillStepIndex];
5279
+ if (!matchesFoldReturnEndpoint(drill.capture))
5280
+ continue;
5281
+ // Widened to a flat (non-array) object response the same way the
5282
+ // structural heuristic is (see findAllObjectArrayFieldsOrWholeObject):
5283
+ // an explicit foldReturn declaration must be able to express a
5284
+ // detail-by-id drill response exactly like the case the heuristic
5285
+ // detects on its own, not just an array field.
5286
+ // Falls through with an empty `[]` path baseline — rather than
5287
+ // bailing here — when the matched drill step's own response holds no
5288
+ // object-array/flat-object candidate, so an intermediate hop that
5289
+ // merely threads a value onward (holding no foldable data itself)
5290
+ // still lets computeFoldChain walk to a later chained step that DOES
5291
+ // hold the real per-item data, exactly as the structural heuristic
5292
+ // now does (see detectDrillDownFoldPlan). Validated after the chain
5293
+ // resolves, below, since `[]` is a valid empty baseline for
5294
+ // computeFoldChain but not a valid final drillArrayPath on its own.
5295
+ const matchResult = resolveSpecMatchedPrimaryItemIndexAlongChain(actions, primaryItems, spec.joinFields, primaryStepIndex, drillStepIndex);
5296
+ if (matchResult === null)
5297
+ continue;
5298
+ const { entryIndex, primaryMatchedItemIndex } = matchResult;
5299
+ // Rooted at `entryIndex` — the step whose request ACTUALLY carries the
5300
+ // join key — not `drillStepIndex`, so an upstream entry hop the join
5301
+ // key was only resolvable through (e.g. a header-threaded token) is
5302
+ // itself part of `chain` and gets re-executed (join-parameterized) per
5303
+ // primary item at runtime, same as every structurally-detected chain.
5304
+ // `drillResultsPath`, when given, still targets `drillStepIndex`'s own
5305
+ // response specifically (the endpoint the spec names); it is otherwise
5306
+ // left for computeFoldChain's own forward richness walk to resolve,
5307
+ // exactly as it does for every step beyond the chain's entry.
5308
+ const drillArrayPath = (() => {
5309
+ if (entryIndex !== drillStepIndex) {
5310
+ return (findObjectArrayFieldOrWholeObject(actions[entryIndex].capture.responseBody)?.path ??
5311
+ null);
5312
+ }
5313
+ if (spec.drillResultsPath === undefined) {
5314
+ return findObjectArrayFieldOrWholeObject(drill.capture.responseBody)?.path ?? null;
5315
+ }
5316
+ const path = spec.drillResultsPath.split(".");
5317
+ return objectItemsAtPath(drill.capture.responseBody, path) ? path : null;
5318
+ })();
5319
+ const { chain, chainArrayPath, chainTerminalIndex } = computeFoldChain(actions, entryIndex, drillArrayPath ?? []);
5320
+ // The chain's resolved terminal must actually hold foldable data —
5321
+ // an intermediate drill step with neither its own candidate NOR a
5322
+ // later chained step that resolves one has nothing to merge, so it
5323
+ // is skipped exactly as the pre-chain null check used to skip it.
5324
+ if (!objectItemsAtPath(actions[chainTerminalIndex].capture.responseBody, chainArrayPath)) {
5325
+ continue;
5326
+ }
5327
+ freshestPlan = {
5328
+ primaryStepIndex,
5329
+ primaryArrayPath,
5330
+ targets: [
5331
+ {
5332
+ joinFields: spec.joinFields,
5333
+ drillStepIndex: entryIndex,
5334
+ drillArrayPath: drillArrayPath ?? [],
5335
+ primaryMatchedItemIndex,
5336
+ chain,
5337
+ chainArrayPath,
5338
+ chainTerminalIndex,
5339
+ },
5340
+ ],
5341
+ };
5342
+ }
5343
+ }
5344
+ return freshestPlan;
5345
+ }
5346
+ /**
5347
+ * Unions a flow-declared `foldReturn` spec's drill-down target into the
5348
+ * structurally-detected plan for the SAME primary array (matched by
5349
+ * `primaryStepIndex` and `primaryArrayPath`), so a spec declaring an
5350
+ * independent target the heuristic missed is not silently discarded just
5351
+ * because the heuristic already resolved something for that primary. A spec
5352
+ * whose own primary/drill pair is entirely independent of every structural
5353
+ * plan — its `primaryStepIndex` and its target chains touch no index any
5354
+ * structural plan already consumes — is appended as a brand-new plan
5355
+ * instead. Only a spec whose primary step is itself already consumed by an
5356
+ * unrelated structural plan's own chain (not its own primary) is left alone,
5357
+ * to avoid folding onto a step that plan already depends on. A spec
5358
+ * re-declaring a `drillStepIndex` the heuristic already found (for the SAME
5359
+ * primary) is skipped, not duplicated.
5360
+ */
5361
+ function mergeSpecPlanOntoSamePrimary(structuralPlans, actions, foldReturnSpec) {
5362
+ if (foldReturnSpec === null)
5363
+ return [...structuralPlans];
5364
+ const specPlan = buildFoldPlanFromSpec(actions, foldReturnSpec);
5365
+ if (specPlan === null)
5366
+ return [...structuralPlans];
5367
+ const samePrimaryPlan = structuralPlans.find((plan) => plan.primaryStepIndex === specPlan.primaryStepIndex &&
5368
+ JSON.stringify(plan.primaryArrayPath) === JSON.stringify(specPlan.primaryArrayPath));
5369
+ if (samePrimaryPlan !== undefined) {
5370
+ return structuralPlans.map((plan) => {
5371
+ if (plan !== samePrimaryPlan)
5372
+ return plan;
5373
+ const existingDrillStepIndexes = new Set(plan.targets.map((target) => target.drillStepIndex));
5374
+ const newTargets = specPlan.targets.filter((target) => !existingDrillStepIndexes.has(target.drillStepIndex));
5375
+ return newTargets.length === 0
5376
+ ? plan
5377
+ : { ...plan, targets: [...plan.targets, ...newTargets] };
5378
+ });
5379
+ }
5380
+ // Keyed by the (primaryStepIndex, primaryArrayPath) pair, not
5381
+ // primaryStepIndex alone — a structural plan only ever consumed ITS OWN
5382
+ // array on that step, not the whole step. A spec whose resultsPath names
5383
+ // a second, structurally-undetected array on that exact same primary step
5384
+ // has already been proven independent by the samePrimaryPlan lookup above
5385
+ // (its primaryArrayPath differs from every structural plan's), so keying
5386
+ // solely on the step index would wrongly treat it as already consumed and
5387
+ // silently drop it.
5388
+ const consumedIndices = new Set();
5389
+ const consumedPrimarySteps = new Set();
5390
+ for (const plan of structuralPlans) {
5391
+ consumedPrimarySteps.add(`${plan.primaryStepIndex}:${JSON.stringify(plan.primaryArrayPath)}`);
5392
+ for (const target of plan.targets) {
5393
+ for (const chainIndex of target.chain)
5394
+ consumedIndices.add(chainIndex);
5395
+ }
5396
+ }
5397
+ const specConsumesOnlyItsOwnIndices = !consumedIndices.has(specPlan.primaryStepIndex) &&
5398
+ !consumedPrimarySteps.has(`${specPlan.primaryStepIndex}:${JSON.stringify(specPlan.primaryArrayPath)}`) &&
5399
+ specPlan.targets.every((target) => target.chain.every((chainIndex) => !consumedIndices.has(chainIndex)));
5400
+ return specConsumesOnlyItsOwnIndices ? [...structuralPlans, specPlan] : [...structuralPlans];
5401
+ }
5402
+ /**
5403
+ * The single fold-plan entry point: {@link detectDrillDownFoldPlan}'s
5404
+ * structural heuristic first, falling back to a flow-declared `foldReturn`
5405
+ * spec when the heuristic finds nothing. `emitMultiStepExecuteHttp` and
5406
+ * `selectEffectiveResponseBody` MUST both resolve through this rather than
5407
+ * calling the detector directly, or the emitted `executeHttp` and its
5408
+ * inferred schema would describe different calls (see
5409
+ * {@link selectEffectiveResponseBody}'s own docstring).
5410
+ *
5411
+ * A multipart step anywhere in a target's fold chain disqualifies only that
5412
+ * target: the fold loop re-issues EVERY chain step's request per item by
5413
+ * re-keying its rendered JSON request template (not just the immediate drill
5414
+ * step's), and a raw `FormData` upload has no such template to re-key — so
5415
+ * that target falls back to ordinary single-call emission instead of
5416
+ * emitting a broken loop, while any other target on the same primary that
5417
+ * has no multipart step in its chain still folds normally. A plan is dropped
5418
+ * from the returned array only when EVERY target for its primary is
5419
+ * multipart-disqualified, leaving nothing left to fold.
5420
+ *
5421
+ * {@link detectDrillDownFoldPlan} can find more than one independent
5422
+ * primary/drill-down pair in a single action sequence, so this returns every
5423
+ * plan that survives multipart disqualification, letting every downstream
5424
+ * emitter/shape-inference caller fold each of them.
5425
+ */
5426
+ /**
5427
+ * Every value ACTUALLY threaded from one dependent-drill-down chain hop's
5428
+ * response into a LATER chain hop's own request/headers — the same overlap
5429
+ * `computeFoldChain`'s `dependsOnChain` check already computes to decide a
5430
+ * step belongs in the chain at all — across every fold target
5431
+ * `detectDrillDownFoldPlan` / `buildFoldPlanFromSpec` resolve. These are the
5432
+ * values `indexStateValues` must index as producible state regardless of
5433
+ * `MIN_STATE_VALUE_LENGTH`, so `compileActionSteps` can thread a
5434
+ * chain-produced join value into the next hop's request even when that
5435
+ * value is a short one (e.g. a bare numeric status token).
5436
+ *
5437
+ * Deliberately NOT "every leaf value on a chain hop's response" — a chain
5438
+ * terminal's response commonly ECHOES the primary item's own join field
5439
+ * back (e.g. `sku` on the folded record), and that echoed value is a
5440
+ * PRIMARY-ITEM field the fold loop's own per-item render already threads
5441
+ * literally (see {@link findThreadedJoinFields}'s docstring); sweeping it
5442
+ * into state-threading's single-earliest-origin index would collapse every
5443
+ * item's distinct value onto whichever item's capture was indexed first.
5444
+ * Restricting to values that themselves reappear in a STRICTLY LATER chain
5445
+ * hop's own request — real cross-call threading, not an echo sitting still
5446
+ * in one response — keeps this exact to the shape state-threading is
5447
+ * actually needed for.
5448
+ *
5449
+ * Runs directly off raw actions (not `resolveFoldPlan`, which needs
5450
+ * `isMultipart` — unavailable before `compileActionSteps` has run) since
5451
+ * fold-plan DETECTION depends only on each action's `capture`.
5452
+ */
5453
+ function collectDependentDrillDownChainValues(actions, foldReturnSpec) {
5454
+ const structuralPlans = detectDrillDownFoldPlan(actions);
5455
+ const specPlan = foldReturnSpec === null ? null : buildFoldPlanFromSpec(actions, foldReturnSpec);
5456
+ const plans = structuralPlans.length > 0 ? structuralPlans : specPlan === null ? [] : [specPlan];
5457
+ const values = new Set();
5458
+ for (const plan of plans) {
5459
+ for (const target of plan.targets) {
5460
+ for (let j = 0; j < target.chain.length; j++) {
5461
+ const priorIndex = target.chain[j];
5462
+ const priorCapture = actions[priorIndex]?.capture;
5463
+ if (!priorCapture)
5464
+ continue;
5465
+ const responseValues = collectResponseLeafValues(priorCapture);
5466
+ const echoedValues = collectRequestValuesIncludingHeaders(priorCapture);
5467
+ for (let k = j + 1; k < target.chain.length; k++) {
5468
+ const laterCapture = actions[target.chain[k]]?.capture;
5469
+ if (!laterCapture)
5470
+ continue;
5471
+ const laterRequestValues = collectRequestValuesIncludingHeaders(laterCapture);
5472
+ for (const v of responseValues) {
5473
+ if (!echoedValues.has(v) && laterRequestValues.has(v))
5474
+ values.add(v);
5475
+ }
5476
+ }
5477
+ }
5478
+ }
5479
+ }
5480
+ return values;
5481
+ }
5482
+ function resolveFoldPlan(actions, foldReturnSpec = null) {
5483
+ const structuralPlans = detectDrillDownFoldPlan(actions);
5484
+ const plans = structuralPlans.length > 0
5485
+ ? mergeSpecPlanOntoSamePrimary(structuralPlans, actions, foldReturnSpec)
5486
+ : (() => {
5487
+ if (foldReturnSpec === null)
5488
+ return [];
5489
+ const specPlan = buildFoldPlanFromSpec(actions, foldReturnSpec);
5490
+ return specPlan === null ? [] : [specPlan];
5491
+ })();
5492
+ return plans.flatMap((plan) => {
5493
+ const targets = plan.targets.filter((target) => !target.chain.some((chainIndex) => actions[chainIndex].isMultipart));
5494
+ return targets.length === 0 ? [] : [{ ...plan, targets }];
5495
+ });
5496
+ }
5497
+ /** Rebuilds `value` with every occurrence of `target` (compared by object
5498
+ * identity) replaced by `replacement`, spreading every ancestor
5499
+ * array/object level so sibling fields and sibling array elements survive
5500
+ * unchanged. Identity, not a path, is what locates the splice point: a
5501
+ * {@link FoldPlan.primaryArrayPath} carrying an {@link ARRAY_WILDCARD_SEGMENT}
5502
+ * names a whole family of per-group arrays, not one splice-able location, so
5503
+ * only the matched item's own object reference (never cloned by
5504
+ * {@link findAllObjectArrayFields}/{@link objectItemsAtPath}, both of which
5505
+ * only filter) pins down where the fold actually lands, regardless of which
5506
+ * group it came from. */
5507
+ function replaceByReference(value, target, replacement) {
5508
+ if (value === target)
5509
+ return replacement;
5510
+ if (Array.isArray(value))
5511
+ return value.map((v) => replaceByReference(v, target, replacement));
5512
+ if (value !== null && typeof value === "object") {
5513
+ return Object.fromEntries(Object.entries(value).map(([k, v]) => [
5514
+ k,
5515
+ replaceByReference(v, target, replacement),
5516
+ ]));
5517
+ }
5518
+ return value;
5519
+ }
5520
+ /**
5521
+ * The value-level counterpart of the per-item loop-and-merge
5522
+ * `emitMultiStepExecuteHttp` emits for a detected {@link FoldPlan}: merges
5523
+ * the captured drill-down item whose `joinFields` match the single captured
5524
+ * primary sample's matched array item — at `primaryMatchedItemIndex`, the
5525
+ * same item `detectDrillDownFoldPlan` built the join key from, not
5526
+ * necessarily index 0 — falling back to the drill array's first item only
5527
+ * when no drill item matches, so schema inference walks the SAME shape the
5528
+ * folded `executeHttp` actually returns at runtime. Reads both arrays at
5529
+ * the plan's OWN paths rather than re-running `findObjectArrayField`, so a
5530
+ * flow-declared `resultsPath` stays authoritative here exactly as it is in
5531
+ * the emitter. Falls back to the unmerged primary body if either capture no
5532
+ * longer resolves an object array — same drift guard as the emitter's own
5533
+ * `throw` at the analogous point, minus the throw, since shape inference
5534
+ * degrading gracefully is preferable to failing a generate run over it.
5535
+ */
5536
+ function foldResponseBodyForShapeInference(actionSteps, foldPlan, initialBody = actionSteps[foldPlan.primaryStepIndex].capture.responseBody) {
5537
+ return foldPlan.targets.reduce((body, target) => {
5538
+ const drillBody = actionSteps[target.chainTerminalIndex].capture.responseBody;
5539
+ const primaryItems = objectItemsAtPath(body, foldPlan.primaryArrayPath);
5540
+ const drillItems = objectItemsAtPath(drillBody, target.chainArrayPath);
5541
+ const matchedItem = primaryItems?.[target.primaryMatchedItemIndex];
5542
+ const drillMatch = drillItems?.find((d) => target.joinFields.every((f) => String(readValueAtPath(d, f.split("."))) ===
5543
+ String(readValueAtPath(matchedItem, f.split("."))))) ?? drillItems?.[0];
5544
+ if (!primaryItems || !matchedItem || !drillMatch)
5545
+ return body;
5546
+ return replaceByReference(body, matchedItem, { ...matchedItem, ...drillMatch });
5547
+ }, initialBody);
5548
+ }
3863
5549
  /**
3864
5550
  * Detects whether a read-only GraphQL primary operation exposes a bounded
3865
5551
  * paging signal: a total/count field in the captured response alongside a
@@ -3945,7 +5631,7 @@ function buildPaginatedGqlExecuteHttpBody(opts) {
3945
5631
  return ` const baseVariables = ${gqlVariablesExpr};
3946
5632
  const PAGE_SIZE = ${pageSize};
3947
5633
  // Bounded so a paging bug (a total that never converges) can't loop forever.
3948
- const MAX_PAGES = 50;
5634
+ const MAX_PAGES = payload.maxPages ?? 50;
3949
5635
  const itemsById = new Map<string, unknown>();
3950
5636
  let skip = 0;
3951
5637
  const page = await getGql(context.baseUrl)(${gqlOperationNameExpr}, ${queryConstName}, ${variablesForCall});
@@ -4008,7 +5694,7 @@ function buildContractChecklist(opts) {
4008
5694
  ].filter((line) => line !== "");
4009
5695
  }
4010
5696
  function emitContractTs(opts) {
4011
- const { siteId, pascal, baseUrl, baseHeaders, minTime, safeRps, hasRateLimitProbeData = false, responseBody, responseBodySamples = [responseBody], gql, gqlQuery, endpointPath, gqlOperationName, gqlVariables, auxFiles, multiStepBody, omitExecuteHttp = false, isSubmissionFlow = false, inputBody, hasMultipartStep = false, discoveredFormFields, fieldOptionsMap, discoveredOptionFields, discoveredRawOptionFields, discoveredAdditionalBodyKeys, discoveredStructuredKeys, payloadFieldNames, headerBindings = [], } = opts;
5697
+ const { siteId, displayName, pascal, baseUrl, baseHeaders, minTime, safeRps, hasRateLimitProbeData = false, responseBody, responseBodySamples = [responseBody], gql, gqlQuery, endpointPath, gqlOperationName, gqlVariables, auxFiles, multiStepBody, omitExecuteHttp = false, isSubmissionFlow = false, inputBody, hasMultipartStep = false, discoveredFormFields, fieldOptionsMap, discoveredOptionFields, discoveredRawOptionFields, discoveredAdditionalBodyKeys, discoveredStructuredKeys, payloadFieldNames, headerBindings = [], unpopulatedDeclaredVariables = [], } = opts;
4012
5698
  // This is the CLIENT-level schema — createHttpClient's default, and the
4013
5699
  // plugin's caller-facing contract (what executeHttp's return value promises
4014
5700
  // its own caller). It does NOT validate any individual call in a multi-step
@@ -4033,6 +5719,7 @@ function emitContractTs(opts) {
4033
5719
  // successful return IS a real signal — z.unknown() would be dishonest
4034
5720
  // in the other direction, hiding a field the flow can actually promise.
4035
5721
  const conditionalFieldNames = gql && gqlQuery ? collectConditionalGraphQLFieldNames(gqlQuery) : undefined;
5722
+ const aggregateUnitBasisFindingsByPath = groupAggregateUnitBasisFindingsByPath(responseBodySamples);
4036
5723
  const responseSchemaExpr = omitExecuteHttp && isSubmissionFlow
4037
5724
  ? `z.object({ verified: z.boolean() })`
4038
5725
  : omitExecuteHttp
@@ -4040,6 +5727,7 @@ function emitContractTs(opts) {
4040
5727
  : inferZodSchemaFromSamples(responseBodySamples, 0, "", {
4041
5728
  conditionalFieldNames,
4042
5729
  looseServerResponse: true,
5730
+ aggregateUnitBasisFindingsByPath,
4043
5731
  });
4044
5732
  // Multi-step flows that include a multipart upload need the binary asset
4045
5733
  // on the payload. ApplicantContactSchema (via ApplicantResumeSchema) already
@@ -4068,6 +5756,12 @@ function emitContractTs(opts) {
4068
5756
  const basePayloadSchemaExpr = inputBody
4069
5757
  ? `ApplicantContactSchema`
4070
5758
  : `z.object({\n query: z.string().min(1),\n})`;
5759
+ // Only the single-endpoint GraphQL read path (a real primary operation, no
5760
+ // multi-step flow) is a candidate for a paging signal — multiStepBody
5761
+ // already owns its own per-call semantics.
5762
+ const paginationSignal = !multiStepBody && gql && gqlOperationName
5763
+ ? detectPaginationSignal(responseBody, gqlVariables)
5764
+ : null;
4071
5765
  // Every field source below (the base extend's own keys, form-schema
4072
5766
  // discovery, browser-flow splicing, option/raw-option enums, additional
4073
5767
  // body keys, and structured keys) is merged into a SINGLE `.extend({...})`
@@ -4080,6 +5774,13 @@ function emitContractTs(opts) {
4080
5774
  const addExtendField = (name, line) => {
4081
5775
  extendFields.set(name, line);
4082
5776
  };
5777
+ // A detected bounded-paging signal means buildPaginatedGqlExecuteHttpBody
5778
+ // will emit a loop bounded by MAX_PAGES — expose that bound as a caller-
5779
+ // overridable payload field, mirroring how PAGE_SIZE is already sourced
5780
+ // from the detected signal.
5781
+ if (paginationSignal) {
5782
+ addExtendField("maxPages", " maxPages: z.number().int().positive().optional(),");
5783
+ }
4083
5784
  // The base extend's own keys — submission flows only.
4084
5785
  if (inputBody) {
4085
5786
  addExtendField("Email", " Email: z.email(),");
@@ -4240,6 +5941,25 @@ function emitContractTs(opts) {
4240
5941
  // that collides with the base extend's own Email/ClickUrl/Answers) collapses
4241
5942
  // to its last-declared line, rather than becoming a second, dupe-prone
4242
5943
  // `.extend()` call chained onto the schema.
5944
+ // A field whose name matches (case-insensitively, the same convention
5945
+ // renderGqlVariablesExpr uses) a declared GraphQL variable that no capture
5946
+ // ever populated has no wiring target in executeHttp — it would replay the
5947
+ // captured frozen value regardless of what the caller sends. Downgrading it
5948
+ // to `.optional()` here, at the single merge point every source funnels
5949
+ // through, keeps the schema honest without special-casing any one source.
5950
+ // Email/ClickUrl/Answers are the public contract every submission-flow
5951
+ // plugin must declare unconditionally (see the basePayloadSchemaExpr
5952
+ // comment above) — a GraphQL mutation that happens to declare an
5953
+ // unpopulated variable with a matching name (e.g. `$email`) must not
5954
+ // downgrade that required base field.
5955
+ const baseContractFieldNames = new Set(["Email", "ClickUrl", "Answers"]);
5956
+ for (const [fieldName, line] of extendFields) {
5957
+ if (inputBody && baseContractFieldNames.has(fieldName))
5958
+ continue;
5959
+ if (unpopulatedDeclaredVariables.some((name) => name.toLowerCase() === fieldName.toLowerCase())) {
5960
+ extendFields.set(fieldName, line.replace(/,\s*$/, ".optional(),"));
5961
+ }
5962
+ }
4243
5963
  const mergedExtension = extendFields.size > 0 ? `.extend({\n${[...extendFields.values()].join("\n")}\n})` : "";
4244
5964
  const payloadSchemaExpr = `${basePayloadSchemaExpr}${mergedExtension}`;
4245
5965
  // basePayloadSchemaExpr's own Answers field always wraps in
@@ -4266,6 +5986,12 @@ function emitContractTs(opts) {
4266
5986
  const caseInsensitiveHeadersImport = hasMultipartStep && !omitExecuteHttp
4267
5987
  ? `import { omitHeaderCaseInsensitive } from "${ENGINE_PKG}/lib/case-insensitive-headers";\n`
4268
5988
  : "";
5989
+ // emitMultiStepExecuteHttp emits a call to this helper whenever more than
5990
+ // one fold plan resolves — the generated plugin package can't reach into
5991
+ // recon-generate.ts's own module scope, so it must import it separately.
5992
+ const mergeFoldedPrimaryBodiesImport = multiStepBody?.includes("mergeFoldedPrimaryBodies(") === true
5993
+ ? `import { mergeFoldedPrimaryBodies } from "${ENGINE_PKG}/lib/merge-folded-primary-bodies";\n`
5994
+ : "";
4269
5995
  // Emit identifier-shaped keys unquoted so Biome's formatter doesn't rewrite
4270
5996
  // the generated file on first lint:fix.
4271
5997
  const headersLiteral = Object.entries(baseHeaders)
@@ -4315,12 +6041,6 @@ const httpClient = createHttpClient({ schema: ${pascal}ResponseSchema, bottlenec
4315
6041
  const gqlVariablesExpr = gqlOperationName
4316
6042
  ? renderGqlVariablesExpr(gqlVariables, payloadFieldNames)
4317
6043
  : "{ q: payload.query }";
4318
- // Only the single-endpoint GraphQL read path (a real primary operation, no
4319
- // multi-step flow) is a candidate for a paging signal — multiStepBody
4320
- // already owns its own per-call semantics.
4321
- const paginationSignal = !multiStepBody && gql && gqlOperationName
4322
- ? detectPaginationSignal(responseBody, gqlVariables)
4323
- : null;
4324
6044
  const executeHttpBody = multiStepBody
4325
6045
  ? multiStepBody
4326
6046
  : paginationSignal
@@ -4414,7 +6134,7 @@ ${executeHttpBody}
4414
6134
 
4415
6135
  ${bottleneckImport}import { z } from "zod/v4";
4416
6136
 
4417
- ${fixtureImport}${applicantContactImport}${caseInsensitiveHeadersImport}${multipartBoolImport}${clientImport}
6137
+ ${fixtureImport}${applicantContactImport}${caseInsensitiveHeadersImport}${mergeFoldedPrimaryBodiesImport}${multipartBoolImport}${clientImport}
4418
6138
  import type { BrowserSession } from "${ENGINE_PKG}/scraper/session";
4419
6139
  import type { SitePlugin, SitePluginContext, SitePluginResult } from "${ENGINE_PKG}/site-plugin";
4420
6140
  import { run${pascal}BrowserFlow } from "@/sites/${siteId}/flows/browser-flow";
@@ -4432,7 +6152,7 @@ ${internalRequestReferenceBlock}${queryConst}${gqlCacheBlock}${fixtureComments}
4432
6152
  ${pluginDocComment}
4433
6153
  export const ${camel}Plugin: SitePlugin<${pascal}Payload, ${pascal}Response> = {
4434
6154
  meta: {
4435
- siteId: ${JSON.stringify(siteId)},
6155
+ siteId: ${JSON.stringify(siteId)},${displayName !== undefined ? `\n displayName: ${JSON.stringify(displayName)},` : ""}
4436
6156
  bodySchema: ${pascal}PayloadSchema,
4437
6157
  responseSchema: ${pascal}ResponseSchema,
4438
6158
  defaultBaseUrl: ${JSON.stringify(baseUrl)},
@@ -5002,15 +6722,29 @@ async function main() {
5002
6722
  return [];
5003
6723
  }
5004
6724
  })();
5005
- const { flowSteps, frameSelector, submitEndpointPattern, submitBodyPattern } = (() => {
6725
+ const { flowSteps, frameSelector, submitEndpointPattern, submitBodyPattern, displayName, foldReturnSpec, } = (() => {
6726
+ const flowFileContents = (() => {
6727
+ try {
6728
+ return (0, node_fs_1.readFileSync)(flowFile, "utf8");
6729
+ }
6730
+ catch {
6731
+ return null;
6732
+ }
6733
+ })();
6734
+ // Parsed off the raw bytes, independently of the steps shape below, so a
6735
+ // valid `foldReturn` still resolves when the rest of the flow file is
6736
+ // degenerate — the two declarations fail independently.
6737
+ const foldReturnSpec = flowFileContents === null ? null : parseFoldReturnSpec(flowFileContents);
5006
6738
  try {
5007
- const raw = JSON.parse((0, node_fs_1.readFileSync)(flowFile, "utf8"));
6739
+ const raw = flowFileContents === null ? null : JSON.parse(flowFileContents);
5008
6740
  if (Array.isArray(raw))
5009
6741
  return {
5010
6742
  flowSteps: raw,
5011
6743
  frameSelector: undefined,
5012
6744
  submitEndpointPattern: null,
5013
6745
  submitBodyPattern: null,
6746
+ displayName: undefined,
6747
+ foldReturnSpec,
5014
6748
  };
5015
6749
  if (raw !== null &&
5016
6750
  typeof raw === "object" &&
@@ -5022,6 +6756,8 @@ async function main() {
5022
6756
  frameSelector: obj.frameSelector,
5023
6757
  submitEndpointPattern: obj.submitEndpointPattern ?? null,
5024
6758
  submitBodyPattern: obj.submitBodyPattern ?? null,
6759
+ displayName: obj.displayName,
6760
+ foldReturnSpec,
5025
6761
  };
5026
6762
  }
5027
6763
  return {
@@ -5029,6 +6765,8 @@ async function main() {
5029
6765
  frameSelector: undefined,
5030
6766
  submitEndpointPattern: null,
5031
6767
  submitBodyPattern: null,
6768
+ displayName: undefined,
6769
+ foldReturnSpec,
5032
6770
  };
5033
6771
  }
5034
6772
  catch {
@@ -5037,6 +6775,8 @@ async function main() {
5037
6775
  frameSelector: undefined,
5038
6776
  submitEndpointPattern: null,
5039
6777
  submitBodyPattern: null,
6778
+ displayName: undefined,
6779
+ foldReturnSpec,
5040
6780
  };
5041
6781
  }
5042
6782
  })();
@@ -5105,7 +6845,9 @@ async function main() {
5105
6845
  // Hoisted so both the primary-operation gate below and rawActionCaptures
5106
6846
  // (further down) read the same computed sequence instead of calling the
5107
6847
  // extractor twice.
5108
- const graphqlActionSequence = gql ? extractGraphQLActionSequence(captures, submitPatterns) : [];
6848
+ const graphqlActionSequence = gql
6849
+ ? extractGraphQLActionSequence(captures, submitPatterns, foldReturnSpec)
6850
+ : [];
5109
6851
  const primaryGraphQLOperation = gql && graphqlActionSequence.length === 0
5110
6852
  ? selectPrimaryGraphQLOperation(captures, flowSteps, vocabulary, process.env, ownBackendHostnames, fallbackDomain)
5111
6853
  : null;
@@ -5156,7 +6898,7 @@ async function main() {
5156
6898
  // heuristic extraction finds.
5157
6899
  const patternedHeuristicActionCaptures = gql
5158
6900
  ? graphqlActionSequence
5159
- : collapseRedundantPatches(extractActionSequence(captures, submitPatterns));
6901
+ : collapseRedundantPatches(extractActionSequence(captures, submitPatterns, foldReturnSpec));
5160
6902
  // The same undercount hazard applies one layer below the manifest: a
5161
6903
  // flow-declared submitEndpointPattern that matches only one section's URL
5162
6904
  // (the natural way to describe "the button that finishes the wizard")
@@ -5169,8 +6911,8 @@ async function main() {
5169
6911
  const unfilteredHeuristicActionCaptures = submitPatterns.endpoint === null && submitPatterns.body === null
5170
6912
  ? patternedHeuristicActionCaptures
5171
6913
  : gql
5172
- ? extractGraphQLActionSequence(captures, null)
5173
- : collapseRedundantPatches(extractActionSequence(captures, null));
6914
+ ? extractGraphQLActionSequence(captures, null, foldReturnSpec)
6915
+ : collapseRedundantPatches(extractActionSequence(captures, null, foldReturnSpec));
5174
6916
  const patternUndercounts = patternedHeuristicActionCaptures.length < unfilteredHeuristicActionCaptures.length;
5175
6917
  if (patternUndercounts) {
5176
6918
  logger.info(`submission selection: ignoring submitEndpointPattern/submitBodyPattern (${patternedHeuristicActionCaptures.length} capture(s)) as an undercount of the unfiltered heuristic action sequence (${unfilteredHeuristicActionCaptures.length} capture(s))`);
@@ -5201,8 +6943,8 @@ async function main() {
5201
6943
  const rawUnfilteredActionCaptures = submitEndpointPattern === null
5202
6944
  ? null
5203
6945
  : gql
5204
- ? extractGraphQLActionSequence(captures, null)
5205
- : collapseRedundantPatches(extractActionSequence(captures, null));
6946
+ ? extractGraphQLActionSequence(captures, null, foldReturnSpec)
6947
+ : collapseRedundantPatches(extractActionSequence(captures, null, foldReturnSpec));
5206
6948
  // Form-schema detection runs BEFORE state-indexing so the field-id/option-id
5207
6949
  // UUIDs can be shielded from indexing — those UUIDs are stable schema
5208
6950
  // anchors that T2/T3 substitution depends on remaining literal in body
@@ -5250,8 +6992,16 @@ async function main() {
5250
6992
  ];
5251
6993
  })();
5252
6994
  const actionCaptureIndices = new Set(actionCaptures.map((a) => a.index));
6995
+ // Resolved off raw actionCaptures — fold-plan DETECTION depends only on
6996
+ // each action's capture, so this runs before compileActionSteps/
6997
+ // indexStateValues even exist — so a short numeric join value threaded
6998
+ // through a dependent-drill-down chain hop still gets indexed as
6999
+ // producible state (see collectDependentDrillDownChainValues).
7000
+ const dependentDrillDownChainValues = actionCaptures.length > 1
7001
+ ? collectDependentDrillDownChainValues(actionCaptures, foldReturnSpec)
7002
+ : new Set();
5253
7003
  const stateIndex = actionCaptures.length > 1
5254
- ? indexStateValues(captures, shieldedUuids, actionCaptureIndices)
7004
+ ? indexStateValues(captures, shieldedUuids, actionCaptureIndices, dependentDrillDownChainValues)
5255
7005
  : new Map();
5256
7006
  const actionSteps = actionCaptures.length > 1 ? compileActionSteps(actionCaptures, stateIndex) : [];
5257
7007
  const isSubmissionFlow = actionSteps.length > 1;
@@ -5363,18 +7113,25 @@ async function main() {
5363
7113
  const multiStepBody = browserFlowOnly
5364
7114
  ? undefined
5365
7115
  : isSubmissionFlow
5366
- ? emitMultiStepExecuteHttp(actionSteps, inputBody, errorSignals, fieldNameMap, discoveredFormFields, fieldOptionsMap, discoveredOptionFields, discoveredRawOptionFields, discoveredAdditionalBodyKeys, baseUrl, baseUrlDerivedHeaders, tenantSubdomainHeaders, formSchema, personaBindings, entryUrlParams, shieldedUuids, selectResolutions, discoveredStructuredKeys, rawCodeFields)
7116
+ ? emitMultiStepExecuteHttp(actionSteps, inputBody, errorSignals, fieldNameMap, discoveredFormFields, fieldOptionsMap, discoveredOptionFields, discoveredRawOptionFields, discoveredAdditionalBodyKeys, baseUrl, baseUrlDerivedHeaders, tenantSubdomainHeaders, formSchema, personaBindings, entryUrlParams, shieldedUuids, selectResolutions, discoveredStructuredKeys, rawCodeFields, foldReturnSpec)
5367
7117
  : undefined;
5368
7118
  const hasMultipartStep = actionSteps.some((s) => s.isMultipart);
5369
7119
  const headerBindings = collectHeaderBindings(actionSteps);
5370
7120
  // Shape inference targets the SAME call executeHttp returns — see
5371
7121
  // selectEffectiveResponseBody — so the two surfaces can't describe different calls.
5372
- const effectiveResponseBody = selectEffectiveResponseBody(isSubmissionFlow, actionSteps, responseBody);
7122
+ const effectiveResponseBody = selectEffectiveResponseBody(isSubmissionFlow, actionSteps, responseBody, foldReturnSpec);
7123
+ // A declared foldReturn that resolves to no plan is a silent no-op otherwise
7124
+ // — the flow author gets the discarding selectReturnAction path with nothing
7125
+ // in the output saying their declaration never applied.
7126
+ if (foldReturnSpec !== null && resolveFoldPlan(actionSteps, foldReturnSpec).length === 0) {
7127
+ logger.warn(`flow declares foldReturn (endpointPattern: ${foldReturnSpec.endpointPattern}, resultsPath: ${foldReturnSpec.resultsPath}, joinFields: ${foldReturnSpec.joinFields.join(", ")}) but no fold plan resolved — no later capture matched the endpoint pattern, resultsPath resolved to no object array, or the matched drill-down is multipart; the drill-down's response will not be folded`);
7128
+ }
5373
7129
  logger.info(`generating plugin for ${siteId} (${gql ? "GraphQL" : browserFlowOnly ? `submission flow, ${actionSteps.length} steps, browser-flow-only (cross-domain hop detected)` : isSubmissionFlow ? `submission flow, ${actionSteps.length} steps` : "single-endpoint REST"}, baseUrl: ${baseUrl})`);
5374
7130
  if (emit === "config") {
5375
7131
  (0, node_fs_1.mkdirSync)(outDir, { recursive: true });
5376
7132
  (0, node_fs_1.writeFileSync)(manifestPath, emitConfigManifest({
5377
7133
  siteId,
7134
+ displayName,
5378
7135
  baseUrl,
5379
7136
  flowSteps,
5380
7137
  vocabulary,
@@ -5408,6 +7165,7 @@ async function main() {
5408
7165
  });
5409
7166
  const contractOpts = {
5410
7167
  siteId,
7168
+ displayName,
5411
7169
  pascal,
5412
7170
  baseUrl,
5413
7171
  // G1+G2: only the static headers (no baseUrl/tenant-subdomain references)
@@ -5445,6 +7203,7 @@ async function main() {
5445
7203
  discoveredStructuredKeys,
5446
7204
  payloadFieldNames: browserFlow.payloadFieldNames,
5447
7205
  headerBindings,
7206
+ unpopulatedDeclaredVariables: primaryGraphQLOperation?.unpopulatedDeclaredVariables ?? [],
5448
7207
  };
5449
7208
  const contractCode = emitContractTs(contractOpts);
5450
7209
  // Fails loudly rather than shipping a flow that requires a URL field it