@claritylabs/cl-sdk 3.0.4 → 3.0.6

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.
package/dist/index.js CHANGED
@@ -9411,6 +9411,150 @@ function cleanText(value, fallback) {
9411
9411
  const text = value?.replace(/\s+/g, " ").trim();
9412
9412
  return text || fallback;
9413
9413
  }
9414
+ function simplifyOrganizerTitle(value, fallback, kind) {
9415
+ const title = cleanText(value, fallback);
9416
+ if (/^declarations\b/i.test(title)) return "Declarations";
9417
+ if (/^policy\s+form\b/i.test(title)) return "Policy Form";
9418
+ if (/^definitions\b/i.test(title)) return "Definitions";
9419
+ if (kind === "page_group" && /^endorsements?\b/i.test(title)) return "Endorsements";
9420
+ const endorsementNumber = title.match(/^endorsement\s+(?:no\.?|number|#)?\s*([A-Z0-9][A-Z0-9.-]*)\b/i)?.[1];
9421
+ if (endorsementNumber) return `Endorsement No. ${endorsementNumber}`;
9422
+ if (kind === "endorsement" && /^endorsements?\s+\d+\s*[–-]\s*\d+\b/i.test(title)) {
9423
+ return title.replace(/[–—]/g, "-").replace(/\s*\(.*/, "").trim();
9424
+ }
9425
+ return title;
9426
+ }
9427
+ function endorsementReference(value) {
9428
+ return value?.match(/\bendorsement\s+(?:no\.?|number|#)?\s*([A-Z0-9][A-Z0-9.-]*)\b/i)?.[1]?.toUpperCase();
9429
+ }
9430
+ function endorsementTitle(value) {
9431
+ const text = cleanText(value, "");
9432
+ const number = text.match(/\bendorsement\s+(?:no\.?|number|#)\s*([A-Z0-9][A-Z0-9.-]*)\b/i)?.[1]?.toUpperCase();
9433
+ return number ? `Endorsement No. ${number}` : void 0;
9434
+ }
9435
+ function rejectsOrganizerGroup(group, children) {
9436
+ if (/^endorsements?\b/i.test(group.title) && group.kind !== "page_group") return true;
9437
+ if (group.kind !== "endorsement") return false;
9438
+ if (/^endorsements?\s+\d+\s*[–-]\s*\d+\b/i.test(group.title)) return true;
9439
+ const childNumbers = new Set(
9440
+ children.map((child) => endorsementReference([child.title, child.description, child.textExcerpt].filter(Boolean).join(" "))).filter((value) => Boolean(value))
9441
+ );
9442
+ return childNumbers.size > 1;
9443
+ }
9444
+ function isEndorsementGroup(node) {
9445
+ return node.kind === "page_group" && /^endorsements?\b/i.test(node.title);
9446
+ }
9447
+ function endorsementGroupNodeId(documentId, parentId) {
9448
+ return [
9449
+ documentId.replace(/[^a-zA-Z0-9_.:-]/g, "_"),
9450
+ "source_node",
9451
+ "page_group",
9452
+ "endorsements",
9453
+ parentId?.replace(/[^a-zA-Z0-9_.:-]/g, "_").slice(0, 48) ?? "root"
9454
+ ].join(":");
9455
+ }
9456
+ function applyEndorsementGrouping(sourceTree) {
9457
+ const relabeledTree = sourceTree.map((node) => {
9458
+ if (node.kind === "document" || isEndorsementGroup(node) || node.kind === "endorsement") return node;
9459
+ const title = endorsementTitle([node.title, node.description, node.textExcerpt].filter(Boolean).join(" "));
9460
+ if (!title) return node;
9461
+ return {
9462
+ ...node,
9463
+ kind: "endorsement",
9464
+ title,
9465
+ description: cleanText(
9466
+ [title, "endorsement", node.pageStart ? `page ${node.pageStart}` : void 0, node.textExcerpt].filter(Boolean).join(" | "),
9467
+ title
9468
+ ),
9469
+ metadata: {
9470
+ ...node.metadata,
9471
+ organizerRepair: "normalize_endorsement_grouping"
9472
+ }
9473
+ };
9474
+ });
9475
+ const byParent = nodesByParent(relabeledTree);
9476
+ const groupsByParent = /* @__PURE__ */ new Map();
9477
+ const endorsementGroupIds = new Set(
9478
+ relabeledTree.filter(isEndorsementGroup).map((node) => node.id)
9479
+ );
9480
+ let nextTree = relabeledTree.map((node) => {
9481
+ if (!isEndorsementGroup(node)) return node;
9482
+ const normalized = {
9483
+ ...node,
9484
+ kind: "page_group",
9485
+ title: "Endorsements",
9486
+ description: cleanText(node.description, "Endorsement forms grouped by source order"),
9487
+ metadata: {
9488
+ ...node.metadata,
9489
+ sourceTreeVersion: "v3",
9490
+ organizer: node.metadata?.organizer ?? "endorsement_grouping"
9491
+ }
9492
+ };
9493
+ groupsByParent.set(node.parentId, normalized);
9494
+ endorsementGroupIds.add(node.id);
9495
+ return normalized;
9496
+ });
9497
+ nextTree = nextTree.map((node) => {
9498
+ if (!endorsementGroupIds.has(node.parentId ?? "")) return node;
9499
+ const title = endorsementTitle([node.title, node.description, node.textExcerpt].filter(Boolean).join(" "));
9500
+ if (!title) return node;
9501
+ return {
9502
+ ...node,
9503
+ kind: "endorsement",
9504
+ title,
9505
+ description: cleanText(
9506
+ [title, "endorsement", node.pageStart ? `page ${node.pageStart}` : void 0, node.textExcerpt].filter(Boolean).join(" | "),
9507
+ title
9508
+ ),
9509
+ metadata: {
9510
+ ...node.metadata,
9511
+ organizerRepair: "normalize_endorsement_grouping"
9512
+ }
9513
+ };
9514
+ });
9515
+ for (const [parentId, children] of byParent) {
9516
+ if (endorsementGroupIds.has(parentId ?? "")) continue;
9517
+ const endorsementChildren = children.filter((child) => child.kind === "endorsement" && !isEndorsementGroup(child));
9518
+ if (endorsementChildren.length < 2) continue;
9519
+ const documentId = endorsementChildren[0].documentId;
9520
+ const pageStarts = endorsementChildren.map((child) => child.pageStart).filter((page) => typeof page === "number");
9521
+ const pageEnds = endorsementChildren.map((child) => child.pageEnd ?? child.pageStart).filter((page) => typeof page === "number");
9522
+ const order = Math.min(...endorsementChildren.map((child) => child.order));
9523
+ const existingGroup = groupsByParent.get(parentId);
9524
+ const groupId = existingGroup?.id ?? endorsementGroupNodeId(documentId, parentId);
9525
+ const groupNode = existingGroup ?? {
9526
+ id: groupId,
9527
+ documentId,
9528
+ parentId,
9529
+ kind: "page_group",
9530
+ title: "Endorsements",
9531
+ description: "Endorsement forms grouped by source order",
9532
+ textExcerpt: void 0,
9533
+ sourceSpanIds: [],
9534
+ pageStart: pageStarts.length ? Math.min(...pageStarts) : void 0,
9535
+ pageEnd: pageEnds.length ? Math.max(...pageEnds) : void 0,
9536
+ bbox: endorsementChildren.flatMap((child) => child.bbox ?? []).slice(0, 12),
9537
+ order,
9538
+ path: "",
9539
+ metadata: { sourceTreeVersion: "v3", organizer: "endorsement_grouping" }
9540
+ };
9541
+ const childSpanIds = [...new Set(endorsementChildren.flatMap((child) => child.sourceSpanIds))];
9542
+ const normalizedGroup = {
9543
+ ...groupNode,
9544
+ sourceSpanIds: groupNode.sourceSpanIds.length ? groupNode.sourceSpanIds : childSpanIds,
9545
+ pageStart: groupNode.pageStart ?? (pageStarts.length ? Math.min(...pageStarts) : void 0),
9546
+ pageEnd: groupNode.pageEnd ?? (pageEnds.length ? Math.max(...pageEnds) : void 0),
9547
+ order
9548
+ };
9549
+ groupsByParent.set(parentId, normalizedGroup);
9550
+ if (!existingGroup) nextTree.push(normalizedGroup);
9551
+ else nextTree = nextTree.map((node) => node.id === normalizedGroup.id ? normalizedGroup : node);
9552
+ nextTree = nextTree.map(
9553
+ (node) => endorsementChildren.some((child) => child.id === node.id) ? { ...node, parentId: groupId, order: node.order + 1e-3 } : node
9554
+ );
9555
+ }
9556
+ return normalizeDocumentSourceTreePaths(nextTree);
9557
+ }
9414
9558
  function compactNode(node, maxText = 700) {
9415
9559
  return {
9416
9560
  id: node.id,
@@ -9498,8 +9642,11 @@ Scope:
9498
9642
  Rules:
9499
9643
  - Use only node IDs from the provided list.
9500
9644
  - Do not invent text, page numbers, source spans, limits, or policy facts.
9501
- - You may relabel existing nodes and group adjacent top-level/page nodes from this batch when they are clearly one form, endorsement, declarations set, schedule, or clause family.
9645
+ - You may relabel existing nodes and group adjacent top-level/page nodes from this batch only when they are clearly one continuous form, one declarations set, one schedule, or one clause family.
9646
+ - Group adjacent separately numbered endorsements under a single generic "Endorsements" page_group parent, with each individual endorsement preserved as its own child node.
9647
+ - Never create rollup titles such as "Endorsements 1-3 (...)" or merge multiple endorsements into one endorsement node.
9502
9648
  - Add concise, human-readable titles to generic text, table, row, and cell nodes when the text makes their role clear.
9649
+ - Keep organizer titles terse. Use the printed heading or a compact canonical title such as "Declarations", "Policy Form", "Definitions", or "Endorsement No. 3"; do not add parenthetical summaries.
9503
9650
  - Groups must list existing childNodeIds only.
9504
9651
  - Keep descriptions short and useful for search.
9505
9652
  - Prefer the document's own form titles, endorsement titles, schedules, declarations headings, and page order over keyword-only guessing.
@@ -9549,17 +9696,20 @@ function applyOrganization(sourceTree, organization) {
9549
9696
  return {
9550
9697
  ...node,
9551
9698
  kind: label.kind ?? node.kind,
9552
- title: cleanText(label.title, node.title),
9699
+ title: simplifyOrganizerTitle(label.title, node.title, label.kind ?? node.kind),
9553
9700
  description: cleanText(label.description, node.description)
9554
9701
  };
9555
9702
  });
9556
9703
  for (const group of organization.groups) {
9557
9704
  const children = group.childNodeIds.map((id2) => byId.get(id2)).filter((node2) => Boolean(node2));
9558
9705
  if (children.length === 0) continue;
9706
+ if (rejectsOrganizerGroup(group, children)) continue;
9559
9707
  const parentId = children[0].parentId;
9560
9708
  if (!children.every((child) => child.parentId === parentId)) continue;
9561
9709
  const documentId = children[0].documentId;
9562
- const id = groupNodeId(documentId, group);
9710
+ const title = simplifyOrganizerTitle(group.title, group.title, group.kind);
9711
+ const description = cleanText(group.description, title);
9712
+ const id = groupNodeId(documentId, { ...group, title });
9563
9713
  if (byId.has(id)) continue;
9564
9714
  const sourceSpanIds = [...new Set(children.flatMap((child) => child.sourceSpanIds))];
9565
9715
  const pageStarts = children.map((child) => child.pageStart).filter((page) => typeof page === "number");
@@ -9570,8 +9720,8 @@ function applyOrganization(sourceTree, organization) {
9570
9720
  documentId,
9571
9721
  parentId,
9572
9722
  kind: group.kind,
9573
- title: group.title,
9574
- description: group.description ?? group.title,
9723
+ title,
9724
+ description,
9575
9725
  textExcerpt: children.map((child) => child.textExcerpt ?? child.description).filter(Boolean).join("\n\n").slice(0, 1600),
9576
9726
  sourceSpanIds,
9577
9727
  pageStart: pageStarts.length ? Math.min(...pageStarts) : void 0,
@@ -9589,7 +9739,7 @@ function applyOrganization(sourceTree, organization) {
9589
9739
  ];
9590
9740
  byId.set(id, node);
9591
9741
  }
9592
- return normalizeDocumentSourceTreePaths(nextTree);
9742
+ return applyEndorsementGrouping(normalizeDocumentSourceTreePaths(nextTree));
9593
9743
  }
9594
9744
  function sourceTreeToOutline(sourceTree) {
9595
9745
  const byParent = /* @__PURE__ */ new Map();
@@ -9808,6 +9958,7 @@ async function runSourceTreeExtraction(params) {
9808
9958
  } catch (error) {
9809
9959
  warnings.push(`Operational profile model pass failed; deterministic profile used (${error instanceof Error ? error.message : String(error)})`);
9810
9960
  }
9961
+ sourceTree = applyEndorsementGrouping(sourceTree);
9811
9962
  const document = materializeDocument({
9812
9963
  id: params.id,
9813
9964
  sourceTree,