@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.mjs CHANGED
@@ -9064,6 +9064,150 @@ function cleanText(value, fallback) {
9064
9064
  const text = value?.replace(/\s+/g, " ").trim();
9065
9065
  return text || fallback;
9066
9066
  }
9067
+ function simplifyOrganizerTitle(value, fallback, kind) {
9068
+ const title = cleanText(value, fallback);
9069
+ if (/^declarations\b/i.test(title)) return "Declarations";
9070
+ if (/^policy\s+form\b/i.test(title)) return "Policy Form";
9071
+ if (/^definitions\b/i.test(title)) return "Definitions";
9072
+ if (kind === "page_group" && /^endorsements?\b/i.test(title)) return "Endorsements";
9073
+ const endorsementNumber = title.match(/^endorsement\s+(?:no\.?|number|#)?\s*([A-Z0-9][A-Z0-9.-]*)\b/i)?.[1];
9074
+ if (endorsementNumber) return `Endorsement No. ${endorsementNumber}`;
9075
+ if (kind === "endorsement" && /^endorsements?\s+\d+\s*[–-]\s*\d+\b/i.test(title)) {
9076
+ return title.replace(/[–—]/g, "-").replace(/\s*\(.*/, "").trim();
9077
+ }
9078
+ return title;
9079
+ }
9080
+ function endorsementReference(value) {
9081
+ return value?.match(/\bendorsement\s+(?:no\.?|number|#)?\s*([A-Z0-9][A-Z0-9.-]*)\b/i)?.[1]?.toUpperCase();
9082
+ }
9083
+ function endorsementTitle(value) {
9084
+ const text = cleanText(value, "");
9085
+ const number = text.match(/\bendorsement\s+(?:no\.?|number|#)\s*([A-Z0-9][A-Z0-9.-]*)\b/i)?.[1]?.toUpperCase();
9086
+ return number ? `Endorsement No. ${number}` : void 0;
9087
+ }
9088
+ function rejectsOrganizerGroup(group, children) {
9089
+ if (/^endorsements?\b/i.test(group.title) && group.kind !== "page_group") return true;
9090
+ if (group.kind !== "endorsement") return false;
9091
+ if (/^endorsements?\s+\d+\s*[–-]\s*\d+\b/i.test(group.title)) return true;
9092
+ const childNumbers = new Set(
9093
+ children.map((child) => endorsementReference([child.title, child.description, child.textExcerpt].filter(Boolean).join(" "))).filter((value) => Boolean(value))
9094
+ );
9095
+ return childNumbers.size > 1;
9096
+ }
9097
+ function isEndorsementGroup(node) {
9098
+ return node.kind === "page_group" && /^endorsements?\b/i.test(node.title);
9099
+ }
9100
+ function endorsementGroupNodeId(documentId, parentId) {
9101
+ return [
9102
+ documentId.replace(/[^a-zA-Z0-9_.:-]/g, "_"),
9103
+ "source_node",
9104
+ "page_group",
9105
+ "endorsements",
9106
+ parentId?.replace(/[^a-zA-Z0-9_.:-]/g, "_").slice(0, 48) ?? "root"
9107
+ ].join(":");
9108
+ }
9109
+ function applyEndorsementGrouping(sourceTree) {
9110
+ const relabeledTree = sourceTree.map((node) => {
9111
+ if (node.kind === "document" || isEndorsementGroup(node) || node.kind === "endorsement") return node;
9112
+ const title = endorsementTitle([node.title, node.description, node.textExcerpt].filter(Boolean).join(" "));
9113
+ if (!title) return node;
9114
+ return {
9115
+ ...node,
9116
+ kind: "endorsement",
9117
+ title,
9118
+ description: cleanText(
9119
+ [title, "endorsement", node.pageStart ? `page ${node.pageStart}` : void 0, node.textExcerpt].filter(Boolean).join(" | "),
9120
+ title
9121
+ ),
9122
+ metadata: {
9123
+ ...node.metadata,
9124
+ organizerRepair: "normalize_endorsement_grouping"
9125
+ }
9126
+ };
9127
+ });
9128
+ const byParent = nodesByParent(relabeledTree);
9129
+ const groupsByParent = /* @__PURE__ */ new Map();
9130
+ const endorsementGroupIds = new Set(
9131
+ relabeledTree.filter(isEndorsementGroup).map((node) => node.id)
9132
+ );
9133
+ let nextTree = relabeledTree.map((node) => {
9134
+ if (!isEndorsementGroup(node)) return node;
9135
+ const normalized = {
9136
+ ...node,
9137
+ kind: "page_group",
9138
+ title: "Endorsements",
9139
+ description: cleanText(node.description, "Endorsement forms grouped by source order"),
9140
+ metadata: {
9141
+ ...node.metadata,
9142
+ sourceTreeVersion: "v3",
9143
+ organizer: node.metadata?.organizer ?? "endorsement_grouping"
9144
+ }
9145
+ };
9146
+ groupsByParent.set(node.parentId, normalized);
9147
+ endorsementGroupIds.add(node.id);
9148
+ return normalized;
9149
+ });
9150
+ nextTree = nextTree.map((node) => {
9151
+ if (!endorsementGroupIds.has(node.parentId ?? "")) return node;
9152
+ const title = endorsementTitle([node.title, node.description, node.textExcerpt].filter(Boolean).join(" "));
9153
+ if (!title) return node;
9154
+ return {
9155
+ ...node,
9156
+ kind: "endorsement",
9157
+ title,
9158
+ description: cleanText(
9159
+ [title, "endorsement", node.pageStart ? `page ${node.pageStart}` : void 0, node.textExcerpt].filter(Boolean).join(" | "),
9160
+ title
9161
+ ),
9162
+ metadata: {
9163
+ ...node.metadata,
9164
+ organizerRepair: "normalize_endorsement_grouping"
9165
+ }
9166
+ };
9167
+ });
9168
+ for (const [parentId, children] of byParent) {
9169
+ if (endorsementGroupIds.has(parentId ?? "")) continue;
9170
+ const endorsementChildren = children.filter((child) => child.kind === "endorsement" && !isEndorsementGroup(child));
9171
+ if (endorsementChildren.length < 2) continue;
9172
+ const documentId = endorsementChildren[0].documentId;
9173
+ const pageStarts = endorsementChildren.map((child) => child.pageStart).filter((page) => typeof page === "number");
9174
+ const pageEnds = endorsementChildren.map((child) => child.pageEnd ?? child.pageStart).filter((page) => typeof page === "number");
9175
+ const order = Math.min(...endorsementChildren.map((child) => child.order));
9176
+ const existingGroup = groupsByParent.get(parentId);
9177
+ const groupId = existingGroup?.id ?? endorsementGroupNodeId(documentId, parentId);
9178
+ const groupNode = existingGroup ?? {
9179
+ id: groupId,
9180
+ documentId,
9181
+ parentId,
9182
+ kind: "page_group",
9183
+ title: "Endorsements",
9184
+ description: "Endorsement forms grouped by source order",
9185
+ textExcerpt: void 0,
9186
+ sourceSpanIds: [],
9187
+ pageStart: pageStarts.length ? Math.min(...pageStarts) : void 0,
9188
+ pageEnd: pageEnds.length ? Math.max(...pageEnds) : void 0,
9189
+ bbox: endorsementChildren.flatMap((child) => child.bbox ?? []).slice(0, 12),
9190
+ order,
9191
+ path: "",
9192
+ metadata: { sourceTreeVersion: "v3", organizer: "endorsement_grouping" }
9193
+ };
9194
+ const childSpanIds = [...new Set(endorsementChildren.flatMap((child) => child.sourceSpanIds))];
9195
+ const normalizedGroup = {
9196
+ ...groupNode,
9197
+ sourceSpanIds: groupNode.sourceSpanIds.length ? groupNode.sourceSpanIds : childSpanIds,
9198
+ pageStart: groupNode.pageStart ?? (pageStarts.length ? Math.min(...pageStarts) : void 0),
9199
+ pageEnd: groupNode.pageEnd ?? (pageEnds.length ? Math.max(...pageEnds) : void 0),
9200
+ order
9201
+ };
9202
+ groupsByParent.set(parentId, normalizedGroup);
9203
+ if (!existingGroup) nextTree.push(normalizedGroup);
9204
+ else nextTree = nextTree.map((node) => node.id === normalizedGroup.id ? normalizedGroup : node);
9205
+ nextTree = nextTree.map(
9206
+ (node) => endorsementChildren.some((child) => child.id === node.id) ? { ...node, parentId: groupId, order: node.order + 1e-3 } : node
9207
+ );
9208
+ }
9209
+ return normalizeDocumentSourceTreePaths(nextTree);
9210
+ }
9067
9211
  function compactNode(node, maxText = 700) {
9068
9212
  return {
9069
9213
  id: node.id,
@@ -9151,8 +9295,11 @@ Scope:
9151
9295
  Rules:
9152
9296
  - Use only node IDs from the provided list.
9153
9297
  - Do not invent text, page numbers, source spans, limits, or policy facts.
9154
- - 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.
9298
+ - 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.
9299
+ - Group adjacent separately numbered endorsements under a single generic "Endorsements" page_group parent, with each individual endorsement preserved as its own child node.
9300
+ - Never create rollup titles such as "Endorsements 1-3 (...)" or merge multiple endorsements into one endorsement node.
9155
9301
  - Add concise, human-readable titles to generic text, table, row, and cell nodes when the text makes their role clear.
9302
+ - 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.
9156
9303
  - Groups must list existing childNodeIds only.
9157
9304
  - Keep descriptions short and useful for search.
9158
9305
  - Prefer the document's own form titles, endorsement titles, schedules, declarations headings, and page order over keyword-only guessing.
@@ -9202,17 +9349,20 @@ function applyOrganization(sourceTree, organization) {
9202
9349
  return {
9203
9350
  ...node,
9204
9351
  kind: label.kind ?? node.kind,
9205
- title: cleanText(label.title, node.title),
9352
+ title: simplifyOrganizerTitle(label.title, node.title, label.kind ?? node.kind),
9206
9353
  description: cleanText(label.description, node.description)
9207
9354
  };
9208
9355
  });
9209
9356
  for (const group of organization.groups) {
9210
9357
  const children = group.childNodeIds.map((id2) => byId.get(id2)).filter((node2) => Boolean(node2));
9211
9358
  if (children.length === 0) continue;
9359
+ if (rejectsOrganizerGroup(group, children)) continue;
9212
9360
  const parentId = children[0].parentId;
9213
9361
  if (!children.every((child) => child.parentId === parentId)) continue;
9214
9362
  const documentId = children[0].documentId;
9215
- const id = groupNodeId(documentId, group);
9363
+ const title = simplifyOrganizerTitle(group.title, group.title, group.kind);
9364
+ const description = cleanText(group.description, title);
9365
+ const id = groupNodeId(documentId, { ...group, title });
9216
9366
  if (byId.has(id)) continue;
9217
9367
  const sourceSpanIds = [...new Set(children.flatMap((child) => child.sourceSpanIds))];
9218
9368
  const pageStarts = children.map((child) => child.pageStart).filter((page) => typeof page === "number");
@@ -9223,8 +9373,8 @@ function applyOrganization(sourceTree, organization) {
9223
9373
  documentId,
9224
9374
  parentId,
9225
9375
  kind: group.kind,
9226
- title: group.title,
9227
- description: group.description ?? group.title,
9376
+ title,
9377
+ description,
9228
9378
  textExcerpt: children.map((child) => child.textExcerpt ?? child.description).filter(Boolean).join("\n\n").slice(0, 1600),
9229
9379
  sourceSpanIds,
9230
9380
  pageStart: pageStarts.length ? Math.min(...pageStarts) : void 0,
@@ -9242,7 +9392,7 @@ function applyOrganization(sourceTree, organization) {
9242
9392
  ];
9243
9393
  byId.set(id, node);
9244
9394
  }
9245
- return normalizeDocumentSourceTreePaths(nextTree);
9395
+ return applyEndorsementGrouping(normalizeDocumentSourceTreePaths(nextTree));
9246
9396
  }
9247
9397
  function sourceTreeToOutline(sourceTree) {
9248
9398
  const byParent = /* @__PURE__ */ new Map();
@@ -9461,6 +9611,7 @@ async function runSourceTreeExtraction(params) {
9461
9611
  } catch (error) {
9462
9612
  warnings.push(`Operational profile model pass failed; deterministic profile used (${error instanceof Error ? error.message : String(error)})`);
9463
9613
  }
9614
+ sourceTree = applyEndorsementGrouping(sourceTree);
9464
9615
  const document = materializeDocument({
9465
9616
  id: params.id,
9466
9617
  sourceTree,