@claritylabs/cl-sdk 3.0.29 → 3.0.31

package/dist/index.js

@@ -3339,6 +3339,88 @@ function firstMatch(nodes, patterns) {
   }
   return void 0;
 }
+var POLICY_NUMBER_PATTERNS = [
+  /\bpolicy\s*(?:number|no\.?|#)\s*:?\s*([A-Z0-9][A-Z0-9,.-]{4,}[A-Z0-9])/i,
+  /\bpolicy\s*[:#]\s*([A-Z0-9][A-Z0-9,.-]{4,}[A-Z0-9])/i
+];
+function policyNumberEvidenceScore(node) {
+  const text = normalizeWhitespace3([node.path, nodeText(node)].filter(Boolean).join(" ")).toLowerCase();
+  let score = 0;
+  if (/\b(policy\s+summary|declarations?|declaration\s+page|schedule)\b/.test(text)) score += 80;
+  if (/\b(plan|policy\s+date|insured\s+person|named\s+insured|insurance\s+amount|benefit\s+amount)\b/.test(text)) score += 35;
+  if (node.kind === "table_row" || node.kind === "table_cell" || node.kind === "text") score += 20;
+  if (node.kind === "page") score += 10;
+  if (typeof node.pageStart === "number" && node.pageStart > 1 && node.pageStart <= 10) score += 20;
+  if (typeof node.pageStart === "number" && node.pageStart === 1) score -= 30;
+  if (/\b(notices?\s+and\s+jacket|policy\s+jacket|front\s+matter|table\s+of\s+contents)\b/.test(text)) score -= 70;
+  if (node.kind === "page_group" || node.kind === "form") score -= 30;
+  return score;
+}
+function policyNumberFromNodes(nodes) {
+  const candidates = nodes.slice(0, 120).flatMap((node) => {
+    const text = nodeText(node);
+    for (const pattern of POLICY_NUMBER_PATTERNS) {
+      const value = cleanValue(text.match(pattern)?.[1]);
+      if (value) return [{ node, value, score: policyNumberEvidenceScore(node) }];
+    }
+    return [];
+  }).sort(
+    (left, right) => right.score - left.score || (left.node.pageStart ?? Number.MAX_SAFE_INTEGER) - (right.node.pageStart ?? Number.MAX_SAFE_INTEGER) || left.node.order - right.node.order
+  );
+  const candidate = candidates[0];
+  return candidate ? valueFromNode(candidate.node, candidate.value, "high") : void 0;
+}
+function compactFactNodes(nodes) {
+  return nodes.filter((node) => {
+    if (node.kind === "document" || node.kind === "page_group" || node.kind === "form") return false;
+    const text = nodeText(node);
+    if (text.length > 900) return false;
+    if (/\b(table of contents|provided solely for your convenience|not to be construed|actual policy issued)\b/i.test(text)) return false;
+    return node.kind === "text" || node.kind === "table_row" || node.kind === "table_cell" || node.kind === "page";
+  });
+}
+function firstCleanMatch(nodes, patterns, clean) {
+  for (const node of compactFactNodes(nodes)) {
+    const text = nodeText(node);
+    for (const pattern of patterns) {
+      const raw = cleanValue(text.match(pattern)?.[1]);
+      if (!raw) continue;
+      const value = clean(raw);
+      if (value) return valueFromNode(node, value, "high");
+    }
+  }
+  return void 0;
+}
+function cleanNamedInsured(value) {
+  const clean = cleanValue(value.replace(/\bborn\s+on\b.*$/i, "").replace(/\bage\s+nearest\b.*$/i, "").replace(/\bbeneficiary\b.*$/i, ""));
+  if (!clean || clean.length > 160) return void 0;
+  if (/^(person|persons)\b/i.test(clean)) return void 0;
+  if (/\b(table of contents|policy wording|provided solely|convenience|not to be construed|actual policy issued)\b/i.test(clean)) {
+    return void 0;
+  }
+  return clean;
+}
+function namedInsuredFromNodes(nodes) {
+  return firstCleanMatch(nodes, [
+    /\b(?:named insured|insured name|insured persons?|insured person)\s*:?\s*(.+?)(?=\s+(?:insurance amount|benefit amount|policy number|policy date|owner|beneficiary|premium|coverage|risk classification|date this)\b|[|;\n]|$)/i,
+    /\b(?:applicant|policyholder)\s*:?\s*(.+?)(?=\s+(?:coverage|policy number|insurer|carrier|premium|effective|expiration)\b|[|;\n]|$)/i
+  ], cleanNamedInsured);
+}
+function cleanInsurer(value) {
+  const clean = cleanValue(value);
+  if (!clean || clean.length > 140) return void 0;
+  if (/^(mean|means|we|us|our)\b/i.test(clean)) return void 0;
+  if (/\b(table of contents|policy wording|provided solely|convenience)\b/i.test(clean)) return void 0;
+  const known = clean.match(/\b(Sun Life Assurance Company of Canada|Manulife|The Manufacturers Life Insurance Company)\b/i)?.[1];
+  return known ?? clean;
+}
+function insurerFromNodes(nodes) {
+  return firstCleanMatch(nodes, [
+    /\bunderwritten by\s+([^|;\n.]{3,160})/i,
+    /\b(?:insurer|carrier|company|security)\s*:?\s*([^|;\n]{3,120})/i,
+    /\b(Sun Life Assurance Company of Canada|Manulife|The Manufacturers Life Insurance Company)\b/i
+  ], cleanInsurer);
+}
 function inferPolicyTypes(nodes) {
   const text = nodes.slice(0, 40).map(nodeText).join(" ").toLowerCase();
   const types = [];
@@ -3792,20 +3874,9 @@ function buildDeterministicOperationalProfile(params) {
   const partial = {
     documentType: inferDocumentType(nodes),
     policyTypes: inferPolicyTypes(nodes),
-    policyNumber: firstMatch(nodes, [
-      /\bpolicy\s*(?:number|no\.?|#)\s*:?\s*([A-Z0-9][A-Z0-9,.-]{4,}[A-Z0-9])/i,
-      /\bpolicy\s*[:#]\s*([A-Z0-9][A-Z0-9,.-]{4,}[A-Z0-9])/i
-    ]),
-    namedInsured: firstMatch(nodes, [
-      /\b(?:named insured|insured name|insured)\s*:?\s*(.+?)(?=\s+(?:coverage|policy number|insurer|carrier|premium|effective|expiration)\b|[|;\n]|$)/i,
-      /\b(?:applicant|policyholder)\s*:?\s*(.+?)(?=\s+(?:coverage|policy number|insurer|carrier|premium|effective|expiration)\b|[|;\n]|$)/i
-    ]),
-    insurer: firstMatch(nodes, [
-      /\b(?:insurer|carrier|company|security)\s*:?\s*([^|;\n]{3,120})/i,
-      /\bunderwritten by\s+([^|;\n]{3,120})/i
-    ]) ?? firstMatch(nodes, [
-      /\b(Manulife)\b/i
-    ]),
+    policyNumber: policyNumberFromNodes(nodes),
+    namedInsured: namedInsuredFromNodes(nodes),
+    insurer: insurerFromNodes(nodes),
     broker: firstMatch(nodes, [
       /\b(?:broker|producer|agent)\s*:?\s*([^|;\n]{3,120})/i
     ]),
@@ -11105,6 +11176,7 @@ Rules:
 - Every returned value must include sourceNodeIds or sourceSpanIds from the provided nodes.
 - If a value is not directly supported, omit it.
 - Prefer declarations, schedules, premium tables, and endorsement schedules over generic policy wording.
+- For life, critical illness, disability, and long-term care policies, keep named benefit units and benefit subconditions as operational facts even when they do not have dollar limits. Examples include death benefit, disability benefit, total disability, catastrophic disability, return of premium, waiver, and conversion options. Put subcondition details in coverages[].limits with kind "other" when they belong under a broader benefit.
 - Treat an endorsement as one coverage unit when it contains a schedule. Do not split an endorsement schedule into generic rows like "Aggregate Limit".
 - For coverage schedules, put each claim, aggregate, sublimit, retention, deductible, and retroactive date values in coverages[].limits with labels and source IDs. Keep the legacy coverages[].limit as the primary display value only.
 - Use coverageOrigin: "endorsement" for endorsement units and "core" for declarations/core policy coverage units.