@claritylabs/cl-sdk 3.1.7 → 3.1.9

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
@@ -2986,878 +2986,6 @@ function buildDocumentSourceTree(sourceSpans, documentId) {
2986
2986
  return normalizeDocumentSourceTreePaths(groupPageContentByTitles([...nodes.values()]));
2987
2987
  }
2988
2988
 
2989
- // src/source/operational-profile.ts
2990
- function normalizeWhitespace3(value) {
2991
- return value.replace(/\s+/g, " ").trim();
2992
- }
2993
- function cleanValue(value) {
2994
- if (!value) return void 0;
2995
- return normalizeWhitespace3(value.replace(/^[\s:;#-]+|[\s;,.]+$/g, ""));
2996
- }
2997
- function cleanCoverageLabel(value) {
2998
- return cleanValue(value)?.replace(/^column\s+\d+\s*:\s*/i, "");
2999
- }
3000
- function moneyValue(value) {
3001
- const clean = cleanValue(value);
3002
- if (!clean) return void 0;
3003
- if (/^\$/.test(clean)) return clean;
3004
- if (/^\d{1,3}(?:,\d{3})*(?:\.\d{2})?$/.test(clean)) return `$${clean}`;
3005
- return clean;
3006
- }
3007
- function premiumValue(value) {
3008
- const clean = cleanValue(value);
3009
- if (!clean) return void 0;
3010
- if (/\$[A-Z0-9]/i.test(clean)) return clean;
3011
- if (/\b(?:CAD|USD)\b/i.test(clean) && /(?:\d|X{2,})/i.test(clean)) return clean;
3012
- if (/^\d{1,3}(?:,\d{3})+\.\d{2}$/.test(clean) || /^\d+\.\d{2}$/.test(clean)) return `$${clean}`;
3013
- if (/^X{2,}(?:,X{3})*(?:\.X{2})$/i.test(clean)) return clean;
3014
- return void 0;
3015
- }
3016
- function nodeText(node) {
3017
- return normalizeWhitespace3([
3018
- node.title,
3019
- node.description,
3020
- node.textExcerpt
3021
- ].filter(Boolean).join(" "));
3022
- }
3023
- function valueFromNode(node, value, confidence = "medium") {
3024
- return {
3025
- value,
3026
- confidence,
3027
- sourceNodeIds: [node.id],
3028
- sourceSpanIds: node.sourceSpanIds
3029
- };
3030
- }
3031
- function valueFromNodes(nodes, value, confidence = "medium", normalizedValue) {
3032
- return {
3033
- value,
3034
- ...normalizedValue ? { normalizedValue } : {},
3035
- confidence,
3036
- ...sourceIds(nodes)
3037
- };
3038
- }
3039
- function firstMatch(nodes, patterns) {
3040
- for (const node of nodes) {
3041
- const text = nodeText(node);
3042
- for (const pattern of patterns) {
3043
- const match = text.match(pattern);
3044
- const value = cleanValue(match?.[1]);
3045
- if (value) return valueFromNode(node, value, "high");
3046
- }
3047
- }
3048
- return void 0;
3049
- }
3050
- var POLICY_NUMBER_PATTERNS = [
3051
- /\bpolicy\s*(?:number|no\.?|#)\s*:?\s*([A-Z0-9][A-Z0-9,.-]{4,}[A-Z0-9])/i,
3052
- /\bpolicy\s*[:#]\s*([A-Z0-9][A-Z0-9,.-]{4,}[A-Z0-9])/i
3053
- ];
3054
- function policyNumberEvidenceScore(node) {
3055
- const text = normalizeWhitespace3([node.path, nodeText(node)].filter(Boolean).join(" ")).toLowerCase();
3056
- let score = 0;
3057
- if (/\b(policy\s+summary|declarations?|declaration\s+page|schedule)\b/.test(text)) score += 80;
3058
- if (/\b(plan|policy\s+date|insured\s+person|named\s+insured|insurance\s+amount|benefit\s+amount)\b/.test(text)) score += 35;
3059
- if (node.kind === "table_row" || node.kind === "table_cell" || node.kind === "text") score += 20;
3060
- if (node.kind === "page") score += 10;
3061
- if (typeof node.pageStart === "number" && node.pageStart > 1 && node.pageStart <= 10) score += 20;
3062
- if (typeof node.pageStart === "number" && node.pageStart === 1) score -= 30;
3063
- if (/\b(notices?\s+and\s+jacket|policy\s+jacket|front\s+matter|table\s+of\s+contents)\b/.test(text)) score -= 70;
3064
- if (node.kind === "page_group" || node.kind === "form") score -= 30;
3065
- return score;
3066
- }
3067
- function policyNumberFromNodes(nodes) {
3068
- const candidates = nodes.slice(0, 120).flatMap((node) => {
3069
- const text = nodeText(node);
3070
- for (const pattern of POLICY_NUMBER_PATTERNS) {
3071
- const value = cleanValue(text.match(pattern)?.[1]);
3072
- if (value) return [{ node, value, score: policyNumberEvidenceScore(node) }];
3073
- }
3074
- return [];
3075
- }).sort(
3076
- (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
3077
- );
3078
- const candidate = candidates[0];
3079
- return candidate ? valueFromNode(candidate.node, candidate.value, "high") : void 0;
3080
- }
3081
- function compactFactNodes(nodes) {
3082
- return nodes.filter((node) => {
3083
- if (node.kind === "document" || node.kind === "page_group" || node.kind === "form") return false;
3084
- const text = nodeText(node);
3085
- if (text.length > 900) return false;
3086
- if (/\b(table of contents|provided solely for your convenience|not to be construed|actual policy issued)\b/i.test(text)) return false;
3087
- return node.kind === "text" || node.kind === "table_row" || node.kind === "table_cell" || node.kind === "page";
3088
- });
3089
- }
3090
- function firstCleanMatch(nodes, patterns, clean) {
3091
- for (const node of compactFactNodes(nodes)) {
3092
- const text = nodeText(node);
3093
- for (const pattern of patterns) {
3094
- const raw = cleanValue(text.match(pattern)?.[1]);
3095
- if (!raw) continue;
3096
- const value = clean(raw);
3097
- if (value) return valueFromNode(node, value, "high");
3098
- }
3099
- }
3100
- return void 0;
3101
- }
3102
- function cleanNamedInsured(value) {
3103
- const clean = cleanValue(value.replace(/\bborn\s+on\b.*$/i, "").replace(/\bage\s+nearest\b.*$/i, "").replace(/\bbeneficiary\b.*$/i, ""));
3104
- if (!clean || clean.length > 160) return void 0;
3105
- if (!/[A-Za-z0-9]/.test(clean)) return void 0;
3106
- if (/^(person|persons)\b/i.test(clean)) return void 0;
3107
- if (/^(insurance amount|benefit amount|policy number|policy date|owner|beneficiary|premium|coverage|risk classification)\b/i.test(clean)) {
3108
- return void 0;
3109
- }
3110
- if (/\b(table of contents|policy wording|provided solely|convenience|not to be construed|actual policy issued)\b/i.test(clean)) {
3111
- return void 0;
3112
- }
3113
- return clean;
3114
- }
3115
- var PARTY_LABEL_PATTERNS = {
3116
- namedInsured: /^(?:item\s*\d+[.)]?\s*)?(?:named insured(?:\s+and\s+address)?|insured name)$/i,
3117
- insurer: /^(?:carrier|insurer|security)$/i,
3118
- broker: /^(?:broker(?:\s+of\s+record)?|producer|agent)$/i
3119
- };
3120
- function partyLabelKind(value) {
3121
- const clean = cleanValue(value.replace(/^\s*column\s+\d+\s*:\s*/i, ""));
3122
- if (!clean) return void 0;
3123
- if (PARTY_LABEL_PATTERNS.namedInsured.test(clean)) return "namedInsured";
3124
- if (PARTY_LABEL_PATTERNS.insurer.test(clean)) return "insurer";
3125
- if (PARTY_LABEL_PATTERNS.broker.test(clean)) return "broker";
3126
- return void 0;
3127
- }
3128
- function isRejectedPartyValue(value) {
3129
- const clean = normalizeWhitespace3(value);
3130
- return /^(?:and address|of record|insurer|carrier|security|broker|producer|agent|the|a|an|is|are|was|were|agrees?|means?|includes?|shall|will)\b/i.test(clean);
3131
- }
3132
- function identityWithoutAddress(value) {
3133
- const clean = cleanValue(value.replace(/\b(?:phone|tel|telephone|email|e-mail)\b.*$/i, "").replace(/\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,}\b.*$/i, ""));
3134
- if (!clean) return void 0;
3135
- const beforeStreet = clean.match(/^(.+?)\s+\d{1,6}\s+[A-Za-z0-9.'-]+(?:\s+[A-Za-z0-9.'-]+){0,5}\s+(?:street|st\.?|avenue|ave\.?|road|rd\.?|drive|dr\.?|lane|ln\.?|boulevard|blvd\.?|suite|ste\.?|floor|fl\.?|way|court|ct\.?)\b/i)?.[1];
3136
- const beforeCityState = clean.match(/^(.+?)\s+[A-Z][A-Za-z.'-]+(?:\s+[A-Z][A-Za-z.'-]+)*,\s*[A-Z]{2}\s+\d{5}(?:-\d{4})?\b/)?.[1];
3137
- return cleanValue(beforeStreet ?? beforeCityState ?? clean);
3138
- }
3139
- function cleanPartyIdentity(value, clean) {
3140
- const raw = cleanValue(value);
3141
- if (!raw || isRejectedPartyValue(raw)) return void 0;
3142
- const identity = identityWithoutAddress(raw);
3143
- const cleaned = identity ? clean(identity) : void 0;
3144
- if (!cleaned || isRejectedPartyValue(cleaned)) return void 0;
3145
- return {
3146
- value: cleaned,
3147
- ...cleaned !== raw ? { normalizedValue: cleaned } : {}
3148
- };
3149
- }
3150
- function partyFromTableRows(nodes, wanted, clean) {
3151
- const children = childMap(nodes);
3152
- for (const row of compactFactNodes(nodes).filter((node) => node.kind === "table_row")) {
3153
- const cells = cellRows(row, children);
3154
- for (const [index, cell] of cells.entries()) {
3155
- const labelKind = partyLabelKind(cell.value) ?? partyLabelKind(cell.label);
3156
- if (labelKind !== wanted) continue;
3157
- const valueCell = cells.slice(index + 1).find((candidate) => !partyLabelKind(candidate.value) && !partyLabelKind(candidate.label));
3158
- if (!valueCell) continue;
3159
- const cleaned = cleanPartyIdentity(valueCell.value, clean);
3160
- if (cleaned) return valueFromNodes([row, valueCell.node], cleaned.value, "high", cleaned.normalizedValue);
3161
- }
3162
- const parts = normalizeWhitespace3(row.textExcerpt ?? row.description ?? nodeText(row)).split(/\s+\|\s+|\t/).map(cleanValue).filter((part) => Boolean(part));
3163
- for (const [index, part] of parts.entries()) {
3164
- const labelKind = partyLabelKind(part);
3165
- if (labelKind !== wanted) continue;
3166
- const cleaned = cleanPartyIdentity(parts[index + 1] ?? "", clean);
3167
- if (cleaned) return valueFromNodes([row], cleaned.value, "high", cleaned.normalizedValue);
3168
- }
3169
- }
3170
- return void 0;
3171
- }
3172
- function namedInsuredFromNodes(nodes) {
3173
- return partyFromTableRows(nodes, "namedInsured", cleanNamedInsured) ?? firstCleanMatch(nodes, [
3174
- /\b(?:named insured|insured name)\s*:?\s*(.+?)(?=\s+(?:insurance amount|benefit amount|policy number|policy date|owner|beneficiary|premium|coverage|risk classification|date this)\b|[|;\n]|$)/i,
3175
- /\b(?: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]|$)/i,
3176
- /\b(?:applicant|policyholder)\s*:?\s*(.+?)(?=\s+(?:coverage|policy number|insurer|carrier|premium|effective|expiration)\b|[|;\n]|$)/i
3177
- ], cleanNamedInsured);
3178
- }
3179
- function cleanInsurer(value) {
3180
- const clean = cleanValue(value);
3181
- if (!clean || clean.length > 140) return void 0;
3182
- if (/^(mean|means|we|us|our|the|a|an|agrees?|shall|will)\b/i.test(clean)) return void 0;
3183
- if (/\b(table of contents|policy wording|provided solely|convenience)\b/i.test(clean)) return void 0;
3184
- const known = clean.match(/\b(Sun Life Assurance Company of Canada|Manulife|The Manufacturers Life Insurance Company)\b/i)?.[1];
3185
- return known ?? clean;
3186
- }
3187
- function insurerFromNodes(nodes) {
3188
- const tableValue = partyFromTableRows(nodes, "insurer", cleanInsurer);
3189
- if (tableValue) return tableValue;
3190
- for (const node of compactFactNodes(nodes)) {
3191
- const text = nodeText(node);
3192
- const value = cleanInsurer(text.match(/\b([A-Z][A-Za-z&.,' -]{2,120}?(?:Insurance|Assurance|Indemnity|Casualty|Underwriting|Mutual|Risk|Reinsurance)\s+Company(?:\s+of\s+[A-Z][A-Za-z .'-]+)?)\s*\(\s*the\s+["']?Insurer["']?\s*\)/i)?.[1] ?? "");
3193
- if (value) return valueFromNode(node, value, "high");
3194
- }
3195
- return firstCleanMatch(nodes, [
3196
- /\bunderwritten by\s+([^|;\n.]{3,160})/i,
3197
- /\b(?:insurer|carrier|company|security)\s*:?\s*([^|;\n]{3,120})/i,
3198
- /\b(Sun Life Assurance Company of Canada|Manulife|The Manufacturers Life Insurance Company)\b/i
3199
- ], cleanInsurer);
3200
- }
3201
- function inferPolicyTypes(nodes) {
3202
- const text = nodes.slice(0, 40).map(nodeText).join(" ").toLowerCase();
3203
- const types = [];
3204
- const add = (pattern, type) => {
3205
- if (pattern.test(text) && !types.includes(type)) types.push(type);
3206
- };
3207
- add(/\b(life insurance|permanent life|term life|whole life|universal life|sun permanent life|sun par protector|manulife par|vitality\s*plus|death benefit)\b/i, "life");
3208
- add(/\b(critical illness|critical illness insurance|covered critical illness|partial benefit payout)\b/i, "critical_illness");
3209
- add(/\b(disability benefit|total disability|catastrophic disability|disability waiver|waiver of premium disability)\b/i, "disability");
3210
- add(/\b(long[-\s]?term care|long term care conversion)\b/i, "long_term_care");
3211
- add(/\b(cyber|network security|privacy liability|data breach)\b/i, "cyber");
3212
- add(/\b(professional liability|errors?\s*&?\s*omissions|e&o)\b/i, "professional_liability");
3213
- add(/\b(commercial general liability|general liability|cgl)\b/i, "general_liability");
3214
- add(/\b(umbrella|excess liability)\b/i, "umbrella");
3215
- add(/\b(workers'? compensation|employers'? liability)\b/i, "workers_comp");
3216
- add(/\b(commercial auto|business auto|automobile liability)\b/i, "commercial_auto");
3217
- add(/\b(commercial property|property coverage|building coverage)\b/i, "commercial_property");
3218
- return types.length ? types : ["other"];
3219
- }
3220
- function inferDocumentType(nodes) {
3221
- const text = nodes.slice(0, 25).map(nodeText).join(" ").toLowerCase();
3222
- if (/\b(quote|proposal|quotation|indication)\b/.test(text) && !/\bpolicy number\b/.test(text)) {
3223
- return "quote";
3224
- }
3225
- return "policy";
3226
- }
3227
- function coverageNameFromRow(text) {
3228
- const labelled = text.match(/\b(coverage part|coverage|line)\s*:?\s*([^|;$]{3,80})/i);
3229
- if (labelled) {
3230
- const value = cleanCoverageLabel(labelled[2]);
3231
- if (!value) return void 0;
3232
- return /^coverage part$/i.test(labelled[1]) ? `Coverage Part ${value}` : value;
3233
- }
3234
- const parts = text.split(/\s+\|\s+| {2,}|\t/).map(cleanCoverageLabel).filter(Boolean);
3235
- const first = parts.find(
3236
- (part) => !/^(limit|limits?|deductible|premium|amount|basis|rate|retroactive|aggregate|each occurrence)$/i.test(part) && !/^\$?[\d,]+/.test(part)
3237
- );
3238
- return cleanCoverageLabel(first);
3239
- }
3240
- function isDeclarationLimitLabel(value) {
3241
- const label = normalizeLabel(value);
3242
- return /\bitem\s*\d+[.)]?\s*limits?\s+of\s+liability\b/.test(label) || /^limits?\s+of\s+liability$/.test(label) || /^policy\s+limits?$/.test(label);
3243
- }
3244
- function declarationCoverageNameFromRow(row, children) {
3245
- const cells = cellRows(row, children);
3246
- const candidates = [
3247
- row.title,
3248
- ...cells.flatMap((cell) => [cell.label, cell.value])
3249
- ];
3250
- if (!candidates.some(isDeclarationLimitLabel)) return void 0;
3251
- return candidates.some((candidate) => /\blimits?\s+of\s+liability\b/i.test(candidate ?? "")) ? "Limits of Liability" : "Policy Limits";
3252
- }
3253
- function limitFromText(text) {
3254
- return moneyValue(
3255
- text.match(/\b(?:limit|liability|aggregate|occurrence|claim)\s*:?\s*(\$?\d[\d,]*(?:\.\d{2})?|\$?\d[\d,]*\s*(?:each|per|aggregate)[^|;]*)/i)?.[1] ?? text.match(/(\$\s?\d[\d,]*(?:\.\d{2})?)(?=.*\b(limit|aggregate|occurrence|claim|liability)\b)/i)?.[1]
3256
- );
3257
- }
3258
- function deductibleFromText(text) {
3259
- return moneyValue(text.match(/\b(?:deductible|retention|sir)\s*:?\s*(\$?\d[\d,]*(?:\.\d{2})?)/i)?.[1]);
3260
- }
3261
- function premiumFromText(text) {
3262
- return premiumValue(text.match(/\b(?:premium|total premium|total cost|amount due)\b(?:\s*(?:is|:|-))?\s*(\$?(?:\d[\d,]*(?:\.\d{2})?|X{2,}(?:,X{3})*(?:\.X{2})?))/i)?.[1]);
3263
- }
3264
- function premiumFromNodes(nodes) {
3265
- for (const node of compactFactNodes(nodes)) {
3266
- const value = premiumFromText(nodeText(node));
3267
- if (value) return valueFromNode(node, value, "high");
3268
- }
3269
- return void 0;
3270
- }
3271
- function moneyAmount(value) {
3272
- const clean = cleanValue(value);
3273
- if (!clean) return void 0;
3274
- const currency = clean.match(/\$\s*([0-9][0-9,]*(?:\.\d+)?)/);
3275
- const percent = clean.match(/\b([0-9][0-9,]*(?:\.\d+)?)\s*%/);
3276
- const explicitNumeric = !/\bitem\s*\d+\b/i.test(clean) && /\b(?:limit|aggregate|claim|occurrence|loss|retention|deductible|sir|premium|amount)\b/i.test(clean) ? clean.match(/\b([0-9][0-9,]*(?:\.\d+)?)\b/) : void 0;
3277
- const match = currency ?? percent ?? explicitNumeric;
3278
- if (!match) return void 0;
3279
- const amount = Number(match[1].replace(/,/g, ""));
3280
- return Number.isFinite(amount) ? amount : void 0;
3281
- }
3282
- function normalizeLabel(value) {
3283
- return normalizeWhitespace3(value ?? "").toLowerCase();
3284
- }
3285
- function isGenericColumnLabel(value) {
3286
- return /^column\s+\d+$/i.test(cleanValue(value) ?? "");
3287
- }
3288
- function isHeaderRow(row) {
3289
- return row.metadata?.isHeader === true || row.metadata?.isHeader === "true";
3290
- }
3291
- var OPERATIONAL_COVERAGE_TERM_KINDS = /* @__PURE__ */ new Set([
3292
- "each_claim_limit",
3293
- "each_occurrence_limit",
3294
- "each_loss_limit",
3295
- "aggregate_limit",
3296
- "sublimit",
3297
- "retention",
3298
- "deductible",
3299
- "retroactive_date",
3300
- "premium",
3301
- "other"
3302
- ]);
3303
- function termKind(label, value) {
3304
- const text = normalizeLabel(`${label} ${value}`);
3305
- if (/\bretroactive\b/.test(text)) return "retroactive_date";
3306
- if (/\b(self[-\s]?insured retention|retention|sir)\b/.test(text)) return "retention";
3307
- if (/\bdeductible\b/.test(text)) return "deductible";
3308
- if (/\bpremium\b/.test(text)) return "premium";
3309
- if (/\bsub[-\s]?limit\b/.test(text)) return "sublimit";
3310
- if (/\baggregate\b/.test(text)) return "aggregate_limit";
3311
- if (/\beach\s+proceeding|per\s+proceeding\b/.test(text)) return "sublimit";
3312
- if (/\beach\s+claim|per\s+claim\b/.test(text)) return "each_claim_limit";
3313
- if (/\beach\s+occurrence|per\s+occurrence\b/.test(text)) return "each_occurrence_limit";
3314
- if (/\beach\s+loss|per\s+loss\b/.test(text)) return "each_loss_limit";
3315
- if (/\blimit\b/.test(text)) return "other";
3316
- return "other";
3317
- }
3318
- function normalizeTermKind(value, label, termValue) {
3319
- return typeof value === "string" && OPERATIONAL_COVERAGE_TERM_KINDS.has(value) ? value : termKind(label, termValue);
3320
- }
3321
- function isValueCell(label, value) {
3322
- const normalizedLabel = normalizeLabel(label);
3323
- const normalizedValue = normalizeLabel(value);
3324
- if (!value || normalizedValue === "\u2014") return false;
3325
- if (/\b(policy\s*(number|no|#)|named insured|insured|carrier|company|broker|producer|agent|coverage|coverage part|description|item|name|form|source)\b/.test(normalizedLabel)) {
3326
- return false;
3327
- }
3328
- return /\b(limit|aggregate|retention|deductible|sir|retroactive|premium|amount)\b/.test(normalizedLabel) || moneyAmount(value) !== void 0 || /\b(full prior acts|none|included|not included|as stated)\b/.test(normalizedValue);
3329
- }
3330
- function isCoverageTermLabel(value) {
3331
- const label = normalizeLabel(value);
3332
- return /\b(limit|aggregate|retention|deductible|sir|retroactive|premium|amount|sub[-\s]?limit)\b/.test(label);
3333
- }
3334
- function isNameCell(label, value) {
3335
- const normalizedLabel = normalizeLabel(label);
3336
- const normalizedValue = normalizeLabel(value);
3337
- if (!value || moneyAmount(value) !== void 0) return false;
3338
- if (/^item\s+\d+[.)]?\s*limits?\s+of\s+liability\b/.test(normalizedValue)) return false;
3339
- if (/\b(coverage|coverage part|insuring agreement|description|item|name)\b/.test(normalizedLabel)) {
3340
- return true;
3341
- }
3342
- if (/\b(sub[-\s]?limit|aggregate(?:\s+policy)?\s+limit|limit\s+of\s+liability)\b/.test(normalizedLabel) && /\b[a-z][a-z0-9&/ -]{2,}\b/.test(normalizedValue) && !/\bcoverage\s+part\s+[a-z]\)?$/.test(normalizedValue)) {
3343
- return true;
3344
- }
3345
- if (/^column\s+1$/.test(normalizedLabel) && !/^(item\s+\d+|nwc-|iso-|cg |il |form\b|page\b)/i.test(normalizedValue)) {
3346
- return true;
3347
- }
3348
- if (/^column\s+\d+$/.test(normalizedLabel) && /\b(coverage|sub[-\s]?limit|liability|expense|part\s+[a-z])\b/.test(normalizedValue)) {
3349
- return true;
3350
- }
3351
- return false;
3352
- }
3353
- function childMap(nodes) {
3354
- const children = /* @__PURE__ */ new Map();
3355
- for (const node of nodes) {
3356
- const group = children.get(node.parentId) ?? [];
3357
- group.push(node);
3358
- children.set(node.parentId, group);
3359
- }
3360
- for (const group of children.values()) group.sort((left, right) => left.order - right.order);
3361
- return children;
3362
- }
3363
- function rawCellRows(row, children) {
3364
- return (children.get(row.id) ?? []).filter((child) => child.kind === "table_cell").map((cell) => ({
3365
- label: cleanValue(cell.title) ?? "Value",
3366
- value: cleanValue(cell.textExcerpt ?? cell.description ?? cell.title) ?? "",
3367
- node: cell
3368
- })).filter((cell) => cell.value);
3369
- }
3370
- function headerLabelsForRow(row, children) {
3371
- if (!row.parentId) return [];
3372
- const labels = [];
3373
- const siblingRows = (children.get(row.parentId) ?? []).filter((candidate) => candidate.kind === "table_row" && candidate.order < row.order).sort((left, right) => left.order - right.order);
3374
- for (const sibling of siblingRows) {
3375
- if (!isHeaderRow(sibling)) continue;
3376
- for (const [index, cell] of rawCellRows(sibling, children).entries()) {
3377
- const label = cleanValue(cell.value) ?? cleanValue(cell.label);
3378
- if (label && !isGenericColumnLabel(label)) labels[index] = label;
3379
- }
3380
- }
3381
- return labels;
3382
- }
3383
- function cellRows(row, children, headerLabels = []) {
3384
- return rawCellRows(row, children).map((cell, index) => ({
3385
- ...cell,
3386
- label: isGenericColumnLabel(cell.label) && headerLabels[index] ? headerLabels[index] : cell.label
3387
- }));
3388
- }
3389
- function directChildren(parent, children) {
3390
- return children.get(parent.id) ?? [];
3391
- }
3392
- function descendants(parent, children) {
3393
- const stack = [...directChildren(parent, children)];
3394
- const result = [];
3395
- while (stack.length > 0) {
3396
- const node = stack.shift();
3397
- result.push(node);
3398
- stack.unshift(...directChildren(node, children));
3399
- }
3400
- return result;
3401
- }
3402
- function ancestry(node, byId) {
3403
- const result = [];
3404
- let current = node.parentId ? byId.get(node.parentId) : void 0;
3405
- while (current) {
3406
- result.push(current);
3407
- current = current.parentId ? byId.get(current.parentId) : void 0;
3408
- }
3409
- return result;
3410
- }
3411
- function endorsementAncestor(node, byId) {
3412
- return ancestry(node, byId).find((ancestor) => ancestor.kind === "endorsement");
3413
- }
3414
- function endorsementNumberFrom(value) {
3415
- return cleanValue(value?.match(/\bendorsement\s+no\.?\s*([0-9A-Z-]+)/i)?.[1]);
3416
- }
3417
- function endorsementNameFrom(node) {
3418
- const candidates = [node.title, node.textExcerpt, node.description, nodeText(node)].filter(Boolean);
3419
- for (const candidate of candidates) {
3420
- const match = candidate.match(/\bendorsement\s+no\.?\s*([0-9A-Z-]+)\s*[—-]\s*(.{3,140}?)(?=\s+This\s+endorsement\b|\.|$)/i);
3421
- if (match) return cleanValue(`Endorsement No. ${match[1]} - ${match[2]}`) ?? node.title;
3422
- }
3423
- return node.title;
3424
- }
3425
- function sourceIds(nodes) {
3426
- return {
3427
- sourceNodeIds: [...new Set(nodes.map((node) => node.id))],
3428
- sourceSpanIds: [...new Set(nodes.flatMap((node) => node.sourceSpanIds))]
3429
- };
3430
- }
3431
- function termFromCell(params) {
3432
- const value = cleanCoverageTermValue(params.value);
3433
- if (!value) return void 0;
3434
- if (isRejectedCoverageTermValue(params.label, value)) return void 0;
3435
- const nodes = [params.row, params.cell].filter((node) => Boolean(node));
3436
- const kind = termKind(params.label, value);
3437
- const amount = moneyAmount(value);
3438
- return {
3439
- kind,
3440
- label: params.label,
3441
- value,
3442
- ...amount !== void 0 && kind !== "retroactive_date" ? { amount } : {},
3443
- ...params.appliesTo ? { appliesTo: params.appliesTo } : {},
3444
- ...sourceIds(nodes)
3445
- };
3446
- }
3447
- function cleanCoverageTermValue(value) {
3448
- return cleanValue(value)?.replace(/\s+\/\s*$/, "").trim();
3449
- }
3450
- function isRejectedCoverageTermValue(label, value) {
3451
- if (/\bshown\s+in\s+item\s*\d+\b/i.test(value) && moneyAmount(value) === void 0) return true;
3452
- if (/\b(does not afford coverage|doesn't afford coverage|no coverage|remains excluded|is excluded|are excluded|shall not cover|will not cover)\b/i.test(value) && moneyAmount(value) === void 0) {
3453
- return true;
3454
- }
3455
- if (/^for:\s*\(\d+\)/i.test(label) && moneyAmount(value) === void 0) return true;
3456
- if (value.length > 80 && /\b(exclusion|excluded|shall not|will not|does not|failure to)\b/i.test(value) && moneyAmount(value) === void 0) return true;
3457
- return false;
3458
- }
3459
- function termsFromRow(row, children) {
3460
- const cells = cellRows(row, children, headerLabelsForRow(row, children));
3461
- if (cells.length > 0) {
3462
- const pairedTerms = [];
3463
- const pairedIndexes = /* @__PURE__ */ new Set();
3464
- for (const [index, cell] of cells.entries()) {
3465
- const next = cells[index + 1];
3466
- if (!next) continue;
3467
- if (!isGenericColumnLabel(cell.label) && isNameCell(cell.label, cell.value)) continue;
3468
- if (!isCoverageTermLabel(cell.value)) continue;
3469
- if (isCoverageTermLabel(next.value) && moneyAmount(next.value) === void 0) continue;
3470
- const term = termFromCell({ row, cell: next.node, label: cell.value, value: next.value });
3471
- if (term) {
3472
- pairedTerms.push(term);
3473
- pairedIndexes.add(index);
3474
- pairedIndexes.add(index + 1);
3475
- }
3476
- }
3477
- const valueTerms = cells.filter((_, index) => !pairedIndexes.has(index)).filter((cell) => isValueCell(cell.label, cell.value)).map((cell) => termFromCell({ row, cell: cell.node, label: cell.label, value: cell.value })).filter((term) => Boolean(term));
3478
- return [...pairedTerms, ...valueTerms];
3479
- }
3480
- const text = normalizeWhitespace3(row.textExcerpt ?? row.description ?? nodeText(row));
3481
- const terms = [];
3482
- for (const part of text.split(/\s+\|\s+/)) {
3483
- const match = part.match(/^([^:]{2,80}):\s*(.+)$/);
3484
- if (!match) continue;
3485
- if (!isValueCell(match[1], match[2])) continue;
3486
- const term = termFromCell({ row, label: cleanValue(match[1]) ?? "Value", value: match[2] });
3487
- if (term) terms.push(term);
3488
- }
3489
- if (terms.length === 0) {
3490
- const limit = limitFromText(text);
3491
- if (limit) {
3492
- const term = termFromCell({ row, label: "Limit", value: limit });
3493
- if (term) terms.push(term);
3494
- }
3495
- const deductible = deductibleFromText(text);
3496
- if (deductible) {
3497
- const term = termFromCell({ row, label: "Deductible", value: deductible });
3498
- if (term) terms.push(term);
3499
- }
3500
- }
3501
- return terms;
3502
- }
3503
- function nameFromRow(row, children) {
3504
- const cells = cellRows(row, children, headerLabelsForRow(row, children));
3505
- const named = cells.find((cell) => isNameCell(cell.label, cell.value));
3506
- if (named) return cleanValue(named.value);
3507
- const declarationName = declarationCoverageNameFromRow(row, children);
3508
- if (declarationName) return declarationName;
3509
- return coverageNameFromRow(row.textExcerpt ?? row.description ?? nodeText(row));
3510
- }
3511
- function legacyLimit(terms) {
3512
- return terms.find(
3513
- (term) => ["each_claim_limit", "each_occurrence_limit", "each_loss_limit", "aggregate_limit", "sublimit", "other"].includes(term.kind)
3514
- )?.value;
3515
- }
3516
- function legacyDeductible(terms) {
3517
- return terms.find((term) => term.kind === "deductible" || term.kind === "retention")?.value;
3518
- }
3519
- function legacyPremium(terms) {
3520
- return terms.find((term) => term.kind === "premium")?.value;
3521
- }
3522
- function retroDate(terms) {
3523
- return terms.find((term) => term.kind === "retroactive_date")?.value;
3524
- }
3525
- function hasCoverageLimitTerm(terms) {
3526
- return terms.some(
3527
- (term) => ["each_claim_limit", "each_occurrence_limit", "each_loss_limit", "aggregate_limit", "sublimit"].includes(term.kind) || term.kind === "other" && /\blimit\b/i.test(term.label) && moneyAmount(term.value) !== void 0
3528
- );
3529
- }
3530
- function isOperationalCoverageRow(coverage) {
3531
- const name = normalizeLabel(coverage.name);
3532
- if (!hasCoverageLimitTerm(coverage.limits)) return false;
3533
- if (/^(item\s+\d+|option:|annual policy premium|total premium|premium and payment)\b/i.test(coverage.name)) return false;
3534
- if (/^coverage part\s+s\b/i.test(coverage.name)) return false;
3535
- if (/^(nwc|iso|cg|il|acord)[-\s]?[a-z0-9]/i.test(coverage.name)) return false;
3536
- if (/\b(forms?|endorsements?|premium|payment|terrorism risk insurance act|tria|erp option|bilateral discovery)\b/i.test(name)) return false;
3537
- if (/\b(incurred in excess|shall erode|subject to|combined defense expenses)\b/i.test(name)) return false;
3538
- if (coverage.name.split(/\s+/).length > 16 && !/\b(coverage|liability|sub[-\s]?limit|aggregate|each\s+(claim|loss|occurrence)|limit)\b/i.test(coverage.name)) {
3539
- return false;
3540
- }
3541
- return /\b(coverage|coverage part|liability|sub[-\s]?limit|aggregate|each\s+(claim|loss|occurrence|proceeding)|bricking|cyber|privacy|media|regulatory|defense|fraud|social engineering|ai\/ml|errors?\s*&?\s*omissions|technology)\b/i.test(coverage.name);
3542
- }
3543
- function uniqueTerms(terms) {
3544
- const seen = /* @__PURE__ */ new Set();
3545
- const result = [];
3546
- for (const term of terms) {
3547
- const key = [term.kind, term.label.toLowerCase(), term.value.toLowerCase(), term.appliesTo ?? ""].join("|");
3548
- if (seen.has(key)) continue;
3549
- seen.add(key);
3550
- result.push(term);
3551
- }
3552
- return result;
3553
- }
3554
- function coverageFromTableRow(row, children, byId) {
3555
- if (isHeaderRow(row)) return void 0;
3556
- const name = nameFromRow(row, children);
3557
- const terms = uniqueTerms(termsFromRow(row, children));
3558
- if (!name || terms.length === 0) return void 0;
3559
- const ids = sourceIds([row, ...directChildren(row, children)]);
3560
- const endorsement = endorsementAncestor(row, byId);
3561
- return {
3562
- name,
3563
- limit: legacyLimit(terms),
3564
- deductible: legacyDeductible(terms),
3565
- premium: legacyPremium(terms),
3566
- retroactiveDate: retroDate(terms),
3567
- formNumber: typeof row.metadata?.formNumber === "string" ? row.metadata.formNumber : void 0,
3568
- sectionRef: endorsement?.title,
3569
- coverageOrigin: endorsement ? "endorsement" : "core",
3570
- endorsementNumber: endorsementNumberFrom(endorsement?.title),
3571
- limits: terms,
3572
- ...ids
3573
- };
3574
- }
3575
- function coverageFromEndorsement(endorsement, children) {
3576
- const rows = descendants(endorsement, children).filter((node) => node.kind === "table_row");
3577
- const terms = uniqueTerms(rows.flatMap(
3578
- (row) => termsFromRow(row, children).map((term) => {
3579
- const appliesTo = nameFromRow(row, children);
3580
- return appliesTo && term.label !== appliesTo ? { ...term, appliesTo } : term;
3581
- })
3582
- ));
3583
- if (terms.length === 0) return void 0;
3584
- const ids = sourceIds([endorsement, ...rows, ...rows.flatMap((row) => directChildren(row, children))]);
3585
- const name = endorsementNameFrom(endorsement);
3586
- return {
3587
- name,
3588
- limit: legacyLimit(terms),
3589
- deductible: legacyDeductible(terms),
3590
- premium: legacyPremium(terms),
3591
- retroactiveDate: retroDate(terms),
3592
- formNumber: typeof endorsement.metadata?.formNumber === "string" ? endorsement.metadata.formNumber : void 0,
3593
- sectionRef: name,
3594
- coverageOrigin: "endorsement",
3595
- endorsementNumber: endorsementNumberFrom(name),
3596
- limits: terms,
3597
- ...ids
3598
- };
3599
- }
3600
- function hasDescendantEndorsementWithTableRows(endorsement, children) {
3601
- return descendants(endorsement, children).some(
3602
- (node) => node.kind === "endorsement" && descendants(node, children).some((descendant) => descendant.kind === "table_row")
3603
- );
3604
- }
3605
- function buildCoverages(nodes) {
3606
- const children = childMap(nodes);
3607
- const byId = new Map(nodes.map((node) => [node.id, node]));
3608
- const coverages = [];
3609
- const seen = /* @__PURE__ */ new Set();
3610
- for (const endorsement of nodes.filter((node) => node.kind === "endorsement")) {
3611
- if (hasDescendantEndorsementWithTableRows(endorsement, children)) continue;
3612
- const coverage = coverageFromEndorsement(endorsement, children);
3613
- if (!coverage) continue;
3614
- const key = [coverage.name.toLowerCase(), coverage.sourceNodeIds.join(",")].join("|");
3615
- if (seen.has(key)) continue;
3616
- seen.add(key);
3617
- coverages.push(coverage);
3618
- }
3619
- const rows = nodes.filter((node) => {
3620
- const text = nodeText(node);
3621
- return (node.kind === "table_row" || node.kind === "schedule" || node.kind === "text") && /\b(coverage|limit|deductible|aggregate|occurrence|liability|premium|retroactive|retention)\b/i.test(text);
3622
- });
3623
- for (const row of rows) {
3624
- if (endorsementAncestor(row, byId)) continue;
3625
- const structured = row.kind === "table_row" ? coverageFromTableRow(row, children, byId) : void 0;
3626
- if (structured) {
3627
- if (!isOperationalCoverageRow(structured)) continue;
3628
- const key2 = [
3629
- structured.name.toLowerCase(),
3630
- structured.limits.map((term) => `${term.kind}:${term.value}`).join(","),
3631
- structured.sourceNodeIds.join(",")
3632
- ].join("|");
3633
- if (seen.has(key2)) continue;
3634
- seen.add(key2);
3635
- coverages.push(structured);
3636
- if (coverages.length >= 60) break;
3637
- continue;
3638
- }
3639
- const text = nodeText(row);
3640
- const name = coverageNameFromRow(text);
3641
- const limit = limitFromText(text);
3642
- const deductible = deductibleFromText(text);
3643
- const premium = premiumFromText(text);
3644
- if (!name || !limit && !deductible && !premium) continue;
3645
- if (!isOperationalCoverageRow({
3646
- name,
3647
- limit,
3648
- deductible,
3649
- premium,
3650
- coverageOrigin: "core",
3651
- limits: [
3652
- ...limit ? [{ kind: "other", label: "Limit", value: limit, sourceNodeIds: [row.id], sourceSpanIds: row.sourceSpanIds }] : [],
3653
- ...deductible ? [{ kind: "deductible", label: "Deductible", value: deductible, sourceNodeIds: [row.id], sourceSpanIds: row.sourceSpanIds }] : [],
3654
- ...premium ? [{ kind: "premium", label: "Premium", value: premium, sourceNodeIds: [row.id], sourceSpanIds: row.sourceSpanIds }] : []
3655
- ],
3656
- sourceNodeIds: [row.id],
3657
- sourceSpanIds: row.sourceSpanIds
3658
- })) continue;
3659
- const key = [name.toLowerCase(), limit, deductible, premium].join("|");
3660
- if (seen.has(key)) continue;
3661
- seen.add(key);
3662
- coverages.push({
3663
- name,
3664
- limit,
3665
- deductible,
3666
- premium,
3667
- formNumber: typeof row.metadata?.formNumber === "string" ? row.metadata.formNumber : void 0,
3668
- coverageOrigin: "core",
3669
- limits: [],
3670
- sourceNodeIds: [row.id],
3671
- sourceSpanIds: row.sourceSpanIds
3672
- });
3673
- if (coverages.length >= 60) break;
3674
- }
3675
- return coverages;
3676
- }
3677
- function buildParties(profile) {
3678
- const parties = [];
3679
- if (profile.namedInsured) {
3680
- parties.push({
3681
- role: "named_insured",
3682
- name: profile.namedInsured.normalizedValue ?? profile.namedInsured.value,
3683
- sourceNodeIds: profile.namedInsured.sourceNodeIds,
3684
- sourceSpanIds: profile.namedInsured.sourceSpanIds
3685
- });
3686
- }
3687
- if (profile.insurer) {
3688
- parties.push({
3689
- role: "insurer",
3690
- name: profile.insurer.normalizedValue ?? profile.insurer.value,
3691
- sourceNodeIds: profile.insurer.sourceNodeIds,
3692
- sourceSpanIds: profile.insurer.sourceSpanIds
3693
- });
3694
- }
3695
- if (profile.broker) {
3696
- parties.push({
3697
- role: "broker",
3698
- name: profile.broker.normalizedValue ?? profile.broker.value,
3699
- sourceNodeIds: profile.broker.sourceNodeIds,
3700
- sourceSpanIds: profile.broker.sourceSpanIds
3701
- });
3702
- }
3703
- return parties;
3704
- }
3705
- function buildEndorsementSupport(nodes) {
3706
- const support = [];
3707
- for (const node of nodes) {
3708
- const text = nodeText(node);
3709
- const add = (kind, status) => {
3710
- if (support.some((item) => item.kind === kind && item.status === status && item.sourceNodeIds.includes(node.id))) {
3711
- return;
3712
- }
3713
- support.push({
3714
- kind,
3715
- status,
3716
- summary: node.textExcerpt ?? node.description,
3717
- sourceNodeIds: [node.id],
3718
- sourceSpanIds: node.sourceSpanIds
3719
- });
3720
- };
3721
- if (/additional insured/i.test(text)) add("additional_insured", /not included|requires endorsement|only by endorsement/i.test(text) ? "requires_review" : "supported");
3722
- if (/waiver of subrogation|subrogation.*waived/i.test(text)) add("waiver_of_subrogation", /not included|requires endorsement|only by endorsement/i.test(text) ? "requires_review" : "supported");
3723
- if (/primary.*non[-\s]?contributory|non[-\s]?contributory/i.test(text)) add("primary_non_contributory", /not included|requires endorsement|only by endorsement/i.test(text) ? "requires_review" : "supported");
3724
- if (/loss payee|mortgagee/i.test(text)) add(/mortgagee/i.test(text) ? "mortgagee" : "loss_payee", "supported");
3725
- if (support.length >= 20) break;
3726
- }
3727
- return support;
3728
- }
3729
- function buildDeterministicOperationalProfile(params) {
3730
- const nodes = params.sourceTree.filter((node) => node.kind !== "document");
3731
- const partial = {
3732
- documentType: inferDocumentType(nodes),
3733
- policyTypes: inferPolicyTypes(nodes),
3734
- policyNumber: policyNumberFromNodes(nodes),
3735
- namedInsured: namedInsuredFromNodes(nodes),
3736
- insurer: insurerFromNodes(nodes),
3737
- broker: firstMatch(nodes, [
3738
- /\b(?:broker|producer|agent)\s*:?\s*([^|;\n]{3,120})/i
3739
- ]),
3740
- effectiveDate: firstMatch(nodes, [
3741
- /\b(?:effective date|policy period from|from)\s*:?\s*([0-9]{1,2}[/-][0-9]{1,2}[/-][0-9]{2,4})/i,
3742
- /\b([0-9]{1,2}[/-][0-9]{1,2}[/-][0-9]{2,4})\s+(?:to|through|-)\s+[0-9]{1,2}[/-][0-9]{1,2}[/-][0-9]{2,4}/i
3743
- ]),
3744
- expirationDate: firstMatch(nodes, [
3745
- /\b(?:expiration date|expiry date|expires|to)\s*:?\s*([0-9]{1,2}[/-][0-9]{1,2}[/-][0-9]{2,4})/i,
3746
- /\b[0-9]{1,2}[/-][0-9]{1,2}[/-][0-9]{2,4}\s+(?:to|through|-)\s+([0-9]{1,2}[/-][0-9]{1,2}[/-][0-9]{2,4})/i
3747
- ]),
3748
- retroactiveDate: firstMatch(nodes, [
3749
- /\bretroactive date\s*:?\s*([0-9]{1,2}[/-][0-9]{1,2}[/-][0-9]{2,4}|full prior acts|none)/i
3750
- ]),
3751
- premium: premiumFromNodes(nodes)
3752
- };
3753
- const coverages = buildCoverages(nodes);
3754
- const coverageTypes = [...new Set(coverages.map((coverage) => coverage.name))];
3755
- const sourceNodeIds = [.../* @__PURE__ */ new Set([
3756
- ...Object.values(partial).flatMap(
3757
- (value) => value && typeof value === "object" && "sourceNodeIds" in value ? value.sourceNodeIds : []
3758
- ),
3759
- ...coverages.flatMap((coverage) => coverage.sourceNodeIds)
3760
- ])];
3761
- const sourceSpanIds = [.../* @__PURE__ */ new Set([
3762
- ...Object.values(partial).flatMap(
3763
- (value) => value && typeof value === "object" && "sourceSpanIds" in value ? value.sourceSpanIds : []
3764
- ),
3765
- ...coverages.flatMap((coverage) => coverage.sourceSpanIds)
3766
- ])];
3767
- return PolicyOperationalProfileSchema.parse({
3768
- ...partial,
3769
- coverageTypes,
3770
- coverages,
3771
- parties: buildParties(partial),
3772
- endorsementSupport: buildEndorsementSupport(nodes),
3773
- sourceNodeIds,
3774
- sourceSpanIds,
3775
- warnings: [
3776
- ...coverages.length === 0 ? ["No source-backed coverage schedule rows were identified deterministically."] : [],
3777
- ...!partial.policyNumber ? ["No source-backed policy number was identified deterministically."] : [],
3778
- ...!partial.namedInsured ? ["No source-backed named insured was identified deterministically."] : []
3779
- ]
3780
- });
3781
- }
3782
- function mergeOperationalProfile(base, candidate, validNodeIds, validSpanIds) {
3783
- const keepIds = (ids, valid) => Array.isArray(ids) ? ids.filter((id) => typeof id === "string" && valid.has(id)) : [];
3784
- const mergeValue = (fallback, next) => {
3785
- if (!next || typeof next !== "object" || Array.isArray(next)) return fallback;
3786
- const record = next;
3787
- const value = typeof record.value === "string" ? cleanValue(record.value) : void 0;
3788
- if (!value) return fallback;
3789
- const sourceNodeIds = keepIds(record.sourceNodeIds, validNodeIds);
3790
- const sourceSpanIds = keepIds(record.sourceSpanIds, validSpanIds);
3791
- if (sourceNodeIds.length === 0 && sourceSpanIds.length === 0) return fallback;
3792
- return {
3793
- value,
3794
- normalizedValue: typeof record.normalizedValue === "string" ? record.normalizedValue : fallback?.normalizedValue,
3795
- confidence: record.confidence === "high" || record.confidence === "low" || record.confidence === "medium" ? record.confidence : "medium",
3796
- sourceNodeIds,
3797
- sourceSpanIds
3798
- };
3799
- };
3800
- const coverages = base.coverages.length > 0 ? base.coverages : Array.isArray(candidate.coverages) ? candidate.coverages.map((coverage) => {
3801
- const record = coverage;
3802
- const name = typeof record.name === "string" ? cleanValue(record.name) : void 0;
3803
- const sourceNodeIds = keepIds(record.sourceNodeIds, validNodeIds);
3804
- const sourceSpanIds = keepIds(record.sourceSpanIds, validSpanIds);
3805
- const limits = Array.isArray(record.limits) ? record.limits.filter(
3806
- (term) => Boolean(term) && typeof term === "object" && !Array.isArray(term)
3807
- ).flatMap((term) => {
3808
- const label = typeof term.label === "string" ? cleanValue(term.label) : void 0;
3809
- const value = typeof term.value === "string" ? cleanValue(term.value) : void 0;
3810
- const sourceNodeIds2 = keepIds(term.sourceNodeIds, validNodeIds);
3811
- const sourceSpanIds2 = keepIds(term.sourceSpanIds, validSpanIds);
3812
- if (!label || !value || sourceNodeIds2.length === 0 && sourceSpanIds2.length === 0) return [];
3813
- return [{
3814
- kind: normalizeTermKind(term.kind, label, value),
3815
- label,
3816
- value,
3817
- amount: typeof term.amount === "number" && Number.isFinite(term.amount) ? term.amount : void 0,
3818
- appliesTo: typeof term.appliesTo === "string" ? term.appliesTo : void 0,
3819
- sourceNodeIds: sourceNodeIds2,
3820
- sourceSpanIds: sourceSpanIds2
3821
- }];
3822
- }) : [];
3823
- return {
3824
- name,
3825
- coverageCode: typeof record.coverageCode === "string" ? cleanValue(record.coverageCode) : void 0,
3826
- limit: typeof record.limit === "string" ? cleanValue(record.limit) : void 0,
3827
- deductible: typeof record.deductible === "string" ? cleanValue(record.deductible) : void 0,
3828
- premium: typeof record.premium === "string" ? cleanValue(record.premium) : void 0,
3829
- retroactiveDate: typeof record.retroactiveDate === "string" ? cleanValue(record.retroactiveDate) : void 0,
3830
- formNumber: typeof record.formNumber === "string" ? cleanValue(record.formNumber) : void 0,
3831
- sectionRef: typeof record.sectionRef === "string" ? cleanValue(record.sectionRef) : void 0,
3832
- coverageOrigin: record.coverageOrigin === "core" || record.coverageOrigin === "endorsement" ? record.coverageOrigin : void 0,
3833
- endorsementNumber: typeof record.endorsementNumber === "string" ? cleanValue(record.endorsementNumber) : void 0,
3834
- limits,
3835
- sourceNodeIds,
3836
- sourceSpanIds
3837
- };
3838
- }).filter((coverage) => coverage.name && (coverage.sourceNodeIds.length > 0 || coverage.sourceSpanIds.length > 0)) : base.coverages;
3839
- return PolicyOperationalProfileSchema.parse({
3840
- ...base,
3841
- documentType: candidate.documentType === "quote" ? "quote" : candidate.documentType === "policy" ? "policy" : base.documentType,
3842
- policyTypes: Array.isArray(candidate.policyTypes) && candidate.policyTypes.length > 0 ? candidate.policyTypes : base.policyTypes,
3843
- policyNumber: mergeValue(base.policyNumber, candidate.policyNumber),
3844
- namedInsured: mergeValue(base.namedInsured, candidate.namedInsured),
3845
- insurer: mergeValue(base.insurer, candidate.insurer),
3846
- broker: mergeValue(base.broker, candidate.broker),
3847
- effectiveDate: mergeValue(base.effectiveDate, candidate.effectiveDate),
3848
- expirationDate: mergeValue(base.expirationDate, candidate.expirationDate),
3849
- retroactiveDate: mergeValue(base.retroactiveDate, candidate.retroactiveDate),
3850
- premium: mergeValue(base.premium, candidate.premium),
3851
- coverageTypes: Array.isArray(candidate.coverageTypes) && candidate.coverageTypes.length > 0 ? candidate.coverageTypes : base.coverageTypes,
3852
- coverages,
3853
- parties: base.parties,
3854
- endorsementSupport: base.endorsementSupport,
3855
- sourceNodeIds: [.../* @__PURE__ */ new Set([...base.sourceNodeIds, ...keepIds(candidate.sourceNodeIds, validNodeIds)])],
3856
- sourceSpanIds: [.../* @__PURE__ */ new Set([...base.sourceSpanIds, ...keepIds(candidate.sourceSpanIds, validSpanIds)])],
3857
- warnings: base.warnings
3858
- });
3859
- }
3860
-
3861
2989
  // src/extraction/pdf.ts
3862
2990
  import {
3863
2991
  PDFDocument,
@@ -9778,6 +8906,199 @@ function groundExtractionMemoryWithSourceSpans(memory, sourceSpans) {
9778
8906
  // src/extraction/source-tree-extractor.ts
9779
8907
  import { z as z42 } from "zod";
9780
8908
 
8909
+ // src/source/operational-profile.ts
8910
+ function normalizeWhitespace3(value) {
8911
+ return value.replace(/\s+/g, " ").trim();
8912
+ }
8913
+ function cleanValue(value) {
8914
+ if (!value) return void 0;
8915
+ return normalizeWhitespace3(value.replace(/^[\s:;#-]+|[\s;,.]+$/g, ""));
8916
+ }
8917
+ var OPERATIONAL_COVERAGE_TERM_KINDS = /* @__PURE__ */ new Set([
8918
+ "each_claim_limit",
8919
+ "each_occurrence_limit",
8920
+ "each_loss_limit",
8921
+ "aggregate_limit",
8922
+ "sublimit",
8923
+ "retention",
8924
+ "deductible",
8925
+ "retroactive_date",
8926
+ "premium",
8927
+ "other"
8928
+ ]);
8929
+ function normalizeTermKind(value) {
8930
+ return typeof value === "string" && OPERATIONAL_COVERAGE_TERM_KINDS.has(value) ? value : "other";
8931
+ }
8932
+ function mergeOperationalProfile(base, candidate, validNodeIds, validSpanIds) {
8933
+ const keepIds = (ids, valid) => Array.isArray(ids) ? ids.filter((id) => typeof id === "string" && valid.has(id)) : [];
8934
+ const mergeValue = (fallback, next) => {
8935
+ if (!next || typeof next !== "object" || Array.isArray(next)) return fallback;
8936
+ const record = next;
8937
+ const value = typeof record.value === "string" ? cleanValue(record.value) : void 0;
8938
+ if (!value) return fallback;
8939
+ const sourceNodeIds2 = keepIds(record.sourceNodeIds, validNodeIds);
8940
+ const sourceSpanIds2 = keepIds(record.sourceSpanIds, validSpanIds);
8941
+ if (sourceNodeIds2.length === 0 && sourceSpanIds2.length === 0) return fallback;
8942
+ return {
8943
+ value,
8944
+ normalizedValue: typeof record.normalizedValue === "string" ? record.normalizedValue : fallback?.normalizedValue,
8945
+ confidence: record.confidence === "high" || record.confidence === "low" || record.confidence === "medium" ? record.confidence : "medium",
8946
+ sourceNodeIds: sourceNodeIds2,
8947
+ sourceSpanIds: sourceSpanIds2
8948
+ };
8949
+ };
8950
+ const policyNumber = mergeValue(base.policyNumber, candidate.policyNumber);
8951
+ const namedInsured = mergeValue(base.namedInsured, candidate.namedInsured);
8952
+ const insurer = mergeValue(base.insurer, candidate.insurer);
8953
+ const broker = mergeValue(base.broker, candidate.broker);
8954
+ const effectiveDate = mergeValue(base.effectiveDate, candidate.effectiveDate);
8955
+ const expirationDate = mergeValue(base.expirationDate, candidate.expirationDate);
8956
+ const retroactiveDate = mergeValue(base.retroactiveDate, candidate.retroactiveDate);
8957
+ const premium = mergeValue(base.premium, candidate.premium);
8958
+ const sourceValues = [
8959
+ policyNumber,
8960
+ namedInsured,
8961
+ insurer,
8962
+ broker,
8963
+ effectiveDate,
8964
+ expirationDate,
8965
+ retroactiveDate,
8966
+ premium
8967
+ ].filter((value) => Boolean(value));
8968
+ const coverages = base.coverages.length > 0 ? base.coverages : Array.isArray(candidate.coverages) ? candidate.coverages.map((coverage) => {
8969
+ const record = coverage;
8970
+ const name = typeof record.name === "string" ? cleanValue(record.name) : void 0;
8971
+ const limits = Array.isArray(record.limits) ? record.limits.filter(
8972
+ (term) => Boolean(term) && typeof term === "object" && !Array.isArray(term)
8973
+ ).flatMap((term) => {
8974
+ const label = typeof term.label === "string" ? cleanValue(term.label) : void 0;
8975
+ const value = typeof term.value === "string" ? cleanValue(term.value) : void 0;
8976
+ const sourceNodeIds3 = keepIds(term.sourceNodeIds, validNodeIds);
8977
+ const sourceSpanIds3 = keepIds(term.sourceSpanIds, validSpanIds);
8978
+ if (!label || !value || sourceNodeIds3.length === 0 && sourceSpanIds3.length === 0) return [];
8979
+ return [{
8980
+ kind: normalizeTermKind(term.kind),
8981
+ label,
8982
+ value,
8983
+ amount: typeof term.amount === "number" && Number.isFinite(term.amount) ? term.amount : void 0,
8984
+ appliesTo: typeof term.appliesTo === "string" ? term.appliesTo : void 0,
8985
+ sourceNodeIds: sourceNodeIds3,
8986
+ sourceSpanIds: sourceSpanIds3
8987
+ }];
8988
+ }) : [];
8989
+ const sourceNodeIds2 = [.../* @__PURE__ */ new Set([
8990
+ ...keepIds(record.sourceNodeIds, validNodeIds),
8991
+ ...limits.flatMap((term) => term.sourceNodeIds)
8992
+ ])];
8993
+ const sourceSpanIds2 = [.../* @__PURE__ */ new Set([
8994
+ ...keepIds(record.sourceSpanIds, validSpanIds),
8995
+ ...limits.flatMap((term) => term.sourceSpanIds)
8996
+ ])];
8997
+ return {
8998
+ name,
8999
+ coverageCode: typeof record.coverageCode === "string" ? cleanValue(record.coverageCode) : void 0,
9000
+ limit: typeof record.limit === "string" ? cleanValue(record.limit) : void 0,
9001
+ deductible: typeof record.deductible === "string" ? cleanValue(record.deductible) : void 0,
9002
+ premium: typeof record.premium === "string" ? cleanValue(record.premium) : void 0,
9003
+ retroactiveDate: typeof record.retroactiveDate === "string" ? cleanValue(record.retroactiveDate) : void 0,
9004
+ formNumber: typeof record.formNumber === "string" ? cleanValue(record.formNumber) : void 0,
9005
+ sectionRef: typeof record.sectionRef === "string" ? cleanValue(record.sectionRef) : void 0,
9006
+ coverageOrigin: record.coverageOrigin === "core" || record.coverageOrigin === "endorsement" ? record.coverageOrigin : void 0,
9007
+ endorsementNumber: typeof record.endorsementNumber === "string" ? cleanValue(record.endorsementNumber) : void 0,
9008
+ limits,
9009
+ sourceNodeIds: sourceNodeIds2,
9010
+ sourceSpanIds: sourceSpanIds2
9011
+ };
9012
+ }).filter((coverage) => coverage.name && (coverage.sourceNodeIds.length > 0 || coverage.sourceSpanIds.length > 0)) : base.coverages;
9013
+ const sourceBackedParty = (role, value) => value ? {
9014
+ role,
9015
+ name: value.normalizedValue ?? value.value,
9016
+ sourceNodeIds: value.sourceNodeIds,
9017
+ sourceSpanIds: value.sourceSpanIds
9018
+ } : void 0;
9019
+ const candidateParties = Array.isArray(candidate.parties) ? candidate.parties.flatMap((party) => {
9020
+ if (!party || typeof party !== "object" || Array.isArray(party)) return [];
9021
+ const record = party;
9022
+ const role = typeof record.role === "string" ? cleanValue(record.role) : void 0;
9023
+ const name = typeof record.name === "string" ? cleanValue(record.name) : void 0;
9024
+ const sourceNodeIds2 = keepIds(record.sourceNodeIds, validNodeIds);
9025
+ const sourceSpanIds2 = keepIds(record.sourceSpanIds, validSpanIds);
9026
+ if (!role || !name || sourceNodeIds2.length === 0 && sourceSpanIds2.length === 0) return [];
9027
+ return [{ role, name, sourceNodeIds: sourceNodeIds2, sourceSpanIds: sourceSpanIds2 }];
9028
+ }) : [];
9029
+ const parties = [
9030
+ ...base.parties,
9031
+ ...candidateParties,
9032
+ sourceBackedParty("named_insured", namedInsured),
9033
+ sourceBackedParty("insurer", insurer),
9034
+ sourceBackedParty("broker", broker)
9035
+ ].filter((party) => Boolean(party)).filter(
9036
+ (party, index, rows) => rows.findIndex(
9037
+ (other) => other.role === party.role && other.name === party.name && other.sourceNodeIds.join(",") === party.sourceNodeIds.join(",") && other.sourceSpanIds.join(",") === party.sourceSpanIds.join(",")
9038
+ ) === index
9039
+ );
9040
+ const endorsementSupport = [
9041
+ ...base.endorsementSupport,
9042
+ ...Array.isArray(candidate.endorsementSupport) ? candidate.endorsementSupport.flatMap((row) => {
9043
+ if (!row || typeof row !== "object" || Array.isArray(row)) return [];
9044
+ const record = row;
9045
+ const kind = typeof record.kind === "string" ? cleanValue(record.kind) : void 0;
9046
+ const summary = typeof record.summary === "string" ? cleanValue(record.summary) : void 0;
9047
+ const status = record.status === "supported" || record.status === "excluded" || record.status === "requires_review" ? record.status : void 0;
9048
+ const sourceNodeIds2 = keepIds(record.sourceNodeIds, validNodeIds);
9049
+ const sourceSpanIds2 = keepIds(record.sourceSpanIds, validSpanIds);
9050
+ if (!kind || !summary || !status || sourceNodeIds2.length === 0 && sourceSpanIds2.length === 0) return [];
9051
+ return [{ kind, status, summary, sourceNodeIds: sourceNodeIds2, sourceSpanIds: sourceSpanIds2 }];
9052
+ }) : []
9053
+ ].filter(
9054
+ (row, index, rows) => rows.findIndex(
9055
+ (other) => other.kind === row.kind && other.status === row.status && other.summary === row.summary && other.sourceNodeIds.join(",") === row.sourceNodeIds.join(",") && other.sourceSpanIds.join(",") === row.sourceSpanIds.join(",")
9056
+ ) === index
9057
+ );
9058
+ const sourceNodeIds = [.../* @__PURE__ */ new Set([
9059
+ ...base.sourceNodeIds,
9060
+ ...keepIds(candidate.sourceNodeIds, validNodeIds),
9061
+ ...sourceValues.flatMap((value) => value.sourceNodeIds),
9062
+ ...coverages.flatMap((coverage) => coverage.sourceNodeIds),
9063
+ ...coverages.flatMap((coverage) => coverage.limits.flatMap((term) => term.sourceNodeIds)),
9064
+ ...parties.flatMap((party) => party.sourceNodeIds),
9065
+ ...endorsementSupport.flatMap((row) => row.sourceNodeIds)
9066
+ ])];
9067
+ const sourceSpanIds = [.../* @__PURE__ */ new Set([
9068
+ ...base.sourceSpanIds,
9069
+ ...keepIds(candidate.sourceSpanIds, validSpanIds),
9070
+ ...sourceValues.flatMap((value) => value.sourceSpanIds),
9071
+ ...coverages.flatMap((coverage) => coverage.sourceSpanIds),
9072
+ ...coverages.flatMap((coverage) => coverage.limits.flatMap((term) => term.sourceSpanIds)),
9073
+ ...parties.flatMap((party) => party.sourceSpanIds),
9074
+ ...endorsementSupport.flatMap((row) => row.sourceSpanIds)
9075
+ ])];
9076
+ const warnings = [
9077
+ ...base.warnings,
9078
+ ...Array.isArray(candidate.warnings) ? candidate.warnings.filter((warning) => typeof warning === "string" && warning.trim().length > 0) : []
9079
+ ];
9080
+ return PolicyOperationalProfileSchema.parse({
9081
+ ...base,
9082
+ documentType: candidate.documentType === "quote" ? "quote" : candidate.documentType === "policy" ? "policy" : base.documentType,
9083
+ policyTypes: Array.isArray(candidate.policyTypes) && candidate.policyTypes.length > 0 ? candidate.policyTypes : base.policyTypes,
9084
+ policyNumber,
9085
+ namedInsured,
9086
+ insurer,
9087
+ broker,
9088
+ effectiveDate,
9089
+ expirationDate,
9090
+ retroactiveDate,
9091
+ premium,
9092
+ coverageTypes: Array.isArray(candidate.coverageTypes) && candidate.coverageTypes.length > 0 ? candidate.coverageTypes : base.coverageTypes,
9093
+ coverages,
9094
+ parties,
9095
+ endorsementSupport,
9096
+ sourceNodeIds,
9097
+ sourceSpanIds,
9098
+ warnings: [...new Set(warnings)]
9099
+ });
9100
+ }
9101
+
9781
9102
  // src/extraction/operational-profile-cleanup.ts
9782
9103
  import { z as z41 } from "zod";
9783
9104
  var OPERATIONAL_COVERAGE_TERM_KINDS2 = [
@@ -11014,8 +10335,8 @@ function applyTitleSectionHierarchy(sourceTree) {
11014
10335
  const current = updates.get(child.id) ?? child;
11015
10336
  const heading = sectionHeadingTitle(current, container);
11016
10337
  if (heading) {
11017
- const descendants2 = byParent.get(child.id) ?? [];
11018
- const hasOwnContent = descendants2.some((descendant) => descendant.kind !== "table_row" && descendant.kind !== "table_cell");
10338
+ const descendants = byParent.get(child.id) ?? [];
10339
+ const hasOwnContent = descendants.some((descendant) => descendant.kind !== "table_row" && descendant.kind !== "table_cell");
11019
10340
  let hasFollowingContent = false;
11020
10341
  for (const nextChild of directPageChildren.slice(index + 1)) {
11021
10342
  const next = updates.get(nextChild.id) ?? nextChild;
@@ -11033,7 +10354,7 @@ function applyTitleSectionHierarchy(sourceTree) {
11033
10354
  parentId: container.id,
11034
10355
  kind: sectionKindForTitle(heading),
11035
10356
  title: heading,
11036
- description: descriptionWithPages(cleanText([heading, "section"].join(" "), heading), [current, ...descendants2]),
10357
+ description: descriptionWithPages(cleanText([heading, "section"].join(" "), heading), [current, ...descendants]),
11037
10358
  metadata: {
11038
10359
  ...current.metadata,
11039
10360
  organizer: "title_section",
@@ -11357,8 +10678,32 @@ ${JSON.stringify(nodes, null, 2)}
11357
10678
 
11358
10679
  Return JSON with labels and groups only.`;
11359
10680
  }
11360
- function buildOperationalProfilePrompt(sourceTree, fallback) {
11361
- const nodes = sourceTree.filter((node) => node.kind !== "document").slice(0, 240).map(compactNode2);
10681
+ function operationalProfilePromptNodes(sourceTree) {
10682
+ return sourceTree.filter((node) => node.kind !== "document").filter((node) => {
10683
+ if (["page_group", "form", "endorsement", "schedule", "table", "table_row", "table_cell"].includes(node.kind)) {
10684
+ return true;
10685
+ }
10686
+ const text = [node.title, node.path, node.description, node.textExcerpt].filter(Boolean).join(" ");
10687
+ return /\b(policy\s*(number|period)|named insured|insurer|carrier|broker|producer|premium|coverage|limit|liability|deductible|retention|retroactive|aggregate|sublimit|sub-limit|endorsement)\b/i.test(text);
10688
+ }).slice(0, 420);
10689
+ }
10690
+ function emptyOperationalProfile() {
10691
+ return {
10692
+ documentType: "policy",
10693
+ policyTypes: ["other"],
10694
+ coverageTypes: [],
10695
+ coverages: [],
10696
+ parties: [],
10697
+ endorsementSupport: [],
10698
+ sourceNodeIds: [],
10699
+ sourceSpanIds: [],
10700
+ warnings: []
10701
+ };
10702
+ }
10703
+ function buildOperationalProfilePrompt(sourceTree) {
10704
+ const nodes = operationalProfilePromptNodes(sourceTree).map(
10705
+ (node) => compactNode2(node, node.kind === "page" || node.kind === "endorsement" ? 900 : 700)
10706
+ );
11362
10707
  return `Extract a source-backed operational profile for an insurance policy or quote.
11363
10708
 
11364
10709
  Return only high-value operational facts needed for policy lists, Q&A, compliance, and certificate generation:
@@ -11375,9 +10720,7 @@ Rules:
11375
10720
  - 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.
11376
10721
  - Use coverageOrigin: "endorsement" for endorsement units and "core" for declarations/core policy coverage units.
11377
10722
  - Do not copy entire policy wording into fields.
11378
-
11379
- Deterministic baseline:
11380
- ${JSON.stringify(fallback, null, 2)}
10723
+ - Extract facts directly from source nodes. There is no deterministic fact baseline.
11381
10724
 
11382
10725
  Source nodes:
11383
10726
  ${JSON.stringify(nodes, null, 2)}
@@ -11550,6 +10893,33 @@ function findCellForContinuation(params) {
11550
10893
  }
11551
10894
  return params.cells[1] ?? params.cells[params.cells.length - 1];
11552
10895
  }
10896
+ function columnLabelStartIndex(rows) {
10897
+ const headerIndex = rows.findIndex(
10898
+ ({ row, cells }) => isSourceTreeHeaderRow(row) && cells.length > 1
10899
+ );
10900
+ return headerIndex >= 0 ? headerIndex : 0;
10901
+ }
10902
+ function metadataWithColumnLabel(metadata, label) {
10903
+ const next = {
10904
+ ...metadata ?? {},
10905
+ columnName: label,
10906
+ visualTableRepairColumnLabel: label
10907
+ };
10908
+ if (metadata?.table && typeof metadata.table === "object" && !Array.isArray(metadata.table)) {
10909
+ next.table = {
10910
+ ...metadata.table,
10911
+ columnName: label
10912
+ };
10913
+ }
10914
+ return next;
10915
+ }
10916
+ function tableCellDescription(cell, label) {
10917
+ const text = tableCellText(cell);
10918
+ return cleanText(
10919
+ text && text.toLowerCase() !== label.toLowerCase() ? `${label} | ${text}` : label,
10920
+ cell.description
10921
+ );
10922
+ }
11553
10923
  function applyVisualTableRepair(sourceTree, repair) {
11554
10924
  if (repair.tables.length === 0) return sourceTree;
11555
10925
  const byId = new Map(sourceTree.map((node) => [node.id, node]));
@@ -11569,8 +10939,10 @@ function applyVisualTableRepair(sourceTree, repair) {
11569
10939
  if (normalized) columnLabels.set(label.columnIndex, normalized);
11570
10940
  }
11571
10941
  if (columnLabels.size > 0) {
11572
- for (const { row, cells } of rows) {
10942
+ const firstLabelRowIndex = columnLabelStartIndex(rows);
10943
+ for (const [rowIndex, { row, cells }] of rows.entries()) {
11573
10944
  if (removeIds.has(row.id)) continue;
10945
+ if (rowIndex < firstLabelRowIndex) continue;
11574
10946
  for (const [fallbackIndex, cell] of cells.entries()) {
11575
10947
  const columnIndex = tableCellColumnIndex(cell, fallbackIndex);
11576
10948
  const label = columnLabels.get(columnIndex);
@@ -11578,10 +10950,8 @@ function applyVisualTableRepair(sourceTree, repair) {
11578
10950
  updates.set(cell.id, {
11579
10951
  ...currentNode(cell.id) ?? cell,
11580
10952
  title: label,
11581
- metadata: {
11582
- ...cell.metadata ?? {},
11583
- visualTableRepairColumnLabel: label
11584
- }
10953
+ description: tableCellDescription(cell, label),
10954
+ metadata: metadataWithColumnLabel(cell.metadata, label)
11585
10955
  });
11586
10956
  }
11587
10957
  }
@@ -12016,11 +11386,8 @@ async function runSourceTreeExtraction(params) {
12016
11386
  });
12017
11387
  sourceTree = visualTableRepair.sourceTree;
12018
11388
  warnings.push(...visualTableRepair.warnings);
12019
- const deterministicProfile = buildDeterministicOperationalProfile({
12020
- sourceTree,
12021
- sourceSpans
12022
- });
12023
- let operationalProfile = deterministicProfile;
11389
+ const emptyProfile = emptyOperationalProfile();
11390
+ let operationalProfile = emptyProfile;
12024
11391
  try {
12025
11392
  const validNodeIds = new Set(sourceTree.map((node) => node.id));
12026
11393
  const validSpanIds = new Set(sourceSpans.map((span) => span.id));
@@ -12029,7 +11396,7 @@ async function runSourceTreeExtraction(params) {
12029
11396
  const response = await safeGenerateObject(
12030
11397
  params.generateObject,
12031
11398
  {
12032
- prompt: buildOperationalProfilePrompt(sourceTree, deterministicProfile),
11399
+ prompt: buildOperationalProfilePrompt(sourceTree),
12033
11400
  schema: OperationalProfilePromptSchema,
12034
11401
  maxTokens: budget.maxTokens,
12035
11402
  taskKind: "extraction_operational_profile",
@@ -12037,7 +11404,7 @@ async function runSourceTreeExtraction(params) {
12037
11404
  providerOptions: params.providerOptions
12038
11405
  },
12039
11406
  {
12040
- fallback: deterministicProfile,
11407
+ fallback: emptyProfile,
12041
11408
  log: params.log
12042
11409
  }
12043
11410
  );
@@ -12048,19 +11415,19 @@ async function runSourceTreeExtraction(params) {
12048
11415
  durationMs: Date.now() - startedAt
12049
11416
  });
12050
11417
  operationalProfile = mergeOperationalProfile(
12051
- deterministicProfile,
11418
+ emptyProfile,
12052
11419
  response.object,
12053
11420
  validNodeIds,
12054
11421
  validSpanIds
12055
11422
  );
12056
11423
  } catch (error) {
12057
- warnings.push(`Operational profile model pass failed; deterministic profile used (${error instanceof Error ? error.message : String(error)})`);
11424
+ warnings.push(`Operational profile model pass failed; coverage rows omitted (${error instanceof Error ? error.message : String(error)})`);
12058
11425
  }
12059
11426
  if (operationalProfile.coverages.length > 0) {
12060
11427
  try {
12061
11428
  const validNodeIds = new Set(sourceTree.map((node) => node.id));
12062
11429
  const validSpanIds = new Set(sourceSpans.map((span) => span.id));
12063
- const budget = params.resolveBudget("extraction_operational_profile", 4096);
11430
+ const budget = params.resolveBudget("extraction_coverage_cleanup", 4096);
12064
11431
  const startedAt = Date.now();
12065
11432
  const response = await safeGenerateObject(
12066
11433
  params.generateObject,
@@ -12068,9 +11435,10 @@ async function runSourceTreeExtraction(params) {
12068
11435
  prompt: buildOperationalProfileCleanupPrompt(sourceTree, operationalProfile),
12069
11436
  schema: OperationalProfileCleanupSchema,
12070
11437
  maxTokens: budget.maxTokens,
12071
- taskKind: "extraction_operational_profile",
11438
+ taskKind: "extraction_coverage_cleanup",
12072
11439
  budgetDiagnostics: budget,
12073
- providerOptions: params.providerOptions
11440
+ providerOptions: params.providerOptions,
11441
+ trace: { phase: "coverage_cleanup", sourceBacked: true }
12074
11442
  },
12075
11443
  {
12076
11444
  fallback: { coverageDecisions: [], warnings: [] },
@@ -12078,7 +11446,7 @@ async function runSourceTreeExtraction(params) {
12078
11446
  }
12079
11447
  );
12080
11448
  localTrack(response.usage, {
12081
- taskKind: "extraction_operational_profile",
11449
+ taskKind: "extraction_coverage_cleanup",
12082
11450
  label: "operational_profile_cleanup",
12083
11451
  maxTokens: budget.maxTokens,
12084
11452
  durationMs: Date.now() - startedAt
@@ -17606,7 +16974,6 @@ export {
17606
16974
  buildConfirmationSummaryPrompt,
17607
16975
  buildConversationMemoryGuidance,
17608
16976
  buildCoverageGapPrompt,
17609
- buildDeterministicOperationalProfile,
17610
16977
  buildDoclingProviderOptions,
17611
16978
  buildDocumentSourceTree,
17612
16979
  buildFieldExplanationPrompt,
@@ -17663,7 +17030,6 @@ export {
17663
17030
  getTemplate,
17664
17031
  isDoclingExtractionInput,
17665
17032
  isFileReference,
17666
- mergeOperationalProfile,
17667
17033
  mergeQuestionAnswers,
17668
17034
  mergeSourceSpans,
17669
17035
  normalizeApplicationQuestionGraph,