/**
 * Copyright (c) 2019-2025 Mark Sandford
 * Licensed under the MIT License. See LICENSE and NOTICE files.
 */
/**
 * Unified Company Entity Resolution
 *
 * Consolidates duplicate name normalization and matching logic from:
 * - fec-entity-resolution.ts (cleanNameForMatching, expandAbbreviation, levenshteinDistance)
 * - influence-chain-analyzer.ts (normalizeOrgName, validateTokenOverlap, levenshteinDistance)
 *
 * Single canonical implementation for cross-API company name matching
 * across EPA, OSHA, CFPB, SEC, FEC, and LDA data sources.
 */
import { IndustrySector } from './industry-taxonomy.js';
export interface ResolvedCompany {
    canonicalName: string;
    normalizedName: string;
    aliases: string[];
    sicCodes: string[];
    naicsCodes: string[];
    sector: IndustrySector | null;
    cik: string | null;
    confidence: number;
}
/**
 * Normalize a company name for matching.
 *
 * Consolidates cleanNameForMatching() (fec-entity-resolution.ts) and
 * normalizeOrgName() (influence-chain-analyzer.ts) into a single pipeline:
 *
 * 1. Trim and uppercase
 * 2. Strip corporate suffixes (word boundary, all occurrences)
 * 3. Normalize & to AND
 * 4. Remove non-word non-space characters
 * 5. Collapse whitespace
 * 6. Expand abbreviations (JNJ -> JOHNSON AND JOHNSON)
 */
export declare function normalizeCompanyName(name: string): string;
/**
 * Compute similarity ratio between two strings (0-1).
 * 1.0 = identical, 0.0 = completely different.
 */
export declare function similarityRatio(a: string, b: string): number;
/**
 * Validate that two normalized names share enough word tokens.
 *
 * Catches false positives that Levenshtein misses. For example:
 * "American Health Association" vs "American Heart Association" = 0.90 Levenshtein
 * But "Health" vs "Heart" = 0.6 token similarity -> correctly rejected.
 *
 * Each token from the shorter name must find a close match (>= 0.75
 * similarity) in the longer name. At least `threshold` fraction of the
 * shorter name's tokens must match.
 */
export declare function validateTokenOverlap(a: string, b: string, threshold?: number): boolean;
/**
 * Check whether two company names refer to the same entity.
 *
 * Pipeline:
 * 1. Normalize both names
 * 2. Check alias table for canonical matches
 * 3. Levenshtein similarity with token overlap validation
 * 4. Optional SIC code cross-validation boost
 */
export declare function companiesMatch(a: string, b: string, context?: {
    sicCodeA?: string;
    sicCodeB?: string;
}): {
    match: boolean;
    confidence: number;
};
/**
 * Resolve a raw company name to its canonical form.
 * Uses alias table lookup and SIC/NAICS cross-validation.
 */
export declare function resolveCompanyName(rawName: string, context?: {
    sicCode?: string;
    naicsCode?: string;
    state?: string;
}): ResolvedCompany | null;
/**
 * Batch resolve company names, deduplicating entries that resolve to the same entity.
 */
export declare function resolveCompanyNames(entries: Array<{
    name: string;
    source: string;
    context?: {
        sicCode?: string;
        naicsCode?: string;
        state?: string;
    };
}>): Map<string, ResolvedCompany>;
//# sourceMappingURL=company-entity-resolver.d.ts.map