/**
 * 4-layer dedup (Task 308).
 *
 * Strict superset of the previous nodeId-only dedup. Catches duplicate
 * representations the typed-edge hook (Task 305) and document-ingest
 * pipeline will inevitably create: slug variants, canonical-name
 * variants, and content-hash duplicates.
 *
 * Order: hits are sorted by descending score FIRST, then walked. The
 * highest-score representative wins on every layer; later entries with
 * the same key on any active layer are dropped. Missing values on a
 * layer (no `slug`, no `canonicalName`, no derivable hash) are not
 * collisions — they simply skip that layer for that row.
 */

import { createHash } from "node:crypto";

export type DedupLayer = "nodeId" | "slug" | "canonicalName" | "contentHash";

export const DEFAULT_LAYERS: DedupLayer[] = [
  "nodeId",
  "slug",
  "canonicalName",
  "contentHash",
];

interface DedupHit {
  nodeId: string;
  properties: Record<string, unknown>;
  score: number;
}

function readSlug(props: Record<string, unknown>): string | null {
  const slug = props["slug"];
  return typeof slug === "string" && slug.length > 0 ? slug : null;
}

function readCanonicalName(props: Record<string, unknown>): string | null {
  const canonical = props["canonicalName"];
  if (typeof canonical === "string" && canonical.length > 0) {
    return canonical.toLowerCase();
  }
  const name = props["name"];
  if (typeof name === "string" && name.length > 0) {
    return name.toLowerCase();
  }
  return null;
}

function readContentHash(props: Record<string, unknown>): string | null {
  const compiled = props["compiledTruth"];
  const content = props["content"];
  const source = (typeof compiled === "string" && compiled.length > 0)
    ? compiled
    : (typeof content === "string" && content.length > 0 ? content : null);
  if (source === null) return null;
  return createHash("sha256").update(source).digest("hex").slice(0, 16);
}

/**
 * Dedup `hits` across the requested layers. Returns a new array with the
 * highest-scoring representative per collision class, in original
 * descending-score order.
 */
export function dedup<T extends DedupHit>(
  hits: T[],
  layers: DedupLayer[] = DEFAULT_LAYERS,
): T[] {
  const sorted = [...hits].sort((a, b) => b.score - a.score);
  const seen = new Map<DedupLayer, Set<string>>();
  for (const layer of layers) seen.set(layer, new Set());
  const out: T[] = [];

  for (const hit of sorted) {
    let collision = false;
    const keys: Array<[DedupLayer, string]> = [];

    for (const layer of layers) {
      let key: string | null;
      switch (layer) {
        case "nodeId":
          key = hit.nodeId;
          break;
        case "slug":
          key = readSlug(hit.properties);
          break;
        case "canonicalName":
          key = readCanonicalName(hit.properties);
          break;
        case "contentHash":
          key = readContentHash(hit.properties);
          break;
      }
      if (key === null) continue;
      if (seen.get(layer)!.has(key)) {
        collision = true;
        break;
      }
      keys.push([layer, key]);
    }

    if (collision) continue;
    for (const [layer, key] of keys) seen.get(layer)!.add(key);
    out.push(hit);
  }

  return out;
}