import type { ParsedArticle } from '../parsers/feed';
import { MAX_ARTICLE_GUID_SIZE, MAX_ARTICLE_LINK_SIZE } from '../utils/constants';
import { parseHttpUrl } from '../utils/validation';

/**
 * The "deduplication identity" of an Article (CONTEXT.md, ADR-0010).
 * `(guid, link)` together — GUID preferred, link fallback. Either side may be
 * null when the source feed omits one.
 */
export interface ExistingIdentity {
  guid: string | null;
  link: string | null;
}

/**
 * Decide which incoming articles are new (not already present) given the set
 * of identities already stored for the same Feed. Implements ADR-0010:
 *
 *   - A candidate is a duplicate if its GUID matches an existing GUID **or**
 *     its link matches an existing link. GUID is preferred because some feeds
 *     rotate links.
 *   - Candidates without a link are dropped (link is required for insertion).
 *   - Empty-string GUIDs are treated as absent (never match).
 *   - Intra-payload dedupe: a second candidate that matches an earlier one in
 *     the same payload is also dropped.
 *
 * Pure: no I/O. Returns a fresh array in payload order.
 */
export type LinkedArticle = ParsedArticle & { link: string };

/** Canonicalize an absolute Article resource identity without truncating it. */
export function canonicalizeArticleLink(link: string | null | undefined): string | undefined {
  if (!link || link.length > MAX_ARTICLE_LINK_SIZE) return undefined;
  const parsed = parseHttpUrl(link);
  if (!parsed.valid || parsed.value.length > MAX_ARTICLE_LINK_SIZE) return undefined;
  return parsed.value;
}

export function partitionNewArticles(candidates: ParsedArticle[], existing: ExistingIdentity[]): LinkedArticle[] {
  const seenGuids = new Set<string>();
  const seenLinks = new Set<string>();
  for (const e of existing) {
    if (e.guid && e.guid.length <= MAX_ARTICLE_GUID_SIZE) seenGuids.add(e.guid);
    const link = canonicalizeArticleLink(e.link);
    if (link) seenLinks.add(link);
  }

  const fresh: LinkedArticle[] = [];
  for (const candidate of candidates) {
    const link = canonicalizeArticleLink(candidate.link);
    if (!link) continue;
    const guid = candidate.guid && candidate.guid.length <= MAX_ARTICLE_GUID_SIZE ? candidate.guid : undefined;
    if (guid && seenGuids.has(guid)) continue;
    if (seenLinks.has(link)) continue;

    const normalized: LinkedArticle = { ...candidate, link };
    if (candidate.guid !== undefined && candidate.guid.length > MAX_ARTICLE_GUID_SIZE) delete normalized.guid;
    fresh.push(normalized);
    if (guid) seenGuids.add(guid);
    seenLinks.add(link);
  }
  return fresh;
}