// Contrat de citations : le modèle ne peut citer que les balises [Sx] réellement fournies.
// Toute balise inconnue est neutralisée et comptée. Les balises valides sont résolues en
// objets cliquables transmis au client.

import type { ContextBlock, RetrievedChunk } from "./search.ts";

export type ResolvedCitation = {
  tag: string; // S1
  index: number; // 1
  courseCode: string | null;
  docTitle: string;
  filename: string;
  path: string;
  refLabel: string; // « Séance 4 — Diapositive 18 »
  refType: string;
  refNumber: number | null;
  title: string;
  excerpt: string;
  chunkId: number;
  documentId: number;
};

export function resolveCitations(text: string, context: ContextBlock): {
  cleaned: string;
  citations: ResolvedCitation[];
  invalidCount: number;
} {
  const valid = new Map(context.sources.map((s) => [s.tag, s.chunk]));
  const used = new Map<string, RetrievedChunk>();
  let invalidCount = 0;

  // Normalise [S1, S3] → [S1][S3], puis vérifie chaque balise.
  let cleaned = text.replace(/\[(S\d+(?:\s*,\s*S\d+)+)\]/g, (_m, group: string) =>
    group
      .split(/\s*,\s*/)
      .map((t: string) => `[${t}]`)
      .join("")
  );
  cleaned = cleaned.replace(/\[S(\d+)\]/g, (m, n: string) => {
    const tag = `S${n}`;
    const chunk = valid.get(tag);
    if (!chunk) {
      invalidCount++;
      return "";
    }
    if (!used.has(tag)) used.set(tag, chunk);
    return m;
  });

  const citations: ResolvedCitation[] = [...used.entries()].map(([tag, c]) => ({
    tag,
    index: parseInt(tag.slice(1), 10),
    courseCode: c.course_code,
    docTitle: c.doc_title,
    filename: c.filename,
    path: c.doc_path,
    refLabel: c.ref_label,
    refType: c.ref_type,
    refNumber: c.ref_number,
    title: c.title,
    excerpt: c.content.slice(0, 700),
    chunkId: c.id,
    documentId: c.document_id,
  }));

  return { cleaned, citations, invalidCount };
}