// Build a path-keyed lookup of per-Produktgruppe per-category
// breakdowns extracted from the source PDF Erläuterungen.
//
// Currently only PG 1601 (Allgemeine Finanzwirtschaft) has parseable
// tabular breakdowns — that's where the major taxes (Gewerbesteuer,
// Grundsteuer, Einkommensteuer, etc.) get itemised. Other PGs have
// either prose-only Erläuterungen or no breakdowns; for those the
// lookup simply has no entries.

import type { BudgetTree, CategoryKey, Flow } from "../data/types.js";

// Teilergebnisplan line numbers map to our CategoryKey + flow side
// per the standard NRW-NKF schema. (Lines 27/28 are internal
// Leistungsbeziehungen and don't map to our category list.)
const LINE_TO_CATEGORY: Record<number, { key: CategoryKey; flow: Flow }> = {
  1: { key: "steuern", flow: "ertraege" },
  2: { key: "zuwendungenAllgemeineUmlagen", flow: "ertraege" },
  3: { key: "sonstigeTransfererträge", flow: "ertraege" },
  4: { key: "öffentlichRechtlicheLeistungsentgelte", flow: "ertraege" },
  5: { key: "privatrechtlicheLeistungsentgelte", flow: "ertraege" },
  6: { key: "kostenerstattungenKostenumlagen", flow: "ertraege" },
  7: { key: "sonstigeOrdentlicheErträge", flow: "ertraege" },
  8: { key: "aktivierteEigenleistungen", flow: "ertraege" },
  9: { key: "bestandsveränderungen", flow: "ertraege" },
  11: { key: "personalaufwendungen", flow: "aufwendungen" },
  12: { key: "versorgungsaufwendungen", flow: "aufwendungen" },
  13: { key: "sachUndDienstleistungen", flow: "aufwendungen" },
  14: { key: "bilanzielleAbschreibungen", flow: "aufwendungen" },
  15: { key: "transferaufwendungen", flow: "aufwendungen" },
  16: { key: "sonstigeOrdentlicheAufwendungen", flow: "aufwendungen" },
  19: { key: "finanzerträge", flow: "ertraege" },
  20: { key: "zinsenFinanzaufwendungen", flow: "aufwendungen" },
  23: { key: "außerordentlicheErträge", flow: "ertraege" },
  24: { key: "außerordentlicheAufwendungen", flow: "aufwendungen" },
};

export interface BreakdownItem {
  name: string;
  slug: string;
  /** € values keyed by year. Sparse — only years the source covers. */
  values: Record<number, number>;
}

/** Lookup keyed by `${flow}/${bereichSlug}/${gruppeSlug}/${categoryKey}` */
export type BreakdownsByPath = Record<string, BreakdownItem[]>;

interface PgSectionWithBreakdowns {
  pgNumber: string;
  name: string;
  beschreibung: string | null;
  erlaeuterungen: string | null;
  breakdowns?: Record<string, Array<{ name: string; values: Record<string, number> }>>;
}

const SLUG_REPLACEMENTS: Array<readonly [RegExp, string]> = [
  [/ä/g, "ae"],
  [/ö/g, "oe"],
  [/ü/g, "ue"],
  [/ß/g, "ss"],
];
function slugify(input: string): string {
  let s = input.toLowerCase();
  for (const [pat, rep] of SLUG_REPLACEMENTS) s = s.replace(pat, rep);
  s = s.replace(/[^a-z0-9]+/g, "-").replace(/^-+|-+$/g, "");
  return s || "n-a";
}

/** Match a PG by name against the budget tree and return its
 *  (bereich, gruppe) pair. Tolerates the same abbreviations as the
 *  pg-notes matcher. */
function matchPgInTree(
  pgName: string,
  tree: BudgetTree
): { bereichSlug: string; gruppeSlug: string } | null {
  const target = pgName.toLowerCase().replace(/[^a-z0-9]/g, "");
  let best: { dist: number; bereich: string; gruppe: string } | null = null;
  for (const bereich of tree.produktbereiche) {
    for (const gruppe of bereich.produktgruppen) {
      const candidate = gruppe.name.toLowerCase().replace(/[^a-z0-9]/g, "");
      if (candidate === target) {
        return { bereichSlug: bereich.slug, gruppeSlug: gruppe.slug };
      }
      // Fuzzy fallback (cheap edit-distance via length symmetric diff).
      let dist = Math.abs(candidate.length - target.length);
      const min = Math.min(candidate.length, target.length);
      for (let i = 0; i < min; i++) {
        if (candidate[i] !== target[i]) dist++;
      }
      if (best === null || dist < best.dist) {
        best = { dist, bereich: bereich.slug, gruppe: gruppe.slug };
      }
    }
  }
  if (best && best.dist / Math.max(target.length, 1) < 0.4) {
    return { bereichSlug: best.bereich, gruppeSlug: best.gruppe };
  }
  return null;
}

export function buildBreakdownsByPath(
  tree: BudgetTree,
  pgSections: Record<string, PgSectionWithBreakdowns>
): BreakdownsByPath {
  const out: BreakdownsByPath = {};

  for (const pg of Object.values(pgSections)) {
    if (!pg.breakdowns || Object.keys(pg.breakdowns).length === 0) continue;
    const place = matchPgInTree(pg.name, tree);
    if (!place) continue;

    for (const [lineNumStr, items] of Object.entries(pg.breakdowns)) {
      const lineNum = parseInt(lineNumStr, 10);
      const map = LINE_TO_CATEGORY[lineNum];
      if (!map) continue;

      // Convert string year keys back to numbers; assemble the lookup
      // key in the same shape consumers will use it.
      const breakdownItems: BreakdownItem[] = items.map((it) => {
        const values: Record<number, number> = {};
        for (const [yStr, v] of Object.entries(it.values)) {
          values[Number(yStr)] = v;
        }
        return {
          name: it.name,
          slug: slugify(it.name),
          values,
        };
      });

      const key = `${map.flow}/${place.bereichSlug}/${place.gruppeSlug}/${map.key}`;
      out[key] = breakdownItems;
    }
  }

  return out;
}