BillTracker/services/subscriptionService.js

'use strict';

const { insertBill, validateBillData } = require('./billsService');

const SUBSCRIPTION_TYPES = [
  'streaming', 'software', 'cloud', 'music', 'news',
  'fitness', 'gaming', 'utilities', 'insurance',
  'food', 'education', 'shopping', 'security', 'other',
];

const MONTHLY_FACTORS = {
  weekly: 52 / 12,
  biweekly: 26 / 12,
  monthly: 1,
  quarterly: 1 / 3,
  annual: 1 / 12,
  annually: 1 / 12,
  irregular: 1,
};

// Fallback keyword list used when catalog lookup finds no match
const TYPE_KEYWORDS = [
  ['streaming', ['netflix', 'hulu', 'disney', 'max', 'paramount', 'peacock', 'youtube tv', 'sling', 'espn', 'fubo', 'starz', 'crunchyroll', 'dazn']],
  ['music',     ['spotify', 'apple music', 'tidal', 'pandora', 'siriusxm', 'soundcloud', 'deezer', 'iheart']],
  ['software',  ['adobe', 'microsoft', 'github', 'notion', 'figma', 'canva', 'openai', 'chatgpt', 'grammarly', 'zoom', 'slack', 'cursor', 'ynab']],
  ['cloud',     ['dropbox', 'icloud', 'google one', 'google storage', 'backblaze', 'box storage']],
  ['news',      ['nyt', 'new york times', 'economist', 'athletic', 'washington post', 'wsj', 'bloomberg', 'substack', 'patreon', 'medium']],
  ['fitness',   ['peloton', 'planet fitness', 'gym', 'fitbit', 'strava', 'headspace', 'calm', 'noom', 'classpass', 'whoop']],
  ['gaming',    ['xbox', 'playstation', 'steam', 'nintendo', 'roblox', 'discord nitro', 'ea play', 'ubisoft']],
  ['utilities', ['verizon', 'at t', 'att', 'comcast', 'xfinity', 'spectrum', 'tmobile', 't mobile']],
  ['insurance', ['insurance', 'geico', 'progressive', 'state farm', 'allstate']],
  ['food',      ['hellofresh', 'blue apron', 'doordash', 'instacart', 'uber eats', 'grubhub', 'factor', 'hungryroot']],
  ['education', ['duolingo', 'masterclass', 'coursera', 'skillshare', 'audible', 'kindle unlimited', 'blinkist']],
  ['shopping',  ['amazon prime', 'walmart plus', 'costco', 'target circle', 'chewy']],
  ['security',  ['nordvpn', 'expressvpn', '1password', 'dashlane', 'norton', 'mcafee', 'surfshark']],
];

// ── Catalog ───────────────────────────────────────────────────────────────────

function loadCatalog(db) {
  try {
    return db.prepare('SELECT id, rank, name, category, subscription_type, domain FROM subscription_catalog ORDER BY rank ASC').all();
  } catch {
    return [];
  }
}

function normalizeCatalogName(value) {
  return String(value || '').toLowerCase().replace(/[^a-z0-9]+/g, ' ').trim();
}

// Given a normalized merchant string, find the best matching catalog entry.
// Matches on service name (normalized) or domain (dot replaced with space).
function lookupCatalog(catalog, merchantText) {
  if (!catalog.length || !merchantText) return null;
  let best = null;
  let bestLen = 0;
  for (const entry of catalog) {
    const nameKey = normalizeCatalogName(entry.name);
    const domainKey = entry.domain ? entry.domain.replace(/\./g, ' ') : '';
    if (nameKey.length >= 3 && merchantText.includes(nameKey) && nameKey.length > bestLen) {
      best = entry;
      bestLen = nameKey.length;
    }
    if (domainKey.length >= 4 && merchantText.includes(domainKey) && domainKey.length > bestLen) {
      best = entry;
      bestLen = domainKey.length;
    }
  }
  return best;
}

// ── Helpers ───────────────────────────────────────────────────────────────────

function normalizeMerchant(value) {
  return String(value || '')
    .toLowerCase()
    .replace(/[^a-z0-9\s]/g, ' ')
    .replace(/\b(pos|debit|card|payment|purchase|recurring|online|inc|llc|co|www)\b/g, ' ')
    .replace(/\s+/g, ' ')
    .trim();
}

function titleCase(value) {
  return String(value || 'Subscription')
    .split(/\s+/)
    .filter(Boolean)
    .map(word => word.charAt(0).toUpperCase() + word.slice(1))
    .join(' ');
}

function inferType(merchantText, catalogEntry) {
  if (catalogEntry?.subscription_type) return catalogEntry.subscription_type;
  const haystack = normalizeMerchant(merchantText);
  for (const [type, words] of TYPE_KEYWORDS) {
    if (words.some(word => haystack.includes(word))) return type;
  }
  return 'other';
}

function monthlyEquivalent(amount, cycleType, billingCycle) {
  const key = String(cycleType || billingCycle || 'monthly').toLowerCase();
  const fallback = String(billingCycle || '').toLowerCase() === 'quarterly'
    ? 'quarterly'
    : String(billingCycle || '').toLowerCase() === 'annually'
    ? 'annual'
    : key;
  const factor = MONTHLY_FACTORS[key] ?? MONTHLY_FACTORS[fallback] ?? 1;
  return Math.round(Number(amount || 0) * factor * 100) / 100;
}

function nextDueDate(bill, now = new Date()) {
  const dueDay = Math.min(Math.max(Number(bill.due_day) || 1, 1), 31);
  const cycle = String(bill.cycle_type || bill.billing_cycle || 'monthly').toLowerCase();
  let date = new Date(now.getFullYear(), now.getMonth(), dueDay);
  if (date < new Date(now.getFullYear(), now.getMonth(), now.getDate())) {
    date = new Date(now.getFullYear(), now.getMonth() + 1, dueDay);
  }
  if (cycle === 'quarterly' || cycle === 'annual') {
    const startMonth = Math.min(Math.max(Number(bill.cycle_day) || 1, 1), 12) - 1;
    const step = cycle === 'quarterly' ? 3 : 12;
    date = new Date(now.getFullYear(), startMonth, dueDay);
    while (date < new Date(now.getFullYear(), now.getMonth(), now.getDate())) {
      date = new Date(date.getFullYear(), date.getMonth() + step, dueDay);
    }
  }
  return date.toISOString().slice(0, 10);
}

function decorateSubscription(bill) {
  const monthly = monthlyEquivalent(bill.expected_amount, bill.cycle_type, bill.billing_cycle);
  return {
    ...bill,
    is_subscription: !!bill.is_subscription,
    active: !!bill.active,
    monthly_equivalent: monthly,
    yearly_equivalent: Math.round(monthly * 12 * 100) / 100,
    next_due_date: nextDueDate(bill),
    subscription_type: bill.subscription_type || inferType(`${bill.name} ${bill.category_name || ''}`, null),
  };
}

function getSubscriptions(db, userId) {
  return db.prepare(`
    SELECT b.*, c.name AS category_name
    FROM bills b
    LEFT JOIN categories c ON c.id = b.category_id AND c.user_id = b.user_id AND c.deleted_at IS NULL
    WHERE b.user_id = ?
      AND b.deleted_at IS NULL
      AND b.is_subscription = 1
    ORDER BY b.active DESC, b.due_day ASC, b.name COLLATE NOCASE ASC
  `).all(userId).map(decorateSubscription);
}

function getSubscriptionSummary(subscriptions) {
  const active = subscriptions.filter(item => item.active);
  const monthlyTotal = active.reduce((sum, item) => sum + Number(item.monthly_equivalent || 0), 0);
  const typeTotals = new Map();
  for (const item of active) {
    const type = item.subscription_type || 'other';
    typeTotals.set(type, (typeTotals.get(type) || 0) + Number(item.monthly_equivalent || 0));
  }
  const topType = [...typeTotals.entries()].sort((a, b) => b[1] - a[1])[0] || null;
  return {
    active_count: active.length,
    paused_count: subscriptions.length - active.length,
    monthly_total: Math.round(monthlyTotal * 100) / 100,
    yearly_total: Math.round(monthlyTotal * 12 * 100) / 100,
    top_type: topType ? { type: topType[0], monthly_total: Math.round(topType[1] * 100) / 100 } : null,
  };
}

function existingBillNames(db, userId) {
  return db.prepare('SELECT name FROM bills WHERE user_id = ? AND deleted_at IS NULL')
    .all(userId)
    .map(row => normalizeMerchant(row.name))
    .filter(Boolean);
}

function dollarsFromTransactionAmount(amount) {
  return Math.round((Math.abs(Number(amount || 0)) / 100) * 100) / 100;
}

function billingCycleForCycleType(cycleType) {
  if (cycleType === 'quarterly') return 'quarterly';
  if (cycleType === 'annual')    return 'annually';
  if (cycleType === 'monthly')   return 'monthly';
  return 'irregular';
}

// ── Recommendations ───────────────────────────────────────────────────────────

function getSubscriptionRecommendations(db, userId) {
  const catalog = loadCatalog(db);
  const existingNames = existingBillNames(db, userId);

  // Scan all transaction sources, not just SimpleFIN
  const rows = db.prepare(`
    SELECT
      t.id, t.amount, t.currency, t.description, t.payee, t.memo, t.category,
      COALESCE(t.posted_date, substr(t.transacted_at, 1, 10)) AS tx_date,
      ds.provider AS data_source_provider,
      ds.name     AS data_source_name
    FROM transactions t
    LEFT JOIN data_sources ds ON ds.id = t.data_source_id AND ds.user_id = t.user_id
    WHERE t.user_id = ?
      AND t.ignored = 0
      AND t.match_status = 'unmatched'
      AND t.amount < 0
      AND COALESCE(t.posted_date, substr(t.transacted_at, 1, 10)) >= date('now', '-420 days')
    ORDER BY tx_date ASC
  `).all(userId);

  // Group by merchant + amount bucket
  const groups = new Map();
  for (const tx of rows) {
    const merchant = normalizeMerchant(tx.payee || tx.description || tx.memo);
    if (!merchant || merchant.length < 3) continue;
    const amount = dollarsFromTransactionAmount(tx.amount);
    if (amount < 1) continue;
    const key = `${merchant}:${Math.round(amount)}`;
    if (!groups.has(key)) {
      groups.set(key, { merchant, amountBucket: Math.round(amount), items: [], catalogEntry: null });
    }
    const group = groups.get(key);
    group.items.push({ ...tx, amount_dollars: amount });
    if (!group.catalogEntry) group.catalogEntry = lookupCatalog(catalog, merchant);
  }

  const recommendations = [];

  for (const group of groups.values()) {
    const { merchant, catalogEntry } = group;

    // Skip if already a known bill
    if (existingNames.some(name => name.includes(merchant) || merchant.includes(name))) continue;

    const sorted = group.items
      .filter(item => item.tx_date)
      .sort((a, b) => String(a.tx_date).localeCompare(String(b.tx_date)));

    if (sorted.length === 0) continue;

    const averageAmount = sorted.reduce((sum, item) => sum + item.amount_dollars, 0) / sorted.length;
    const maxDelta = sorted.length > 1
      ? Math.max(...sorted.map(item => Math.abs(item.amount_dollars - averageAmount)))
      : 0;
    const last = sorted[sorted.length - 1];

    // ── Tier 1: catalog match with 1 occurrence (possible subscription) ──────
    if (catalogEntry && sorted.length === 1) {
      const confidence = 62;
      recommendations.push(buildRecommendation({
        merchant, catalogEntry, sorted, averageAmount, maxDelta, last,
        cycleType: 'monthly', avgGap: 30, confidence, tier: 'possible',
      }));
      continue;
    }

    // ── Tier 2: 2+ occurrences — pattern detection ────────────────────────────
    if (sorted.length < 2) continue;

    const gaps = [];
    for (let i = 1; i < sorted.length; i++) {
      gaps.push(Math.round(
        (new Date(`${sorted[i].tx_date}T00:00:00`) - new Date(`${sorted[i - 1].tx_date}T00:00:00`)) / 86400000
      ));
    }
    const avgGap = gaps.reduce((sum, g) => sum + g, 0) / gaps.length;
    const cycleType = avgGap >= 320 ? 'annual'
      : avgGap >= 75 ? 'quarterly'
      : avgGap >= 10 && avgGap <= 18 ? 'biweekly'
      : avgGap <= 9 ? 'weekly'
      : 'monthly';

    if (cycleType === 'monthly'   && (avgGap < 24 || avgGap > 38))  continue;
    if (cycleType === 'quarterly' && (avgGap < 75 || avgGap > 105)) continue;
    if (maxDelta > Math.max(3, averageAmount * 0.18)) continue;

    // Confidence: catalog match raises the floor and ceiling
    let confidence;
    if (catalogEntry) {
      confidence = Math.min(99, 68 + sorted.length * 8 + (maxDelta <= 1 ? 8 : 0));
    } else {
      confidence = Math.min(96, 58 + sorted.length * 9 + (maxDelta <= 1 ? 10 : 0));
    }

    const tier = catalogEntry ? 'confirmed' : 'pattern';
    recommendations.push(buildRecommendation({
      merchant, catalogEntry, sorted, averageAmount, maxDelta, last,
      cycleType, avgGap, confidence, tier,
    }));
  }

  return recommendations
    .sort((a, b) => b.confidence - a.confidence || b.occurrence_count - a.occurrence_count)
    .slice(0, 20);
}

function buildRecommendation({ merchant, catalogEntry, sorted, averageAmount, maxDelta, last, cycleType, avgGap, confidence, tier }) {
  const name = catalogEntry ? catalogEntry.name : titleCase(merchant);
  const subscriptionType = inferType(merchant, catalogEntry);

  const reasons = [];
  if (catalogEntry)        reasons.push(`Matches known service: ${catalogEntry.name}`);
  if (sorted.length > 1)  reasons.push(`${sorted.length} similar charges`);
  if (sorted.length > 1)  reasons.push(`About ${Math.round(avgGap)} days apart`);
  reasons.push(`${last.currency || 'USD'} ${averageAmount.toFixed(2)} average`);

  return {
    id: Buffer.from(`${merchant}:${Math.round(averageAmount)}:${last.tx_date}`).toString('base64url'),
    name,
    subscription_type: subscriptionType,
    expected_amount: Math.round(averageAmount * 100) / 100,
    monthly_equivalent: monthlyEquivalent(averageAmount, cycleType, cycleType),
    cycle_type: cycleType,
    billing_cycle: billingCycleForCycleType(cycleType),
    due_day: Number(String(last.tx_date).slice(8, 10)) || 1,
    last_seen_date: last.tx_date,
    occurrence_count: sorted.length,
    confidence,
    tier,
    catalog_match: catalogEntry ? { id: catalogEntry.id, name: catalogEntry.name, category: catalogEntry.category } : null,
    transaction_ids: sorted.map(item => item.id),
    merchant,
    source: last.data_source_name || 'Transaction history',
    reasons,
  };
}

function createSubscriptionFromRecommendation(db, userId, payload = {}) {
  const seenDate = payload.last_seen_date || new Date().toISOString().slice(0, 10);
  const source = payload.catalog_match
    ? 'catalog_match'
    : 'simplefin_recommendation';

  const draft = {
    name: payload.name,
    category_id: payload.category_id || null,
    due_day: payload.due_day,
    expected_amount: payload.expected_amount,
    billing_cycle: billingCycleForCycleType(payload.cycle_type || 'monthly'),
    cycle_type: payload.cycle_type || 'monthly',
    cycle_day: payload.cycle_type === 'annual' || payload.cycle_type === 'quarterly'
      ? String(new Date(`${seenDate}T00:00:00`).getMonth() + 1)
      : String(payload.due_day || 1),
    is_subscription: 1,
    subscription_type: SUBSCRIPTION_TYPES.includes(payload.subscription_type) ? payload.subscription_type : 'other',
    reminder_days_before: 3,
    subscription_source: source,
    subscription_detected_at: new Date().toISOString(),
    notes: payload.merchant ? `Detected from recurring merchant: ${payload.merchant}` : null,
  };

  const validation = validateBillData(draft);
  if (validation.errors.length > 0) {
    const err = new Error(validation.errors[0].message);
    err.field = validation.errors[0].field;
    err.status = 400;
    throw err;
  }

  const created = insertBill(db, userId, validation.normalized);
  const ids = Array.isArray(payload.transaction_ids)
    ? payload.transaction_ids.map(id => Number(id)).filter(Number.isInteger).slice(0, 50)
    : [];
  if (ids.length > 0) {
    const update = db.prepare(`
      UPDATE transactions
      SET matched_bill_id = ?, match_status = 'matched', updated_at = CURRENT_TIMESTAMP
      WHERE user_id = ? AND id = ? AND ignored = 0
    `);
    for (const id of ids) update.run(created.id, userId, id);
  }

  return decorateSubscription(created);
}

module.exports = {
  SUBSCRIPTION_TYPES,
  createSubscriptionFromRecommendation,
  decorateSubscription,
  getSubscriptionRecommendations,
  getSubscriptionSummary,
  getSubscriptions,
  lookupCatalog,
  loadCatalog,
  monthlyEquivalent,
};