fix: parse gathered price labels

price-watch/latest-report.md

@@ -2,6 +2,14 @@
 
 Checked: 2026-04-30
 
+## Data normalization update
+
+- The gathered data has been normalized into `price-watch/history.jsonl` with 40 parseable price points across hotels, golf, nightlife, excursions, itineraries, and budget options.
+- Each normalized price point now carries a stable seed key, numeric price, source, booking type, price basis, and included/excluded components when applicable.
+- Costco package result keys for Dreams, Zoetry, and Hard Rock are now mapped to the seeded hotel cards instead of remaining orphaned under source-specific keys.
+- Package quotes are now labeled separately from standalone bookings. Costco and Apple Vacations are treated as package pricing, while KAYAK hotel results are treated as standalone hotel rates.
+- Budget and itinerary rows are included as calculated planning baselines until the next full live refresh recalculates them from fresh component prices.
+
 ## Biggest price changes
 
 - Palmilla Golf Club: current public rates are $194.53-$291.80 for 18 holes depending on date/time window, versus the seed planning floor of about $130. That is up about $64.53 at the low end and as much as $161.80 at prime time. Source: [Cabo Villas](https://www.cabovillas.com/golf/palmilla)
@@ -75,9 +83,9 @@ Checked: 2026-04-30
 - Balanced track is still the cleanest value story, but the current public golf pricing makes the "balanced" number more sensitive to tee-time selection.
 - Splurge track is still viable, but the live golf and nightlife signals are now firmly in premium territory, so the high-end itinerary should be modeled with a little extra buffer.
 - On the Costco package screen, `Dreams` is the cheapest visible option for the searched dates, with `Breathless` mid-pack and `Secrets` the priciest visible live option among the shown results.
-- For the other travel sites, I am withholding a budget call until we have date-matched quotes rather than generic floors.
+- Date-matched Apple Vacations and KAYAK hotel prices are now included in the tracked data. CheapCaribbean still needs an exact-date results pass before its values should be counted.
 
 ## Notes
 
 - Public activity pricing remained usable for the bachelor-trip mix: Cabo Adventures ATV still shows the required $25 entrance fee and $35 damage waiver on the live page, and Cabo Villas sailing prices are still publicly posted.
-- This was the first run, so there is no prior history row to compare against.
+- Earlier non-date-matched generic floors remain excluded from tracking unless a source exposes an exact date-matched quote.

server.js

@@ -107,6 +107,9 @@ const HISTORY_KEY_ALIASES = {
   'costco-secrets': 'hotel-secrets',
   'costco-corazon': 'hotel-corazon',
   'costco-pacifica': 'hotel-pacifica',
+  'costco-dreams': 'hotel-dreams-los-cabos',
+  'costco-zoetry': 'hotel-zoetry-casa-del-mar',
+  'costco-hard-rock': 'hotel-hard-rock-los-cabos',
 };
 
 function toTextList(value) {
@@ -192,11 +195,82 @@ function extractNumericPrice(point) {
   for (const candidate of candidates) {
     if (typeof candidate === 'number' && Number.isFinite(candidate)) return candidate;
     if (typeof candidate === 'string') {
-      const parsed = Number(candidate.replace(/[^0-9.-]/g, ''));
+      const parsed = parseNumericPriceFromText(candidate);
       if (Number.isFinite(parsed)) return parsed;
     }
   }
 
+  const textCandidates = [
+    point.displayPrice,
+    point.displayLabel,
+    point.priceLabel,
+    point.label,
+    point.note,
+    point.description,
+  ].filter(Boolean);
+
+  for (const candidate of textCandidates) {
+    const parsed = parseNumericPriceFromText(candidate, point.priceBasis || point.price_basis || point.unit);
+    if (Number.isFinite(parsed)) return parsed;
+  }
+
+  return null;
+}
+
+function parseNumericPriceFromText(value, priceBasis = '') {
+  if (typeof value !== 'string') return null;
+  const normalized = value.replace(/\s+/g, ' ').trim();
+  if (!normalized || /\b(no|not)\s+(fresh\s+)?(price|rates?|available|counted|visible|captured)\b/i.test(normalized)) {
+    return null;
+  }
+
+  const matches = [...normalized.matchAll(/\$?\s*([0-9]{1,3}(?:,[0-9]{3})*(?:\.[0-9]{1,2})?|[0-9]+(?:\.[0-9]{1,2})?)/g)]
+    .map((match) => ({
+      value: Number(match[1].replace(/,/g, '')),
+      index: match.index || 0,
+    }))
+    .filter((match) => Number.isFinite(match.value));
+
+  if (!matches.length) return null;
+
+  const basis = normalizeKey(priceBasis);
+  if (basis === 'totalpackage' || basis === 'pergroup') {
+    return matches.at(-1).value;
+  }
+
+  const travelerMatch = matches.find((match) => (
+    /per\s+(traveler|person|guest|adult|round|table|night)|pp|\/night/i.test(normalized.slice(match.index, match.index + 80))
+  ));
+  if (travelerMatch) return travelerMatch.value;
+
+  return matches[0].value;
+}
+
+function inferPriceBasis(point, defaults = {}) {
+  if (point.priceBasis || point.price_basis || point.unit || defaults.priceBasis) {
+    return point.priceBasis || point.price_basis || point.unit || defaults.priceBasis;
+  }
+
+  const haystack = [
+    point.displayPrice,
+    point.displayLabel,
+    point.priceLabel,
+    point.label,
+    point.note,
+    point.description,
+  ].filter(Boolean).join(' ').toLowerCase();
+
+  if (haystack.includes('/night') || haystack.includes('per night')) return 'perNight';
+  if (haystack.includes('per traveler')) return 'perTraveler';
+  if (haystack.includes('per person') || /\bpp\b/.test(haystack)) return 'perPerson';
+  if (haystack.includes('per round')) return 'perRound';
+  if (haystack.includes('per table')) return 'perTable';
+  if (haystack.includes('total') || haystack.includes('package')) return 'totalPackage';
+
+  const bookingType = normalizeBookingType(inferBookingType(point, defaults));
+  if (bookingType === 'package') return 'perTraveler';
+  if (bookingType === 'calculated') return 'perPerson';
+
   return null;
 }
 
@@ -242,7 +316,7 @@ function loadPriceHistoryState() {
       sourceKey: normalizeKey(point.sourceKey || point.sourceId || point.source || point.sourceLabel || point.vendor || defaults.sourceKey || 'unknown-source'),
       sourceUrl: point.sourceUrl || point.url || defaults.sourceUrl || null,
       bookingType: normalizeBookingType(inferBookingType(point, defaults)),
-      priceBasis: point.priceBasis || point.price_basis || point.unit || defaults.priceBasis || null,
+      priceBasis: inferPriceBasis(point, defaults),
       includedComponents: toTextList(point.includedComponents || point.includesComponents || point.componentsIncluded || defaults.includedComponents),
       excludedComponents: toTextList(point.excludedComponents || point.componentsExcluded || defaults.excludedComponents),
       origin: point.origin || point.originAirport || defaults.origin || null,