Feature: Search

Property search is a UK-only feature that queries the MySQL database directly (not the REST API). All search logic lives in src/features/search/.

Directory Structure

src/features/search/
  actions/
    get-properties.ts          # Main property search query
    get-area-property-counts.ts # Area-level property count queries
  api/
  components/
  hooks/
  types/
  utils/

How Search Works

Direct MySQL Queries

Search bypasses the REST API and queries the houses table directly via src/lib/db.ts. This is a server action (runs server-side only).

The main query in get-properties.ts:

1. Builds a WHERE clause from search params (city, bedrooms, bathrooms, map bounds)
2. JOINs with bills_prices_selectable (from the bills database) to calculate total price including bills
3. JOINs with bills_included_agents to account for agents that include certain bills
4. Applies HAVING conditions for price range filters (since total price is a calculated column)
5. Orders results based on sort preference
6. Paginates with LIMIT/OFFSET

Search Parameters

type SearchParams = {
  city: string;
  page?: string;
  minBedrooms?: string;
  maxBedrooms?: string;
  minBathrooms?: string;
  maxBathrooms?: string;
  minPrice?: string;
  maxPrice?: string;
  sort?: string;         // "recommended" | "price-low" | "price-high" | "newest" | "oldest"
  limit?: number;        // default: 24
  bounds?: {             // map viewport bounds
    west: number;
    south: number;
    east: number;
    north: number;
  };
};

Sort Options

Total Price Calculation

Total price = rent per week + bills cost. The bills cost is calculated in SQL by joining with the bills pricing table and accounting for which utilities the agent already includes:

• If agent includes both electric and gas, energy cost = 0
• If agent includes only gas, energy cost = electric only
• If agent includes only electric, energy cost = total energy minus electric
• Water and WiFi are added unless the agent includes them

Caching Strategy

Two unstable_cache wrappers handle caching:

• getCachedProperties(city) — Used when no filters are applied. Cache key: ["properties"], revalidation: 3 hours.
• getCachedPropertiesWithFilters(...) — Used when filters are present. Cache key: ["properties-filtered"], revalidation: 3 hours.

Both share the "properties" cache tag for on-demand revalidation.

Map-bounds queries are never cached — when bounds is provided, fetchPropertiesFromDb() is called directly.

Area Property Counts

get-area-property-counts.ts provides getAreaPropertyCounts(city, areas):

• Takes a city name and an array of areas with bounding boxes
• Runs parallel COUNT(*) queries for each area
• Results are cached via unstable_cache with the same 3-hour TTL and "properties" tag

Response Shape

type PropertiesResponse = {
  houses: House[];
  totalHouses: number;
  totalPages: number;
  currentPage: number;
  filters: Filters;
};

Each House object includes id, title, address, city, images (parsed from JSON), price_pw, total_price (rent + bills), bedrooms, bathrooms, lat, lng, tier, and timestamps.