# update_listings.R
# Weekly update: pull active listings from RentCast within 1 mile of
# St. Andrews center, clip to plat boundary, save data/listings.rds.
# Input:  .Renviron (RENTCAST_API_KEY)
#         data/plats/plats.shp
# Output: data/listings.rds

library(httr2)
library(jsonlite)
library(sf)
library(dplyr)

# ── API key ───────────────────────────────────────────────────────────────────
api_key <- Sys.getenv("RENTCAST_API_KEY")
if (identical(api_key, "")) stop("Missing RENTCAST_API_KEY in .Renviron")

# ── Pull listings ─────────────────────────────────────────────────────────────
resp <- request("https://api.rentcast.io/v1/listings/sale") |>
  req_headers(
    "Accept"    = "application/json",
    "X-Api-Key" = api_key
  ) |>
  req_url_query(
    latitude  = 27.076199,
    longitude = -82.362253,
    radius    = 1,
    status    = "Active",
    limit     = 500
  ) |>
  req_retry(max_tries = 3) |>
  req_perform()

if (resp_status(resp) >= 400)
  stop("RentCast API error: HTTP ", resp_status(resp), "\n", resp_body_string(resp))

listings_raw <- jsonlite::fromJSON(resp_body_string(resp), flatten = TRUE)
cat("Fetched:", nrow(listings_raw), "listings\n")

# ── Clip to St. Andrews plat boundary ────────────────────────────────────────
plats <- sf::st_read("./data/plats/plats.shp", quiet = TRUE) |>
  sf::st_union()

listings_sf <- sf::st_as_sf(
  listings_raw,
  coords = c("longitude", "latitude"),
  crs    = 4326,
  remove = FALSE
)

in_plat  <- lengths(sf::st_within(listings_sf, plats)) > 0
listings <- listings_raw[in_plat, ]
cat("After plat clip:", nrow(listings), "listings\n")

# ── Override RentCast coordinates with building footprint geometry ────────────
# RentCast geocoding is approximate. Our owners data uses building centroids
# from Sarasota County GIS footprints — far more accurate. Match on house
# number + street name and substitute when found.
owners_sf <- readRDS("./data/owners.rds")

# Extract house number from owners location field (e.g. "878   CHALMERS DR, Venice FL")
owners_coords <- owners_sf |>
  mutate(
    house_num  = trimws(sub("^(\\d+).*", "\\1", location)),
    street_raw = trimws(sub("^\\d+\\s+(.*),.*$", "\\1", location)),
    match_key  = paste(house_num, toupper(street_raw))
  ) |>
  select(match_key, owner_1, geom) |>
  distinct(match_key, .keep_all = TRUE)

# Extract house number + street from RentCast address
# e.g. "878 Chalmers Dr, Unit 878, Venice, FL 34293" -> "878 CHALMERS DR"
listings <- listings |>
  mutate(
    house_num  = sub("^(\\d+)\\s.*", "\\1", addressLine1),
    street_raw = gsub("[^A-Za-z ]", "", sub("^\\d+\\s+(\\S+\\s+\\S+).*", "\\1", addressLine1)),
    match_key  = paste(house_num, toupper(trimws(street_raw)))
  )

matched <- merge(listings, owners_coords, by = "match_key", all.x = TRUE)

# For matched rows, replace RentCast lat/lng with footprint centroid coords
has_geom <- !is.na(matched$geom)
if (any(has_geom)) {
  coords <- sf::st_coordinates(sf::st_as_sf(matched[has_geom, ], sf_column_name = "geom"))
  matched$longitude[has_geom] <- coords[, "X"]
  matched$latitude[has_geom]  <- coords[, "Y"]
  cat("Coordinates corrected from building footprints:", sum(has_geom), "listing(s)\n")
}

listings <- matched

# ── Select and clean columns ──────────────────────────────────────────────────
listings <- listings |>
  transmute(
    listed_date    = as.Date(listedDate),
    address        = formattedAddress,
    owner_1,
    sqft           = as.numeric(squareFootage),
    price          = as.numeric(price),
    price_per_sqft = round(price / sqft, 0),
    latitude,
    longitude
  ) |>
  arrange(desc(listed_date))

cat("Listings saved:", nrow(listings), "\n")
saveRDS(listings, "./data/listings.rds")