Add deployment pipeline and clean up repo
- Add Dockerfile, docker-compose.yml, .dockerignore, .env (port 3842) - Add Caddyfile.snippet for analytics gateway import pattern - Add .gitea/workflows/deploy.yaml for act_runner SSH deploy - Untrack sensitive/data files (SCPA xlsx, owners.rds) - Add renv lockfile and infrastructure files - Reorganize data-raw scripts and add SarasotaCounty boundary data - Move www assets to www/images/, add docs PDFs
This commit is contained in:
82
data-raw/update_owners.R
Normal file
82
data-raw/update_owners.R
Normal file
@@ -0,0 +1,82 @@
|
||||
# update_owners.R
|
||||
# Weekly update script. Reads fresh SCPA property data, joins to stable
|
||||
# geometry lookup by account_number, saves data/owners.rds.
|
||||
# Only the SCPA xlsx needs to be replaced to refresh ownership data.
|
||||
# Input: data-raw/property/SCPA Public.xlsx (replace weekly)
|
||||
# data-raw/addresses/geometry_lookup.rds (static)
|
||||
# Output: data/owners.rds
|
||||
|
||||
library(readxl)
|
||||
library(janitor)
|
||||
library(dplyr)
|
||||
library(stringr)
|
||||
library(sf)
|
||||
|
||||
subdivisions <- c(
|
||||
"8120", "8113", "8171", "8195", "8221",
|
||||
"8163", "8240", "8159", "8149", "8110", "8254", "8215", "8143"
|
||||
)
|
||||
|
||||
# load geometry lookup (static) ----
|
||||
geometry_lookup <- readRDS("./data-raw/addresses/geometry_lookup.rds")
|
||||
|
||||
# download fresh scpa data ----
|
||||
download.file(
|
||||
url = "https://www.sc-pa.com/downloads/SCPA%20Public.xlsx",
|
||||
destfile = "./data-raw/property/SCPA Public.xlsx",
|
||||
mode = "wb"
|
||||
)
|
||||
|
||||
# load and clean scpa data ----
|
||||
owners_raw <-
|
||||
readxl::read_xlsx(
|
||||
path = "./data-raw/property/SCPA Public.xlsx",
|
||||
n_max = Inf,
|
||||
.name_repair = ~janitor::make_clean_names(.x)
|
||||
) |>
|
||||
filter(subdivision %in% subdivisions) |>
|
||||
rename(
|
||||
situs_address = situs_address_property_address,
|
||||
homestead = homestead_exemption_yes_or_no
|
||||
) |>
|
||||
filter(!is.na(situs_address)) |>
|
||||
filter(!grepl("^0", situs_address)) |>
|
||||
mutate(
|
||||
# extract clean street address (before multiple spaces / unit suffix)
|
||||
label = str_trim(str_extract(situs_address, "^\\d+\\s+\\S+\\s+\\S+")),
|
||||
location = paste0(label, ", Venice FL")
|
||||
) |>
|
||||
select(account_number, owner_1, owner_2, subdivision, homestead, label, location)
|
||||
|
||||
# join to geometry ----
|
||||
owners <- owners_raw |>
|
||||
inner_join(geometry_lookup, by = "account_number") |>
|
||||
st_as_sf(sf_column_name = "geom")
|
||||
|
||||
# report any unmatched records ----
|
||||
n_unmatched <- nrow(owners_raw) - nrow(owners)
|
||||
if (n_unmatched > 0) {
|
||||
cat("WARNING:", n_unmatched, "records had no matching geometry and were dropped.\n")
|
||||
missing <- anti_join(owners_raw, st_drop_geometry(geometry_lookup), by = "account_number")
|
||||
print(missing)
|
||||
}
|
||||
|
||||
# report most recent sale date in st. andrews ----
|
||||
latest_sale <-
|
||||
readxl::read_xlsx(
|
||||
path = "./data-raw/property/SCPA Public.xlsx",
|
||||
n_max = Inf,
|
||||
.name_repair = ~janitor::make_clean_names(.x)
|
||||
) |>
|
||||
filter(subdivision %in% subdivisions) |>
|
||||
select(account_number, owner_1, contains("sale")) |>
|
||||
filter(!is.na(account_number)) |>
|
||||
mutate(last_sale_date = as.Date(last_sale_date, format = "%m/%d/%Y")) |>
|
||||
arrange(desc(last_sale_date)) |>
|
||||
head(1)
|
||||
|
||||
cat("Owners written:", nrow(owners), "\n")
|
||||
attr(owners, "last_sale_date") <- latest_sale$last_sale_date
|
||||
saveRDS(owners, "./data/owners.rds")
|
||||
cat("Saved to data/owners.rds\n")
|
||||
cat("Most recent sale date:", format(latest_sale$last_sale_date, "%B %d, %Y"), "\n")
|
||||
Reference in New Issue
Block a user