R code to import all years of county-level building permit data and lightly clean

permits-import.r

library("tidyverse")
library("rvest")

## Import ----
### The directory location
url_loc <- "www2.census.gov/econ/bps/County/"

files <- rvest::read_html("https://www2.census.gov/econ/bps/County/") |>
  rvest::html_elements("a") |>
  rvest::html_text2()

### Just take files containing  `coYYYYa``
fname_stubs <- files[str_detect(files, "(co)\\d{4}(:?a)")]
# remove 2022 update file
fname_stubs <- fname_stubs[-34]

### Add year
fname_year <- substr(fname_stubs, start = 3, stop = 6)

### Clean up the URLs
fnames <- paste0("https://", url_loc, fname_stubs)
fnames <- set_names(fnames, fname_year)

### Import
cpb <- fnames |>
  map(~ read_csv(., skip = 1)) |>
  list_rbind(names_to = "year") |>
  rename(
    survey = Date,
    state_fips = State,
    county_fips = County,
    region = "Code...4",
    division = "Code...5",
    county_name = Name,
    bldg_1unit = "Bldgs...7",
    units_1unit = "Units...8",
    val_1unit = "Value...9",
    bldg_2unit = "Bldgs...10",
    units_2unit = "Units...11",
    val_2unit = "Value...12",
    bldg_34unit = "Bldgs...13",
    units_34unit = "Units...14",
    val_34unit = "Value...15",
    bldg_5unit = "Bldgs...16",
    units_5unit = "Units...17",
    val_5unit = "Value...18",
    bldg_1unit_rep = "Bldgs...19",
    units_1unit_rep = "Units...20",
    val_1unit_rep = "Value...21",
    bldg_2unit_rep = "Bldgs...22",
    units_2unit_rep = "Units...23",
    val_2unit_rep = "Value...24",
    bldg_34unit_rep = "Bldgs...25",
    units_34unit_rep = "Units...26",
    val_34unit_rep = "Value...27",
    bldg_5unit_rep = "Bldgs...28",
    units_5unit_rep = "Units...29",
    val_5unit_rep = "Value...30"
  )