Sveriges äldsta byggnader

get_buildings.R

library(sf)
library(dplyr)
library(stringr)
library(rvest)

download.file("https://gis-services.metria.se/raafeed/files/gpc/BU.BuildingsGPKG.zip",
              destfile = "buildingsGPKG.zip")

unzip("BuildingsGPKG.zip", files = "BU.Buildings.gpkg")

buildings <- st_read("shiny-app/BU.Buildings.gpkg")

get_building_data <- function(url){
  
  resp <- read_html(url)
  
  url_id <- stringr::str_extract(url, "[:digit:]+$")
  
  table <- resp |> 
    html_table()
  
  if(length(table) < 2){
    df <- tibble::tibble(
      names = c("Namn",
                "Nybyggnadsår",
                "Län",
                "Kommun",
                "Landskap",
                "Socken",
                "Församling",
                "Stift",
                "Adress",
                "Historisk kategori",
                "Nuvarande kategori"),
      values = NA,
      id = url_id
    )
  } else {
    
    df <- table[[2]]
    
    df$id <- url_id
    
    colnames(df) <-  c("names", "values", "id")
  }
  
  df |> 
    tidyr::pivot_wider(
      names_from = names, values_from = values
    ) |> 
    janitor::clean_names()
}

get_building_data(url)
library(purrr)
building_data <- map_df(buildings$url, get_building_data)

buildings_complete <- dplyr::bind_cols(
  buildings,
  building_data
)

#tranform from sweref to wsg84

buildings_complete <- buildings_complete |> sf::st_transform(crs = 4326)

coordinates <- buildings_complete |> 
  sf::st_coordinates() |> 
  as_tibble() |> 
  rename(
    lng = X,
    lat = Y
  )

buildings <- buildings_complete |> 
  as_tibble() |> 
  select(-geom) |> 
  bind_cols(coordinates) |> 
  mutate(
    byggnadsar_start = str_extract(nybyggnadsar, "[:digit:]+"),
    sitename = if_else(str_detect(sitename, ";"), str_extract(sitename, ".*(?=;)"), sitename),
    sitename = if_else(is.na(sitename), "Okänd", sitename),
    byggnadsar_klart = str_extract(nybyggnadsar, " [:digit:]+"),
    historisk_kategori_smal = str_extract(historisk_kategori, "[:alpha:]+"),
    across(byggnadsar_start:byggnadsar_klart, as.numeric),
    across(c("sitename", "namn"), ~stringr::str_trim(stringr::str_to_title(.)))
  )

#buildings <- readr::read_csv("shiny-app/data/buildings_complete.csv")
#write to csv
#buildings_complete <- readr::read_rds("shiny-app/data/buildings_complete.RDS")

buildings$sitename <- buildings_complete$sitename

buildings <- buildings |>
  mutate(
    #across(c("sitename", "namn"), ~stringr::str_trim(stringr::str_to_title(.))),
    sitename = stringr::str_trim(stringr::str_to_title(sitename)),
    sitename = if_else(str_detect(sitename, ";"), str_extract(sitename, "[^;]*"), sitename),
    sitename = if_else(is.na(sitename), "Okänd", sitename)) 

buildings |> 
  mutate(
    #across(c("sitename", "namn"), ~stringr::str_trim(stringr::str_to_title(.))),
    sitename = stringr::str_trim(stringr::str_to_title(sitename)),
    sitename = if_else(str_detect(sitename, ";"), str_extract(sitename, "[^;]*"), sitename),
    sitename = if_else(is.na(sitename), "Okänd", sitename)) |> 
  filter(str_detect(sitename, ";")) |> 
  pull(sitename)

readr::write_csv(buildings, "buildings_complete.csv")