R/geocode_google.R

In lvaudor/mixr: Functions to simplify text mining with multiple languages

#' Geocodes one location through the Google API
#' @param stringlocation a location string
#' @param info the variables to inform: by default, c("lat","lng").
#' Information can be e.g. "locality","country","country_code","formatted_address".
#' It can also be any type supported by the Google Map geocoding API:
#' see [Address types](https://developers.google.com/maps/documentation/geocoding/intro#Types).
#' @return tibble with additional columns corresponding to the info required
#' @export
#' @importFrom rlang .data
#' @examples
#' geocode_google("Grande Rue de la Guillotière, Lyon",
#'  info=c("lat",
#'         "lng",
#'         "country_code",
#'         "locality",
#'         "administrative_area_level_1"))
geocode_google=function(stringlocation,info=c("lat","lng")){
  if(stringlocation==""|is.na(stringlocation)){return(make_empty_result(info=info))}
  url_base <- "https://maps.googleapis.com/maps/api/geocode/json?"
  # API response
  url_query <- stringlocation %>%
    stringr::str_trim() %>%
    stringr::str_replace_all(" +","+") %>%
    utils::URLencode(reserved = FALSE) %>%
    c(address = .) %>%
    c(.,key = Sys.getenv("GOOGLE_KEY")) %>%
    stringr::str_c(names(.),., sep = "=",collapse = "&") %>%
    stringr::str_c(url_base,.)
  repcontent=httr::GET(url_query) %>%
    httr::content()
  if(repcontent$status=="ZERO_RESULTS"){
    result=make_empty_result(info)
  }
  if(repcontent$status!="ZERO_RESULTS"){
    represult=repcontent$results[[1]]
    if(any(info %in% c("lat","lng"))){
      mylat=represult$geometry$location$lat
      mylng=represult$geometry$location$lng
    }
    address_components=represult$address_components %>%
      purrr::map_df(tibble::as_tibble) %>%
      dplyr::mutate(types=unlist(.data$types))
    result=dplyr::bind_rows(
      address_components,
      tibble::tibble(long_name=represult$formatted_address,
                     short_name=NA,
                     types="formatted_address"),
      tibble::tibble(long_name=address_components %>%
                       dplyr::filter(.data$types=="country") %>%
                       dplyr::pull(.data$short_name),
                     short_name=NA_character_,
                     types="country_code")) %>%
      dplyr::select(name=.data$long_name,
                    types=.data$types) %>%
      dplyr::right_join(tibble::tibble(types=info),
                        by="types") %>%
      dplyr::select(.data$types,
                    .data$name) %>%
      tibble::deframe() %>%
      dplyr::bind_rows() %>%
      dplyr::mutate(stringlocation=stringlocation) %>%
      dplyr::select(.data$stringlocation,
                    dplyr::everything())
    if(any(info=="lng")){
      result=result %>%
        dplyr::mutate(lng=mylng) %>%
        dplyr::select(.data$stringlocation,
                      .data$lng,
                      dplyr::everything())
    }
    if(any(info=="lat")){
      result=result %>%
        dplyr::mutate(lat=mylat) %>%
        dplyr::select(.data$stringlocation,
                      .data$lat,
                      dplyr::everything())
    }
    return(result)
  }
}