⭐ Add new find_nearest_coord as a new function

NAMESPACE

@@ -1,6 +1,7 @@
 # Generated by roxygen2: do not edit by hand
 
 export(detect_date_format)
+export(find_nearest_coord)
 export(get_multi_ona_data)
 export(get_ona_data)
 export(get_updated_ona_data)
@@ -13,6 +14,7 @@ import(doFuture)
 import(dplyr)
 import(foreach)
 import(future)
+import(sf)
 importFrom(cli,cli_alert_warning)
 importFrom(cli,cli_h2)
 importFrom(cli,cli_text)

R/find_nearest_coord.R

@@ -0,0 +1,58 @@
+#' Find Nearest Coordinate from shapefile Dataset
+#'
+#' This function takes a pair of coordinates (longitude and latitude) and finds
+#' the nearest coordinate from the `shp_global` dataset.
+#'
+#' The shapefile dataset should contain longitude and latitude information
+#' with columns named `CENTER_LON` and `CENTER_LAT`. The function calculates the
+#' distance between the target point and all points in the dataset, returning
+#' the nearest coordinate.
+#'
+#' @param lon Numeric. Longitude of the target point.
+#' @param lat Numeric. Latitude of the target point.
+#' @param shapefile An optional `sf` object containing the shapefile data. 
+#'      Defaults to `poliprep::shp_global`.
+#' @param level_to_return The admin level to return.
+#' @return A vector of admin level names containing the nearest coordinate from
+#'   the provided shapefile.
+#' @examples
+#' # Example usage
+#' # Assuming `poliprep::shp_global` is already loaded in the environment
+#' nearest_coord <- find_nearest_coord(15.28172, -4.271301)
+#' print(nearest_coord)
+#' @export
+# Define the function
+find_nearest_coord <- function(lon, lat, 
+                               shapefile = NULL,
+                               level_to_return = "adm2") {
+
+  # Use default shapefile if none provided
+  if (is.null(shapefile)) {
+    shapefile <- poliprep::shp_global
+  }
+
+  # Prepare the `shp_global` dataset
+  res <- shapefile |>
+    dplyr::filter(ENDDATE > as.Date("9900-12-31")) |> 
+    dplyr::rename(adm0 = ADM0_NAME,
+                  adm1 = ADM1_NAME,
+                  adm2 = ADM2_NAME) |> 
+    dplyr::mutate(
+      CENTER_LON = as.numeric(CENTER_LON),
+      CENTER_LAT = as.numeric(CENTER_LAT)
+    ) |>
+    sf::st_as_sf(coords = c("CENTER_LON", "CENTER_LAT"), crs = 4326)
+
+  # Define the target point
+  target_point <- sf::st_sfc(
+    sf::st_point(c(lon, lat)), crs = 4326
+  ) |>
+    sf::st_sf()
+
+  # Find the nearest coordinate
+  distances <- sf::st_distance(target_point, res)
+  nearest_index <- which.min(distances)
+  nearest_coord <- res[nearest_index, ][[level_to_return]]
+
+  return(as.vector(nearest_coord))
+}

R/globals.R

@@ -8,7 +8,8 @@ globalVariables(
     "distance", "longname_corrected", "longname_corrected2",
     "match_rank", "matched_names",  "form_id_num", "last_date_in_chunk", "x",
     "date_last_updated", "form_id", "name_of_creator", "created_time", 
-    "user_name",  "ADM0_NAME", "ADM1_NAME", "ADM2_NAME", "ENDDATE"
+    "user_name",  "ADM0_NAME", "ADM1_NAME", "ADM2_NAME", "ENDDATE",
+    "CENTER_LAT", "CENTER_LON"
     )
   )
 

R/prep_geonames.R

@@ -616,9 +616,9 @@ construct_geo_names <- function(data, level0, level1, level2) {
 #'        \code{"soundex"}.
 #' @param stratify Logical; if TRUE, performs cleaning stratified by
 #'        admin levels to maintain hierarchical consistency.
-#' @param user_name name of user to save onto the cache file. 
-#'        Very important for future auditing if cache file is being used 
-#'        by many. Make sure it's recognisable. 
+#' @param user_name Full name of user to save onto the cache file. 
+#'        Very important for future auditing specially if cache file is being used 
+#'        by many.
 #' @param non_interactive Whether to use the interactivity or not. 
 #' @details
 #' The function performs the following steps:

tests/testthat/test-find_nearest_coord.R

@@ -0,0 +1,22 @@
+
+# set up mocek shapefile
+mock_data <- data.frame(
+  CENTER_LON = c(-4.222, -3.2, -24.5),
+  CENTER_LAT = c(15.02222, 43.2, 23.5),
+  ENDDATE = c("9999-12-31 01:00:00", "9999-12-31 01:00:00", 
+              "9999-12-31 01:00:00"),
+  ADM0_NAME = c("Mock Area 1", "Mock Area 2", "Mock Area 3"),
+  ADM1_NAME = c("Mock Area 1", "Mock Area 2", "Mock Area 3"),
+  ADM2_NAME = c("Mock Area 1", "Mock Area 2", "Mock Area 3") 
+) 
+
+# Test cases
+testthat::test_that("find_nearest_coord returns correct area", {
+  result <- find_nearest_coord(shapefile = mock_data,  -24.0, 15.9)
+  testthat::expect_equal(result, "Mock Area 3")
+
+  result2 <- find_nearest_coord(
+    lon = -24.0324, lat = 15.93214, level_to_return = "adm2")
+  testthat::expect_equal(result2, "RIBERIA BRAVA")
+
+})