Add function for #154

R/get.R

@@ -34,12 +34,6 @@
 #'   Check Details and Examples in [oe_match()] to understand why this parameter
 #'   is important. Ignored if `place` is not a character vector since the
 #'   matching is performed through a spatial operation.
-#' @param interactive_ask Boolean. If `TRUE` the function creates an interactive
-#'   menu in case the best match is further than `max_string_dist`, otherwise it
-#'   fails with `stop()`. Check details and examples in [oe_match()] to
-#'   understand why this parameter is important. Ignored if `place` is not a
-#'   character vector since the matching is performed through a spatial
-#'   operation.
 #' @param download_directory Where to download the file containing the OSM data?
 #'   By default this is equal to [oe_download_directory()], which is equal to
 #'   [`tempdir()`] and it changes each time you restart R. You can set a
@@ -128,6 +122,9 @@
 #' west_yorkshire = oe_get("West Yorkshire", quiet = FALSE)
 #' # If you run it again, the function will not download the file or convert it
 #' west_yorkshire = oe_get("West Yorkshire", quiet = FALSE)
+#' # Match with place name
+#' oe_get("Milan", quiet = FALSE) # Warning: the .pbf file is 400MB
+#'
 #' # Match with coordinates (any EPSG)
 #' milan_duomo = sf::st_sfc(sf::st_point(c(1514924, 5034552)), crs = 3003)
 #' oe_get(milan_duomo, quiet = FALSE) # Warning: the .pbf file is 400MB
@@ -147,7 +144,6 @@ oe_get = function(
   provider = "geofabrik",
   match_by = "name",
   max_string_dist = 1,
-  interactive_ask = FALSE,
   download_directory = oe_download_directory(),
   force_download = FALSE,
   max_file_size = 5e+8,
@@ -175,7 +171,6 @@ oe_get = function(
     provider = provider,
     match_by = match_by,
     max_string_dist = max_string_dist,
-    interactive_ask = interactive_ask,
     quiet = quiet
   )
 

R/match.R

@@ -67,11 +67,6 @@
 #' # (in which case crs = 4326 is assumed)
 #' oe_match(c(9.1916, 45.4650)) # Milan, Duomo using CRS = 4326
 #'
-#' # Check interactive_ask:
-#' if (interactive()) {
-#'     oe_match("London", interactive_ask = TRUE)
-#' }
-#'
 #' # It returns a warning since Berin is matched both with Benin and Berlin
 #' oe_match("Berin", quiet = FALSE)
 oe_match = function(place, ...) {
@@ -205,7 +200,6 @@ oe_match.character = function(
   quiet = FALSE,
   match_by = "name",
   max_string_dist = 1,
-  interactive_ask = FALSE,
   ...
   ) {
   # For the moment we support only length-one character vectors
@@ -250,6 +244,7 @@ oe_match.character = function(
     ignore.case = TRUE
   )
   best_match_id = which(matching_dists == min(matching_dists, na.rm = TRUE))
+
   if (length(best_match_id) > 1L) {
     warning(
       "The input place was matched with multiple geographical zones: ",
@@ -265,36 +260,74 @@ oe_match.character = function(
   # Check if the best match is still too far
   high_distance = matching_dists[best_match_id, 1] > max_string_dist
 
+  # If the approximate string distance between the best match is greater than
+  # the max_string_dist threshold, then:
   if (isTRUE(high_distance)) {
-    if (isFALSE(quiet) || isTRUE(interactive_ask)) {
+
+    # 1. Raise a message
+    if (isFALSE(quiet)) {
       message(
-        "No exact matching found for place = ", place, ". ",
-        "Best match is ", best_matched_place[[match_by]], "."
+        "No exact match found for place = ", place, " and provider = ",
+        provider, ". ", "Best match is ", best_matched_place[[match_by]], ".",
+        " \nChecking the other providers."
       )
     }
-    if (interactive() && isTRUE(interactive_ask)) {
-      continue = utils::menu(
-        choices = c("Yes", "No"),
-        title = "Do you confirm that this is the right match?"
-      )
-      # since the options are Yes/No, then Yes == 1L
-      if (continue != 1L) {
-        stop("Search for a closer match in the chosen provider's database.",
-             call. = FALSE
+
+    # 2. Check the other providers and, if there is an exact match, just return
+    # the matched value from that other provider:
+    other_providers = setdiff(oe_available_providers(), provider)
+    exact_match = FALSE
+    for (other_provider in other_providers) {
+      if (match_by %!in% colnames(load_provider_data(other_provider))) {
+        next
+      }
+      all_match_by = load_provider_data(other_provider)[[match_by]]
+
+      if (any(tolower(place) == tolower(all_match_by))) {
+        exact_match = TRUE
+        break
+      }
+    }
+
+    if (exact_match) {
+      if (isFALSE(quiet)) {
+        message(
+          "I found an exact string match using provider = ", other_provider,
+          " so I'm going to return that. "
         )
       }
-    } else {
-      stop(
-        "String distance between best match and the input place is ",
-        matching_dists[best_match_id, 1],
-        ", while the maximum threshold distance is equal to ",
-        max_string_dist,
-        ". You should increase the max_string_dist parameter, ",
-        "look for a closer match in the chosen provider database",
-        " or consider using a different match_by variable.",
-        call. = FALSE
+
+      return(
+       oe_match(
+         place = place,
+         provider = other_provider,
+         match_by = match_by,
+         quiet = TRUE,
+         max_string_dist = max_string_dist
+        )
+      )
+    }
+
+    # 3. Otherwise, we can use oe_search to look for the lat/long coordinates of the input place
+    if (isFALSE(quiet)) {
+      message(
+        "No exact match found in any OSM provider data.",
+        " Searching for the location online."
       )
     }
+
+    place_online = oe_search(place = place)
+    # I added Sys.sleep(1) since the usage policty of OSM nominatim (see
+    # https://operations.osmfoundation.org/policies/nominatim/) requires max 1
+    # request per second.
+    Sys.sleep(1)
+    return(
+      oe_match(
+        place = sf::st_geometry(place_online),
+        provider = provider,
+        quiet = quiet
+      )
+    )
   }
 
   if (isFALSE(quiet)) {
@@ -367,7 +400,7 @@ oe_match_pattern = function(
 
   # Then we extract only the elements of the match_by_column that match the
   # input pattern.
-  match_ID = grep(pattern, match_by_column)
+  match_ID = grep(pattern, match_by_column, ignore.case = TRUE)
 
   # If full_row is TRUE than return the corresponding row of provider_data,
   # otherwise just the matched pattern.

R/search.R

@@ -0,0 +1,25 @@
+#' Search for a place and return an sf data frame locating it
+#'
+#' This (at the moment internal) function provides a simple interface to the
+#' [nominatim](https://nominatim.openstreetmap.org) service for finding the
+#' geographical location of place names.
+#'
+#' @param place Text string containing the name of a place the location of
+#'   which is to be found, such as `"Leeds"` or `"Milan"`.
+#' @param base_url The URL of the nominatim server to use. The main
+#'   open server hosted by OpenStreetMap is the default.
+#' @param destfile The name of the destination file where the output
+#'   of the search query, a `.geojson` file, should be saved.
+#' @param ... Extra arguments that are passed to `sf::st_read`.
+oe_search = function(
+  place,
+  base_url = "https://nominatim.openstreetmap.org",
+  destfile = tempfile(fileext = ".geojson"),
+  ...
+  ) {
+  # See https://nominatim.org/release-docs/develop/api/Overview/ for more
+  # details realted to the URL
+  u = paste0(base_url, "/search?q=", place, "&limit=1&format=geojson")
+  utils::download.file(url = u, destfile = destfile, quiet = TRUE)
+  sf::st_read(destfile, quiet = TRUE, ...)
+}

R/update.R

@@ -96,12 +96,14 @@ oe_update = function(
 
   # Check if the .gpkg files should be deleted
   if (isTRUE(delete_gpkg)) {
-    cat("The .gpkg files are going to be removed.\n")
+    if (isFALSE(quiet)) {
+      cat("The .gpkg files are going to be removed.\n")
+    }
     file.remove(
       file.path(download_directory, grep("\\.gpkg", all_files, value = TRUE))
     )
     if (isFALSE(quiet)) {
-      cat("\nThe .gpkg files in download_directory were removed.\n")
+      cat("The .gpkg files in download_directory were removed.\n")
     }
   }
 
@@ -169,6 +171,11 @@ oe_update = function(
       id = paste0("us/", id)
     }
 
+    # Print a message
+    if (isFALSE(quiet)) {
+      message("The function is processing the file ", file, ".")
+    }
+
     # Update the .osm.pbf files, skipping the vectortranslate step
     oe_get(
       place = id,
@@ -178,6 +185,7 @@ oe_update = function(
       download_only = TRUE,
       skip_vectortranslate = TRUE,
       max_file_size = max_file_size,
+      quiet = quiet,
       ...
     )
   }

tests/testthat/test-find.R

@@ -2,15 +2,17 @@ test_that("oe_find: simplest example works", {
   its_leeds_find <- oe_find(
     "ITS Leeds",
     provider = "test",
-    download_if_missing = TRUE
+    download_if_missing = TRUE,
+    quiet = TRUE
   )
   expect_type(its_leeds_find, "character")
 
   file.remove(its_leeds_find)
   its_leeds_find <- oe_find(
     "ITS Leeds",
     provider = "test",
-    download_if_missing = TRUE
+    download_if_missing = TRUE,
+    quiet = TRUE
   )
   expect_type(its_leeds_find, "character")
   expect_length(its_leeds_find, 2)

tests/testthat/test-get.R

@@ -1,5 +1,5 @@
 test_that("oe_get: simplest examples work", {
-  expect_s3_class(oe_get("ITS Leeds", provider = "test"), "sf")
+  expect_s3_class(oe_get("ITS Leeds", provider = "test", quiet = TRUE), "sf")
 })
 
 test_that("vectortranslate is not skipped if force_download is TRUE", {
@@ -17,13 +17,16 @@ test_that("vectortranslate is not skipped if force_download is TRUE", {
     "-lco", "GEOMETRY_NAME=geometry",
     "-where", "highway IN ('service')",
     "lines"
-  ))
+    ),
+    quiet = TRUE
+  )
 
   # Download it again
   its_leeds <- oe_get(
     "ITS Leeds",
     download_directory = my_tempdir,
     force_vectortranslate = TRUE,
+    quiet = TRUE
   )
 
   expect_gte(nrow(its_leeds), nrow(small_its_leeds))

tests/testthat/test-match.R

@@ -50,12 +50,14 @@ test_that("oe_match: numeric input", {
   expect_match(oe_match(c(9.1916, 45.4650))$url, "italy")
 })
 
-test_that("oe_match: different providers, match_by or max_string dist args", {
+test_that("oe_match: different providers, match_by or max_string_dist args", {
   expect_error(oe_match("Italy", provider = "XXX"))
   expect_error(oe_match("Italy", match_by = "XXX"))
   expect_match(oe_match("RU", match_by = "iso3166_1_alpha2")$url, "russia")
 
-  expect_error(oe_match("Isle Wight"))
+  # expect_null(oe_match("Isle Wight"))
+  # The previous test was removed in #155 since now oe_match calls nominatim servers in
+  # case it doesn't find an exact match, so it should never return NULL
   expect_match(oe_match("Isle Wight", max_string_dist = 3)$url, "isle-of-wight")
   expect_message(oe_match("London", max_string_dist = 3, quiet = FALSE))