rewrite get_regional_data so it uses Google when there is no country-specific class #406

NAMESPACE

@@ -43,6 +43,7 @@ export(test_cleaning)
 export(test_download)
 export(test_processing)
 export(test_return)
+export(try_regional_data)
 importFrom(R6,R6Class)
 importFrom(countrycode,countrycode)
 importFrom(countrycode,countryname)

R/try_regional_data.R

@@ -0,0 +1,125 @@
+#'  Try for regional-level data
+#'
+#' @description Provides an interface to source specific classes which
+#' support regional level data, and where these do not exist, seeks regional
+#' level data from JHU or google. For simple use cases this allows downloading
+#' clean, standardised, regional-level COVID-19 data sets. Internally this uses
+#' the `DataClass()` parent class which allows documented downloading, cleaning,
+#' and processing. Optionally all steps of data processing can be returned
+#' along with the functions used for processing but by default just the
+#' finalised processed data is returned. See the examples for some potential
+#' use cases and the links to lower level functions for more details and
+#' options.
+#'
+#' @param country A character string specifying the country to get data from.
+#' Not case dependent. Name should be the English name. For a list of
+#' options use `get_available_datasets()`.
+#' @param include_level_2_regions `r lifecycle::badge("deprecated")` Boolean. If TRUE, returns data stratified by
+#'  level 2 regions. If FALSE, stratified by Level 1. Note that Level 2 region
+#'  data is not always available. In these cases the user will get a warning
+#'  and the Level 1 data will be returned.
+#' @param localise_regions `r lifecycle::badge("deprecated")` Logical, defaults to TRUE. Should region names be localised.
+#' @inheritParams return_data
+#' @inheritParams initialise_dataclass
+#' @return A tibble with data related to cases, deaths, hospitalisations,
+#' recoveries and testing stratified by regions within the given country.
+#' @importFrom lifecycle deprecated is_present deprecate_warn
+#' @family interface
+#' @seealso [Italy()], [UK()]
+#' @export
+#' @examples
+#' \dontrun{
+#' # set up a data cache
+#' start_using_memoise()
+#'
+#' # download data for Italy
+#' try_regional_data("italy")
+#'
+#' # return totals for Italy with no localisation
+#' try_regional_data("italy", localise = FALSE, totals = TRUE)
+#'
+#' # download data for the UK but return the class
+#' uk <- try_regional_data("United Kingdom", class = TRUE)
+#' uk
+#'
+#' # return UK data from the class object]
+#' uk$return()
+#' }
+try_regional_data <- function(country, level = "1", totals = FALSE,
+                              localise = TRUE, steps = FALSE,
+                              class = FALSE, verbose = TRUE, regions,
+                              include_level_2_regions = deprecated(),
+                              localise_regions = deprecated(),
+                              ...) {
+  if (is_present(include_level_2_regions)) {
+    deprecate_warn(
+      "0.9.0",
+      "covidregionaldata::try_regional_data(include_level_2_regions = )", "covidregionaldata::try_regional_data(level = )"
+    )
+    if (include_level_2_regions) {
+      level <- "1"
+    } else {
+      level <- "2"
+    }
+  }
+
+  if (is_present(localise_regions)) {
+    deprecate_warn(
+      "0.9.0",
+      "covidregionaldata::try_regional_data(localise_regions = )", "covidregionaldata::try_regional_data(localise = )"
+    )
+    localise <- localise_regions
+  }
+
+  # construct short hand options
+  title_country <- str_to_title(country)
+  nospace <- str_replace_all(title_country, " ", "")
+  targets <- c(
+    title_country, toupper(title_country), nospace, toupper(nospace)
+  )
+
+  # check we have data for desired class
+  datasets <- covidregionaldata::get_available_datasets("regional")
+  target_class <- bind_rows(
+    filter(datasets, map_lgl(.data$class, ~ any(str_detect(., targets)))),
+    filter(datasets, map_lgl(.data$origin, ~ any(str_detect(., targets))))
+  ) %>%
+    distinct()
+
+  if (nrow(target_class) != 0) {
+    # check data availability and initiate country class if available
+    region_class <- initialise_dataclass(
+      class = country, level = level, regions = regions,
+      totals = totals, localise = localise,
+      verbose = verbose, steps = steps, get = TRUE,
+      type = "regional", ...
+    )
+
+    return(
+      return_data(region_class, class = class)
+    )
+  } else {
+    google_sources <- csv_reader(Google$public_fields$common_data_urls$index)
+
+    google_inputs <-
+      filter(google_sources, map_lgl(.data$country_name, ~ any(str_detect(., targets))))
+    if (nrow(google_inputs) != 0) {
+      message_verbose(verbose = verbose,
+                      "No country-specific data class found. Getting data from Google for ",
+                      country)
+      google_class <- initialise_dataclass(
+        class = "Google", level = sprintf("%d", as.integer(level)+1),
+        totals = totals, localise = TRUE,
+        verbose = verbose, steps = steps,
+        regions = country, get = TRUE,
+        type = "national", ...
+      )
+      return_data(google_class, class = class)
+    } else {
+      stop(
+        "No data available for ", country, ", including from Google.\n",
+        "See get_available_datasets(type = c(\"regional\")) for supported datasets."
+      )
+    }
+  }
+}

R/utils.R

@@ -131,7 +131,7 @@ reset_cache <- function() {
 
 #' Control data return
 #'
-#' @description Controls data return for `get_reigonal_data` and
+#' @description Controls data return for `get_regional_data` and
 #' `get_national_data`
 #' @param obj A Class based on a `DataClass`
 #' @param class Logical, defaults to FALSE. If TRUE returns the