feat: support for core languages and additional data cleaning

R/wb.R

@@ -20,15 +20,17 @@ wb_lang <- function() {
 #' World Bank lending type data
 #'
 #' @param type character(1) lending type to query. Default is "all".
+#' @param lang character(1) language to query. Default is "en".
 #' @export
 #' @examples
 #' \dontrun{
 #' wb_lending_type()
 #' }
-wb_lending_type <- function(type = "all") {
+wb_lending_type <- function(type = "all", lang = "en") {
   stopifnot(is.character(type) && length(type) == 1)
+  lang <- match.arg(lang, c("en", "fr", "es", "ar", "zh"))
 
-  resource <- sprintf("lendingType/%s", type)
+  resource <- sprintf("%s/lendingType/%s", lang, type)
   res <- worldbank(resource, \(resp) {
     data <- resp_body_json(resp)[[2]]
     data.frame(
@@ -43,15 +45,17 @@ wb_lending_type <- function(type = "all") {
 #' World Bank income level data
 #'
 #' @param income character(1) income level to query. Default is "all".
+#' @param lang character(1) language to query. Default is "en".
 #' @export
 #' @examples
 #' \dontrun{
 #' wb_income_lvl()
 #' }
-wb_income_lvl <- function(income = "all") {
+wb_income_lvl <- function(income = "all", lang = "en") {
   stopifnot(is.character(income) && length(income) == 1)
+  lang <- match.arg(lang, c("en", "fr", "es", "ar", "zh"))
 
-  resource <- sprintf("incomeLevel/%s", income)
+  resource <- sprintf("%s/incomeLevel/%s", lang, income)
   res <- worldbank(resource, \(resp) {
     data <- resp_body_json(resp)[[2]]
     data.frame(
@@ -66,108 +70,112 @@ wb_income_lvl <- function(income = "all") {
 #' World Bank source data
 #'
 #' @param source character(1) source to query. Default is "all".
-#' @param page integer(1) page number to query. Default is NULL.
+#' @param lang character(1) language to query. Default is "en".
 #' @export
 #' @examples
 #' \dontrun{
 #' wb_source()
 #' }
-wb_source <- function(source = "all", page = NULL) {
+wb_source <- function(source = "all", lang = "en") {
   stopifnot(is.character(source) && length(source) == 1)
-  stopifnot(is_page(page))
+  lang <- match.arg(lang, c("en", "fr", "es", "ar", "zh"))
 
-  resource <- sprintf("source/%s", source)
+  resource <- sprintf("%s/source/%s", lang, source)
   res <- worldbank(resource, \(resp) {
     data <- resp_body_json(resp)[[2]]
     data.frame(
-      id = map_chr(data, "id"),
-      last_updated = map_chr(data, "lastupdated"),
-      name = map_chr(data, "name"),
+      id = map_chr(data, "id") |> as.integer(),
+      last_updated = map_chr(data, "lastupdated") |> as.Date(),
+      name = map_chr(data, "name") |> na_if_empty(),
       code = map_chr(data, "code"),
       description = map_chr(data, "description") |> na_if_empty(),
       url = map_chr(data, "url") |> na_if_empty(),
-      data_availability = map_chr(data, "dataavailability"),
-      metadata_availability = map_chr(data, "metadataavailability"),
-      concepts = map_chr(data, "concepts")
+      data_availability = map_chr(data, "dataavailability") |> to_logical(),
+      metadata_availability = map_chr(data, "metadataavailability") |>
+        to_logical(),
+      concepts = map_chr(data, "concepts") |> as.integer()
     )
-  }, page = page)
+  }, page = 1)
   as_tibble(res)
 }
 
 #' World Bank topic data
 #'
 #' @param topic character(1) topic to query. Default is "all".
-#' @param page integer(1) page number to query. Default is NULL.
+#' @param lang character(1) language to query. Default is "en".
 #' @export
 #' @examples
 #' \dontrun{
 #' wb_topic()
 #' }
-wb_topic <- function(topic = "all", page = NULL) {
+wb_topic <- function(topic = "all", lang = "en") {
   stopifnot(is.character(topic) && length(topic) == 1)
-  stopifnot(is_page(page))
+  lang <- match.arg(lang, c("en", "fr", "es", "ar", "zh"))
 
-  resource <- sprintf("topic/%s", topic)
+  resource <- sprintf("%s/topic/%s", lang, topic)
   res <- worldbank(resource, \(resp) {
     data <- resp_body_json(resp)[[2]]
     data.frame(
-      id = map_chr(data, "id"),
+      id = map_chr(data, "id") |> as.integer(),
       value = map_chr(data, "value"),
       source_note = map_chr(data, "sourceNote")
     )
-  }, page = page)
+  }, page = 1)
   as_tibble(res)
 }
 
 #' World Bank region data
 #'
 #' @param region character(1) region to query. Default is "all".
-#' @param page integer(1) page number to query. Default is NULL.
+#' @param lang character(1) language to query. Default is "en".
 #' @export
 #' @examples
 #' \dontrun{
 #' wb_region()
 #' }
-wb_region <- function(region = "all", page = NULL) {
+wb_region <- function(region = "all", lang = "en") {
   stopifnot(is.character(region) && length(region) == 1)
-  stopifnot(is_page(page))
+  lang <- match.arg(lang, c("en", "fr", "es", "ar", "zh"))
 
-  resource <- sprintf("region/%s", region)
+  resource <- sprintf("%s/region/%s", lang, region)
   res <- worldbank(resource, \(resp) {
     data <- resp_body_json(resp)[[2]]
     data.frame(
       id = map_chr(data, "id") |> na_if_empty(),
       code = map_chr(data, "code"),
       iso2code = map_chr(data, "iso2code"),
-      name = map_chr(data, "name")
+      name = map_chr(data, "name") |> na_if_empty() |> trimws()
     )
-  }, page = page)
+  }, page = 1)
   as_tibble(res)
 }
 
 #' World Bank country data
 #'
 #' @param country character(1) country to query.
-#' @param page integer(1) page number to query. Default is NULL.
+#' @param lang character(1) language to query. Default is "en".
 #' @export
 #' @examples
 #' \dontrun{
 #' wb_country()
 #' }
-wb_country <- function(country = "all", page = NULL) {
+wb_country <- function(country = "all", lang = "en") {
   stopifnot(is.character(country) && length(country) == 1)
+  lang <- match.arg(lang, c("en", "fr", "es", "ar", "zh"))
   stopifnot(is_page(page))
 
-  resource <- sprintf("country/%s", country)
+  resource <- sprintf("%s/country/%s", lang, country)
   res <- worldbank(resource, \(resp) {
     data <- resp_body_json(resp)[[2]]
     data.frame(
       country_id = map_chr(data, "id"),
       country_code = map_chr(data, "iso2Code"),
-      country_name = map_chr(data, "name"),
+      country_name = map_chr(data, "name") |> na_if_empty(),
       region_id = map_chr(data, \(x) x$region$id),
       region_code = map_chr(data, \(x) x$region$iso2code),
-      region_value = map_chr(data, \(x) x$region$value) |> trimws(),
+      region_value = map_chr(data, \(x) x$region$value) |>
+        trimws() |>
+        na_if_empty(),
       admin_region_id = map_chr(data, \(x) x$adminregion$id) |> na_if_empty(),
       admin_region_code = map_chr(data, \(x) x$adminregion$iso2code) |>
         na_if_empty(),
@@ -179,29 +187,32 @@ wb_country <- function(country = "all", page = NULL) {
       lending_type_id = map_chr(data, \(x) x$lendingType$id) |> na_if_empty(),
       lending_type_code = map_chr(data, \(x) x$lendingType$iso2code) |>
         na_if_empty(),
-      lending_type_value = map_chr(data, \(x) x$lendingType$value),
+      lending_type_value = map_chr(data, \(x) x$lendingType$value) |>
+        na_if_empty(),
       capital_city = map_chr(data, "capitalCity") |> na_if_empty(),
       longitude = map_chr(data, "longitude") |> na_if_empty() |> as.numeric(),
       latitude = map_chr(data, "latitude") |> na_if_empty() |> as.numeric()
     )
-  }, page = page)
+  }, page = 1)
   as_tibble(res)
 }
 
 #' World Bank indicator data
 #'
 #' @param indicator character(1) indicator to query. Default is "all".
+#' @param lang character(1) language to query. Default is "en".
 #' @param page integer(1) page number to query. Default is NULL.
 #' @export
 #' @examples
 #' \dontrun{
 #' wb_indicator()
 #' }
-wb_indicator <- function(indicator = "all", page = NULL) {
+wb_indicator <- function(indicator = "all", lang = "en", page = NULL) {
   stopifnot(is.character(indicator) && length(indicator) == 1)
+  lang <- match.arg(lang, c("en", "fr", "es", "ar", "zh"))
   stopifnot(is_page(page))
 
-  resource <- sprintf("indicator/%s", indicator)
+  resource <- sprintf("%s/indicator/%s", lang, indicator)
   res <- worldbank(resource, \(resp) {
     data <- resp_body_json(resp)[[2]]
     data.frame(
@@ -236,6 +247,7 @@ wb_indicator <- function(indicator = "all", page = NULL) {
 #'
 #' @param indicator character(1) indicator to query.
 #' @param country character(1) country to query.
+#' @param lang character(1) language to query. Default is "en".
 #' @param page integer(1) page number to query. Default is NULL.
 #' @export
 #' @examples
@@ -244,14 +256,16 @@ wb_indicator <- function(indicator = "all", page = NULL) {
 #' }
 wb_country_indicator <- function(indicator = "NY.GDP.MKTP.CD",
                                  country = "US",
+                                 lang = "en",
                                  page = NULL) {
   stopifnot(is.character(indicator) && length(indicator) == 1)
   stopifnot(
     is.character(country) && length(country) == 1 && nchar(country) %in% 2:3
   )
+  lang <- match.arg(lang, c("en", "fr", "es", "ar", "zh"))
   stopifnot(is_page(page))
 
-  resource <- sprintf("country/%s/indicator/%s", country, indicator)
+  resource <- sprintf("%s/country/%s/indicator/%s", lang, country, indicator)
   res <- worldbank(resource, \(resp) {
     data <- resp_body_json(resp)[[2]]
     data <- lapply(data, \(x) {
@@ -271,7 +285,11 @@ wb_country_indicator <- function(indicator = "NY.GDP.MKTP.CD",
         decimal = x$decimal
       )
     })
-    do.call(rbind, data)
+    data <- do.call(rbind, data)
+    data$year <- as.integer(data$year)
+    data$unit <- na_if_empty(data$unit)
+    data$obs_status <- na_if_empty(data$obs_status)
+    data
   }, page = page)
   as_tibble(res)
 }