Support tidyselect expressions in cis_read columns

Author: hmeleiro

DESCRIPTION

@@ -24,7 +24,8 @@ Imports:
     jsonlite,
     rappdirs,
     rlang,
-    tibble
+    tibble,
+    tidyselect
 Suggests:
     devtools,
     pkgdown,

R/read.R

@@ -28,7 +28,8 @@ cis_open <- function() {
 #' @param fecha_min Minimum survey date, included. `NULL` means no lower bound.
 #' @param fecha_max Maximum survey date, included. `NULL` means no upper bound.
 #' @param estudios Optional vector of study codes.
-#' @param cols Optional character vector of columns to select.
+#' @param cols Optional tidyselect expression or character vector of columns to
+#'   select.
 #' @param keep_core_cols If `TRUE`, always include `estudio`, `fecha`, `genero`,
 #'   and `edad` when `cols` is supplied.
 #' @param collect If `TRUE`, return a tibble in memory. If `FALSE`, return a lazy
@@ -47,6 +48,8 @@ cis_open <- function() {
 #'   cols = c("estudio", "fecha", "genero", "edad", "idv", "recuerdo")
 #' )
 #'
+#' cis_read(cols = dplyr::starts_with("val_"))
+#'
 #' cis_read(fecha_min = "2020-01-01", collect = FALSE) |>
 #'   dplyr::count(estudio) |>
 #'   dplyr::collect()
@@ -57,6 +60,7 @@ cis_read <- function(fecha_min = NULL,
                      cols = NULL,
                      keep_core_cols = TRUE,
                      collect = TRUE) {
+    cols <- rlang::enquo(cols)
     cis_check_bool(keep_core_cols, "keep_core_cols")
     cis_check_bool(collect, "collect")
 
@@ -66,10 +70,6 @@ cis_read <- function(fecha_min = NULL,
         cli::cli_abort("{.arg fecha_min} cannot be later than {.arg fecha_max}.")
     }
 
-    if (!is.null(cols) && (!is.character(cols) || anyNA(cols))) {
-        cli::cli_abort("{.arg cols} must be a character vector of column names.")
-    }
-
     ds <- cis_open()
     available_cols <- names(ds)
     selected_cols <- cis_selected_cols(cols, keep_core_cols, available_cols)
@@ -97,10 +97,10 @@ cis_read <- function(fecha_min = NULL,
 }
 
 cis_selected_cols <- function(cols, keep_core_cols, available_cols) {
-    if (is.null(cols)) {
+    if (rlang::quo_is_missing(cols) || identical(rlang::quo_get_expr(cols), NULL)) {
         return(NULL)
     }
-    selected <- unique(cols)
+    selected <- cis_eval_cols(cols, available_cols)
     if (keep_core_cols) {
         selected <- unique(c(.cis_core_cols, selected))
     }
@@ -115,6 +115,36 @@ cis_selected_cols <- function(cols, keep_core_cols, available_cols) {
     selected
 }
 
+cis_eval_cols <- function(cols, available_cols) {
+    expr <- rlang::quo_get_expr(cols)
+    value <- NULL
+    if (is.character(expr) || rlang::is_symbol(expr)) {
+        value <- tryCatch(
+            rlang::eval_tidy(cols),
+            error = function(e) NULL
+        )
+    }
+    if (is.character(value)) {
+        if (anyNA(value)) {
+            cli::cli_abort("{.arg cols} must not contain missing values.")
+        }
+        return(unique(value))
+    }
+
+    data <- stats::setNames(rep(list(logical()), length(available_cols)), available_cols)
+    selected <- tryCatch(
+        tidyselect::eval_select(cols, data = data, allow_rename = FALSE),
+        error = function(e) {
+            cli::cli_abort(c(
+                "Could not evaluate {.arg cols} as a tidyselect expression.",
+                "x" = conditionMessage(e),
+                "i" = "Use column names, helpers such as {.code dplyr::starts_with()}, or {.code dplyr::all_of()}."
+            ))
+        }
+    )
+    names(selected)
+}
+
 #' List available CIS columns
 #'
 #' @return A character vector with column names.

man/cis_read.Rd

@@ -15,7 +15,9 @@ create_test_parquet <- function(path = file.path(local_test_cache(), "cis-longit
         genero = c("Mujer", "Hombre", "Mujer"),
         edad = c(35L, 47L, 52L),
         idv = c("PSOE", "PP", "SUMAR"),
-        recuerdo = c("PSOE", "PP", "UP")
+        recuerdo = c("PSOE", "PP", "UP"),
+        val_min_1 = c(4.5, 5.2, 6.1),
+        val_min_2 = c(3.8, 4.1, 5.0)
     )
     arrow::write_parquet(df, path)
     path

man/cislongitudinal-package.Rd

@@ -7,6 +7,28 @@ test_that("cis_read includes core columns when requested", {
     expect_named(df, c("estudio", "fecha", "genero", "edad", "idv"))
 })
 
+test_that("cis_read accepts tidyselect helpers", {
+    dir <- local_test_cache()
+    create_test_parquet(file.path(dir, "cis-longitudinal.parquet"))
+
+    df <- cis_read(
+        fecha_min = "2023-01-01",
+        cols = c(dplyr::starts_with("val_min"))
+    )
+
+    expect_named(df, c("estudio", "fecha", "genero", "edad", "val_min_1", "val_min_2"))
+})
+
+test_that("cis_read accepts character vectors through all_of", {
+    dir <- local_test_cache()
+    create_test_parquet(file.path(dir, "cis-longitudinal.parquet"))
+    cols <- c("idv", "recuerdo")
+
+    df <- cis_read(cols = dplyr::all_of(cols), keep_core_cols = FALSE)
+
+    expect_named(df, cols)
+})
+
 test_that("cis_read fails with missing columns", {
     dir <- local_test_cache()
     create_test_parquet(file.path(dir, "cis-longitudinal.parquet"))
@@ -60,6 +82,6 @@ test_that("cis_info reports row counts and manifest dates when available", {
 
     info <- cis_info()
     expect_equal(info$rows, 3)
-    expect_equal(info$columns, 6)
+    expect_equal(info$columns, 8)
     expect_equal(info$manifest_updated_at, "2026-06-30T10:27:15+0200")
 })