Make compute_parquet and compute_csv generic S3 functions

.gitignore

@@ -3,3 +3,5 @@
 /docs
 inst/doc
 README_cache/
+*.tar.gz
+*.Rcheck/

NAMESPACE

@@ -21,6 +21,10 @@ S3method(auto_copy,duckplyr_df)
 S3method(collect,duckplyr_df)
 S3method(collect,prudent_duckplyr_df)
 S3method(compute,duckplyr_df)
+S3method(compute_csv,data.frame)
+S3method(compute_csv,duckplyr_df)
+S3method(compute_parquet,data.frame)
+S3method(compute_parquet,duckplyr_df)
 S3method(count,duckplyr_df)
 S3method(cross_join,duckplyr_df)
 S3method(distinct,duckplyr_df)

R/compute_csv.R

@@ -1,19 +1,16 @@
 #' Compute results to a CSV file
 #'
-#' For a duckplyr frame, this function executes the query
-#' and stores the results in a CSV file,
-#' without converting it to an R data frame.
+#' This is a generic function that executes a query
+#' and stores the results in a CSV file.
+#' For a duckplyr frame, the materialization occurs outside of R.
 #' The result is a duckplyr frame that can be used with subsequent dplyr verbs.
-#' This function can also be used as a CSV writer for regular data frames.
 #'
 #' @inheritParams rlang::args_dots_empty
-#' @inheritParams compute.duckplyr_df
-#' @inheritParams compute_parquet
-#' @param options A list of additional options to pass to create the storage format,
-#'   see <https://duckdb.org/docs/sql/statements/copy.html#csv-options>
-#'   for details.
+#' @param x A data frame or lazy data frame.
+#' @param path The path of the CSV file to create.
+#' @param ... Additional arguments passed to methods.
 #'
-#' @return A duckplyr frame.
+#' @return A data frame (the class may vary based on the input).
 #'
 #' @export
 #' @examples
@@ -24,7 +21,18 @@
 #' df <- compute_csv(df, path)
 #' readLines(path)
 #' @seealso [compute_parquet()], [compute.duckplyr_df()], [dplyr::collect()]
-compute_csv <- function(x, path, ..., prudence = NULL, options = NULL) {
+compute_csv <- function(x, path, ...) {
+  UseMethod("compute_csv")
+}
+
+#' @inheritParams compute.duckplyr_df
+#' @param options A list of additional options to pass to create the storage format,
+#'   see <https://duckdb.org/docs/sql/statements/copy.html#csv-options>
+#'   for details.
+#'
+#' @rdname compute_csv
+#' @export
+compute_csv.duckplyr_df <- function(x, path, ..., prudence = NULL, options = NULL) {
   check_dots_empty()
 
   if (is.null(options)) {
@@ -46,3 +54,10 @@ compute_csv <- function(x, path, ..., prudence = NULL, options = NULL) {
 
   read_csv_duckdb(path, prudence = prudence)
 }
+
+#' @rdname compute_csv
+#' @export
+compute_csv.data.frame <- function(x, path, ..., prudence = NULL, options = NULL) {
+  x <- as_duckdb_tibble(x)
+  compute_csv.duckplyr_df(x, path, ..., prudence = prudence, options = options)
+}

R/compute_parquet.R

@@ -1,20 +1,16 @@
 #' Compute results to a Parquet file
 #'
-#' For a duckplyr frame, this function executes the query
-#' and stores the results in a Parquet file,
-#' without converting it to an R data frame.
+#' This is a generic function that executes a query
+#' and stores the results in a Parquet file.
+#' For a duckplyr frame, the materialization occurs outside of R.
 #' The result is a duckplyr frame that can be used with subsequent dplyr verbs.
-#' This function can also be used as a Parquet writer for regular data frames.
 #'
 #' @inheritParams rlang::args_dots_empty
-#' @inheritParams compute.duckplyr_df
-#' @param x A duckplyr frame.
+#' @param x A data frame or lazy data frame.
 #' @param path The path of the Parquet file to create.
-#' @param options A list of additional options to pass to create the Parquet file,
-#'   see <https://duckdb.org/docs/sql/statements/copy.html#parquet-options>
-#'   for details.
+#' @param ... Additional arguments passed to methods.
 #'
-#' @return A duckplyr frame.
+#' @return A data frame (the class may vary based on the input).
 #'
 #' @export
 #' @examples
@@ -25,7 +21,18 @@
 #' df <- compute_parquet(df, path)
 #' explain(df)
 #' @seealso [compute_csv()], [compute.duckplyr_df()], [dplyr::collect()]
-compute_parquet <- function(x, path, ..., prudence = NULL, options = NULL) {
+compute_parquet <- function(x, path, ...) {
+  UseMethod("compute_parquet")
+}
+
+#' @inheritParams compute.duckplyr_df
+#' @param options A list of additional options to pass to create the Parquet file,
+#'   see <https://duckdb.org/docs/sql/statements/copy.html#parquet-options>
+#'   for details.
+#'
+#' @rdname compute_parquet
+#' @export
+compute_parquet.duckplyr_df <- function(x, path, ..., prudence = NULL, options = NULL) {
   check_dots_empty()
 
   if (is.null(options)) {
@@ -47,3 +54,10 @@ compute_parquet <- function(x, path, ..., prudence = NULL, options = NULL) {
 
   read_parquet_duckdb(path, prudence = prudence)
 }
+
+#' @rdname compute_parquet
+#' @export
+compute_parquet.data.frame <- function(x, path, ..., prudence = NULL, options = NULL) {
+  x <- as_duckdb_tibble(x)
+  compute_parquet.duckplyr_df(x, path, ..., prudence = prudence, options = options)
+}

tests/testthat/test-compute_csv.R

@@ -15,3 +15,19 @@ test_that("compute_csv() prudence", {
   expect_true(is_prudent_duckplyr_df(out))
   expect_identical(collect(out), as_tibble(df))
 })
+
+test_that("compute_csv() is a generic function", {
+  expect_true(is.function(compute_csv))
+  m <- methods("compute_csv")
+  expect_true(any(grepl("compute_csv.duckplyr_df", m)))
+  expect_true(any(grepl("compute_csv.data.frame", m)))
+})
+
+test_that("compute_csv() with duckplyr_df", {
+  df <- duckdb_tibble(x = c(1, 2))
+  withr::defer(unlink("test_duck.csv"))
+  out <- compute_csv(df, path = "test_duck.csv")
+
+  expect_identical(collect(out), collect(df))
+  expect_true(inherits(out, "duckplyr_df"))
+})

tests/testthat/test-compute_parquet.R

@@ -16,3 +16,19 @@ test_that("compute_parquet() with options", {
   expect_identical(out, as_duckdb_tibble(df))
   expect_false(is_prudent_duckplyr_df(out))
 })
+
+test_that("compute_parquet() is a generic function", {
+  expect_true(is.function(compute_parquet))
+  m <- methods("compute_parquet")
+  expect_true(any(grepl("compute_parquet.duckplyr_df", m)))
+  expect_true(any(grepl("compute_parquet.data.frame", m)))
+})
+
+test_that("compute_parquet() with duckplyr_df", {
+  df <- duckdb_tibble(x = c(1, 2))
+  withr::defer(unlink("test_duck.parquet"))
+  out <- compute_parquet(df, path = "test_duck.parquet")
+
+  expect_identical(collect(out), collect(df))
+  expect_true(inherits(out, "duckplyr_df"))
+})