$sink_*() and $write_*() functions invisibly return the input

NEWS.md

@@ -205,6 +205,8 @@
   - Export the `Duration` datatype (#955).
   - New active binding `<Series>$struct$fields` (#1002).
   - rust-polars is updated to 0.39.0 (#937, #1034).
+  - All `$write_*()` and `$sink_*()` functions now invisibly return the input
+    data (#1039).
 
 
 ### Bug fixes

R/dataframe__frame.R

@@ -1970,7 +1970,7 @@ DataFrame_write_csv = function(
   ) |>
     unwrap("in $write_csv():")
 
-  invisible(NULL)
+  invisible(self)
 }
 
 
@@ -2009,7 +2009,7 @@ DataFrame_write_ipc = function(
   ) |>
     unwrap("in $write_ipc():")
 
-  invisible(NULL)
+  invisible(self)
 }
 
 
@@ -2044,7 +2044,7 @@ DataFrame_write_parquet = function(
   ) |>
     unwrap("in $write_parquet():")
 
-  invisible(NULL)
+  invisible(self)
 }
 
 #' Write to JSON file
@@ -2075,7 +2075,7 @@ DataFrame_write_json = function(
   .pr$DataFrame$write_json(self, file, pretty, row_oriented) |>
     unwrap("in $write_json():")
 
-  invisible(NULL)
+  invisible(self)
 }
 
 #' Write to NDJSON file
@@ -2096,7 +2096,7 @@ DataFrame_write_ndjson = function(file) {
   .pr$DataFrame$write_ndjson(self, file) |>
     unwrap("in $write_ndjson():")
 
-  invisible(NULL)
+  invisible(self)
 }
 
 #' @inherit LazyFrame_rolling title description params details

R/lazyframe__lazy.R

@@ -596,6 +596,7 @@ LazyFrame_collect_in_background = function() {
 #' @inheritParams LazyFrame_collect
 #'
 #' @rdname IO_sink_parquet
+#' @return Invisibly returns the input LazyFrame
 #'
 #' @examples
 #' # sink table 'mtcars' from mem to parquet
@@ -655,8 +656,9 @@ LazyFrame_sink_parquet = function(
       data_pagesize_limit,
       maintain_order
     ) |>
-    unwrap("in $sink_parquet()") |>
-    invisible()
+    unwrap("in $sink_parquet()")
+
+  invisible(self)
 }
 
 
@@ -674,6 +676,8 @@ LazyFrame_sink_parquet = function(
 #' @inheritParams LazyFrame_group_by
 #' @inheritParams DataFrame_unique
 #'
+#' @inherit LazyFrame_sink_parquet return
+#'
 #' @rdname IO_sink_ipc
 #'
 #' @examples
@@ -726,8 +730,9 @@ LazyFrame_sink_ipc = function(
       compression %||% "uncompressed",
       maintain_order
     ) |>
-    unwrap("in $sink_ipc()") |>
-    invisible()
+    unwrap("in $sink_ipc()")
+
+  invisible(self)
 }
 
 
@@ -743,6 +748,7 @@ LazyFrame_sink_ipc = function(
 #' @inheritParams LazyFrame_group_by
 #' @inheritParams DataFrame_unique
 #'
+#' @inherit LazyFrame_sink_parquet return
 #' @rdname IO_sink_csv
 #'
 #' @examples
@@ -817,8 +823,9 @@ LazyFrame_sink_csv = function(
       quote_style,
       maintain_order
     ) |>
-    unwrap("in $sink_csv()") |>
-    invisible()
+    unwrap("in $sink_csv()")
+
+  invisible(self)
 }
 
 
@@ -834,6 +841,7 @@ LazyFrame_sink_csv = function(
 #' @inheritParams LazyFrame_group_by
 #' @inheritParams DataFrame_unique
 #'
+#' @inherit LazyFrame_sink_parquet return
 #' @rdname IO_sink_ndjson
 #'
 #' @examples
@@ -880,8 +888,9 @@ LazyFrame_sink_ndjson = function(
       path,
       maintain_order
     ) |>
-    unwrap("in $sink_ndjson()") |>
-    invisible()
+    unwrap("in $sink_ndjson()")
+
+  invisible(self)
 }
 
 

tests/testthat/test-csv-write.R

@@ -13,6 +13,11 @@ test_that("write_csv: path works", {
   )
 })
 
+test_that("write_csv returns the input data", {
+  x = dat_pl$write_csv(temp_out)
+  expect_identical(x$to_list(), dat_pl$to_list())
+})
+
 test_that("write_csv: null_values works", {
   expect_grepl_error(
     dat_pl$write_csv(temp_out, null_values = NULL)

tests/testthat/test-ipc.R

@@ -67,3 +67,10 @@ patrick::with_parameters_test_that("write and read Apache Arrow file",
   compression = list(NULL, "uncompressed", "lz4", "zstd"),
   .test_name = compression
 )
+
+test_that("write_ipc returns the input data", {
+  dat = pl$DataFrame(mtcars)
+  tmpf = tempfile(fileext = ".arrow")
+  x = dat$write_ipc(tmpf)
+  expect_identical(x$to_list(), dat$to_list())
+})

tests/testthat/test-json_write.R

@@ -23,3 +23,10 @@ test_that("write_ndjson: path works", {
     ignore_attr = TRUE # rownames are lost when writing / reading from json
   )
 })
+
+test_that("write_ndjson returns the input data", {
+  dat = pl$DataFrame(mtcars)
+  tmpf = tempfile(fileext = ".arrow")
+  x = dat$write_ndjson(tmpf)
+  expect_identical(x$to_list(), dat$to_list())
+})

tests/testthat/test-parquet.R

@@ -84,3 +84,10 @@ test_that("throw error if invalid compression is passed", {
     "Failed to set parquet compression method"
   )
 })
+
+test_that("write_parquet returns the input data", {
+  dat = pl$DataFrame(mtcars)
+  tmpf = tempfile()
+  x = dat$write_parquet(tmpf)
+  expect_identical(x$to_list(), dat$to_list())
+})

tests/testthat/test-sink_stream.R

@@ -7,9 +7,13 @@ test_that("Test sinking data to parquet file", {
   expect_grepl_error(lf$sink_parquet(tmpf, compression = "rar"))
   lf$sink_parquet(tmpf)
   expect_equal(pl$scan_parquet(tmpf)$collect()$to_data_frame(), rdf)
+
+  # return the input data
+  x = lf$sink_parquet(tmpf)
+  expect_identical(x$collect()$to_list(), lf$collect()$to_list())
 })
 
-test_that("Test sinking data to parquet file", {
+test_that("Test sinking data to IPC file", {
   tmpf = tempfile()
   on.exit(unlink(tmpf))
   lf$sink_ipc(tmpf)
@@ -51,6 +55,10 @@ test_that("Test sinking data to parquet file", {
       expect_identical(rdf_in_bg$to_data_frame(), rdf)
     }
   )
+
+  # return the input data
+  x = lf$sink_ipc(tmpf)
+  expect_identical(x$collect()$to_list(), lf$collect()$to_list())
 })
 
 
@@ -92,6 +100,10 @@ test_that("sink_csv works", {
     dat[, c("drat", "mpg")],
     ignore_attr = TRUE # ignore row names
   )
+
+  # return the input data
+  x = dat_pl$sink_csv(temp_out)
+  expect_identical(x$collect()$to_list(), dat_pl$collect()$to_list())
 })
 
 test_that("sink_csv: null_values works", {
@@ -231,6 +243,11 @@ test_that("sink_csv: float_precision works", {
 
 test_that("sink_ndjson works", {
   temp_out = tempfile(fileext = ".json")
-  pl$LazyFrame(mtcars)$head(15)$select(pl$col("drat", "mpg"))$sink_ndjson(temp_out)
+  dat = pl$LazyFrame(mtcars)$head(15)$select(pl$col("drat", "mpg"))
+  dat$sink_ndjson(temp_out)
   expect_snapshot_file(temp_out)
+
+  # return the input data
+  x = dat$sink_ndjson(temp_out)
+  expect_identical(x$collect()$to_list(), dat$collect()$to_list())
 })