S3 methods for DataFrame and LazyFrame: simplest (#107)

* S3 methods for DataFrame and LazyFrame: simplest

* @nord tags

* remove upper case duplicates (docs warning)

* s3: patrick + series + agg

* docs

* vignette

* vignette

* vignette

* prefix, maintain_order not sort, drop cliffhanger

* responses to Grant's comments on Get Started vignette

* typo

---------

Co-authored-by: sorhawell <[email protected]>

DESCRIPTION

@@ -62,6 +62,7 @@ Collate:
     'pkg-nanoarrow.R'
     'rlang.R'
     'rust_result.R'
+    's3_methods.R'
     'series__series.R'
     'translation.R'
     'vctrs.R'

NAMESPACE

@@ -81,8 +81,29 @@ S3method(.DollarNames,WhenThen)
 S3method(.DollarNames,WhenThenThen)
 S3method(.DollarNames,method_environment)
 S3method(as.data.frame,DataFrame)
+S3method(as.data.frame,LazyFrame)
+S3method(as.matrix,DataFrame)
+S3method(as.matrix,LazyFrame)
 S3method(c,Series)
+S3method(dim,DataFrame)
+S3method(dim,LazyFrame)
+S3method(head,DataFrame)
+S3method(head,LazyFrame)
+S3method(length,DataFrame)
+S3method(length,LazyFrame)
 S3method(length,Series)
+S3method(max,DataFrame)
+S3method(max,LazyFrame)
+S3method(max,Series)
+S3method(mean,DataFrame)
+S3method(mean,LazyFrame)
+S3method(median,DataFrame)
+S3method(median,LazyFrame)
+S3method(min,DataFrame)
+S3method(min,LazyFrame)
+S3method(min,Series)
+S3method(names,DataFrame)
+S3method(names,LazyFrame)
 S3method(print,DataFrame)
 S3method(print,Expr)
 S3method(print,GroupBy)
@@ -96,6 +117,11 @@ S3method(print,Series)
 S3method(print,When)
 S3method(print,WhenThen)
 S3method(print,WhenThenThen)
+S3method(sum,DataFrame)
+S3method(sum,LazyFrame)
+S3method(sum,Series)
+S3method(tail,DataFrame)
+S3method(tail,LazyFrame)
 export("%**%")
 export("%**%.Expr")
 export(.pr)
@@ -104,10 +130,15 @@ export(GroupBy_agg)
 export(GroupBy_as_data_frame)
 export(LazyFrame_print)
 export(csv_reader)
+export(ncol.DataFrame)
+export(ncol.LazyFrame)
+export(nrow.DataFrame)
+export(nrow.LazyFrame)
 export(pl)
 export(read_csv_)
 export(scan_parquet)
 export(unwrap)
+importFrom(stats,median)
 importFrom(utils,.DollarNames)
 importFrom(utils,capture.output)
 importFrom(utils,download.file)

R/s3_methods.R

@@ -0,0 +1,136 @@
+#' @export
+#' @noRd
+head.DataFrame = function(x, n = 6L, ...) x$limit(n = n)
+
+#' @export
+#' @noRd
+head.LazyFrame = head.DataFrame
+
+#' @export
+#' @noRd
+tail.DataFrame = function(x, n = 6L, ...) x$tail(n = n)
+
+#' @export
+#' @noRd
+tail.LazyFrame = tail.DataFrame
+
+#' @export
+#' @noRd
+dim.DataFrame = function(x, ...) x$shape
+
+#' @export
+#' @noRd
+dim.LazyFrame = function(x, ...) x$collect()$shape
+
+#' @export
+#' @noRd
+length.DataFrame = function(x, ...) x$width
+
+#' @export
+#' @noRd
+length.Series = function(x, ...) x$len()
+
+#' @export
+#' @noRd
+length.LazyFrame = function(x, ...) x$collect()$width
+
+#' The Number of Rows of a DataFrame 
+#' @param x DataFrame
+#' @return Integer
+#' @export
+nrow.DataFrame = function(x) x$height
+
+#' The Number of Rows of a LazyFrame 
+#' @param x LazyFrame
+#' @return Integer
+#' @export
+nrow.LazyFrame = function(x) x$collect()$height
+
+#' The Number of Columns of a DataFrame 
+#' @param x DataFrame
+#' @return Integer
+#' @export
+ncol.DataFrame = function(x) x$height
+
+#' The Number of Columns of a LazyFrame 
+#' @param x LazyFrame
+#' @return Integer
+#' @export
+ncol.LazyFrame = function(x) x$collect()$height
+
+#' @export
+#' @noRd
+names.DataFrame = function(x) x$columns
+
+# TODO: inefficient to collect, but attribute is missing
+#' @export
+#' @noRd
+names.LazyFrame = function(x) x$collect()$columns
+
+# TODO: inefficient to collect, but attribute is missing
+#' @export
+#' @noRd
+as.data.frame.LazyFrame = function(x, ...) x$collect()$as_data_frame(...)
+
+#' @export
+#' @noRd
+as.matrix.DataFrame = function(x, ...) as.matrix(x$as_data_frame(...))
+
+# TODO: inefficient to collect, but attribute is missing
+#' @export
+#' @noRd
+as.matrix.LazyFrame = function(x, ...) as.matrix(x$collect()$as_data_frame(...))
+
+#' @export
+#' @noRd
+mean.DataFrame = function(x, ...) x$mean()
+
+#' @export
+#' @noRd
+mean.LazyFrame = function(x, ...) x$mean()
+
+#' @export
+#' @importFrom stats median
+#' @noRd
+median.DataFrame = function(x, ...) x$median()
+
+#' @export
+#' @importFrom stats median
+#' @noRd
+median.LazyFrame = function(x, ...) x$median()
+
+#' @export
+#' @noRd
+min.DataFrame = function(x, ...) x$min()
+
+#' @export
+#' @noRd
+min.LazyFrame = function(x, ...) x$min()
+
+#' @export
+#' @noRd
+min.Series = function(x, ...) x$min()
+
+#' @export
+#' @noRd
+max.DataFrame = function(x, ...) x$max()
+
+#' @export
+#' @noRd
+max.LazyFrame = function(x, ...) x$max()
+
+#' @export
+#' @noRd
+max.Series = function(x, ...) x$max()
+
+#' @export
+#' @noRd
+sum.DataFrame = function(x, ...) x$sum()
+
+#' @export
+#' @noRd
+sum.LazyFrame = function(x, ...) x$sum()
+
+#' @export
+#' @noRd
+sum.Series = function(x, ...) x$sum()

tests/testthat/test-expr.R

@@ -1012,8 +1012,8 @@ test_that("sort_by", {
     expect_grepl_error(pl$lit(1:4)$sort_by(1)$to_r(),"Expected length\\: 4")
     expect_grepl_error(pl$lit(1:4)$sort_by("blop")$to_r(),"column 'blop' not available in schema")
     expect_grepl_error(pl$lit(1:4)$sort_by("blop")$to_r(),"column 'blop' not available in schema")
-    expect_grepl_error(pl$lit(1:4)$sort_by(df)$to_r(),"of sequence not convertable into an Expr")
-    expect_grepl_error(pl$lit(1:4)$sort_by(df)$to_r(),"of sequence not convertable into an Expr")
+    expect_grepl_error(pl$lit(1:4)$sort_by(df)$to_r(),"not convertable into.* Expr")
+    expect_grepl_error(pl$lit(1:4)$sort_by(df)$to_r(),"not convertable into.* Expr")
 
   #this test is minimal, if polars give better documentation on behaviour, expand the test.
 })

tests/testthat/test-s3_methods.R

@@ -0,0 +1,74 @@
+make_cases <- function() {
+  tibble::tribble(
+    ~ .test_name, ~ pola,   ~ base,
+    "mean",       "mean",   mean,
+    "median",     "median", stats::median,
+    "min",        "min",    min,
+    "max",        "max",    max,
+    "sum",        "sum",    sum,
+  )
+}
+patrick::with_parameters_test_that("aggregations",
+    {
+        d = pl$DataFrame(mtcars)
+        w = d[[pola]]()$as_data_frame()
+        x = base(d)$as_data_frame()
+        y = base(d$lazy())$collect()$as_data_frame()
+        z = data.frame(t(sapply(mtcars, base)))
+        expect_equal(w, x, ignore_attr = TRUE)
+        expect_equal(w, y, ignore_attr = TRUE)
+        expect_equal(w, z, ignore_attr = TRUE)
+    },
+    .cases = make_cases()
+)
+
+
+make_cases <- function() {
+  tibble::tribble(
+    ~ .test_name, ~ FUN,
+    "head",       head,
+    "tail",       tail,
+    "nrow",       nrow,
+    "ncol",       ncol,
+    "length",     length,
+    "as.matrix",  as.matrix,
+    "names",      names,
+  )
+}
+patrick::with_parameters_test_that("inspection",
+    {
+        d = pl$DataFrame(mtcars)
+        x = FUN(mtcars)
+        y = FUN(d)
+        z = FUN(d$lazy())
+        if (inherits(y, "DataFrame")) y = y$as_data_frame()
+        if (inherits(z, "LazyFrame")) z = z$collect()$as_data_frame()
+        expect_equal(x, y, ignore_attr = TRUE)
+        expect_equal(x, z, ignore_attr = TRUE)
+    },
+    .cases = make_cases()
+)
+
+
+make_cases <- function() {
+  tibble::tribble(
+    ~ .test_name, ~ pola,   ~ base,
+    "length",     "len",    length,
+    "min",        "min",    min,
+    "max",        "max",    max,
+    "sum",        "sum",    sum,
+  )
+}
+patrick::with_parameters_test_that("Series",
+    {
+        d = pl$Series(mtcars$mpg)
+        x = base(mtcars$mpg)
+        y = base(d)
+        z = d[[pola]]()
+        if (inherits(y, "Series")) y = y$to_r_vector()
+        if (inherits(z, "Series")) z = z$to_r_vector()
+        expect_equal(x, y, ignore_attr = TRUE)
+        expect_equal(x, z, ignore_attr = TRUE)
+    },
+    .cases = make_cases()
+)