feat: add a new option to export decimal as character

pola-rs · Sep 13, 2024 · 28eae28 · 28eae28
1 parent aa36918
commit 28eae28
Show file tree

Hide file tree

Showing 14 changed files with 117 additions and 10 deletions.
diff --git a/R/000-wrappers.R b/R/000-wrappers.R
@@ -1406,9 +1406,9 @@ class(`PlRLazyGroupBy`) <- "PlRLazyGroupBy__bundle"
 }
 
 `PlRSeries_to_r_vector` <- function(self) {
-  function(`ensure_vector`, `int64`, `struct`, `as_clock_class`, `ambiguous`, `non_existent`, `local_time_zone`) {
+  function(`ensure_vector`, `int64`, `struct`, `decimal`, `as_clock_class`, `ambiguous`, `non_existent`, `local_time_zone`) {
     `ambiguous` <- .savvy_extract_ptr(`ambiguous`, "PlRExpr")
-    .Call(savvy_PlRSeries_to_r_vector__impl, `self`, `ensure_vector`, `int64`, `struct`, `as_clock_class`, `ambiguous`, `non_existent`, `local_time_zone`)
+    .Call(savvy_PlRSeries_to_r_vector__impl, `self`, `ensure_vector`, `int64`, `struct`, `decimal`, `as_clock_class`, `ambiguous`, `non_existent`, `local_time_zone`)
   }
 }
 

diff --git a/R/dataframe-s3-base.R b/R/dataframe-s3-base.R
@@ -38,6 +38,7 @@ as.list.polars_data_frame <- function(
     as_series = FALSE,
     int64 = "double",
     struct = "dataframe",
+    decimal = "double",
     as_clock_class = FALSE,
     ambiguous = "raise",
     non_existent = "raise") {
@@ -47,6 +48,7 @@ as.list.polars_data_frame <- function(
     x$to_r_list(
       int64 = int64,
       struct = struct,
+      decimal = decimal,
       as_clock_class = as_clock_class,
       ambiguous = ambiguous,
       non_existent = non_existent
@@ -69,13 +71,15 @@ as.list.polars_data_frame <- function(
 as.data.frame.polars_data_frame <- function(
     x, ...,
     int64 = "double",
+    decimal = "double",
     as_clock_class = FALSE,
     ambiguous = "raise",
     non_existent = "raise") {
   x$to_struct()$to_r_vector(
     ensure_vector = FALSE,
     int64 = int64,
     struct = "dataframe",
+    decimal = decimal,
     as_clock_class = as_clock_class,
     ambiguous = ambiguous,
     non_existent = non_existent

diff --git a/R/dataframe-s3-tibble.R b/R/dataframe-s3-tibble.R
@@ -25,13 +25,15 @@ as_tibble.polars_data_frame <- function(
     x, ...,
     .name_repair = "check_unique",
     int64 = "double",
+    decimal = "double",
     as_clock_class = FALSE,
     ambiguous = "raise",
     non_existent = "raise") {
   x$to_struct()$to_r_vector(
     ensure_vector = FALSE,
     int64 = int64,
     struct = "tibble",
+    decimal = decimal,
     as_clock_class = as_clock_class,
     ambiguous = ambiguous,
     non_existent = non_existent

diff --git a/R/dataframe-to_r_list.R b/R/dataframe-to_r_list.R
@@ -22,6 +22,7 @@ dataframe__to_r_list <- function(
     int64 = "double",
     as_clock_class = FALSE,
     struct = "dataframe",
+    decimal = "double",
     ambiguous = "raise",
     non_existent = "raise") {
   wrap({
@@ -30,8 +31,9 @@ dataframe__to_r_list <- function(
     self$to_struct()$to_r_vector(
       ensure_vector = TRUE,
       int64 = int64,
-      as_clock_class = as_clock_class,
       struct = struct,
+      decimal = decimal,
+      as_clock_class = as_clock_class,
       ambiguous = ambiguous,
       non_existent = non_existent
     )

diff --git a/R/series-to_r_vector.R b/R/series-to_r_vector.R
@@ -53,6 +53,10 @@
 #' - `"dataframe"` (default): Convert to the R's [data.frame] class.
 #' - `"tibble"`: Convert to the [tibble][tibble::tbl_df] class.
 #'   If the [tibble][tibble::tibble-package] package is not installed, a warning will be shown.
+#' @param decimal Determine how to convert Polars' Decimal type values to R type.
+#' One of the followings:
+#' - `"double"` (default): Convert to the R's [double] type.
+#' - `"character"`: Convert to the R's [character] type.
 #' @param as_clock_class A logical value indicating whether to export datetimes and duration as
 #' the [clock][clock::clock] package's classes.
 #' - `FALSE` (default): Duration values are exported as [difftime]
@@ -157,6 +161,7 @@ series__to_r_vector <- function(
     ensure_vector = FALSE,
     int64 = "double",
     struct = "dataframe",
+    decimal = "double",
     as_clock_class = FALSE,
     ambiguous = "raise",
     non_existent = "raise") {
@@ -185,6 +190,7 @@ series__to_r_vector <- function(
       ensure_vector = ensure_vector,
       int64 = int64,
       struct = struct,
+      decimal = decimal,
       as_clock_class = as_clock_class,
       ambiguous = ambiguous,
       non_existent = non_existent,

diff --git a/man/dataframe__to_r_list.Rd b/man/dataframe__to_r_list.Rd
diff --git a/man/s3_as.data.frame.Rd b/man/s3_as.data.frame.Rd
diff --git a/man/s3_as.list.Rd b/man/s3_as.list.Rd
diff --git a/man/s3_as_tibble.Rd b/man/s3_as_tibble.Rd
diff --git a/man/series__to_r_vector.Rd b/man/series__to_r_vector.Rd
diff --git a/src/init.c b/src/init.c
@@ -874,8 +874,8 @@ SEXP savvy_PlRSeries_new_i64_from_clock_pair__impl(SEXP c_arg__name, SEXP c_arg_
     return handle_result(res);
 }
 
-SEXP savvy_PlRSeries_to_r_vector__impl(SEXP self__, SEXP c_arg__ensure_vector, SEXP c_arg__int64, SEXP c_arg__struct, SEXP c_arg__as_clock_class, SEXP c_arg__ambiguous, SEXP c_arg__non_existent, SEXP c_arg__local_time_zone) {
-    SEXP res = savvy_PlRSeries_to_r_vector__ffi(self__, c_arg__ensure_vector, c_arg__int64, c_arg__struct, c_arg__as_clock_class, c_arg__ambiguous, c_arg__non_existent, c_arg__local_time_zone);
+SEXP savvy_PlRSeries_to_r_vector__impl(SEXP self__, SEXP c_arg__ensure_vector, SEXP c_arg__int64, SEXP c_arg__struct, SEXP c_arg__decimal, SEXP c_arg__as_clock_class, SEXP c_arg__ambiguous, SEXP c_arg__non_existent, SEXP c_arg__local_time_zone) {
+    SEXP res = savvy_PlRSeries_to_r_vector__ffi(self__, c_arg__ensure_vector, c_arg__int64, c_arg__struct, c_arg__decimal, c_arg__as_clock_class, c_arg__ambiguous, c_arg__non_existent, c_arg__local_time_zone);
     return handle_result(res);
 }
 
@@ -1139,7 +1139,7 @@ static const R_CallMethodDef CallEntries[] = {
     {"savvy_PlRSeries_new_binary__impl", (DL_FUNC) &savvy_PlRSeries_new_binary__impl, 2},
     {"savvy_PlRSeries_new_series_list__impl", (DL_FUNC) &savvy_PlRSeries_new_series_list__impl, 3},
     {"savvy_PlRSeries_new_i64_from_clock_pair__impl", (DL_FUNC) &savvy_PlRSeries_new_i64_from_clock_pair__impl, 4},
-    {"savvy_PlRSeries_to_r_vector__impl", (DL_FUNC) &savvy_PlRSeries_to_r_vector__impl, 8},
+    {"savvy_PlRSeries_to_r_vector__impl", (DL_FUNC) &savvy_PlRSeries_to_r_vector__impl, 9},
     {"savvy_PlRSeries_print__impl", (DL_FUNC) &savvy_PlRSeries_print__impl, 1},
     {"savvy_PlRSeries_struct_unnest__impl", (DL_FUNC) &savvy_PlRSeries_struct_unnest__impl, 1},
     {"savvy_PlRSeries_struct_fields__impl", (DL_FUNC) &savvy_PlRSeries_struct_fields__impl, 1},

diff --git a/src/rust/api.h b/src/rust/api.h
@@ -182,7 +182,7 @@ SEXP savvy_PlRSeries_new_single_binary__ffi(SEXP c_arg__name, SEXP c_arg__values
 SEXP savvy_PlRSeries_new_binary__ffi(SEXP c_arg__name, SEXP c_arg__values);
 SEXP savvy_PlRSeries_new_series_list__ffi(SEXP c_arg__name, SEXP c_arg__values, SEXP c_arg__strict);
 SEXP savvy_PlRSeries_new_i64_from_clock_pair__ffi(SEXP c_arg__name, SEXP c_arg__left, SEXP c_arg__right, SEXP c_arg__precision);
-SEXP savvy_PlRSeries_to_r_vector__ffi(SEXP self__, SEXP c_arg__ensure_vector, SEXP c_arg__int64, SEXP c_arg__struct, SEXP c_arg__as_clock_class, SEXP c_arg__ambiguous, SEXP c_arg__non_existent, SEXP c_arg__local_time_zone);
+SEXP savvy_PlRSeries_to_r_vector__ffi(SEXP self__, SEXP c_arg__ensure_vector, SEXP c_arg__int64, SEXP c_arg__struct, SEXP c_arg__decimal, SEXP c_arg__as_clock_class, SEXP c_arg__ambiguous, SEXP c_arg__non_existent, SEXP c_arg__local_time_zone);
 SEXP savvy_PlRSeries_print__ffi(SEXP self__);
 SEXP savvy_PlRSeries_struct_unnest__ffi(SEXP self__);
 SEXP savvy_PlRSeries_struct_fields__ffi(SEXP self__);

diff --git a/src/rust/src/series/export.rs b/src/rust/src/series/export.rs
@@ -20,6 +20,13 @@ enum StructConversion {
     Tibble,
 }
 
+#[derive(Debug, Clone, EnumString)]
+#[strum(serialize_all = "lowercase")]
+enum DecimalConversion {
+    Character,
+    Double,
+}
+
 // `vctrs::unspecified` like function
 fn vctrs_unspecified_sexp(n: usize) -> Sexp {
     let mut sexp = OwnedLogicalSexp::new(n).unwrap();
@@ -51,6 +58,7 @@ impl PlRSeries {
         ensure_vector: bool,
         int64: &str,
         r#struct: &str,
+        decimal: &str,
         as_clock_class: bool,
         ambiguous: &PlRExpr,
         non_existent: &str,
@@ -69,6 +77,11 @@ impl PlRSeries {
                 "Argument `struct` must be one of ('dataframe', 'tibble')".to_string(),
             )
         })?;
+        let decimal = DecimalConversion::try_from(decimal).map_err(|_| {
+            savvy::Error::from(
+                "Argument `decimal` must be one of ('character', 'double')".to_string(),
+            )
+        })?;
         let ambiguous = ambiguous.inner.clone();
         let non_existent = <Wrap<NonExistent>>::try_from(non_existent)?.0;
 
@@ -77,6 +90,7 @@ impl PlRSeries {
             ensure_vector: bool,
             int64: Int64Conversion,
             r#struct: StructConversion,
+            decimal: DecimalConversion,
             as_clock_class: bool,
             ambiguous: Expr,
             non_existent: NonExistent,
@@ -128,6 +142,7 @@ impl PlRSeries {
                             false,
                             int64.clone(),
                             r#struct.clone(),
+                            decimal.clone(),
                             as_clock_class,
                             ambiguous.clone(),
                             non_existent,
@@ -145,6 +160,7 @@ impl PlRSeries {
                                     false,
                                     int64.clone(),
                                     r#struct.clone(),
+                                    decimal.clone(),
                                     as_clock_class,
                                     ambiguous.clone(),
                                     non_existent,
@@ -168,6 +184,7 @@ impl PlRSeries {
                             false,
                             int64.clone(),
                             r#struct.clone(),
+                            decimal.clone(),
                             as_clock_class,
                             ambiguous.clone(),
                             non_existent,
@@ -185,6 +202,7 @@ impl PlRSeries {
                                     false,
                                     int64.clone(),
                                     r#struct.clone(),
+                                    decimal.clone(),
                                     as_clock_class,
                                     ambiguous.clone(),
                                     non_existent,
@@ -237,9 +255,14 @@ impl PlRSeries {
                         }
                     }
                 }
-                DataType::Decimal(_, _) => Ok(<Sexp>::from(Wrap(
-                    series.cast(&DataType::Float64).unwrap().f64().unwrap(),
-                ))),
+                DataType::Decimal(_, _) => match decimal {
+                    DecimalConversion::Character => Ok(<Sexp>::from(Wrap(
+                        series.cast(&DataType::String).unwrap().str().unwrap(),
+                    ))),
+                    DecimalConversion::Double => Ok(<Sexp>::from(Wrap(
+                        series.cast(&DataType::Float64).unwrap().f64().unwrap(),
+                    ))),
+                },
                 DataType::String => Ok(<Sexp>::from(Wrap(series.str().unwrap()))),
                 DataType::Struct(_) => {
                     let df = series
@@ -269,6 +292,7 @@ impl PlRSeries {
                                 false,
                                 int64.clone(),
                                 r#struct.clone(),
+                                decimal.clone(),
                                 as_clock_class,
                                 ambiguous.clone(),
                                 non_existent,
@@ -297,6 +321,7 @@ impl PlRSeries {
             ensure_vector,
             int64,
             r#struct,
+            decimal,
             as_clock_class,
             ambiguous,
             non_existent,

diff --git a/tests/testthat/test-series-to_r_vector.R b/tests/testthat/test-series-to_r_vector.R
@@ -119,6 +119,34 @@ test_that("struct argument warning and error", {
   )
 })
 
+patrick::with_parameters_test_that(
+  "decimal conversion",
+  .cases = {
+    tibble::tribble(
+      ~.test_name, ~type, ~expected_out,
+      "double", "double", c(NA, 1, 0.1),
+      "character", "character", c(NA, "1.000", "0.100"),
+    )
+  },
+  code = {
+    series_decimal <- as_polars_series(c(NA, 1, 0.1))$cast(pl$Decimal(5, 3))
+
+    out <- series_decimal$to_r_vector(decimal = type)
+    expect_identical(out, expected_out)
+  }
+)
+
+test_that("decimal argument error", {
+  expect_error(
+    as_polars_series(1)$to_r_vector(decimal = TRUE),
+    r"(Argument `decimal` must be character)"
+  )
+  expect_error(
+    as_polars_series(1)$to_r_vector(decimal = "foo"),
+    r"(must be one of \('character', 'double'\))"
+  )
+})
+
 patrick::with_parameters_test_that("datetime conversion to clock classes",
   .cases = {
     skip_if_not_installed("clock")