Merge pull request #46 from Tmonster/add_parquet_write_support

R/cpp11.R

@@ -196,6 +196,10 @@ rapi_execute <- function(stmt, arrow, integer64) {
   .Call(`_duckdb_rapi_execute`, stmt, arrow, integer64)
 }
 
+rapi_rel_to_parquet <- function(rel, file_name) {
+  invisible(.Call(`_duckdb_rapi_rel_to_parquet`, rel, file_name))
+}
+
 rapi_adbc_init_func <- function() {
   .Call(`_duckdb_rapi_adbc_init_func`)
 }

R/relational.R

@@ -399,3 +399,5 @@ rel_from_table_function <- function(con, function_name, positional_parameters =
     rapi_rel_from_table_function(con@conn_ref, function_name, positional_parameters, named_parameters)
 }
 
+rel_to_parquet <- rapi_rel_to_parquet
+

src/cpp11.cpp

@@ -356,6 +356,14 @@ extern "C" SEXP _duckdb_rapi_execute(SEXP stmt, SEXP arrow, SEXP integer64) {
     return cpp11::as_sexp(rapi_execute(cpp11::as_cpp<cpp11::decay_t<duckdb::stmt_eptr_t>>(stmt), cpp11::as_cpp<cpp11::decay_t<bool>>(arrow), cpp11::as_cpp<cpp11::decay_t<bool>>(integer64)));
   END_CPP11
 }
+// statement.cpp
+void rapi_rel_to_parquet(duckdb::rel_extptr_t rel, std::string file_name);
+extern "C" SEXP _duckdb_rapi_rel_to_parquet(SEXP rel, SEXP file_name) {
+  BEGIN_CPP11
+    rapi_rel_to_parquet(cpp11::as_cpp<cpp11::decay_t<duckdb::rel_extptr_t>>(rel), cpp11::as_cpp<cpp11::decay_t<std::string>>(file_name));
+    return R_NilValue;
+  END_CPP11
+}
 // utils.cpp
 SEXP rapi_adbc_init_func();
 extern "C" SEXP _duckdb_rapi_adbc_init_func() {
@@ -416,6 +424,7 @@ static const R_CallMethodDef CallEntries[] = {
     {"_duckdb_rapi_rel_sql",                 (DL_FUNC) &_duckdb_rapi_rel_sql,                 2},
     {"_duckdb_rapi_rel_to_altrep",           (DL_FUNC) &_duckdb_rapi_rel_to_altrep,           1},
     {"_duckdb_rapi_rel_to_df",               (DL_FUNC) &_duckdb_rapi_rel_to_df,               1},
+    {"_duckdb_rapi_rel_to_parquet",          (DL_FUNC) &_duckdb_rapi_rel_to_parquet,          2},
     {"_duckdb_rapi_rel_to_sql",              (DL_FUNC) &_duckdb_rapi_rel_to_sql,              1},
     {"_duckdb_rapi_rel_tostring",            (DL_FUNC) &_duckdb_rapi_rel_tostring,            1},
     {"_duckdb_rapi_rel_union_all",           (DL_FUNC) &_duckdb_rapi_rel_union_all,           2},

src/statement.cpp

@@ -357,3 +357,7 @@ bool FetchArrowChunk(ChunkScanState &scan_state, ClientProperties options, Appen
 		return duckdb_execute_R_impl(result, integer64);
 	}
 }
+
+[[cpp11::register]] void rapi_rel_to_parquet(duckdb::rel_extptr_t rel, std::string file_name) {
+    rel->rel->WriteParquet(file_name);
+}

tests/testthat/test_parquet.R

@@ -11,3 +11,40 @@ test_that("parquet reader works with the binary as string flag", {
   expect_true(res[1] == "VARCHAR")
   dbDisconnect(con, shutdown = TRUE)
 })
+
+test_that("duckdb_write_parquet() works as expected", {
+  con <- dbConnect(duckdb())
+
+  tf <- tempfile()
+
+  # write to parquet
+  iris_rel <- rel_from_df(con, iris)
+  rel_to_parquet(iris_rel, tf)
+
+  res_rel <- rel_from_table_function(con, 'read_parquet', list(tf))
+  res_df <- rel_to_altrep(res_rel)
+  res_df$Species <- as.factor(res_df$Species)
+  expect_true(identical(res_df, iris))
+
+
+  # nulls
+  iris_na <- iris
+  iris_na[[2]][42] <- NA
+
+  iris_na_rel <- duckdb:::rel_from_df(con, iris_na)
+  duckdb:::rel_to_parquet(iris_na_rel, tf)
+
+  res_rel <- duckdb:::rel_from_table_function(con, 'read_parquet', list(tf))
+  res_df <- duckdb:::rel_to_altrep(res_rel)
+  res_df$Species <- as.factor(res_df$Species)
+  expect_true(identical(res_df, iris_na))
+})
+
+test_that("duckdb rel_to_parquet() throws error with no file name", {
+  con <- dbConnect(duckdb())
+
+  # write to parquet
+  iris_rel <- rel_from_df(con, iris)
+  expect_error(rel_to_parquet(iris_rel, ""))
+})
+

tests/testthat/test_read.R

@@ -109,4 +109,4 @@ test_that("duckdb_read_csv() works as expected", {
   expect_true(identical(res, iris))
 
   dbDisconnect(con, shutdown = TRUE)
-})
+})