feat(output-formats): add support for to_parquet_dir

ibis/backends/__init__.py

@@ -473,6 +473,45 @@ def to_parquet(
                 for batch in batch_reader:
                     writer.write_batch(batch)
 
+    @util.experimental
+    def to_parquet_dir(
+        self,
+        expr: ir.Table,
+        path: str | Path,
+        *,
+        params: Mapping[ir.Scalar, Any] | None = None,
+        **kwargs: Any,
+    ) -> None:
+        """Write the results of executing the given expression to a parquet file in a directory.
+
+        This method is eager and will execute the associated expression
+        immediately.
+
+        Parameters
+        ----------
+        expr
+            The ibis expression to execute and persist to parquet.
+        path
+            The data source. A string or Path to the directory where the parquet file will be written.
+        params
+            Mapping of scalar parameter expressions to value.
+        **kwargs
+            Additional keyword arguments passed to pyarrow.dataset.write_dataset
+
+        https://arrow.apache.org/docs/python/generated/pyarrow.dataset.write_dataset.html
+
+        """
+        self._import_pyarrow()
+        import pyarrow.dataset as ds
+
+        dir_path = Path(path)
+
+        # by default write_dataset creates the directory
+        with expr.to_pyarrow_batches(params=params) as batch_reader:
+            ds.write_dataset(
+                batch_reader, base_dir=dir_path, format="parquet", **kwargs
+            )
+
     @util.experimental
     def to_csv(
         self,

ibis/backends/pyspark/__init__.py

@@ -1286,7 +1286,7 @@
             df = df.write.format(format)
             for k, v in (options or {}).items():
                 df = df.option(k, v)
-            df.save(path)
+            df.save(os.fspath(path))
             return None
         sq = df.writeStream.format(format)
         sq = sq.option("path", os.fspath(path))

ibis/backends/tests/test_export.py

@@ -215,6 +215,31 @@
     )
 
 
+def test_table_to_parquet_dir(tmp_path, backend, awards_players):
+    outparquet_dir = tmp_path / "out"
+
+    if backend.name() == "pyspark":
+        # pyspark already writes more than one file
+        awards_players.to_parquet_dir(outparquet_dir)
+    else:
+        # max_ force pyarrow to write more than one parquet file
+        awards_players.to_parquet_dir(
+            outparquet_dir, max_rows_per_file=3000, max_rows_per_group=3000
+        )
+
+    parquet_files = sorted(
+        outparquet_dir.glob("*.parquet"),
+        key=lambda path: int(path.with_suffix("").name.split("-")[1]),
+    )
+
+    df_list = [pd.read_parquet(file) for file in parquet_files]
+    df = pd.concat(df_list).reset_index(drop=True)
+
+    backend.assert_frame_equal(
+        awards_players.to_pandas().fillna(pd.NA), df.fillna(pd.NA)
+    )
+
+
 @pytest.mark.notimpl(
     ["duckdb"],
     reason="cannot inline WriteOptions objects",

ibis/expr/types/core.py

@@ -604,6 +604,37 @@ def to_parquet(
         """
         self._find_backend(use_default=True).to_parquet(self, path, **kwargs)
 
+    @experimental
+    def to_parquet_dir(
+        self,
+        path: str | Path,
+        *,
+        params: Mapping[ir.Scalar, Any] | None = None,
+        **kwargs: Any,
+    ) -> None:
+        """Write the results of executing the given expression to a parquet file in a directory.
+
+        This method is eager and will execute the associated expression
+        immediately.
+
+        Parameters
+        ----------
+        path
+            The data source. A string or Path to the directory where the parquet file will be written.
+        filename_prefix
+            The prefix name of the parquet file. What is before `.parquet`.
+        write_batches
+            If True writes each batch into a different file.
+        params
+            Mapping of scalar parameter expressions to value.
+        **kwargs
+            Additional keyword arguments passed to pyarrow.parquet.ParquetWriter
+
+        https://arrow.apache.org/docs/python/generated/pyarrow.parquet.ParquetWriter.html
+
+        """
+        self._find_backend(use_default=True).to_parquet_dir(self, path, **kwargs)
+
     @experimental
     def to_csv(
         self,