Skip to content

Add float, boolean and array indexing support to Java Variant extraction #23711

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
nartal1 wants to merge 3 commits into
base: main
Choose a base branch
from
+326 −14
Open

Add float, boolean and array indexing support to Java Variant extraction #23711

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
e7645f8
Support float and boolean Variant extraction in Java
nartal1 Aug 18, 2026
45e7a14
Merge branch 'main' into support-float-bool
nartal1 Aug 19, 2026
7897476
address review comments - add tests
nartal1 Aug 22, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 26 additions & 6 deletions java/src/main/java/ai/rapids/cudf/VariantUtils.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,8 @@ public class VariantUtils {
// cpp/include/cudf/io/experimental/variant.hpp and
// cpp/src/io/parquet/experimental/variant_extract.cu:is_variant_castable.
private static final List<DType> SUPPORTED_TYPES = Arrays.asList(
DType.STRING, DType.INT8, DType.INT16, DType.INT32, DType.INT64);
DType.STRING, DType.INT8, DType.INT16, DType.INT32, DType.INT64,
DType.FLOAT32, DType.FLOAT64, DType.BOOL8);

private VariantUtils() {}

Expand All @@ -38,8 +39,11 @@ private static void validateTargetType(DType targetType) {
*
* @param variantStruct Variant materialization: STRUCT(metadata LIST&lt;UINT8&gt;,
* value LIST&lt;UINT8&gt;, optional shredded children...)
* @param path JSONPath-like path accepted by cuDF's Variant extractor. Paths are expected to
* be ASCII object-field paths like {@code x}, {@code $.x}, or {@code $.x.y}.
* @param path JSONPath-like path accepted by cuDF's Variant extractor. Object-field steps use
* dot notation and zero-based array-index steps use bracket notation, for example
* {@code x}, {@code $.x.y}, {@code $[0]}, or {@code $.a[0].b}. Missing fields,
* out-of-bounds indices, and container mismatches produce null rows. Wildcards,
* negative indices, and quoted names inside brackets are not supported.
* @return LIST&lt;UINT8&gt; column of raw encoded Variant values
*/
public static ColumnVector getVariantFieldValue(ColumnView variantStruct, String path) {
Expand All @@ -50,8 +54,11 @@ public static ColumnVector getVariantFieldValue(ColumnView variantStruct, String

/**
* Decode raw Variant-encoded value bytes into {@code targetType}. Supported target types are
* {@link DType#STRING}, {@link DType#INT8}, {@link DType#INT16}, {@link DType#INT32}, and
* {@link DType#INT64}.
* {@link DType#STRING}, {@link DType#INT8}, {@link DType#INT16}, {@link DType#INT32},
* {@link DType#INT64}, {@link DType#FLOAT32}, {@link DType#FLOAT64}, and {@link DType#BOOL8}.
* Decoding requires the encoded physical type to exactly match {@code targetType}; no numeric
* conversions are performed. Input nulls, encoded Variant nulls, and physical-type mismatches
* produce null output rows.
*/
public static ColumnVector castVariantValue(ColumnView valueBytes, DType targetType) {
Objects.requireNonNull(valueBytes, "valueBytes");
Expand All @@ -63,7 +70,20 @@ public static ColumnVector castVariantValue(ColumnView valueBytes, DType targetT
/**
* Extract a Variant field and decode it into {@code targetType} in one native call.
* Supported target types are {@link DType#STRING}, {@link DType#INT8}, {@link DType#INT16},
* {@link DType#INT32}, and {@link DType#INT64}.
* {@link DType#INT32}, {@link DType#INT64}, {@link DType#FLOAT32}, {@link DType#FLOAT64}, and
* {@link DType#BOOL8}.
* Decoding requires the encoded physical type to exactly match {@code targetType}; no numeric
* conversions are performed. Missing fields, input nulls, encoded Variant nulls, and
* physical-type mismatches produce null output rows.
*
* @param variantStruct Variant materialization: STRUCT(metadata LIST&lt;UINT8&gt;,
* value LIST&lt;UINT8&gt;, optional shredded children...)
* @param path JSONPath-like path accepted by cuDF's Variant extractor. Object-field steps use
* dot notation and zero-based array-index steps use bracket notation, for example
* {@code x}, {@code $.x.y}, {@code $[0]}, or {@code $.a[0].b}. Missing fields,
* out-of-bounds indices, and container mismatches produce null rows. Wildcards,
* negative indices, and quoted names inside brackets are not supported.
* @param targetType decoded output type
*/
public static ColumnVector extractVariantField(
ColumnView variantStruct, String path, DType targetType) {
Expand Down
Loading
Loading