(iceberg-cpp) branch main updated: fix(row): apply array offsets in ArrowArray wrappers (#1039)

[ga...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

wgtmac pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/iceberg-cpp.git


The following commit(s) were added to refs/heads/main by this push:
     new 08d2f863 fix(row): apply array offsets in ArrowArray wrappers (#1039)
08d2f863 is described below

commit 08d2f863b3c921674c8cca9b532c84143d05619b
Author: YinZheng-Sun <57...@users.noreply.github.com>
AuthorDate: Wed Oct 21 15:47:42 2026 +0800

    fix(row): apply array offsets in ArrowArray wrappers (#1039)
    
    ArrowArrayStructLike, ArrowArrayArrayLike and ArrowArrayMapLike ignored the
    offset of the wrapped ArrowArray, so sliced inputs returned wrong rows:
    
    - Struct children are not shifted by the parent's offset in the Arrow C data
      interface, so GetField must add the struct's own offset.
    - ArrowArrayViewListChildOffset and the raw map offsets buffer are indexed
      without the array's offset, so list/map rows must add it before lookup.
    - Map key/value children are not shifted by the entries struct's offset.
    
    Add tests for sliced struct, list and map arrays.
    
    Co-authored-by: yinzhengsun <yi...@tencent.com>
---
 src/iceberg/row/arrow_array_wrapper.cc |  21 ++++--
 src/iceberg/test/struct_like_test.cc   | 119 +++++++++++++++++++++++++++++++++
 2 files changed, 133 insertions(+), 7 deletions(-)

diff --git a/src/iceberg/row/arrow_array_wrapper.cc b/src/iceberg/row/arrow_array_wrapper.cc
index 0d6d48c1..5a9cb089 100644
--- a/src/iceberg/row/arrow_array_wrapper.cc
+++ b/src/iceberg/row/arrow_array_wrapper.cc
@@ -134,7 +134,8 @@ class ArrowArrayStructLike::Impl {
     const ArrowArray* child_array = array_.get().children[pos];
     const ArrowArrayView* child_view = array_view_.children[pos];
 
-    return ExtractValue(child_schema, child_array, child_view, row_index_);
+    return ExtractValue(child_schema, child_array, child_view,
+                        array_view_.offset + row_index_);
   }
 
   size_t num_fields() const { return static_cast<size_t>(schema_.n_children); }
@@ -252,8 +253,9 @@ class ArrowArrayArrayLike::Impl {
                              array_.get().length);
     }
 
-    offset_ = ArrowArrayViewListChildOffset(&array_view_, row_index_);
-    length_ = ArrowArrayViewListChildOffset(&array_view_, row_index_ + 1) - offset_;
+    const int64_t row = array_view_.offset + row_index_;
+    offset_ = ArrowArrayViewListChildOffset(&array_view_, row);
+    length_ = ArrowArrayViewListChildOffset(&array_view_, row + 1) - offset_;
     return {};
   }
 
@@ -312,7 +314,7 @@ class ArrowArrayMapLike::Impl {
     const ArrowArrayView* keys_view = array_view_.children[0]->children[0];
 
     return ExtractValue(keys_schema, keys_array, keys_view,
-                        offset_ + static_cast<int64_t>(pos));
+                        EntryIndex(static_cast<int64_t>(pos)));
   }
 
   Result<Scalar> GetValue(size_t pos) const {
@@ -326,7 +328,7 @@ class ArrowArrayMapLike::Impl {
     const ArrowArrayView* values_view = array_view_.children[0]->children[1];
 
     return ExtractValue(values_schema, values_array, values_view,
-                        offset_ + static_cast<int64_t>(pos));
+                        EntryIndex(static_cast<int64_t>(pos)));
   }
 
   size_t size() const { return static_cast<size_t>(length_); }
@@ -365,12 +367,17 @@ class ArrowArrayMapLike::Impl {
     // XXX: ArrowArrayViewListChildOffset does not work for map types.
     // We need to directly access the offsets buffer instead.
     auto* offsets_buffer = array_view_.buffer_views[1].data.as_int32;
-    offset_ = offsets_buffer[row_index_];
-    length_ = offsets_buffer[row_index_ + 1] - offset_;
+    const int64_t row = array_view_.offset + row_index_;
+    offset_ = offsets_buffer[row];
+    length_ = offsets_buffer[row + 1] - offset_;
 
     return {};
   }
 
+  int64_t EntryIndex(int64_t pos) const {
+    return array_view_.children[0]->offset + offset_ + pos;
+  }
+
   ArrowArrayView array_view_;
   internal::ArrowArrayViewGuard array_view_guard_{&array_view_};
 
diff --git a/src/iceberg/test/struct_like_test.cc b/src/iceberg/test/struct_like_test.cc
index a5786011..e04d2c08 100644
--- a/src/iceberg/test/struct_like_test.cc
+++ b/src/iceberg/test/struct_like_test.cc
@@ -22,9 +22,12 @@
 #include <array>
 #include <memory>
 #include <optional>
+#include <string>
 #include <utility>
 #include <vector>
 
+#include <arrow/array.h>
+#include <arrow/buffer.h>
 #include <arrow/c/bridge.h>
 #include <arrow/json/from_string.h>
 #include <arrow/type.h>
@@ -663,4 +666,120 @@ TEST(ArrowArrayStructLike, Accessor) {
   }
 }
 
+TEST(ArrowArrayStructLike, SlicedStruct) {
+  auto person_type =
+      ::arrow::struct_({::arrow::field("name", ::arrow::utf8(), /*nullable=*/false)});
+  auto root_type = ::arrow::struct_(
+      {::arrow::field("id", ::arrow::int64(), /*nullable=*/false),
+       ::arrow::field("person", person_type, /*nullable=*/false),
+       ::arrow::field("tags", ::arrow::list(::arrow::int32()), /*nullable=*/false)});
+
+  auto arrow_array = ::arrow::json::ArrayFromJSONString(root_type, R"([
+    {"id": 1, "person": {"name": "Alice"}, "tags": [1]},
+    {"id": 2, "person": {"name": "Bob"}, "tags": [2, 3]},
+    {"id": 3, "person": {"name": "Carol"}, "tags": []}])")
+                         .ValueOrDie();
+  auto sliced = arrow_array->Slice(1);
+
+  ArrowSchema c_schema;
+  ArrowArray c_array;
+  internal::ArrowSchemaGuard schema_guard(&c_schema);
+  internal::ArrowArrayGuard array_guard(&c_array);
+  ASSERT_TRUE(::arrow::ExportType(*root_type, &c_schema).ok());
+  ASSERT_TRUE(::arrow::ExportArray(*sliced, &c_array).ok());
+  ASSERT_EQ(c_array.offset, 1);
+
+  ICEBERG_UNWRAP_OR_FAIL(auto struct_like, ArrowArrayStructLike::Make(c_schema, c_array));
+
+  constexpr int64_t kNumRows = 2;
+  std::array<int64_t, kNumRows> ids = {2, 3};
+  std::array<std::string, kNumRows> names = {"Bob", "Carol"};
+  std::array<std::vector<int32_t>, kNumRows> tags = {std::vector<int32_t>{2, 3},
+                                                     std::vector<int32_t>{}};
+
+  for (int64_t i = 0; i < kNumRows; ++i) {
+    ASSERT_THAT(struct_like->Reset(i), IsOk());
+    EXPECT_SCALAR_EQ(struct_like->GetField(0), int64_t, ids[i]);
+
+    ICEBERG_UNWRAP_OR_FAIL(auto person_scalar, struct_like->GetField(1));
+    auto person = std::get<std::shared_ptr<StructLike>>(person_scalar);
+    EXPECT_SCALAR_EQ(person->GetField(0), std::string_view, names[i]);
+
+    ICEBERG_UNWRAP_OR_FAIL(auto tags_scalar, struct_like->GetField(2));
+    auto tags_like = std::get<std::shared_ptr<ArrayLike>>(tags_scalar);
+    ASSERT_EQ(tags_like->size(), tags[i].size());
+    for (size_t j = 0; j < tags[i].size(); ++j) {
+      EXPECT_SCALAR_EQ(tags_like->GetElement(j), int32_t, tags[i][j]);
+    }
+  }
+}
+
+TEST(ArrowArrayStructLike, SlicedList) {
+  auto list_type = ::arrow::list(::arrow::int32());
+  auto arrow_array =
+      ::arrow::json::ArrayFromJSONString(list_type, R"([[1], [2, 3], [4]])").ValueOrDie();
+  auto sliced = arrow_array->Slice(1);
+
+  ArrowSchema c_schema;
+  ArrowArray c_array;
+  internal::ArrowSchemaGuard schema_guard(&c_schema);
+  internal::ArrowArrayGuard array_guard(&c_array);
+  ASSERT_TRUE(::arrow::ExportType(*list_type, &c_schema).ok());
+  ASSERT_TRUE(::arrow::ExportArray(*sliced, &c_array).ok());
+
+  ICEBERG_UNWRAP_OR_FAIL(auto array_like, ArrowArrayArrayLike::Make(c_schema, c_array));
+  ASSERT_EQ(array_like->size(), 2);
+  EXPECT_SCALAR_EQ(array_like->GetElement(0), int32_t, 2);
+  EXPECT_SCALAR_EQ(array_like->GetElement(1), int32_t, 3);
+
+  ASSERT_THAT(array_like->Reset(1), IsOk());
+  ASSERT_EQ(array_like->size(), 1);
+  EXPECT_SCALAR_EQ(array_like->GetElement(0), int32_t, 4);
+}
+
+TEST(ArrowArrayStructLike, SlicedMap) {
+  auto map_type = std::make_shared<::arrow::MapType>(
+      ::arrow::field("key", ::arrow::utf8(), /*nullable=*/false),
+      ::arrow::field("value", ::arrow::int32(), /*nullable=*/false));
+  auto entries_source =
+      ::arrow::json::ArrayFromJSONString(map_type, R"([[["X", 0], ["Foo", 1],
+                                                        ["Bar", 2], ["Baz", 3]]])")
+          .ValueOrDie();
+  auto entries =
+      std::static_pointer_cast<::arrow::MapArray>(entries_source)->values()->Slice(1);
+  auto offsets = ::arrow::Buffer::FromVector(std::vector<int32_t>{0, 0, 2, 3, 3});
+  auto map_data = ::arrow::ArrayData::Make(map_type, /*length=*/4, {nullptr, offsets},
+                                           {entries->data()}, /*null_count=*/0);
+  auto sliced = ::arrow::MakeArray(map_data)->Slice(1);
+  ASSERT_TRUE(sliced->ValidateFull().ok());
+
+  ArrowSchema c_schema;
+  ArrowArray c_array;
+  internal::ArrowSchemaGuard schema_guard(&c_schema);
+  internal::ArrowArrayGuard array_guard(&c_array);
+  ASSERT_TRUE(::arrow::ExportType(*map_type, &c_schema).ok());
+  ASSERT_TRUE(::arrow::ExportArray(*sliced, &c_array).ok());
+  ASSERT_EQ(c_array.offset, 1);
+  ASSERT_EQ(c_array.children[0]->offset, 1);
+
+  ICEBERG_UNWRAP_OR_FAIL(auto map_like, ArrowArrayMapLike::Make(c_schema, c_array));
+
+  constexpr int64_t kNumRows = 3;
+  std::array<std::vector<std::pair<std::string, int32_t>>, kNumRows> expected_maps = {
+      std::vector<std::pair<std::string, int32_t>>{{"Foo", 1}, {"Bar", 2}},
+      std::vector<std::pair<std::string, int32_t>>{{"Baz", 3}},
+      std::vector<std::pair<std::string, int32_t>>{},
+  };
+
+  for (int64_t i = 0; i < kNumRows; ++i) {
+    ASSERT_THAT(map_like->Reset(i), IsOk());
+    const auto& expected_map = expected_maps[i];
+    ASSERT_EQ(map_like->size(), expected_map.size());
+    for (size_t j = 0; j < expected_map.size(); ++j) {
+      EXPECT_SCALAR_EQ(map_like->GetKey(j), std::string_view, expected_map[j].first);
+      EXPECT_SCALAR_EQ(map_like->GetValue(j), int32_t, expected_map[j].second);
+    }
+  }
+}
+
 }  // namespace iceberg