(arrow-adbc) branch main updated: fix(c/validation): compare widened ingestion values (#4778)

[lidavidm]

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

lidavidm pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/arrow-adbc.git


The following commit(s) were added to refs/heads/main by this push:
     new 86667c4d7 fix(c/validation): compare widened ingestion values (#4778)
86667c4d7 is described below

commit 86667c4d7fea767b7aeceb5639147cc6c4861f54
Author: Fredrik Fornwall <fred...@fornwall.net>
AuthorDate: Mon Sep 14 02:37:55 2026 +0200

    fix(c/validation): compare widened ingestion values (#4778)
    
    Compare ingestion values and nulls after lossless numeric widening and
    string/binary normalization, fixing checks skipped when Arrow types
    differ.
    
    The `CompareArray` comparison function seems to handle the comparisons
    just fine here.
    
    Signed-off-by: Fredrik Fornwall <fred...@fornwall.net>
---
 c/validation/adbc_validation_statement.cc | 40 +++++++++++++++++++++++++++++--
 1 file changed, 38 insertions(+), 2 deletions(-)

diff --git a/c/validation/adbc_validation_statement.cc 
b/c/validation/adbc_validation_statement.cc
index 11c802b23..e77ad0588 100644
--- a/c/validation/adbc_validation_statement.cc
+++ b/c/validation/adbc_validation_statement.cc
@@ -85,6 +85,43 @@ void StatementTest::TestRelease() {
   ASSERT_EQ(NULL, statement.private_data);
 }
 
+namespace {
+
+bool CanCompareIngestValues(ArrowType input, ArrowType output) {
+  if (input == output) return true;
+
+  switch (input) {
+    case NANOARROW_TYPE_BOOL:
+      return output == NANOARROW_TYPE_INT8 || output == NANOARROW_TYPE_INT16 ||
+             output == NANOARROW_TYPE_INT32 || output == NANOARROW_TYPE_INT64;
+    case NANOARROW_TYPE_INT8:
+    case NANOARROW_TYPE_UINT8:
+      return output == NANOARROW_TYPE_INT16 || output == NANOARROW_TYPE_INT32 
||
+             output == NANOARROW_TYPE_INT64;
+    case NANOARROW_TYPE_INT16:
+    case NANOARROW_TYPE_UINT16:
+      return output == NANOARROW_TYPE_INT32 || output == NANOARROW_TYPE_INT64;
+    case NANOARROW_TYPE_INT32:
+    case NANOARROW_TYPE_UINT32:
+      return output == NANOARROW_TYPE_INT64;
+    case NANOARROW_TYPE_HALF_FLOAT:
+      return output == NANOARROW_TYPE_FLOAT || output == NANOARROW_TYPE_DOUBLE;
+    case NANOARROW_TYPE_FLOAT:
+      return output == NANOARROW_TYPE_DOUBLE;
+    case NANOARROW_TYPE_LARGE_STRING:
+    case NANOARROW_TYPE_STRING_VIEW:
+      return output == NANOARROW_TYPE_STRING;
+    case NANOARROW_TYPE_LARGE_BINARY:
+    case NANOARROW_TYPE_BINARY_VIEW:
+    case NANOARROW_TYPE_FIXED_SIZE_BINARY:
+      return output == NANOARROW_TYPE_BINARY;
+    default:
+      return false;
+  }
+}
+
+}  // namespace
+
 template <typename CType>
 void StatementTest::TestSqlIngestType(SchemaField field,
                                       const std::vector<std::optional<CType>>& 
values,
@@ -175,8 +212,7 @@ void StatementTest::TestSqlIngestType(SchemaField field,
     ASSERT_EQ(values.size(), reader.array->length);
     ASSERT_EQ(1, reader.array->n_children);
 
-    if (round_trip_field.type == field.type) {
-      // XXX: for now we can't compare values; we would need casting
+    if (CanCompareIngestValues(field.type, round_trip_field.type)) {
       ASSERT_NO_FATAL_FAILURE(
           CompareArray<CType>(reader.array_view->children[0], values));
     }