concatenate should ensure null slots contain an empty string view

bkietz · Oct 10, 2023 · 8bececb · 8bececb
1 parent 15e5846
commit 8bececb
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 3 deletions.
diff --git a/cpp/src/arrow/array/concatenate.cc b/cpp/src/arrow/array/concatenate.cc
@@ -43,6 +43,7 @@
 #include "arrow/util/int_util_overflow.h"
 #include "arrow/util/logging.h"
 #include "arrow/util/ree_util.h"
+#include "arrow/visit_data_inline.h"
 #include "arrow/visit_type_inline.h"
 
 namespace arrow {
@@ -256,6 +257,16 @@ class ConcatenateImpl {
       }
     }
 
+    if (out_->buffers[0] != nullptr) {
+      i = in_[0]->length;
+      VisitNullBitmapInline(
+          out_->buffers[0]->data(), in_[0]->length, out_->length, out_->null_count,
+          [&] { ++i; },
+          [&] {
+            views[i++] = {};  // overwrite views under null bits with an empty view
+          });
+    }
+
     out_->buffers[1] = std::move(view_buffer);
     return Status::OK();
   }

diff --git a/cpp/src/arrow/visit_data_inline.h b/cpp/src/arrow/visit_data_inline.h
@@ -176,9 +176,7 @@ struct ArraySpanInlineVisitor<T, enable_if_binary_view_like<T>> {
     auto* data_buffers = arr.GetVariadicBuffers().data();
     VisitBitBlocksVoid(
         arr.buffers[0].data, arr.offset, arr.length,
-        [&](int64_t index) {
-          valid_func(util::FromBinaryView(s[index], data_buffers));
-        },
+        [&](int64_t index) { valid_func(util::FromBinaryView(s[index], data_buffers)); },
         std::forward<NullFunc>(null_func));
   }
 };