From 701fdd77b4622ebd83bd52c22cb2229c953a47e9 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@clickhouse.com>
Date: Fri, 16 Sep 2022 08:12:28 +0200
Subject: [PATCH] Fix wrong implementation of getDataAt for multidimensional
 arrays

---
 src/Columns/ColumnArray.cpp                    | 18 +++++++++++++-----
 ...ultidimensional_array_get_data_at.reference |  7 +++++++
 ...2423_multidimensional_array_get_data_at.sql |  7 +++++++
 3 files changed, 27 insertions(+), 5 deletions(-)
 create mode 100644 tests/queries/0_stateless/02423_multidimensional_array_get_data_at.reference
 create mode 100644 tests/queries/0_stateless/02423_multidimensional_array_get_data_at.sql

diff --git a/src/Columns/ColumnArray.cpp b/src/Columns/ColumnArray.cpp
index 7bddfc14707..960f0fa183d 100644
--- a/src/Columns/ColumnArray.cpp
+++ b/src/Columns/ColumnArray.cpp
@@ -151,22 +151,30 @@ void ColumnArray::get(size_t n, Field & res) const
 
 StringRef ColumnArray::getDataAt(size_t n) const
 {
+    assert(n < size());
+
     /** Returns the range of memory that covers all elements of the array.
       * Works for arrays of fixed length values.
       * For arrays of strings and arrays of arrays, the resulting chunk of memory may not be one-to-one correspondence with the elements,
       *  since it contains only the data laid in succession, but not the offsets.
       */
 
-    size_t offset_of_first_elem = offsetAt(n);
-    StringRef first = getData().getDataAtWithTerminatingZero(offset_of_first_elem);
-
     size_t array_size = sizeAt(n);
-    if (array_size == 0)
-        return StringRef(first.data, 0);
 
+    if (array_size == 0)
+        return StringRef(nullptr, 0);
+
+    size_t offset_of_first_elem = offsetAt(n);
     size_t offset_of_last_elem = getOffsets()[n] - 1;
+
+    StringRef first = getData().getDataAtWithTerminatingZero(offset_of_first_elem);
     StringRef last = getData().getDataAtWithTerminatingZero(offset_of_last_elem);
 
+    if (first.empty())
+        return last;
+    if (last.empty())
+        return first;
+
     return StringRef(first.data, last.data + last.size - first.data);
 }
 
diff --git a/tests/queries/0_stateless/02423_multidimensional_array_get_data_at.reference b/tests/queries/0_stateless/02423_multidimensional_array_get_data_at.reference
new file mode 100644
index 00000000000..eaf629aa9d9
--- /dev/null
+++ b/tests/queries/0_stateless/02423_multidimensional_array_get_data_at.reference
@@ -0,0 +1,7 @@
+!
+
+Hello
+
+Hello\0
+World\0
+
diff --git a/tests/queries/0_stateless/02423_multidimensional_array_get_data_at.sql b/tests/queries/0_stateless/02423_multidimensional_array_get_data_at.sql
new file mode 100644
index 00000000000..9bf7c9b49c2
--- /dev/null
+++ b/tests/queries/0_stateless/02423_multidimensional_array_get_data_at.sql
@@ -0,0 +1,7 @@
+SELECT formatRow('RawBLOB', [[[33]], []]);
+SELECT formatRow('RawBLOB', [[[]], []]);
+SELECT formatRow('RawBLOB', [[[[[[[0x48, 0x65, 0x6c, 0x6c, 0x6f]]]]]], []]);
+SELECT formatRow('RawBLOB', []::Array(Array(Nothing)));
+SELECT formatRow('RawBLOB', [[], [['Hello']]]);
+SELECT formatRow('RawBLOB', [[['World']], []]);
+SELECT formatRow('RawBLOB', []::Array(String));