From 919bbf57a7bca538f4a75c0e63af10555b75aa44 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@clickhouse.com>
Date: Tue, 16 Jul 2024 05:07:39 +0200
Subject: [PATCH] Fix hashing of empty tuples

---
 src/Functions/FunctionsHashing.h              | 20 ++++++++++++----
 .../03205_hashing_empty_tuples.reference      | 22 ++++++++++++++++++
 .../03205_hashing_empty_tuples.sql            | 23 +++++++++++++++++++
 3 files changed, 60 insertions(+), 5 deletions(-)
 create mode 100644 tests/queries/0_stateless/03205_hashing_empty_tuples.reference
 create mode 100644 tests/queries/0_stateless/03205_hashing_empty_tuples.sql

diff --git a/src/Functions/FunctionsHashing.h b/src/Functions/FunctionsHashing.h
index 27717ea3611..95c54ac9528 100644
--- a/src/Functions/FunctionsHashing.h
+++ b/src/Functions/FunctionsHashing.h
@@ -1184,7 +1184,7 @@ private:
 
         if (icolumn->size() != vec_to.size())
             throw Exception(ErrorCodes::LOGICAL_ERROR, "Argument column '{}' size {} doesn't match result column size {} of function {}",
-                    icolumn->getName(), icolumn->size(), vec_to.size(), getName());
+                icolumn->getName(), icolumn->size(), vec_to.size(), getName());
 
         if constexpr (Keyed)
             if (key_cols.size() != vec_to.size() && key_cols.size() != 1)
@@ -1223,6 +1223,9 @@ private:
         else executeGeneric<first>(key_cols, icolumn, vec_to);
     }
 
+    /// Return a fixed random-looking magic number when input is empty.
+    static constexpr auto filler = 0xe28dbde7fe22e41c;
+
     void executeForArgument(const KeyColumnsType & key_cols, const IDataType * type, const IColumn * column, typename ColumnVector<ToType>::Container & vec_to, bool & is_first) const
     {
         /// Flattening of tuples.
@@ -1231,6 +1234,11 @@ private:
             const auto & tuple_columns = tuple->getColumns();
             const DataTypes & tuple_types = typeid_cast<const DataTypeTuple &>(*type).getElements();
             size_t tuple_size = tuple_columns.size();
+
+            if (0 == tuple_size && is_first)
+                for (auto & hash : vec_to)
+                    hash = static_cast<ToType>(filler);
+
             for (size_t i = 0; i < tuple_size; ++i)
                 executeForArgument(key_cols, tuple_types[i].get(), tuple_columns[i].get(), vec_to, is_first);
         }
@@ -1239,6 +1247,11 @@ private:
             const auto & tuple_columns = tuple_const->getColumns();
             const DataTypes & tuple_types = typeid_cast<const DataTypeTuple &>(*type).getElements();
             size_t tuple_size = tuple_columns.size();
+
+            if (0 == tuple_size && is_first)
+                for (auto & hash : vec_to)
+                    hash = static_cast<ToType>(filler);
+
             for (size_t i = 0; i < tuple_size; ++i)
             {
                 auto tmp = ColumnConst::create(tuple_columns[i], column->size());
@@ -1300,10 +1313,7 @@ public:
             constexpr size_t first_data_argument = Keyed;
 
             if (arguments.size() <= first_data_argument)
-            {
-                /// Return a fixed random-looking magic number when input is empty
-                vec_to.assign(input_rows_count, static_cast<ToType>(0xe28dbde7fe22e41c));
-            }
+                vec_to.assign(input_rows_count, static_cast<ToType>(filler));
 
             KeyColumnsType key_cols{};
             if constexpr (Keyed)
diff --git a/tests/queries/0_stateless/03205_hashing_empty_tuples.reference b/tests/queries/0_stateless/03205_hashing_empty_tuples.reference
new file mode 100644
index 00000000000..e24b5809aee
--- /dev/null
+++ b/tests/queries/0_stateless/03205_hashing_empty_tuples.reference
@@ -0,0 +1,22 @@
+16324913028386710556
+16324913028386710556
+5049034479224883533
+7385293435322750976
+12248912094175844631
+5049034479224883533
+5887129541803688833
+5887129541803688833
+13747979201178469747
+5887129541803688833
+15520217392480966957
+16324913028386710556
+16324913028386710556
+5049034479224883533
+7385293435322750976
+12248912094175844631
+5049034479224883533
+5887129541803688833
+5887129541803688833
+13747979201178469747
+5887129541803688833
+15520217392480966957
diff --git a/tests/queries/0_stateless/03205_hashing_empty_tuples.sql b/tests/queries/0_stateless/03205_hashing_empty_tuples.sql
new file mode 100644
index 00000000000..4a97f30ced3
--- /dev/null
+++ b/tests/queries/0_stateless/03205_hashing_empty_tuples.sql
@@ -0,0 +1,23 @@
+SELECT sipHash64(());
+SELECT sipHash64((), ());
+SELECT sipHash64((), 1);
+SELECT sipHash64(1, ());
+SELECT sipHash64(1, (), 1);
+SELECT sipHash64((), 1, ());
+SELECT sipHash64((), (1, 2));
+SELECT sipHash64((), (1, 2));
+SELECT sipHash64((1, 2), ());
+SELECT sipHash64((), (1, 2), ());
+SELECT sipHash64((1, 2), (), (3, 4));
+
+SELECT sipHash64(materialize(()));
+SELECT sipHash64(materialize(()), materialize(()));
+SELECT sipHash64(materialize(()), 1);
+SELECT sipHash64(1, materialize(()));
+SELECT sipHash64(1, materialize(()), 1);
+SELECT sipHash64((), 1, materialize(()));
+SELECT sipHash64(materialize(()), (1, 2));
+SELECT sipHash64(materialize(()), (1, 2));
+SELECT sipHash64((1, 2), materialize(()));
+SELECT sipHash64(materialize(()), (1, 2), ());
+SELECT sipHash64((1, 2), materialize(()), (3, 4));