From 99cd25d70e066a6a612799c027955b9cee89ed86 Mon Sep 17 00:00:00 2001
From: lgbo-ustc <lgbo.ustc@gmail.com>
Date: Mon, 28 Feb 2022 20:51:33 +0800
Subject: [PATCH] add new table function: hive()

---
 src/Storages/Hive/HiveFile.cpp                |  2 ++
 src/Storages/Hive/HiveFile.h                  |  9 +++++++--
 src/TableFunctions/CMakeLists.txt             |  4 +++-
 src/TableFunctions/registerTableFunctions.cpp |  4 ++++
 src/TableFunctions/registerTableFunctions.h   |  4 ++++
 tests/integration/test_hive_query/test.py     | 14 ++++++++++++++
 6 files changed, 34 insertions(+), 3 deletions(-)

diff --git a/src/Storages/Hive/HiveFile.cpp b/src/Storages/Hive/HiveFile.cpp
index b0cfa9809e1..dffcca61a9c 100644
--- a/src/Storages/Hive/HiveFile.cpp
+++ b/src/Storages/Hive/HiveFile.cpp
@@ -8,8 +8,10 @@
 #include <arrow/api.h>
 #include <arrow/io/api.h>
 #include <arrow/status.h>
+#include <arrow/filesystem/filesystem.h>
 #include <orc/OrcFile.hh>
 #include <orc/Reader.hh>
+#include <orc/Statistics.hh>
 #include <parquet/arrow/reader.h>
 #include <parquet/file_reader.h>
 #include <parquet/statistics.h>
diff --git a/src/Storages/Hive/HiveFile.h b/src/Storages/Hive/HiveFile.h
index 63cca2562eb..6d2ba29ba0f 100644
--- a/src/Storages/Hive/HiveFile.h
+++ b/src/Storages/Hive/HiveFile.h
@@ -7,8 +7,6 @@
 #include <memory>
 
 #include <boost/algorithm/string/join.hpp>
-#include <orc/Statistics.hh>
-#include <arrow/filesystem/filesystem.h>
 
 #include <Core/Field.h>
 #include <Core/Block.h>
@@ -18,6 +16,8 @@
 namespace orc
 {
 class Reader;
+class Statistics;
+class ColumnStatistics;
 }
 
 namespace parquet
@@ -36,6 +36,11 @@ namespace io
     class RandomAccessFile;
 }
 
+namespace fs
+{
+    class FileSystem;
+}
+
 class Buffer;
 }
 
diff --git a/src/TableFunctions/CMakeLists.txt b/src/TableFunctions/CMakeLists.txt
index 576d1ea23ff..10f4ba8c4c8 100644
--- a/src/TableFunctions/CMakeLists.txt
+++ b/src/TableFunctions/CMakeLists.txt
@@ -1,9 +1,11 @@
 include("${ClickHouse_SOURCE_DIR}/cmake/dbms_glob_sources.cmake")
 add_headers_and_sources(clickhouse_table_functions .)
+add_headers_and_sources(clickhouse_table_functions Hive)
 
 list(REMOVE_ITEM clickhouse_table_functions_sources ITableFunction.cpp TableFunctionFactory.cpp)
 list(REMOVE_ITEM clickhouse_table_functions_headers ITableFunction.h TableFunctionFactory.h)
 
 add_library(clickhouse_table_functions ${clickhouse_table_functions_sources})
-target_link_libraries(clickhouse_table_functions PRIVATE clickhouse_parsers clickhouse_storages_system dbms)
+target_link_libraries(clickhouse_table_functions PRIVATE clickhouse_parsers clickhouse_storages_system dbms ch_contrib::hivemetastore ch_contrib::hdfs)
+
 
diff --git a/src/TableFunctions/registerTableFunctions.cpp b/src/TableFunctions/registerTableFunctions.cpp
index ed08972e74d..9e09fac665a 100644
--- a/src/TableFunctions/registerTableFunctions.cpp
+++ b/src/TableFunctions/registerTableFunctions.cpp
@@ -31,6 +31,10 @@ void registerTableFunctions()
     registerTableFunctionHDFSCluster(factory);
 #endif
 
+#if USE_HIVE
+    registerTableFunctionHive(factory);
+#endif
+
     registerTableFunctionODBC(factory);
     registerTableFunctionJDBC(factory);
 
diff --git a/src/TableFunctions/registerTableFunctions.h b/src/TableFunctions/registerTableFunctions.h
index 72ca185f656..e39d21cb580 100644
--- a/src/TableFunctions/registerTableFunctions.h
+++ b/src/TableFunctions/registerTableFunctions.h
@@ -29,6 +29,10 @@ void registerTableFunctionHDFS(TableFunctionFactory & factory);
 void registerTableFunctionHDFSCluster(TableFunctionFactory & factory);
 #endif
 
+#if USE_HIVE
+void registerTableFunctionHive(TableFunctionFactory & factory);
+#endif
+
 void registerTableFunctionODBC(TableFunctionFactory & factory);
 void registerTableFunctionJDBC(TableFunctionFactory & factory);
 
diff --git a/tests/integration/test_hive_query/test.py b/tests/integration/test_hive_query/test.py
index a68ae0b066d..093dd74ca64 100644
--- a/tests/integration/test_hive_query/test.py
+++ b/tests/integration/test_hive_query/test.py
@@ -107,6 +107,20 @@ def test_parquet_groupby_with_cache(started_cluster):
 2021-11-16	2
 """
     assert result == expected_result
+
+def test_parquet_groupby_by_hive_function(started_cluster):
+    logging.info('Start testing groupby ...')
+    node = started_cluster.instances['h0_0_0']
+    result = node.query("""
+    SELECT day, count(*) FROM hive('thrift://hivetest:9083', 'test', 'demo', '`id` Nullable(String), `score` Nullable(Int32), `day` Nullable(String)', 'day') group by day order by day
+            """)
+    expected_result = """2021-11-01	1
+2021-11-05	2
+2021-11-11	1
+2021-11-16	2
+"""
+    assert result == expected_result
+    
 def test_cache_read_bytes(started_cluster):
     node = started_cluster.instances['h0_0_0']
     node.query("set input_format_parquet_allow_missing_columns = true")