Implemented MySQL column comments support

src/Databases/MySQL/DatabaseMySQL.cpp

@@ -232,7 +232,7 @@ void DatabaseMySQL::fetchLatestTablesStructureIntoCache(
             wait_update_tables_name.emplace_back(table_modification_time.first);
     }
 
-    std::map<String, NamesAndTypesList> tables_and_columns = fetchTablesColumnsList(wait_update_tables_name, local_context);
+    std::map<String, ColumnsDescription> tables_and_columns = fetchTablesColumnsList(wait_update_tables_name, local_context);
 
     for (const auto & table_and_columns : tables_and_columns)
     {
@@ -296,7 +296,7 @@ std::map<String, UInt64> DatabaseMySQL::fetchTablesWithModificationTime(ContextP
     return tables_with_modification_time;
 }
 
-std::map<String, NamesAndTypesList>
+std::map<String, ColumnsDescription>
 DatabaseMySQL::fetchTablesColumnsList(const std::vector<String> & tables_name, ContextPtr local_context) const
 {
     const auto & settings = local_context->getSettingsRef();

src/Databases/MySQL/DatabaseMySQL.h

@@ -8,6 +8,7 @@
 #include <Core/MultiEnum.h>
 #include <Core/NamesAndTypes.h>
 #include <Common/ThreadPool.h>
+#include <Storages/ColumnsDescription.h>
 #include <Databases/DatabasesCommon.h>
 #include <Databases/MySQL/ConnectionMySQLSettings.h>
 #include <Parsers/ASTCreateQuery.h>
@@ -111,7 +112,7 @@ private:
 
     std::map<String, UInt64> fetchTablesWithModificationTime(ContextPtr local_context) const;
 
-    std::map<String, NamesAndTypesList> fetchTablesColumnsList(const std::vector<String> & tables_name, ContextPtr context) const;
+    std::map<String, ColumnsDescription> fetchTablesColumnsList(const std::vector<String> & tables_name, ContextPtr context) const;
 
     void destroyLocalCacheExtraTables(const std::map<String, UInt64> & tables_with_modification_time) const;
 

src/Databases/MySQL/FetchTablesColumnsList.cpp

@@ -40,14 +40,14 @@ String toQueryStringWithQuote(const std::vector<String> & quote_list)
 namespace DB
 {
 
-std::map<String, NamesAndTypesList> fetchTablesColumnsList(
+std::map<String, ColumnsDescription> fetchTablesColumnsList(
         mysqlxx::PoolWithFailover & pool,
         const String & database_name,
         const std::vector<String> & tables_name,
         const Settings & settings,
         MultiEnum<MySQLDataTypesSupport> type_support)
 {
-    std::map<String, NamesAndTypesList> tables_and_columns;
+    std::map<String, ColumnsDescription> tables_and_columns;
 
     if (tables_name.empty())
         return tables_and_columns;
@@ -62,6 +62,7 @@ std::map<String, NamesAndTypesList> fetchTablesColumnsList(
         { std::make_shared<DataTypeUInt64>(),   "length" },
         { std::make_shared<DataTypeUInt64>(),   "precision" },
         { std::make_shared<DataTypeUInt64>(),   "scale" },
+        { std::make_shared<DataTypeString>(),   "column_comment" },
     };
 
     WriteBufferFromOwnString query;
@@ -72,8 +73,9 @@ std::map<String, NamesAndTypesList> fetchTablesColumnsList(
              " IS_NULLABLE = 'YES' AS is_nullable,"
              " COLUMN_TYPE LIKE '%unsigned' AS is_unsigned,"
              " CHARACTER_MAXIMUM_LENGTH AS length,"
-             " NUMERIC_PRECISION as numeric_precision,"
+             " NUMERIC_PRECISION AS numeric_precision,"
-             " IF(ISNULL(NUMERIC_SCALE), DATETIME_PRECISION, NUMERIC_SCALE) AS scale" // we know DATETIME_PRECISION as a scale in CH
+             " IF(ISNULL(NUMERIC_SCALE), DATETIME_PRECISION, NUMERIC_SCALE) AS scale," // we know DATETIME_PRECISION as a scale in CH
+             " COLUMN_COMMENT AS column_comment"
              " FROM INFORMATION_SCHEMA.COLUMNS"
              " WHERE ";
 
@@ -94,21 +96,24 @@ std::map<String, NamesAndTypesList> fetchTablesColumnsList(
         const auto & char_max_length_col = *block.getByPosition(5).column;
         const auto & precision_col = *block.getByPosition(6).column;
         const auto & scale_col = *block.getByPosition(7).column;
+        const auto & column_comment_col = *block.getByPosition(8).column;
 
         size_t rows = block.rows();
         for (size_t i = 0; i < rows; ++i)
         {
             String table_name = table_name_col[i].safeGet<String>();
-            tables_and_columns[table_name].emplace_back(
+            tables_and_columns[table_name].add(
-                    column_name_col[i].safeGet<String>(),
+                    ColumnDescription(
-                    convertMySQLDataType(
+                        column_name_col[i].safeGet<String>(),
-                            type_support,
+                        convertMySQLDataType(
-                            column_type_col[i].safeGet<String>(),
+                                type_support,
-                            settings.external_table_functions_use_nulls && is_nullable_col[i].safeGet<UInt64>(),
+                                column_type_col[i].safeGet<String>(),
-                            is_unsigned_col[i].safeGet<UInt64>(),
+                                settings.external_table_functions_use_nulls && is_nullable_col[i].safeGet<UInt64>(),
-                            char_max_length_col[i].safeGet<UInt64>(),
+                                is_unsigned_col[i].safeGet<UInt64>(),
-                            precision_col[i].safeGet<UInt64>(),
+                                char_max_length_col[i].safeGet<UInt64>(),
-                            scale_col[i].safeGet<UInt64>()));
+                                precision_col[i].safeGet<UInt64>(),
+                                scale_col[i].safeGet<UInt64>()),
+                        column_comment_col[i].safeGet<String>()));
         }
     }
     return tables_and_columns;

src/Databases/MySQL/FetchTablesColumnsList.h

@@ -7,8 +7,8 @@
 
 #include <common/types.h>
 #include <Core/MultiEnum.h>
-#include <Core/NamesAndTypes.h>
 #include <Core/SettingsEnums.h>
+#include <Storages/ColumnsDescription.h>
 
 #include <map>
 #include <vector>
@@ -17,7 +17,7 @@
 namespace DB
 {
 
-std::map<String, NamesAndTypesList> fetchTablesColumnsList(
+std::map<String, ColumnsDescription> fetchTablesColumnsList(
         mysqlxx::PoolWithFailover & pool,
         const String & database_name,
         const std::vector<String> & tables_name,

src/Interpreters/MySQL/InterpretersMySQLDDLQuery.cpp

@@ -123,7 +123,6 @@ static ColumnsDescription createColumnsDescription(const NamesAndTypesList & col
             throw Exception("Columns of different size provided.", ErrorCodes::LOGICAL_ERROR);
 
     ColumnsDescription columns_description;
-    ColumnDescription column_description;
 
     for (
         auto [column_name_and_type, declare_column_ast] = std::tuple{columns_name_and_type.begin(), columns_definition->children.begin()};
@@ -139,11 +138,7 @@ static ColumnsDescription createColumnsDescription(const NamesAndTypesList & col
                 if (options->changes.count("comment"))
                     comment = options->changes.at("comment")->as<ASTLiteral>()->value.safeGet<String>();
 
-        column_description.name = column_name_and_type->name;
+        columns_description.add(ColumnDescription(column_name_and_type->name, column_name_and_type->type, comment));
-        column_description.type = column_name_and_type->type;
-        if (!comment.empty())
-            column_description.comment = std::move(comment);
-        columns_description.add(column_description);
     }
 
     return columns_description;

src/Storages/ColumnsDescription.cpp

@@ -43,8 +43,8 @@ namespace ErrorCodes
     extern const int LOGICAL_ERROR;
 }
 
-ColumnDescription::ColumnDescription(String name_, DataTypePtr type_)
+ColumnDescription::ColumnDescription(String name_, DataTypePtr type_, String comment_)
-    : name(std::move(name_)), type(std::move(type_))
+    : name(std::move(name_)), type(std::move(type_)), comment(std::move(comment_))
 {
 }
 

src/Storages/ColumnsDescription.h

@@ -39,7 +39,7 @@ struct ColumnDescription
     ColumnDescription() = default;
     ColumnDescription(ColumnDescription &&) = default;
     ColumnDescription(const ColumnDescription &) = default;
-    ColumnDescription(String name_, DataTypePtr type_);
+    ColumnDescription(String name_, DataTypePtr type_, String comment_ = "");
 
     bool operator==(const ColumnDescription & other) const;
     bool operator!=(const ColumnDescription & other) const { return !(*this == other); }

src/TableFunctions/TableFunctionMySQL.cpp

@@ -87,7 +87,7 @@ ColumnsDescription TableFunctionMySQL::getActualTableStructure(ContextPtr contex
         throw Exception("MySQL table " + (remote_database_name.empty() ? "" : (backQuote(remote_database_name) + "."))
             + backQuote(remote_table_name) + " doesn't exist.", ErrorCodes::UNKNOWN_TABLE);
 
-    return ColumnsDescription{columns->second};
+    return columns->second;
 }
 
 StoragePtr TableFunctionMySQL::executeImpl(

tests/integration/test_mysql_database_engine/test.py

@@ -167,6 +167,25 @@ def test_bad_arguments_for_mysql_database_engine(started_cluster):
         assert 'Database engine MySQL requested literal argument.' in str(exception.value)
         mysql_node.query("DROP DATABASE test_bad_arguments")
 
+def test_column_comments_for_mysql_database_engine(started_cluster):
+    with contextlib.closing(MySQLNodeInstance('root', 'clickhouse', started_cluster.mysql_ip, started_cluster.mysql_port)) as mysql_node:
+        mysql_node.query("DROP DATABASE IF EXISTS test_database")
+        mysql_node.query("CREATE DATABASE test_database DEFAULT CHARACTER SET 'utf8'")
+
+        clickhouse_node.query(
+            "CREATE DATABASE test_database ENGINE = MySQL('mysql57:3306', 'test_database', 'root', 'clickhouse')")
+        assert 'test_database' in clickhouse_node.query('SHOW DATABASES')
+
+        mysql_node.query(
+            "CREATE TABLE `test_database`.`test_table` ( `id` int(11) NOT NULL, PRIMARY KEY (`id`), `test` int COMMENT 'test comment') ENGINE=InnoDB;")
+        assert 'test comment' in clickhouse_node.query('DESCRIBE TABLE `test_database`.`test_table`')
+
+        mysql_node.query("ALTER TABLE `test_database`.`test_table` ADD COLUMN `add_column` int(11) COMMENT 'add_column comment'")
+        assert 'add_column comment' in clickhouse_node.query(
+            "SELECT comment FROM system.columns WHERE table = 'test_table' AND database = 'test_database'")
+        
+        mysql_node.query("DROP DATABASE test_database")
+        
 
 def test_data_types_support_level_for_mysql_database_engine(started_cluster):
     with contextlib.closing(MySQLNodeInstance('root', 'clickhouse', started_cluster.mysql_ip, started_cluster.mysql_port)) as mysql_node: