Merge pull request #70957 from bigo-sg/tostring-fix

Function cast support Spark text output

src/Core/FormatFactorySettings.h

@@ -1242,6 +1242,9 @@ Set the quoting rule for identifiers in SHOW CREATE query
 )", 0) \
     DECLARE(IdentifierQuotingStyle, show_create_query_identifier_quoting_style, IdentifierQuotingStyle::Backticks, R"(
 Set the quoting style for identifiers in SHOW CREATE query
+)", 0) \
+    DECLARE(String, composed_data_type_output_format_mode, "default", R"(
+Set composed data type output format mode, default or spark.
 )", 0) \
 
 // End of FORMAT_FACTORY_SETTINGS

src/Core/SettingsChangesHistory.cpp

@@ -67,6 +67,7 @@ static std::initializer_list<std::pair<ClickHouseVersion, SettingsChangesHistory
             {"max_bytes_ratio_before_external_group_by", 0., 0., "New setting."},
             {"max_bytes_ratio_before_external_sort", 0., 0., "New setting."},
             {"use_async_executor_for_materialized_views", false, false, "New setting."},
+            {"composed_data_type_output_format_mode", "default", "default", "New setting"},
             {"http_response_headers", "", "", "New setting."},
         }
     },

src/DataTypes/Serializations/SerializationArray.cpp

@@ -401,7 +401,7 @@ void SerializationArray::deserializeBinaryBulkWithMultipleStreams(
 
 
 template <typename Writer>
-static void serializeTextImpl(const IColumn & column, size_t row_num, WriteBuffer & ostr, Writer && write_nested)
+static void serializeTextImpl(const IColumn & column, size_t row_num, const FormatSettings & settings, WriteBuffer & ostr, Writer && write_nested)
 {
     const ColumnArray & column_array = assert_cast<const ColumnArray &>(column);
     const ColumnArray::Offsets & offsets = column_array.getOffsets();
@@ -412,10 +412,14 @@ static void serializeTextImpl(const IColumn & column, size_t row_num, WriteBuffe
     const IColumn & nested_column = column_array.getData();
 
     writeChar('[', ostr);
-    for (size_t i = offset; i < next_offset; ++i)
+
+    if (next_offset != offset)
+        write_nested(nested_column, offset);
+    for (size_t i = offset + 1; i < next_offset; ++i)
     {
-        if (i != offset)
+        writeChar(',', ostr);
-            writeChar(',', ostr);
+        if (settings.composed_data_type_output_format_mode == "spark")
+            writeChar(' ', ostr);
         write_nested(nested_column, i);
     }
     writeChar(']', ostr);
@@ -520,10 +524,13 @@ static ReturnType deserializeTextImpl(IColumn & column, ReadBuffer & istr, Reade
 
 void SerializationArray::serializeText(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const
 {
-    serializeTextImpl(column, row_num, ostr,
+    serializeTextImpl(column, row_num, settings, ostr,
         [&](const IColumn & nested_column, size_t i)
         {
-            nested->serializeTextQuoted(nested_column, i, ostr, settings);
+            if (settings.composed_data_type_output_format_mode == "spark")
+                nested->serializeText(nested_column, i, ostr, settings);
+            else
+                nested->serializeTextQuoted(nested_column, i, ostr, settings);
         });
 }
 

src/DataTypes/Serializations/SerializationMap.cpp

@@ -90,6 +90,7 @@ template <typename KeyWriter, typename ValueWriter>
 void SerializationMap::serializeTextImpl(
     const IColumn & column,
     size_t row_num,
+    const FormatSettings & settings,
     WriteBuffer & ostr,
     KeyWriter && key_writer,
     ValueWriter && value_writer) const
@@ -104,15 +105,31 @@ void SerializationMap::serializeTextImpl(
     size_t next_offset = offsets[row_num];
 
     writeChar('{', ostr);
-    for (size_t i = offset; i < next_offset; ++i)
+    if (offset != next_offset)
     {
-        if (i != offset)
+        key_writer(ostr, key, nested_tuple.getColumn(0), offset);
-            writeChar(',', ostr);
+        if (settings.composed_data_type_output_format_mode == "spark")
-
+            writeString(std::string_view(" -> "), ostr);
-        key_writer(ostr, key, nested_tuple.getColumn(0), i);
+        else
-        writeChar(':', ostr);
+            writeChar(':', ostr);
-        value_writer(ostr, value, nested_tuple.getColumn(1), i);
+        value_writer(ostr, value, nested_tuple.getColumn(1), offset);
     }
+    if (settings.composed_data_type_output_format_mode == "spark")
+        for (size_t i = offset + 1; i < next_offset; ++i)
+        {
+            writeString(std::string_view(", "), ostr);
+            key_writer(ostr, key, nested_tuple.getColumn(0), i);
+            writeString(std::string_view(" -> "), ostr);
+            value_writer(ostr, value, nested_tuple.getColumn(1), i);
+        }
+    else
+        for (size_t i = offset + 1; i < next_offset; ++i)
+        {
+            writeChar(',', ostr);
+            key_writer(ostr, key, nested_tuple.getColumn(0), i);
+            writeChar(':', ostr);
+            value_writer(ostr, value, nested_tuple.getColumn(1), i);
+        }
     writeChar('}', ostr);
 }
 
@@ -221,10 +238,13 @@ void SerializationMap::serializeText(const IColumn & column, size_t row_num, Wri
 {
     auto writer = [&settings](WriteBuffer & buf, const SerializationPtr & subcolumn_serialization, const IColumn & subcolumn, size_t pos)
     {
-        subcolumn_serialization->serializeTextQuoted(subcolumn, pos, buf, settings);
+        if (settings.composed_data_type_output_format_mode == "spark")
+            subcolumn_serialization->serializeText(subcolumn, pos, buf, settings);
+        else
+            subcolumn_serialization->serializeTextQuoted(subcolumn, pos, buf, settings);
     };
 
-    serializeTextImpl(column, row_num, ostr, writer, writer);
+    serializeTextImpl(column, row_num, settings, ostr, writer, writer);
 }
 
 void SerializationMap::deserializeText(IColumn & column, ReadBuffer & istr, const FormatSettings & settings, bool whole) const
@@ -266,7 +286,7 @@ bool SerializationMap::tryDeserializeText(IColumn & column, ReadBuffer & istr, c
 
 void SerializationMap::serializeTextJSON(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const
 {
-    serializeTextImpl(column, row_num, ostr,
+    serializeTextImpl(column, row_num, settings, ostr,
         [&settings](WriteBuffer & buf, const SerializationPtr & subcolumn_serialization, const IColumn & subcolumn, size_t pos)
         {
             /// We need to double-quote all keys (including integers) to produce valid JSON.

src/DataTypes/Serializations/SerializationMap.h

@@ -70,7 +70,7 @@ public:
 
 private:
     template <typename KeyWriter, typename ValueWriter>
-    void serializeTextImpl(const IColumn & column, size_t row_num, WriteBuffer & ostr, KeyWriter && key_writer, ValueWriter && value_writer) const;
+    void serializeTextImpl(const IColumn & column, size_t row_num, const FormatSettings & settings, WriteBuffer & ostr, KeyWriter && key_writer, ValueWriter && value_writer) const;
 
     template <typename ReturnType = void, typename Reader>
     ReturnType deserializeTextImpl(IColumn & column, ReadBuffer & istr, Reader && reader) const;

src/DataTypes/Serializations/SerializationTuple.cpp

@@ -137,12 +137,25 @@ void SerializationTuple::deserializeBinary(IColumn & column, ReadBuffer & istr,
 void SerializationTuple::serializeText(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const
 {
     writeChar('(', ostr);
-    for (size_t i = 0; i < elems.size(); ++i)
+    if (!elems.empty())
     {
-        if (i != 0)
+        if (settings.composed_data_type_output_format_mode == "spark")
-            writeChar(',', ostr);
+            elems[0]->serializeText(extractElementColumn(column, 0), row_num, ostr, settings);
-        elems[i]->serializeTextQuoted(extractElementColumn(column, i), row_num, ostr, settings);
+        else
+            elems[0]->serializeTextQuoted(extractElementColumn(column, 0), row_num, ostr, settings);
     }
+    if (settings.composed_data_type_output_format_mode == "spark")
+        for (size_t i = 1; i < elems.size(); ++i)
+        {
+            writeString(std::string_view(", "), ostr);
+            elems[i]->serializeText(extractElementColumn(column, i), row_num, ostr, settings);
+        }
+    else
+        for (size_t i = 1; i < elems.size(); ++i)
+        {
+            writeChar(',', ostr);
+            elems[i]->serializeTextQuoted(extractElementColumn(column, i), row_num, ostr, settings);
+        }
     writeChar(')', ostr);
 }
 

src/Formats/FormatFactory.cpp

@@ -251,6 +251,7 @@ FormatSettings getFormatSettings(const ContextPtr & context, const Settings & se
     format_settings.values.deduce_templates_of_expressions = settings[Setting::input_format_values_deduce_templates_of_expressions];
     format_settings.values.interpret_expressions = settings[Setting::input_format_values_interpret_expressions];
     format_settings.values.escape_quote_with_quote = settings[Setting::output_format_values_escape_quote_with_quote];
+    format_settings.composed_data_type_output_format_mode = settings[Setting::composed_data_type_output_format_mode];
     format_settings.with_names_use_header = settings[Setting::input_format_with_names_use_header];
     format_settings.with_types_use_header = settings[Setting::input_format_with_types_use_header];
     format_settings.write_statistics = settings[Setting::output_format_write_statistics];

src/Formats/FormatSettings.h

@@ -38,6 +38,7 @@ struct FormatSettings
     bool try_infer_variant = false;
 
     bool seekable_read = true;
+    String composed_data_type_output_format_mode = "default";
     UInt64 max_rows_to_read_for_schema_inference = 25000;
     UInt64 max_bytes_to_read_for_schema_inference = 32 * 1024 * 1024;
 

tests/queries/0_stateless/03259_to_string_spark_format.reference

@@ -0,0 +1,16 @@
+-- array format --
+[\'1\']
+[1, 2, abc, \'1\']
+[1, 2, abc, \'1\']
+[1, 2, abc, \'1\']
+[1, 2, abc, \'1\']
+-- map format --
+{1343 -> fe, afe -> fefe}
+{1343 -> fe, afe -> fefe}
+{1343 -> fe, afe -> fefe}
+{1343 -> fe, afe -> fefe}
+-- tuple format --
+(1, 3, abc)
+(1, 3, abc)
+(1, 3, abc)
+(1, 3, abc)

tests/queries/0_stateless/03259_to_string_spark_format.sql

@@ -0,0 +1,18 @@
+SELECT '-- array format --';
+SELECT CAST(array('\'1\'') , 'String') SETTINGS composed_data_type_output_format_mode = 'spark';
+SELECT CAST([materialize('1'), '2', 'abc', '\'1\''], 'String') SETTINGS composed_data_type_output_format_mode = 'spark';
+SELECT CAST([materialize('1'), materialize('2'), 'abc', '\'1\''], 'String') SETTINGS composed_data_type_output_format_mode = 'spark';
+SELECT CAST([materialize('1'), materialize('2'), materialize('abc'), '\'1\''], 'String') SETTINGS composed_data_type_output_format_mode = 'spark';
+SELECT CAST([materialize('1'), materialize('2'), materialize('abc'), materialize('\'1\'')], 'String') SETTINGS composed_data_type_output_format_mode = 'spark';
+
+SELECT '-- map format --';
+SELECT toString(map('1343', 'fe', 'afe', 'fefe')) SETTINGS composed_data_type_output_format_mode = 'spark';
+SELECT toString(map(materialize('1343'), materialize('fe'), 'afe', 'fefe')) SETTINGS composed_data_type_output_format_mode = 'spark';
+SELECT toString(map(materialize('1343'), materialize('fe'), materialize('afe'), 'fefe')) SETTINGS composed_data_type_output_format_mode = 'spark';
+SELECT toString(map(materialize('1343'), materialize('fe'), materialize('afe'), materialize('fefe'))) SETTINGS composed_data_type_output_format_mode = 'spark';
+
+SELECT '-- tuple format --';
+SELECT toString(('1', '3', 'abc')) SETTINGS composed_data_type_output_format_mode = 'spark';
+SELECT toString((materialize('1'), '3', 'abc')) SETTINGS composed_data_type_output_format_mode = 'spark';
+SELECT toString((materialize('1'), materialize('3'), 'abc')) SETTINGS composed_data_type_output_format_mode = 'spark';
+SELECT toString((materialize('1'), materialize('3'), materialize('abc'))) SETTINGS composed_data_type_output_format_mode = 'spark';