Merge pull request #1034 from ClickHouse/json_string_column

feat: json string column

Author: ernado

compress/_golden/data_compressed_lz4hc.hex

@@ -0,0 +1,3 @@
+00000000  34 39 ad b3 8d 96 d2 87  bb 3b aa 1e 3f 4b 64 f5  |49.......;..?Kd.|
+00000010  82 1d 00 00 00 af 00 00  00 8f 48 65 6c 6c 6f 21  |..........Hello!|
+00000020  0a 48 07 00 8d 70 48 65  6c 6c 6f 21 0a           |.H...pHello!.|

compress/_golden/data_compressed_lz4hc.raw

@@ -0,0 +1,6 @@
+00000000  01 00 00 00 00 00 00 00  11 7b 22 78 22 3a 20 35  |.........{"x": 5|
+00000010  2c 20 22 79 22 3a 20 31  30 7d 1b 7b 22 61 22 3a  |, "y": 10}.{"a":|
+00000020  20 22 74 65 73 74 22 2c  20 22 62 22 3a 20 22 74  | "test", "b": "t|
+00000030  65 73 74 32 22 7d 21 7b  22 61 22 3a 20 22 6f 62  |est2"}!{"a": "ob|
+00000040  6a 20 74 65 73 74 22 2c  20 22 62 22 3a 20 7b 22  |j test", "b": {"|
+00000050  63 22 3a 20 32 30 7d 7d                           |c": 20}}|

proto/_golden/col_json_str.hex

@@ -0,0 +1,6 @@
+00000000  01 00 00 00 00 00 00 00  11 7b 22 78 22 3a 20 35  |.........{"x": 5|
+00000010  2c 20 22 79 22 3a 20 31  30 7d 1b 7b 22 61 22 3a  |, "y": 10}.{"a":|
+00000020  20 22 74 65 73 74 22 2c  20 22 62 22 3a 20 22 74  | "test", "b": "t|
+00000030  65 73 74 32 22 7d 21 7b  22 61 22 3a 20 22 6f 62  |est2"}!{"a": "ob|
+00000040  6a 20 74 65 73 74 22 2c  20 22 62 22 3a 20 7b 22  |j test", "b": {"|
+00000050  63 22 3a 20 32 30 7d 7d                           |c": 20}}|

proto/_golden/col_json_str.raw

@@ -0,0 +1,6 @@
+00000000  01 00 00 00 00 00 00 00  11 7b 22 78 22 3a 20 35  |.........{"x": 5|
+00000010  2c 20 22 79 22 3a 20 31  30 7d 1b 7b 22 61 22 3a  |, "y": 10}.{"a":|
+00000020  20 22 74 65 73 74 22 2c  20 22 62 22 3a 20 22 74  | "test", "b": "t|
+00000030  65 73 74 32 22 7d 21 7b  22 61 22 3a 20 22 6f 62  |est2"}!{"a": "ob|
+00000040  6a 20 74 65 73 74 22 2c  20 22 62 22 3a 20 7b 22  |j test", "b": {"|
+00000050  63 22 3a 20 32 30 7d 7d                           |c": 20}}|

proto/_golden/col_json_str_bytes.hex

@@ -0,0 +1,158 @@
+package proto
+
+import (
+	"github.com/go-faster/errors"
+)
+
+const JSONStringSerializationVersion uint64 = 1
+
+// ColJSONStr represents String column.
+//
+// Use ColJSONBytes for []bytes ColumnOf implementation.
+type ColJSONStr struct {
+	str ColStr
+}
+
+// Append string to column.
+func (c *ColJSONStr) Append(v string) {
+	c.str.Append(v)
+}
+
+// AppendBytes append byte slice as string to column.
+func (c *ColJSONStr) AppendBytes(v []byte) {
+	c.str.AppendBytes(v)
+}
+
+func (c *ColJSONStr) AppendArr(v []string) {
+	c.str.AppendArr(v)
+}
+
+// Compile-time assertions for ColJSONStr.
+var (
+	_ ColInput          = ColJSONStr{}
+	_ ColResult         = (*ColJSONStr)(nil)
+	_ Column            = (*ColJSONStr)(nil)
+	_ ColumnOf[string]  = (*ColJSONStr)(nil)
+	_ Arrayable[string] = (*ColJSONStr)(nil)
+)
+
+// Type returns ColumnType of JSON.
+func (ColJSONStr) Type() ColumnType {
+	return ColumnTypeJSON
+}
+
+// Rows returns count of rows in column.
+func (c ColJSONStr) Rows() int {
+	return c.str.Rows()
+}
+
+// Reset resets data in row, preserving capacity for efficiency.
+func (c *ColJSONStr) Reset() {
+	c.str.Reset()
+}
+
+// EncodeColumn encodes String rows to *Buffer.
+func (c ColJSONStr) EncodeColumn(b *Buffer) {
+	b.PutUInt64(JSONStringSerializationVersion)
+
+	c.str.EncodeColumn(b)
+}
+
+// WriteColumn writes JSON rows to *Writer.
+func (c ColJSONStr) WriteColumn(w *Writer) {
+	w.ChainBuffer(func(b *Buffer) {
+		b.PutUInt64(JSONStringSerializationVersion)
+	})
+
+	c.str.WriteColumn(w)
+}
+
+// ForEach calls f on each string from column.
+func (c ColJSONStr) ForEach(f func(i int, s string) error) error {
+	return c.str.ForEach(f)
+}
+
+// First returns the first row of the column.
+func (c ColJSONStr) First() string {
+	return c.str.First()
+}
+
+// Row returns row with number i.
+func (c ColJSONStr) Row(i int) string {
+	return c.str.Row(i)
+}
+
+// RowBytes returns row with number i as byte slice.
+func (c ColJSONStr) RowBytes(i int) []byte {
+	return c.str.RowBytes(i)
+}
+
+// ForEachBytes calls f on each string from column as byte slice.
+func (c ColJSONStr) ForEachBytes(f func(i int, b []byte) error) error {
+	return c.str.ForEachBytes(f)
+}
+
+// DecodeColumn decodes String rows from *Reader.
+func (c *ColJSONStr) DecodeColumn(r *Reader, rows int) error {
+	jsonSerializationVersion, err := r.UInt64()
+	if err != nil {
+		return errors.Wrap(err, "failed to read json serialization version")
+	}
+
+	if jsonSerializationVersion != JSONStringSerializationVersion {
+		return errors.Errorf("received invalid JSON string serialization version %d. Setting \"output_format_native_write_json_as_string\" must be enabled.", jsonSerializationVersion)
+	}
+
+	return c.str.DecodeColumn(r, rows)
+}
+
+// LowCardinality returns LowCardinality(JSON).
+func (c *ColJSONStr) LowCardinality() *ColLowCardinality[string] {
+	return c.str.LowCardinality()
+}
+
+// Array is helper that creates Array(JSON).
+func (c *ColJSONStr) Array() *ColArr[string] {
+	return c.str.Array()
+}
+
+// Nullable is helper that creates Nullable(JSON).
+func (c *ColJSONStr) Nullable() *ColNullable[string] {
+	return c.str.Nullable()
+}
+
+// ColJSONBytes is ColJSONStr wrapper to be ColumnOf for []byte.
+type ColJSONBytes struct {
+	ColJSONStr
+}
+
+// Row returns row with number i.
+func (c ColJSONBytes) Row(i int) []byte {
+	return c.RowBytes(i)
+}
+
+// Append byte slice to column.
+func (c *ColJSONBytes) Append(v []byte) {
+	c.AppendBytes(v)
+}
+
+// AppendArr append slice of byte slices to column.
+func (c *ColJSONBytes) AppendArr(v [][]byte) {
+	for _, s := range v {
+		c.Append(s)
+	}
+}
+
+// Array is helper that creates Array(JSON).
+func (c *ColJSONBytes) Array() *ColArr[[]byte] {
+	return &ColArr[[]byte]{
+		Data: c,
+	}
+}
+
+// Nullable is helper that creates Nullable(JSON).
+func (c *ColJSONBytes) Nullable() *ColNullable[[]byte] {
+	return &ColNullable[[]byte]{
+		Values: c,
+	}
+}

proto/_golden/col_json_str_bytes.raw

@@ -0,0 +1,150 @@
+package proto
+
+import (
+	"bytes"
+	"io"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+
+	"github.com/ClickHouse/ch-go/internal/gold"
+)
+
+var testJSONValues = []string{
+	"{\"x\": 5, \"y\": 10}",
+	"{\"a\": \"test\", \"b\": \"test2\"}",
+	"{\"a\": \"obj test\", \"b\": {\"c\": 20}}",
+}
+
+func TestColJSONBytes(t *testing.T) {
+	testColumn(t, "json_bytes", func() ColumnOf[[]byte] {
+		return new(ColJSONBytes)
+	}, []byte(testJSONValues[0]), []byte(testJSONValues[1]), []byte(testJSONValues[2]))
+}
+
+func TestColJSONStr_AppendBytes(t *testing.T) {
+	var data ColJSONStr
+
+	data.AppendBytes([]byte(testJSONValues[0]))
+	data.AppendBytes([]byte(testJSONValues[1]))
+	data.AppendBytes([]byte(testJSONValues[2]))
+
+	var buf Buffer
+	data.EncodeColumn(&buf)
+
+	t.Run("Golden", func(t *testing.T) {
+		gold.Bytes(t, buf.Buf, "col_json_str_bytes")
+	})
+	t.Run("Ok", func(t *testing.T) {
+		br := bytes.NewReader(buf.Buf)
+		r := NewReader(br)
+
+		var dec ColJSONStr
+		require.NoError(t, dec.DecodeColumn(r, 3))
+		require.Equal(t, data, dec)
+
+		t.Run("ForEach", func(t *testing.T) {
+			var output []string
+			f := func(i int, s string) error {
+				output = append(output, s)
+				return nil
+			}
+			require.NoError(t, dec.ForEach(f))
+			require.Equal(t, testJSONValues, output)
+		})
+	})
+}
+
+func TestColJSONStr_EncodeColumn(t *testing.T) {
+	var data ColJSONStr
+
+	input := testJSONValues
+	rows := len(input)
+	for _, s := range input {
+		data.Append(s)
+	}
+
+	var buf Buffer
+	data.EncodeColumn(&buf)
+
+	t.Run("Golden", func(t *testing.T) {
+		gold.Bytes(t, buf.Buf, "col_json_str")
+	})
+	t.Run("Ok", func(t *testing.T) {
+		br := bytes.NewReader(buf.Buf)
+		r := NewReader(br)
+
+		var dec ColJSONStr
+		require.NoError(t, dec.DecodeColumn(r, rows))
+		require.Equal(t, data, dec)
+
+		t.Run("ForEach", func(t *testing.T) {
+			var output []string
+			f := func(i int, s string) error {
+				output = append(output, s)
+				return nil
+			}
+			require.NoError(t, dec.ForEach(f))
+			require.Equal(t, input, output)
+		})
+	})
+	t.Run("EOF", func(t *testing.T) {
+		r := NewReader(bytes.NewReader(nil))
+
+		var dec ColJSONStr
+		require.ErrorIs(t, dec.DecodeColumn(r, rows), io.EOF)
+	})
+}
+
+func BenchmarkColJSONStr_DecodeColumn(b *testing.B) {
+	const rows = 1_000
+	var data ColJSONStr
+	for i := 0; i < rows; i++ {
+		data.Append("{\"x\": 5}")
+	}
+
+	var buf Buffer
+	data.EncodeColumn(&buf)
+
+	br := bytes.NewReader(buf.Buf)
+	r := NewReader(br)
+
+	var dec ColJSONStr
+	if err := dec.DecodeColumn(r, rows); err != nil {
+		b.Fatal(err)
+	}
+
+	b.SetBytes(int64(len(buf.Buf)))
+	b.ResetTimer()
+	b.ReportAllocs()
+
+	for i := 0; i < b.N; i++ {
+		br.Reset(buf.Buf)
+		r.raw.Reset(br)
+		dec.Reset()
+
+		if err := dec.DecodeColumn(r, rows); err != nil {
+			b.Fatal(err)
+		}
+	}
+}
+
+func BenchmarkColJSONStr_EncodeColumn(b *testing.B) {
+	const rows = 1_000
+	var data ColJSONStr
+	for i := 0; i < rows; i++ {
+		data.Append("{\"x\": 5}")
+	}
+
+	var buf Buffer
+	data.EncodeColumn(&buf)
+
+	b.SetBytes(int64(len(buf.Buf)))
+	b.ResetTimer()
+	b.ReportAllocs()
+
+	for i := 0; i < b.N; i++ {
+		buf.Reset()
+		data.EncodeColumn(&buf)
+	}
+}

proto/_golden/column_of_json_bytes.hex

@@ -250,6 +250,7 @@ const (
 	ColumnTypePoint          ColumnType = "Point"
 	ColumnTypeInterval       ColumnType = "Interval"
 	ColumnTypeNothing        ColumnType = "Nothing"
+	ColumnTypeJSON           ColumnType = "JSON"
 )
 
 // colWrap wraps Column with type t.