Merge pull request #551 from egor-ryashin/arrow-v12

Migrate Arrow to v12

common/common.go

@@ -9,8 +9,9 @@ import (
 	"strconv"
 	"strings"
 
-	"github.com/apache/arrow/go/arrow"
-	"github.com/apache/arrow/go/arrow/array"
+	"github.com/apache/arrow/go/v12/arrow"
+	"github.com/apache/arrow/go/v12/arrow/array"
+
 	"github.com/xitongsys/parquet-go/parquet"
 )
 
@@ -530,7 +531,7 @@ func DeepCopy(src, dst interface{}) {
 	return
 }
 
-//Get key tag map for map
+// Get key tag map for map
 func GetKeyTagMap(src *Tag) *Tag {
 	res := NewTag()
 	res.InName = "Key"
@@ -548,7 +549,7 @@ func GetKeyTagMap(src *Tag) *Tag {
 	return res
 }
 
-//Get value tag map for map
+// Get value tag map for map
 func GetValueTagMap(src *Tag) *Tag {
 	res := NewTag()
 	res.InName = "Value"
@@ -566,7 +567,7 @@ func GetValueTagMap(src *Tag) *Tag {
 	return res
 }
 
-//Convert string to a golang variable name
+// Convert string to a golang variable name
 func StringToVariableName(str string) string {
 	ln := len(str)
 	if ln <= 0 {
@@ -588,7 +589,7 @@ func StringToVariableName(str string) string {
 	return name
 }
 
-//Convert the first letter of a string to uppercase
+// Convert the first letter of a string to uppercase
 func HeadToUpper(str string) string {
 	ln := len(str)
 	if ln <= 0 {
@@ -928,7 +929,7 @@ func (table decimalStringFuncTable) MinMaxSize(minVal interface{}, maxVal interf
 	return Min(table, minVal, val), Max(table, maxVal, val), int32(len(val.(string)))
 }
 
-//Get the size of a parquet value
+// Get the size of a parquet value
 func SizeOf(val reflect.Value) int64 {
 	var size int64
 	switch val.Type().Kind() {
@@ -977,17 +978,17 @@ func ReformPathStr(pathStr string) string {
 	return strings.ReplaceAll(pathStr, ".", "\x01")
 }
 
-//Convert path slice to string
+// Convert path slice to string
 func PathToStr(path []string) string {
 	return strings.Join(path, PAR_GO_PATH_DELIMITER)
 }
 
-//Convert string to path slice
+// Convert string to path slice
 func StrToPath(str string) []string {
 	return strings.Split(str, PAR_GO_PATH_DELIMITER)
 }
 
-//Get the pathStr index in a path
+// Get the pathStr index in a path
 func PathStrIndex(str string) int {
 	return len(strings.Split(str, PAR_GO_PATH_DELIMITER))
 }
@@ -1030,7 +1031,7 @@ func TransposeTable(table [][]interface{}) [][]interface{} {
 //
 // If `col` contains Null value but `field` is not marked as Nullable this
 // results in an error.
-func ArrowColToParquetCol(field arrow.Field, col array.Interface) (
+func ArrowColToParquetCol(field arrow.Field, col arrow.Array) (
 	[]interface{}, error) {
 	recs := make([]interface{}, col.Len())
 	switch field.Type.(type) {

example/arrow_to_parquet.go

@@ -5,9 +5,9 @@ import (
 	"log"
 	"time"
 
-	"github.com/apache/arrow/go/arrow"
-	"github.com/apache/arrow/go/arrow/array"
-	"github.com/apache/arrow/go/arrow/memory"
+	"github.com/apache/arrow/go/v12/arrow"
+	"github.com/apache/arrow/go/v12/arrow/array"
+	"github.com/apache/arrow/go/v12/arrow/memory"
 	"github.com/xitongsys/parquet-go-source/local"
 	"github.com/xitongsys/parquet-go/reader"
 	"github.com/xitongsys/parquet-go/writer"

go.mod

@@ -3,13 +3,13 @@ module github.com/xitongsys/parquet-go
 go 1.16
 
 require (
-	github.com/apache/arrow/go/arrow v0.0.0-20200730104253-651201b0f516
-	github.com/apache/thrift v0.14.2
+	github.com/apache/arrow/go/v12 v12.0.1
+	github.com/apache/thrift v0.16.0
 	github.com/aws/aws-sdk-go v1.30.19
 	github.com/goccy/go-reflect v1.2.0
-	github.com/golang/snappy v0.0.3
-	github.com/klauspost/compress v1.13.1
-	github.com/pierrec/lz4/v4 v4.1.8
-	github.com/stretchr/testify v1.7.0
+	github.com/golang/snappy v0.0.4
+	github.com/klauspost/compress v1.15.9
+	github.com/pierrec/lz4/v4 v4.1.15
+	github.com/stretchr/testify v1.8.0
 	github.com/xitongsys/parquet-go-source v0.0.0-20200817004010-026bad9b25d0
 )

schema/arrow.go

@@ -3,7 +3,7 @@ package schema
 import (
 	"fmt"
 
-	"github.com/apache/arrow/go/arrow"
+	"github.com/apache/arrow/go/v12/arrow"
 	"github.com/xitongsys/parquet-go/common"
 	"github.com/xitongsys/parquet-go/parquet"
 )

schema/arrow_test.go

@@ -3,7 +3,7 @@ package schema
 import (
 	"testing"
 
-	"github.com/apache/arrow/go/arrow"
+	"github.com/apache/arrow/go/v12/arrow"
 	"github.com/stretchr/testify/assert"
 )
 

writer/arrow.go

@@ -3,8 +3,7 @@ package writer
 import (
 	"fmt"
 
-	"github.com/apache/arrow/go/arrow"
-	"github.com/apache/arrow/go/arrow/array"
+	"github.com/apache/arrow/go/v12/arrow"
 	"github.com/xitongsys/parquet-go/common"
 	"github.com/xitongsys/parquet-go/layout"
 	"github.com/xitongsys/parquet-go/marshal"
@@ -25,10 +24,10 @@ type ArrowWriter struct {
 	ParquetWriter
 }
 
-//NewArrowWriter creates arrow schema parquet writer given the native
-//arrow schema, parquet file writer which contains the parquet file in
-//which we will write the record along with the number of parallel threads
-//which will write in the file.
+// NewArrowWriter creates arrow schema parquet writer given the native
+// arrow schema, parquet file writer which contains the parquet file in
+// which we will write the record along with the number of parallel threads
+// which will write in the file.
 func NewArrowWriter(arrowSchema *arrow.Schema, pfile source.ParquetFile,
 	np int64) (*ArrowWriter, error) {
 	var err error
@@ -61,7 +60,7 @@ func NewArrowWriter(arrowSchema *arrow.Schema, pfile source.ParquetFile,
 // The function transforms the data from the record, which the go arrow library
 // gives as array of columns, to array of rows which the parquet-go library
 // can understand as it does not accepts data by columns, but rather by rows.
-func (w *ArrowWriter) WriteArrow(record array.Record) error {
+func (w *ArrowWriter) WriteArrow(record arrow.Record) error {
 	table := make([][]interface{}, 0)
 	for i, column := range record.Columns() {
 		columnFromRecord, err := common.ArrowColToParquetCol(