1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
|
// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package array
import (
"bytes"
"fmt"
"strings"
"sync/atomic"
"github.com/apache/arrow-go/v18/arrow"
"github.com/apache/arrow-go/v18/arrow/internal/debug"
"github.com/apache/arrow-go/v18/arrow/memory"
"github.com/apache/arrow-go/v18/internal/json"
)
// RecordReader reads a stream of records.
type RecordReader interface {
Retain()
Release()
Schema() *arrow.Schema
Next() bool
Record() arrow.Record
Err() error
}
// simpleRecords is a simple iterator over a collection of records.
type simpleRecords struct {
refCount int64
schema *arrow.Schema
recs []arrow.Record
cur arrow.Record
}
// NewRecordReader returns a simple iterator over the given slice of records.
func NewRecordReader(schema *arrow.Schema, recs []arrow.Record) (RecordReader, error) {
rs := &simpleRecords{
refCount: 1,
schema: schema,
recs: recs,
cur: nil,
}
for _, rec := range rs.recs {
rec.Retain()
}
for _, rec := range recs {
if !rec.Schema().Equal(rs.schema) {
rs.Release()
return nil, fmt.Errorf("arrow/array: mismatch schema")
}
}
return rs, nil
}
// Retain increases the reference count by 1.
// Retain may be called simultaneously from multiple goroutines.
func (rs *simpleRecords) Retain() {
atomic.AddInt64(&rs.refCount, 1)
}
// Release decreases the reference count by 1.
// When the reference count goes to zero, the memory is freed.
// Release may be called simultaneously from multiple goroutines.
func (rs *simpleRecords) Release() {
debug.Assert(atomic.LoadInt64(&rs.refCount) > 0, "too many releases")
if atomic.AddInt64(&rs.refCount, -1) == 0 {
if rs.cur != nil {
rs.cur.Release()
}
for _, rec := range rs.recs {
rec.Release()
}
rs.recs = nil
}
}
func (rs *simpleRecords) Schema() *arrow.Schema { return rs.schema }
func (rs *simpleRecords) Record() arrow.Record { return rs.cur }
func (rs *simpleRecords) Next() bool {
if len(rs.recs) == 0 {
return false
}
if rs.cur != nil {
rs.cur.Release()
}
rs.cur = rs.recs[0]
rs.recs = rs.recs[1:]
return true
}
func (rs *simpleRecords) Err() error { return nil }
// simpleRecord is a basic, non-lazy in-memory record batch.
type simpleRecord struct {
refCount int64
schema *arrow.Schema
rows int64
arrs []arrow.Array
}
// NewRecord returns a basic, non-lazy in-memory record batch.
//
// NewRecord panics if the columns and schema are inconsistent.
// NewRecord panics if rows is larger than the height of the columns.
func NewRecord(schema *arrow.Schema, cols []arrow.Array, nrows int64) arrow.Record {
rec := &simpleRecord{
refCount: 1,
schema: schema,
rows: nrows,
arrs: make([]arrow.Array, len(cols)),
}
copy(rec.arrs, cols)
for _, arr := range rec.arrs {
arr.Retain()
}
if rec.rows < 0 {
switch len(rec.arrs) {
case 0:
rec.rows = 0
default:
rec.rows = int64(rec.arrs[0].Len())
}
}
err := rec.validate()
if err != nil {
rec.Release()
panic(err)
}
return rec
}
func (rec *simpleRecord) SetColumn(i int, arr arrow.Array) (arrow.Record, error) {
if i < 0 || i >= len(rec.arrs) {
return nil, fmt.Errorf("arrow/array: column index out of range [0, %d): got=%d", len(rec.arrs), i)
}
if arr.Len() != int(rec.rows) {
return nil, fmt.Errorf("arrow/array: mismatch number of rows in column %q: got=%d, want=%d",
rec.schema.Field(i).Name,
arr.Len(), rec.rows,
)
}
f := rec.schema.Field(i)
if !arrow.TypeEqual(f.Type, arr.DataType()) {
return nil, fmt.Errorf("arrow/array: column %q type mismatch: got=%v, want=%v",
f.Name,
arr.DataType(), f.Type,
)
}
arrs := make([]arrow.Array, len(rec.arrs))
copy(arrs, rec.arrs)
arrs[i] = arr
return NewRecord(rec.schema, arrs, rec.rows), nil
}
func (rec *simpleRecord) validate() error {
if rec.rows == 0 && len(rec.arrs) == 0 {
return nil
}
if len(rec.arrs) != rec.schema.NumFields() {
return fmt.Errorf("arrow/array: number of columns/fields mismatch")
}
for i, arr := range rec.arrs {
f := rec.schema.Field(i)
if int64(arr.Len()) < rec.rows {
return fmt.Errorf("arrow/array: mismatch number of rows in column %q: got=%d, want=%d",
f.Name,
arr.Len(), rec.rows,
)
}
if !arrow.TypeEqual(f.Type, arr.DataType()) {
return fmt.Errorf("arrow/array: column %q type mismatch: got=%v, want=%v",
f.Name,
arr.DataType(), f.Type,
)
}
}
return nil
}
// Retain increases the reference count by 1.
// Retain may be called simultaneously from multiple goroutines.
func (rec *simpleRecord) Retain() {
atomic.AddInt64(&rec.refCount, 1)
}
// Release decreases the reference count by 1.
// When the reference count goes to zero, the memory is freed.
// Release may be called simultaneously from multiple goroutines.
func (rec *simpleRecord) Release() {
debug.Assert(atomic.LoadInt64(&rec.refCount) > 0, "too many releases")
if atomic.AddInt64(&rec.refCount, -1) == 0 {
for _, arr := range rec.arrs {
arr.Release()
}
rec.arrs = nil
}
}
func (rec *simpleRecord) Schema() *arrow.Schema { return rec.schema }
func (rec *simpleRecord) NumRows() int64 { return rec.rows }
func (rec *simpleRecord) NumCols() int64 { return int64(len(rec.arrs)) }
func (rec *simpleRecord) Columns() []arrow.Array { return rec.arrs }
func (rec *simpleRecord) Column(i int) arrow.Array { return rec.arrs[i] }
func (rec *simpleRecord) ColumnName(i int) string { return rec.schema.Field(i).Name }
// NewSlice constructs a zero-copy slice of the record with the indicated
// indices i and j, corresponding to array[i:j].
// The returned record must be Release()'d after use.
//
// NewSlice panics if the slice is outside the valid range of the record array.
// NewSlice panics if j < i.
func (rec *simpleRecord) NewSlice(i, j int64) arrow.Record {
arrs := make([]arrow.Array, len(rec.arrs))
for ii, arr := range rec.arrs {
arrs[ii] = NewSlice(arr, i, j)
}
defer func() {
for _, arr := range arrs {
arr.Release()
}
}()
return NewRecord(rec.schema, arrs, j-i)
}
func (rec *simpleRecord) String() string {
o := new(strings.Builder)
fmt.Fprintf(o, "record:\n %v\n", rec.schema)
fmt.Fprintf(o, " rows: %d\n", rec.rows)
for i, col := range rec.arrs {
fmt.Fprintf(o, " col[%d][%s]: %v\n", i, rec.schema.Field(i).Name, col)
}
return o.String()
}
func (rec *simpleRecord) MarshalJSON() ([]byte, error) {
arr := RecordToStructArray(rec)
defer arr.Release()
return arr.MarshalJSON()
}
// RecordBuilder eases the process of building a Record, iteratively, from
// a known Schema.
type RecordBuilder struct {
refCount int64
mem memory.Allocator
schema *arrow.Schema
fields []Builder
}
// NewRecordBuilder returns a builder, using the provided memory allocator and a schema.
func NewRecordBuilder(mem memory.Allocator, schema *arrow.Schema) *RecordBuilder {
b := &RecordBuilder{
refCount: 1,
mem: mem,
schema: schema,
fields: make([]Builder, schema.NumFields()),
}
for i := 0; i < schema.NumFields(); i++ {
b.fields[i] = NewBuilder(b.mem, schema.Field(i).Type)
}
return b
}
// Retain increases the reference count by 1.
// Retain may be called simultaneously from multiple goroutines.
func (b *RecordBuilder) Retain() {
atomic.AddInt64(&b.refCount, 1)
}
// Release decreases the reference count by 1.
func (b *RecordBuilder) Release() {
debug.Assert(atomic.LoadInt64(&b.refCount) > 0, "too many releases")
if atomic.AddInt64(&b.refCount, -1) == 0 {
for _, f := range b.fields {
f.Release()
}
b.fields = nil
}
}
func (b *RecordBuilder) Schema() *arrow.Schema { return b.schema }
func (b *RecordBuilder) Fields() []Builder { return b.fields }
func (b *RecordBuilder) Field(i int) Builder { return b.fields[i] }
func (b *RecordBuilder) Reserve(size int) {
for _, f := range b.fields {
f.Reserve(size)
}
}
// NewRecord creates a new record from the memory buffers and resets the
// RecordBuilder so it can be used to build a new record.
//
// The returned Record must be Release()'d after use.
//
// NewRecord panics if the fields' builder do not have the same length.
func (b *RecordBuilder) NewRecord() arrow.Record {
cols := make([]arrow.Array, len(b.fields))
rows := int64(0)
defer func(cols []arrow.Array) {
for _, col := range cols {
if col == nil {
continue
}
col.Release()
}
}(cols)
for i, f := range b.fields {
cols[i] = f.NewArray()
irow := int64(cols[i].Len())
if i > 0 && irow != rows {
panic(fmt.Errorf("arrow/array: field %d has %d rows. want=%d", i, irow, rows))
}
rows = irow
}
return NewRecord(b.schema, cols, rows)
}
// UnmarshalJSON for record builder will read in a single object and add the values
// to each field in the recordbuilder, missing fields will get a null and unexpected
// keys will be ignored. If reading in an array of records as a single batch, then use
// a structbuilder and use RecordFromStruct.
func (b *RecordBuilder) UnmarshalJSON(data []byte) error {
dec := json.NewDecoder(bytes.NewReader(data))
// should start with a '{'
t, err := dec.Token()
if err != nil {
return err
}
if delim, ok := t.(json.Delim); !ok || delim != '{' {
return fmt.Errorf("record should start with '{', not %s", t)
}
keylist := make(map[string]bool)
for dec.More() {
keyTok, err := dec.Token()
if err != nil {
return err
}
key := keyTok.(string)
if keylist[key] {
return fmt.Errorf("key %s shows up twice in row to be decoded", key)
}
keylist[key] = true
indices := b.schema.FieldIndices(key)
if len(indices) == 0 {
var extra interface{}
if err := dec.Decode(&extra); err != nil {
return err
}
continue
}
if err := b.fields[indices[0]].UnmarshalOne(dec); err != nil {
return err
}
}
for i := 0; i < b.schema.NumFields(); i++ {
if !keylist[b.schema.Field(i).Name] {
b.fields[i].AppendNull()
}
}
return nil
}
var (
_ arrow.Record = (*simpleRecord)(nil)
_ RecordReader = (*simpleRecords)(nil)
)
|