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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443
|
package hdfs
import (
"crypto/md5"
"errors"
"fmt"
"io"
"os"
"time"
hdfs "github.com/colinmarc/hdfs/v2/internal/protocol/hadoop_hdfs"
"github.com/colinmarc/hdfs/v2/internal/transfer"
"google.golang.org/protobuf/proto"
)
// A FileReader represents an existing file or directory in HDFS. It implements
// io.Reader, io.ReaderAt, io.Seeker, and io.Closer, and can only be used for
// reads. For writes, see FileWriter and Client.Create.
type FileReader struct {
client *Client
name string
info os.FileInfo
blocks []*hdfs.LocatedBlockProto
blockReader *transfer.BlockReader
deadline time.Time
offset int64
readdirLast string
closed bool
}
// Open returns an FileReader which can be used for reading.
func (c *Client) Open(name string) (*FileReader, error) {
info, err := c.getFileInfo(name)
if err != nil {
return nil, &os.PathError{"open", name, interpretException(err)}
}
return &FileReader{
client: c,
name: name,
info: info,
closed: false,
}, nil
}
// Name returns the name of the file.
func (f *FileReader) Name() string {
return f.info.Name()
}
// Stat returns the FileInfo structure describing file.
func (f *FileReader) Stat() os.FileInfo {
return f.info
}
// SetDeadline sets the deadline for future Read, ReadAt, and Checksum calls. A
// zero value for t means those calls will not time out.
func (f *FileReader) SetDeadline(t time.Time) error {
f.deadline = t
if f.blockReader != nil {
return f.blockReader.SetDeadline(t)
}
// Return the error at connection time.
return nil
}
// Checksum returns HDFS's internal "MD5MD5CRC32C" checksum for a given file.
//
// Internally to HDFS, it works by calculating the MD5 of all the CRCs (which
// are stored alongside the data) for each block, and then calculating the MD5
// of all of those.
func (f *FileReader) Checksum() ([]byte, error) {
if f.info.IsDir() {
return nil, &os.PathError{
"checksum",
f.name,
errors.New("is a directory"),
}
}
if f.blocks == nil {
err := f.getBlocks()
if err != nil {
return nil, err
}
}
// Hadoop calculates this by writing the checksums out to a byte array, which
// is automatically padded with zeroes out to the next power of 2
// (with a minimum of 32)... and then takes the MD5 of that array, including
// the zeroes. This is pretty shady business, but we want to track
// the 'hadoop fs -checksum' behavior if possible.
paddedLength := 32
totalLength := 0
checksum := md5.New()
for _, block := range f.blocks {
d, err := f.client.wrapDatanodeDial(f.client.options.DatanodeDialFunc,
block.GetBlockToken())
if err != nil {
return nil, err
}
cr := &transfer.ChecksumReader{
Block: block,
UseDatanodeHostname: f.client.options.UseDatanodeHostname,
DialFunc: d,
}
err = cr.SetDeadline(f.deadline)
if err != nil {
return nil, err
}
blockChecksum, err := cr.ReadChecksum()
if err != nil {
return nil, err
}
checksum.Write(blockChecksum)
totalLength += len(blockChecksum)
if paddedLength < totalLength {
paddedLength *= 2
}
}
checksum.Write(make([]byte, paddedLength-totalLength))
return checksum.Sum(nil), nil
}
// Seek implements io.Seeker.
//
// The seek is virtual - it starts a new block read at the new position.
func (f *FileReader) Seek(offset int64, whence int) (int64, error) {
if f.closed {
return 0, io.ErrClosedPipe
}
var off int64
switch whence {
case io.SeekStart:
off = offset
case io.SeekCurrent:
off = f.offset + offset
case io.SeekEnd:
off = f.info.Size() + offset
default:
return f.offset, fmt.Errorf("invalid whence: %d", whence)
}
if off < 0 || off > f.info.Size() {
return f.offset, fmt.Errorf("invalid resulting offset: %d", off)
}
if f.offset != off {
f.offset = off
if f.blockReader != nil {
// If the seek is within the next few chunks, it's much more
// efficient to throw away a few bytes than to reconnect and start
// a read at the new offset.
err := f.blockReader.Skip(f.offset - f.blockReader.Offset)
if err == nil {
return f.offset, nil
}
// It isn't possible to seek within the current block, so reset such
// that we can connect to the new block.
f.blockReader.Close()
f.blockReader = nil
}
}
return f.offset, nil
}
// Read implements io.Reader.
func (f *FileReader) Read(b []byte) (int, error) {
if f.closed {
return 0, io.ErrClosedPipe
}
if f.info.IsDir() {
return 0, &os.PathError{
"read",
f.name,
errors.New("is a directory"),
}
}
if f.offset >= f.info.Size() {
return 0, io.EOF
}
if len(b) == 0 {
return 0, nil
}
if f.blocks == nil {
err := f.getBlocks()
if err != nil {
return 0, err
}
}
for {
if f.blockReader == nil {
err := f.getNewBlockReader()
if err != nil {
return 0, err
}
}
n, err := f.blockReader.Read(b)
f.offset += int64(n)
if err != nil && err != io.EOF {
f.blockReader.Close()
f.blockReader = nil
return n, err
} else if n > 0 {
return n, nil
}
err = f.blockReader.Close()
f.blockReader = nil
if err != nil {
return n, err
}
}
}
// ReadAt implements io.ReaderAt.
func (f *FileReader) ReadAt(b []byte, off int64) (int, error) {
if f.closed {
return 0, io.ErrClosedPipe
}
if off < 0 {
return 0, &os.PathError{"readat", f.name, errors.New("negative offset")}
}
_, err := f.Seek(off, 0)
if err != nil {
return 0, err
}
n, err := io.ReadFull(f, b)
// For some reason, os.File.ReadAt returns io.EOF in this case instead of
// io.ErrUnexpectedEOF.
if err == io.ErrUnexpectedEOF {
err = io.EOF
}
return n, err
}
// Readdir reads the contents of the directory associated with file and returns
// a slice of up to n os.FileInfo values, as would be returned by Stat, in
// directory order. Subsequent calls on the same file will yield further
// os.FileInfos.
//
// If n > 0, Readdir returns at most n os.FileInfo values. In this case, if
// Readdir returns an empty slice, it will return a non-nil error explaining
// why. At the end of a directory, the error is io.EOF.
//
// If n <= 0, Readdir returns all the os.FileInfo from the directory in a single
// slice. In this case, if Readdir succeeds (reads all the way to the end of
// the directory), it returns the slice and a nil error. If it encounters an
// error before the end of the directory, Readdir returns the os.FileInfo read
// until that point and a non-nil error.
//
// The os.FileInfo values returned will not have block location attached to
// the struct returned by Sys(). To fetch that information, make a separate
// call to Stat.
//
// Note that making multiple calls to Readdir with a smallish n (as you might do
// with the os version) is slower than just requesting everything at once.
// That's because HDFS has no mechanism for limiting the number of entries
// returned; whatever extra entries it returns are simply thrown away.
func (f *FileReader) Readdir(n int) ([]os.FileInfo, error) {
if f.closed {
return nil, io.ErrClosedPipe
}
if !f.info.IsDir() {
return nil, &os.PathError{
"readdir",
f.name,
errors.New("the file is not a directory"),
}
}
if n <= 0 {
f.readdirLast = ""
}
res := make([]os.FileInfo, 0)
for {
batch, remaining, err := f.readdir()
if err != nil {
return nil, &os.PathError{"readdir", f.name, interpretException(err)}
}
if len(batch) > 0 {
f.readdirLast = batch[len(batch)-1].Name()
}
res = append(res, batch...)
if remaining == 0 || (n > 0 && len(res) >= n) {
break
}
}
if n > 0 {
if len(res) == 0 {
return nil, io.EOF
}
if len(res) > n {
res = res[:n]
f.readdirLast = res[len(res)-1].Name()
}
}
return res, nil
}
func (f *FileReader) readdir() ([]os.FileInfo, int, error) {
req := &hdfs.GetListingRequestProto{
Src: proto.String(f.name),
StartAfter: []byte(f.readdirLast),
NeedLocation: proto.Bool(false),
}
resp := &hdfs.GetListingResponseProto{}
err := f.client.namenode.Execute("getListing", req, resp)
if err != nil {
return nil, 0, err
} else if resp.GetDirList() == nil {
return nil, 0, os.ErrNotExist
}
list := resp.GetDirList().GetPartialListing()
res := make([]os.FileInfo, 0, len(list))
for _, status := range list {
res = append(res, newFileInfo(status, ""))
}
remaining := int(resp.GetDirList().GetRemainingEntries())
return res, remaining, nil
}
// Readdirnames reads and returns a slice of names from the directory f.
//
// If n > 0, Readdirnames returns at most n names. In this case, if Readdirnames
// returns an empty slice, it will return a non-nil error explaining why. At the
// end of a directory, the error is io.EOF.
//
// If n <= 0, Readdirnames returns all the names from the directory in a single
// slice. In this case, if Readdirnames succeeds (reads all the way to the end
// of the directory), it returns the slice and a nil error. If it encounters an
// error before the end of the directory, Readdirnames returns the names read
// until that point and a non-nil error.
func (f *FileReader) Readdirnames(n int) ([]string, error) {
if f.closed {
return nil, io.ErrClosedPipe
}
fis, err := f.Readdir(n)
if err != nil {
return nil, err
}
names := make([]string, 0, len(fis))
for _, fi := range fis {
names = append(names, fi.Name())
}
return names, nil
}
// Close implements io.Closer.
func (f *FileReader) Close() error {
f.closed = true
if f.blockReader != nil {
f.blockReader.Close()
}
return nil
}
func (f *FileReader) getBlocks() error {
req := &hdfs.GetBlockLocationsRequestProto{
Src: proto.String(f.name),
Offset: proto.Uint64(0),
Length: proto.Uint64(uint64(f.info.Size())),
}
resp := &hdfs.GetBlockLocationsResponseProto{}
err := f.client.namenode.Execute("getBlockLocations", req, resp)
if err != nil {
return err
}
f.blocks = resp.GetLocations().GetBlocks()
return nil
}
func (f *FileReader) getNewBlockReader() error {
off := uint64(f.offset)
for _, block := range f.blocks {
start := block.GetOffset()
end := start + block.GetB().GetNumBytes()
if start <= off && off < end {
dialFunc, err := f.client.wrapDatanodeDial(
f.client.options.DatanodeDialFunc,
block.GetBlockToken())
if err != nil {
return err
}
f.blockReader = &transfer.BlockReader{
ClientName: f.client.namenode.ClientName,
Block: block,
Offset: int64(off - start),
UseDatanodeHostname: f.client.options.UseDatanodeHostname,
DialFunc: dialFunc,
}
return f.SetDeadline(f.deadline)
}
}
return errors.New("invalid offset")
}
|