sdk/go/arvados/fs_collection.go

   1 // Copyright (C) The Arvados Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: Apache-2.0
   4
   5 package arvados
   6
   7 import (
   8         "encoding/json"
   9         "fmt"
  10         "io"
  11         "log"
  12         "os"
  13         "path"
  14         "regexp"
  15         "sort"
  16         "strconv"
  17         "strings"
  18         "sync"
  19         "time"
  20 )
  21
  22 var maxBlockSize = 1 << 26
  23
  24 // A CollectionFileSystem is a FileSystem that can be serialized as a
  25 // manifest and stored as a collection.
  26 type CollectionFileSystem interface {
  27         FileSystem
  28
  29         // Flush all file data to Keep and return a snapshot of the
  30         // filesystem suitable for saving as (Collection)ManifestText.
  31         // Prefix (normally ".") is a top level directory, effectively
  32         // prepended to all paths in the returned manifest.
  33         MarshalManifest(prefix string) (string, error)
  34 }
  35
  36 type collectionFileSystem struct {
  37         fileSystem
  38         uuid string
  39 }
  40
  41 // FileSystem returns a CollectionFileSystem for the collection.
  42 func (c *Collection) FileSystem(client apiClient, kc keepClient) (CollectionFileSystem, error) {
  43         var modTime time.Time
  44         if c.ModifiedAt == nil {
  45                 modTime = time.Now()
  46         } else {
  47                 modTime = *c.ModifiedAt
  48         }
  49         fs := &collectionFileSystem{
  50                 uuid: c.UUID,
  51                 fileSystem: fileSystem{
  52                         fsBackend: keepBackend{apiClient: client, keepClient: kc},
  53                 },
  54         }
  55         root := &dirnode{
  56                 fs: fs,
  57                 treenode: treenode{
  58                         fileinfo: fileinfo{
  59                                 name:    ".",
  60                                 mode:    os.ModeDir | 0755,
  61                                 modTime: modTime,
  62                         },
  63                         inodes: make(map[string]inode),
  64                 },
  65         }
  66         root.SetParent(root, ".")
  67         if err := root.loadManifest(c.ManifestText); err != nil {
  68                 return nil, err
  69         }
  70         backdateTree(root, modTime)
  71         fs.root = root
  72         return fs, nil
  73 }
  74
  75 func backdateTree(n inode, modTime time.Time) {
  76         switch n := n.(type) {
  77         case *filenode:
  78                 n.fileinfo.modTime = modTime
  79         case *dirnode:
  80                 n.fileinfo.modTime = modTime
  81                 for _, n := range n.inodes {
  82                         backdateTree(n, modTime)
  83                 }
  84         }
  85 }
  86
  87 func (fs *collectionFileSystem) newNode(name string, perm os.FileMode, modTime time.Time) (node inode, err error) {
  88         if name == "" || name == "." || name == ".." {
  89                 return nil, ErrInvalidArgument
  90         }
  91         if perm.IsDir() {
  92                 return &dirnode{
  93                         fs: fs,
  94                         treenode: treenode{
  95                                 fileinfo: fileinfo{
  96                                         name:    name,
  97                                         mode:    perm | os.ModeDir,
  98                                         modTime: modTime,
  99                                 },
 100                                 inodes: make(map[string]inode),
 101                         },
 102                 }, nil
 103         } else {
 104                 return &filenode{
 105                         fs: fs,
 106                         fileinfo: fileinfo{
 107                                 name:    name,
 108                                 mode:    perm & ^os.ModeDir,
 109                                 modTime: modTime,
 110                         },
 111                 }, nil
 112         }
 113 }
 114
 115 func (fs *collectionFileSystem) Sync() error {
 116         log.Printf("cfs.Sync()")
 117         if fs.uuid == "" {
 118                 return nil
 119         }
 120         txt, err := fs.MarshalManifest(".")
 121         if err != nil {
 122                 log.Printf("WARNING: (collectionFileSystem)Sync() failed: %s", err)
 123                 return err
 124         }
 125         coll := &Collection{
 126                 UUID:         fs.uuid,
 127                 ManifestText: txt,
 128         }
 129         err = fs.RequestAndDecode(nil, "PUT", "arvados/v1/collections/"+fs.uuid, fs.UpdateBody(coll), map[string]interface{}{"select": []string{"uuid"}})
 130         if err != nil {
 131                 log.Printf("WARNING: (collectionFileSystem)Sync() failed: %s", err)
 132         }
 133         return err
 134 }
 135
 136 func (fs *collectionFileSystem) MarshalManifest(prefix string) (string, error) {
 137         fs.fileSystem.root.Lock()
 138         defer fs.fileSystem.root.Unlock()
 139         return fs.fileSystem.root.(*dirnode).marshalManifest(prefix)
 140 }
 141
 142 // filenodePtr is an offset into a file that is (usually) efficient to
 143 // seek to. Specifically, if filenode.repacked==filenodePtr.repacked
 144 // then
 145 // filenode.segments[filenodePtr.segmentIdx][filenodePtr.segmentOff]
 146 // corresponds to file offset filenodePtr.off. Otherwise, it is
 147 // necessary to reexamine len(filenode.segments[0]) etc. to find the
 148 // correct segment and offset.
 149 type filenodePtr struct {
 150         off        int64
 151         segmentIdx int
 152         segmentOff int
 153         repacked   int64
 154 }
 155
 156 // seek returns a ptr that is consistent with both startPtr.off and
 157 // the current state of fn. The caller must already hold fn.RLock() or
 158 // fn.Lock().
 159 //
 160 // If startPtr is beyond EOF, ptr.segment* will indicate precisely
 161 // EOF.
 162 //
 163 // After seeking:
 164 //
 165 //     ptr.segmentIdx == len(filenode.segments) // i.e., at EOF
 166 //     ||
 167 //     filenode.segments[ptr.segmentIdx].Len() > ptr.segmentOff
 168 func (fn *filenode) seek(startPtr filenodePtr) (ptr filenodePtr) {
 169         ptr = startPtr
 170         if ptr.off < 0 {
 171                 // meaningless anyway
 172                 return
 173         } else if ptr.off >= fn.fileinfo.size {
 174                 ptr.segmentIdx = len(fn.segments)
 175                 ptr.segmentOff = 0
 176                 ptr.repacked = fn.repacked
 177                 return
 178         } else if ptr.repacked == fn.repacked {
 179                 // segmentIdx and segmentOff accurately reflect
 180                 // ptr.off, but might have fallen off the end of a
 181                 // segment
 182                 if ptr.segmentOff >= fn.segments[ptr.segmentIdx].Len() {
 183                         ptr.segmentIdx++
 184                         ptr.segmentOff = 0
 185                 }
 186                 return
 187         }
 188         defer func() {
 189                 ptr.repacked = fn.repacked
 190         }()
 191         if ptr.off >= fn.fileinfo.size {
 192                 ptr.segmentIdx, ptr.segmentOff = len(fn.segments), 0
 193                 return
 194         }
 195         // Recompute segmentIdx and segmentOff.  We have already
 196         // established fn.fileinfo.size > ptr.off >= 0, so we don't
 197         // have to deal with edge cases here.
 198         var off int64
 199         for ptr.segmentIdx, ptr.segmentOff = 0, 0; off < ptr.off; ptr.segmentIdx++ {
 200                 // This would panic (index out of range) if
 201                 // fn.fileinfo.size were larger than
 202                 // sum(fn.segments[i].Len()) -- but that can't happen
 203                 // because we have ensured fn.fileinfo.size is always
 204                 // accurate.
 205                 segLen := int64(fn.segments[ptr.segmentIdx].Len())
 206                 if off+segLen > ptr.off {
 207                         ptr.segmentOff = int(ptr.off - off)
 208                         break
 209                 }
 210                 off += segLen
 211         }
 212         return
 213 }
 214
 215 // filenode implements inode.
 216 type filenode struct {
 217         parent   inode
 218         fs       FileSystem
 219         fileinfo fileinfo
 220         segments []segment
 221         // number of times `segments` has changed in a
 222         // way that might invalidate a filenodePtr
 223         repacked int64
 224         memsize  int64 // bytes in memSegments
 225         sync.RWMutex
 226         nullnode
 227 }
 228
 229 // caller must have lock
 230 func (fn *filenode) appendSegment(e segment) {
 231         fn.segments = append(fn.segments, e)
 232         fn.fileinfo.size += int64(e.Len())
 233 }
 234
 235 func (fn *filenode) SetParent(p inode, name string) {
 236         fn.Lock()
 237         defer fn.Unlock()
 238         fn.parent = p
 239         fn.fileinfo.name = name
 240 }
 241
 242 func (fn *filenode) Parent() inode {
 243         fn.RLock()
 244         defer fn.RUnlock()
 245         return fn.parent
 246 }
 247
 248 func (fn *filenode) FS() FileSystem {
 249         return fn.fs
 250 }
 251
 252 // Read reads file data from a single segment, starting at startPtr,
 253 // into p. startPtr is assumed not to be up-to-date. Caller must have
 254 // RLock or Lock.
 255 func (fn *filenode) Read(p []byte, startPtr filenodePtr) (n int, ptr filenodePtr, err error) {
 256         ptr = fn.seek(startPtr)
 257         if ptr.off < 0 {
 258                 err = ErrNegativeOffset
 259                 return
 260         }
 261         if ptr.segmentIdx >= len(fn.segments) {
 262                 err = io.EOF
 263                 return
 264         }
 265         n, err = fn.segments[ptr.segmentIdx].ReadAt(p, int64(ptr.segmentOff))
 266         if n > 0 {
 267                 ptr.off += int64(n)
 268                 ptr.segmentOff += n
 269                 if ptr.segmentOff == fn.segments[ptr.segmentIdx].Len() {
 270                         ptr.segmentIdx++
 271                         ptr.segmentOff = 0
 272                         if ptr.segmentIdx < len(fn.segments) && err == io.EOF {
 273                                 err = nil
 274                         }
 275                 }
 276         }
 277         return
 278 }
 279
 280 func (fn *filenode) Size() int64 {
 281         fn.RLock()
 282         defer fn.RUnlock()
 283         return fn.fileinfo.Size()
 284 }
 285
 286 func (fn *filenode) FileInfo() os.FileInfo {
 287         fn.RLock()
 288         defer fn.RUnlock()
 289         return fn.fileinfo
 290 }
 291
 292 func (fn *filenode) Truncate(size int64) error {
 293         fn.Lock()
 294         defer fn.Unlock()
 295         return fn.truncate(size)
 296 }
 297
 298 func (fn *filenode) truncate(size int64) error {
 299         if size == fn.fileinfo.size {
 300                 return nil
 301         }
 302         fn.repacked++
 303         if size < fn.fileinfo.size {
 304                 ptr := fn.seek(filenodePtr{off: size})
 305                 for i := ptr.segmentIdx; i < len(fn.segments); i++ {
 306                         if seg, ok := fn.segments[i].(*memSegment); ok {
 307                                 fn.memsize -= int64(seg.Len())
 308                         }
 309                 }
 310                 if ptr.segmentOff == 0 {
 311                         fn.segments = fn.segments[:ptr.segmentIdx]
 312                 } else {
 313                         fn.segments = fn.segments[:ptr.segmentIdx+1]
 314                         switch seg := fn.segments[ptr.segmentIdx].(type) {
 315                         case *memSegment:
 316                                 seg.Truncate(ptr.segmentOff)
 317                                 fn.memsize += int64(seg.Len())
 318                         default:
 319                                 fn.segments[ptr.segmentIdx] = seg.Slice(0, ptr.segmentOff)
 320                         }
 321                 }
 322                 fn.fileinfo.size = size
 323                 return nil
 324         }
 325         for size > fn.fileinfo.size {
 326                 grow := size - fn.fileinfo.size
 327                 var seg *memSegment
 328                 var ok bool
 329                 if len(fn.segments) == 0 {
 330                         seg = &memSegment{}
 331                         fn.segments = append(fn.segments, seg)
 332                 } else if seg, ok = fn.segments[len(fn.segments)-1].(*memSegment); !ok || seg.Len() >= maxBlockSize {
 333                         seg = &memSegment{}
 334                         fn.segments = append(fn.segments, seg)
 335                 }
 336                 if maxgrow := int64(maxBlockSize - seg.Len()); maxgrow < grow {
 337                         grow = maxgrow
 338                 }
 339                 seg.Truncate(seg.Len() + int(grow))
 340                 fn.fileinfo.size += grow
 341                 fn.memsize += grow
 342         }
 343         return nil
 344 }
 345
 346 // Write writes data from p to the file, starting at startPtr,
 347 // extending the file size if necessary. Caller must have Lock.
 348 func (fn *filenode) Write(p []byte, startPtr filenodePtr) (n int, ptr filenodePtr, err error) {
 349         if startPtr.off > fn.fileinfo.size {
 350                 if err = fn.truncate(startPtr.off); err != nil {
 351                         return 0, startPtr, err
 352                 }
 353         }
 354         ptr = fn.seek(startPtr)
 355         if ptr.off < 0 {
 356                 err = ErrNegativeOffset
 357                 return
 358         }
 359         for len(p) > 0 && err == nil {
 360                 cando := p
 361                 if len(cando) > maxBlockSize {
 362                         cando = cando[:maxBlockSize]
 363                 }
 364                 // Rearrange/grow fn.segments (and shrink cando if
 365                 // needed) such that cando can be copied to
 366                 // fn.segments[ptr.segmentIdx] at offset
 367                 // ptr.segmentOff.
 368                 cur := ptr.segmentIdx
 369                 prev := ptr.segmentIdx - 1
 370                 var curWritable bool
 371                 if cur < len(fn.segments) {
 372                         _, curWritable = fn.segments[cur].(*memSegment)
 373                 }
 374                 var prevAppendable bool
 375                 if prev >= 0 && fn.segments[prev].Len() < maxBlockSize {
 376                         _, prevAppendable = fn.segments[prev].(*memSegment)
 377                 }
 378                 if ptr.segmentOff > 0 && !curWritable {
 379                         // Split a non-writable block.
 380                         if max := fn.segments[cur].Len() - ptr.segmentOff; max <= len(cando) {
 381                                 // Truncate cur, and insert a new
 382                                 // segment after it.
 383                                 cando = cando[:max]
 384                                 fn.segments = append(fn.segments, nil)
 385                                 copy(fn.segments[cur+1:], fn.segments[cur:])
 386                         } else {
 387                                 // Split cur into two copies, truncate
 388                                 // the one on the left, shift the one
 389                                 // on the right, and insert a new
 390                                 // segment between them.
 391                                 fn.segments = append(fn.segments, nil, nil)
 392                                 copy(fn.segments[cur+2:], fn.segments[cur:])
 393                                 fn.segments[cur+2] = fn.segments[cur+2].Slice(ptr.segmentOff+len(cando), -1)
 394                         }
 395                         cur++
 396                         prev++
 397                         seg := &memSegment{}
 398                         seg.Truncate(len(cando))
 399                         fn.memsize += int64(len(cando))
 400                         fn.segments[cur] = seg
 401                         fn.segments[prev] = fn.segments[prev].Slice(0, ptr.segmentOff)
 402                         ptr.segmentIdx++
 403                         ptr.segmentOff = 0
 404                         fn.repacked++
 405                         ptr.repacked++
 406                 } else if curWritable {
 407                         if fit := int(fn.segments[cur].Len()) - ptr.segmentOff; fit < len(cando) {
 408                                 cando = cando[:fit]
 409                         }
 410                 } else {
 411                         if prevAppendable {
 412                                 // Shrink cando if needed to fit in
 413                                 // prev segment.
 414                                 if cangrow := maxBlockSize - fn.segments[prev].Len(); cangrow < len(cando) {
 415                                         cando = cando[:cangrow]
 416                                 }
 417                         }
 418
 419                         if cur == len(fn.segments) {
 420                                 // ptr is at EOF, filesize is changing.
 421                                 fn.fileinfo.size += int64(len(cando))
 422                         } else if el := fn.segments[cur].Len(); el <= len(cando) {
 423                                 // cando is long enough that we won't
 424                                 // need cur any more. shrink cando to
 425                                 // be exactly as long as cur
 426                                 // (otherwise we'd accidentally shift
 427                                 // the effective position of all
 428                                 // segments after cur).
 429                                 cando = cando[:el]
 430                                 copy(fn.segments[cur:], fn.segments[cur+1:])
 431                                 fn.segments = fn.segments[:len(fn.segments)-1]
 432                         } else {
 433                                 // shrink cur by the same #bytes we're growing prev
 434                                 fn.segments[cur] = fn.segments[cur].Slice(len(cando), -1)
 435                         }
 436
 437                         if prevAppendable {
 438                                 // Grow prev.
 439                                 ptr.segmentIdx--
 440                                 ptr.segmentOff = fn.segments[prev].Len()
 441                                 fn.segments[prev].(*memSegment).Truncate(ptr.segmentOff + len(cando))
 442                                 fn.memsize += int64(len(cando))
 443                                 ptr.repacked++
 444                                 fn.repacked++
 445                         } else {
 446                                 // Insert a segment between prev and
 447                                 // cur, and advance prev/cur.
 448                                 fn.segments = append(fn.segments, nil)
 449                                 if cur < len(fn.segments) {
 450                                         copy(fn.segments[cur+1:], fn.segments[cur:])
 451                                         ptr.repacked++
 452                                         fn.repacked++
 453                                 } else {
 454                                         // appending a new segment does
 455                                         // not invalidate any ptrs
 456                                 }
 457                                 seg := &memSegment{}
 458                                 seg.Truncate(len(cando))
 459                                 fn.memsize += int64(len(cando))
 460                                 fn.segments[cur] = seg
 461                                 cur++
 462                                 prev++
 463                         }
 464                 }
 465
 466                 // Finally we can copy bytes from cando to the current segment.
 467                 fn.segments[ptr.segmentIdx].(*memSegment).WriteAt(cando, ptr.segmentOff)
 468                 n += len(cando)
 469                 p = p[len(cando):]
 470
 471                 ptr.off += int64(len(cando))
 472                 ptr.segmentOff += len(cando)
 473                 if ptr.segmentOff >= maxBlockSize {
 474                         fn.pruneMemSegments()
 475                 }
 476                 if fn.segments[ptr.segmentIdx].Len() == ptr.segmentOff {
 477                         ptr.segmentOff = 0
 478                         ptr.segmentIdx++
 479                 }
 480
 481                 fn.fileinfo.modTime = time.Now()
 482         }
 483         return
 484 }
 485
 486 // Write some data out to disk to reduce memory use. Caller must have
 487 // write lock.
 488 func (fn *filenode) pruneMemSegments() {
 489         // TODO: async (don't hold Lock() while waiting for Keep)
 490         // TODO: share code with (*dirnode)sync()
 491         // TODO: pack/flush small blocks too, when fragmented
 492         for idx, seg := range fn.segments {
 493                 seg, ok := seg.(*memSegment)
 494                 if !ok || seg.Len() < maxBlockSize {
 495                         continue
 496                 }
 497                 locator, _, err := fn.FS().PutB(seg.buf)
 498                 if err != nil {
 499                         // TODO: stall (or return errors from)
 500                         // subsequent writes until flushing
 501                         // starts to succeed
 502                         continue
 503                 }
 504                 fn.memsize -= int64(seg.Len())
 505                 fn.segments[idx] = storedSegment{
 506                         kc:      fn.FS(),
 507                         locator: locator,
 508                         size:    seg.Len(),
 509                         offset:  0,
 510                         length:  seg.Len(),
 511                 }
 512         }
 513 }
 514
 515 type dirnode struct {
 516         fs *collectionFileSystem
 517         treenode
 518 }
 519
 520 func (dn *dirnode) FS() FileSystem {
 521         return dn.fs
 522 }
 523
 524 func (dn *dirnode) Child(name string, replace func(inode) (inode, error)) (inode, error) {
 525         if dn == dn.fs.rootnode() && name == ".arvados#collection" {
 526                 gn := &getternode{Getter: func() ([]byte, error) {
 527                         var coll Collection
 528                         var err error
 529                         coll.ManifestText, err = dn.fs.MarshalManifest(".")
 530                         if err != nil {
 531                                 return nil, err
 532                         }
 533                         data, err := json.Marshal(&coll)
 534                         if err == nil {
 535                                 data = append(data, '\n')
 536                         }
 537                         return data, err
 538                 }}
 539                 gn.SetParent(dn, name)
 540                 return gn, nil
 541         }
 542         return dn.treenode.Child(name, replace)
 543 }
 544
 545 // sync flushes in-memory data (for all files in the tree rooted at
 546 // dn) to persistent storage. Caller must hold dn.Lock().
 547 func (dn *dirnode) sync() error {
 548         type shortBlock struct {
 549                 fn  *filenode
 550                 idx int
 551         }
 552         var pending []shortBlock
 553         var pendingLen int
 554
 555         flush := func(sbs []shortBlock) error {
 556                 if len(sbs) == 0 {
 557                         return nil
 558                 }
 559                 block := make([]byte, 0, maxBlockSize)
 560                 for _, sb := range sbs {
 561                         block = append(block, sb.fn.segments[sb.idx].(*memSegment).buf...)
 562                 }
 563                 locator, _, err := dn.fs.PutB(block)
 564                 if err != nil {
 565                         return err
 566                 }
 567                 off := 0
 568                 for _, sb := range sbs {
 569                         data := sb.fn.segments[sb.idx].(*memSegment).buf
 570                         sb.fn.segments[sb.idx] = storedSegment{
 571                                 kc:      dn.fs,
 572                                 locator: locator,
 573                                 size:    len(block),
 574                                 offset:  off,
 575                                 length:  len(data),
 576                         }
 577                         off += len(data)
 578                         sb.fn.memsize -= int64(len(data))
 579                 }
 580                 return nil
 581         }
 582
 583         names := make([]string, 0, len(dn.inodes))
 584         for name := range dn.inodes {
 585                 names = append(names, name)
 586         }
 587         sort.Strings(names)
 588
 589         for _, name := range names {
 590                 fn, ok := dn.inodes[name].(*filenode)
 591                 if !ok {
 592                         continue
 593                 }
 594                 fn.Lock()
 595                 defer fn.Unlock()
 596                 for idx, seg := range fn.segments {
 597                         seg, ok := seg.(*memSegment)
 598                         if !ok {
 599                                 continue
 600                         }
 601                         if seg.Len() > maxBlockSize/2 {
 602                                 if err := flush([]shortBlock{{fn, idx}}); err != nil {
 603                                         return err
 604                                 }
 605                                 continue
 606                         }
 607                         if pendingLen+seg.Len() > maxBlockSize {
 608                                 if err := flush(pending); err != nil {
 609                                         return err
 610                                 }
 611                                 pending = nil
 612                                 pendingLen = 0
 613                         }
 614                         pending = append(pending, shortBlock{fn, idx})
 615                         pendingLen += seg.Len()
 616                 }
 617         }
 618         return flush(pending)
 619 }
 620
 621 // caller must have read lock.
 622 func (dn *dirnode) marshalManifest(prefix string) (string, error) {
 623         var streamLen int64
 624         type filepart struct {
 625                 name   string
 626                 offset int64
 627                 length int64
 628         }
 629         var fileparts []filepart
 630         var subdirs string
 631         var blocks []string
 632
 633         if err := dn.sync(); err != nil {
 634                 return "", err
 635         }
 636
 637         names := make([]string, 0, len(dn.inodes))
 638         for name, node := range dn.inodes {
 639                 names = append(names, name)
 640                 node.Lock()
 641                 defer node.Unlock()
 642         }
 643         sort.Strings(names)
 644
 645         for _, name := range names {
 646                 switch node := dn.inodes[name].(type) {
 647                 case *dirnode:
 648                         subdir, err := node.marshalManifest(prefix + "/" + name)
 649                         if err != nil {
 650                                 return "", err
 651                         }
 652                         subdirs = subdirs + subdir
 653                 case *filenode:
 654                         if len(node.segments) == 0 {
 655                                 fileparts = append(fileparts, filepart{name: name})
 656                                 break
 657                         }
 658                         for _, seg := range node.segments {
 659                                 switch seg := seg.(type) {
 660                                 case storedSegment:
 661                                         if len(blocks) > 0 && blocks[len(blocks)-1] == seg.locator {
 662                                                 streamLen -= int64(seg.size)
 663                                         } else {
 664                                                 blocks = append(blocks, seg.locator)
 665                                         }
 666                                         next := filepart{
 667                                                 name:   name,
 668                                                 offset: streamLen + int64(seg.offset),
 669                                                 length: int64(seg.length),
 670                                         }
 671                                         if prev := len(fileparts) - 1; prev >= 0 &&
 672                                                 fileparts[prev].name == name &&
 673                                                 fileparts[prev].offset+fileparts[prev].length == next.offset {
 674                                                 fileparts[prev].length += next.length
 675                                         } else {
 676                                                 fileparts = append(fileparts, next)
 677                                         }
 678                                         streamLen += int64(seg.size)
 679                                 default:
 680                                         // This can't happen: we
 681                                         // haven't unlocked since
 682                                         // calling sync().
 683                                         panic(fmt.Sprintf("can't marshal segment type %T", seg))
 684                                 }
 685                         }
 686                 default:
 687                         panic(fmt.Sprintf("can't marshal inode type %T", node))
 688                 }
 689         }
 690         var filetokens []string
 691         for _, s := range fileparts {
 692                 filetokens = append(filetokens, fmt.Sprintf("%d:%d:%s", s.offset, s.length, manifestEscape(s.name)))
 693         }
 694         if len(filetokens) == 0 {
 695                 return subdirs, nil
 696         } else if len(blocks) == 0 {
 697                 blocks = []string{"d41d8cd98f00b204e9800998ecf8427e+0"}
 698         }
 699         return manifestEscape(prefix) + " " + strings.Join(blocks, " ") + " " + strings.Join(filetokens, " ") + "\n" + subdirs, nil
 700 }
 701
 702 func (dn *dirnode) loadManifest(txt string) error {
 703         var dirname string
 704         streams := strings.Split(txt, "\n")
 705         if streams[len(streams)-1] != "" {
 706                 return fmt.Errorf("line %d: no trailing newline", len(streams))
 707         }
 708         streams = streams[:len(streams)-1]
 709         segments := []storedSegment{}
 710         for i, stream := range streams {
 711                 lineno := i + 1
 712                 var anyFileTokens bool
 713                 var pos int64
 714                 var segIdx int
 715                 segments = segments[:0]
 716                 for i, token := range strings.Split(stream, " ") {
 717                         if i == 0 {
 718                                 dirname = manifestUnescape(token)
 719                                 continue
 720                         }
 721                         if !strings.Contains(token, ":") {
 722                                 if anyFileTokens {
 723                                         return fmt.Errorf("line %d: bad file segment %q", lineno, token)
 724                                 }
 725                                 toks := strings.SplitN(token, "+", 3)
 726                                 if len(toks) < 2 {
 727                                         return fmt.Errorf("line %d: bad locator %q", lineno, token)
 728                                 }
 729                                 length, err := strconv.ParseInt(toks[1], 10, 32)
 730                                 if err != nil || length < 0 {
 731                                         return fmt.Errorf("line %d: bad locator %q", lineno, token)
 732                                 }
 733                                 segments = append(segments, storedSegment{
 734                                         locator: token,
 735                                         size:    int(length),
 736                                         offset:  0,
 737                                         length:  int(length),
 738                                 })
 739                                 continue
 740                         } else if len(segments) == 0 {
 741                                 return fmt.Errorf("line %d: bad locator %q", lineno, token)
 742                         }
 743
 744                         toks := strings.SplitN(token, ":", 3)
 745                         if len(toks) != 3 {
 746                                 return fmt.Errorf("line %d: bad file segment %q", lineno, token)
 747                         }
 748                         anyFileTokens = true
 749
 750                         offset, err := strconv.ParseInt(toks[0], 10, 64)
 751                         if err != nil || offset < 0 {
 752                                 return fmt.Errorf("line %d: bad file segment %q", lineno, token)
 753                         }
 754                         length, err := strconv.ParseInt(toks[1], 10, 64)
 755                         if err != nil || length < 0 {
 756                                 return fmt.Errorf("line %d: bad file segment %q", lineno, token)
 757                         }
 758                         name := dirname + "/" + manifestUnescape(toks[2])
 759                         fnode, err := dn.createFileAndParents(name)
 760                         if err != nil {
 761                                 return fmt.Errorf("line %d: cannot use path %q: %s", lineno, name, err)
 762                         }
 763                         // Map the stream offset/range coordinates to
 764                         // block/offset/range coordinates and add
 765                         // corresponding storedSegments to the filenode
 766                         if pos > offset {
 767                                 // Can't continue where we left off.
 768                                 // TODO: binary search instead of
 769                                 // rewinding all the way (but this
 770                                 // situation might be rare anyway)
 771                                 segIdx, pos = 0, 0
 772                         }
 773                         for next := int64(0); segIdx < len(segments); segIdx++ {
 774                                 seg := segments[segIdx]
 775                                 next = pos + int64(seg.Len())
 776                                 if next <= offset || seg.Len() == 0 {
 777                                         pos = next
 778                                         continue
 779                                 }
 780                                 if pos >= offset+length {
 781                                         break
 782                                 }
 783                                 var blkOff int
 784                                 if pos < offset {
 785                                         blkOff = int(offset - pos)
 786                                 }
 787                                 blkLen := seg.Len() - blkOff
 788                                 if pos+int64(blkOff+blkLen) > offset+length {
 789                                         blkLen = int(offset + length - pos - int64(blkOff))
 790                                 }
 791                                 fnode.appendSegment(storedSegment{
 792                                         kc:      dn.fs,
 793                                         locator: seg.locator,
 794                                         size:    seg.size,
 795                                         offset:  blkOff,
 796                                         length:  blkLen,
 797                                 })
 798                                 if next > offset+length {
 799                                         break
 800                                 } else {
 801                                         pos = next
 802                                 }
 803                         }
 804                         if segIdx == len(segments) && pos < offset+length {
 805                                 return fmt.Errorf("line %d: invalid segment in %d-byte stream: %q", lineno, pos, token)
 806                         }
 807                 }
 808                 if !anyFileTokens {
 809                         return fmt.Errorf("line %d: no file segments", lineno)
 810                 } else if len(segments) == 0 {
 811                         return fmt.Errorf("line %d: no locators", lineno)
 812                 } else if dirname == "" {
 813                         return fmt.Errorf("line %d: no stream name", lineno)
 814                 }
 815         }
 816         return nil
 817 }
 818
 819 // only safe to call from loadManifest -- no locking
 820 func (dn *dirnode) createFileAndParents(path string) (fn *filenode, err error) {
 821         var node inode = dn
 822         names := strings.Split(path, "/")
 823         basename := names[len(names)-1]
 824         if !permittedName(basename) {
 825                 err = fmt.Errorf("invalid file part %q in path %q", basename, path)
 826                 return
 827         }
 828         for _, name := range names[:len(names)-1] {
 829                 switch name {
 830                 case "", ".":
 831                         continue
 832                 case "..":
 833                         if node == dn {
 834                                 // can't be sure parent will be a *dirnode
 835                                 return nil, ErrInvalidArgument
 836                         }
 837                         node = node.Parent()
 838                         continue
 839                 }
 840                 node, err = node.Child(name, func(child inode) (inode, error) {
 841                         if child == nil {
 842                                 child, err := node.FS().newNode(name, 0755|os.ModeDir, node.Parent().FileInfo().ModTime())
 843                                 if err != nil {
 844                                         return nil, err
 845                                 }
 846                                 child.SetParent(node, name)
 847                                 return child, nil
 848                         } else if !child.IsDir() {
 849                                 return child, ErrFileExists
 850                         } else {
 851                                 return child, nil
 852                         }
 853                 })
 854                 if err != nil {
 855                         return
 856                 }
 857         }
 858         _, err = node.Child(basename, func(child inode) (inode, error) {
 859                 switch child := child.(type) {
 860                 case nil:
 861                         child, err = node.FS().newNode(basename, 0755, node.FileInfo().ModTime())
 862                         if err != nil {
 863                                 return nil, err
 864                         }
 865                         child.SetParent(node, basename)
 866                         fn = child.(*filenode)
 867                         return child, nil
 868                 case *filenode:
 869                         fn = child
 870                         return child, nil
 871                 case *dirnode:
 872                         return child, ErrIsDirectory
 873                 default:
 874                         return child, ErrInvalidArgument
 875                 }
 876         })
 877         return
 878 }
 879
 880 type segment interface {
 881         io.ReaderAt
 882         Len() int
 883         // Return a new segment with a subsection of the data from this
 884         // one. length<0 means length=Len()-off.
 885         Slice(off int, length int) segment
 886 }
 887
 888 type memSegment struct {
 889         buf []byte
 890 }
 891
 892 func (me *memSegment) Len() int {
 893         return len(me.buf)
 894 }
 895
 896 func (me *memSegment) Slice(off, length int) segment {
 897         if length < 0 {
 898                 length = len(me.buf) - off
 899         }
 900         buf := make([]byte, length)
 901         copy(buf, me.buf[off:])
 902         return &memSegment{buf: buf}
 903 }
 904
 905 func (me *memSegment) Truncate(n int) {
 906         if n > cap(me.buf) {
 907                 newsize := 1024
 908                 for newsize < n {
 909                         newsize = newsize << 2
 910                 }
 911                 newbuf := make([]byte, n, newsize)
 912                 copy(newbuf, me.buf)
 913                 me.buf = newbuf
 914         } else {
 915                 // Zero unused part when shrinking, in case we grow
 916                 // and start using it again later.
 917                 for i := n; i < len(me.buf); i++ {
 918                         me.buf[i] = 0
 919                 }
 920         }
 921         me.buf = me.buf[:n]
 922 }
 923
 924 func (me *memSegment) WriteAt(p []byte, off int) {
 925         if off+len(p) > len(me.buf) {
 926                 panic("overflowed segment")
 927         }
 928         copy(me.buf[off:], p)
 929 }
 930
 931 func (me *memSegment) ReadAt(p []byte, off int64) (n int, err error) {
 932         if off > int64(me.Len()) {
 933                 err = io.EOF
 934                 return
 935         }
 936         n = copy(p, me.buf[int(off):])
 937         if n < len(p) {
 938                 err = io.EOF
 939         }
 940         return
 941 }
 942
 943 type storedSegment struct {
 944         kc      fsBackend
 945         locator string
 946         size    int // size of stored block (also encoded in locator)
 947         offset  int // position of segment within the stored block
 948         length  int // bytes in this segment (offset + length <= size)
 949 }
 950
 951 func (se storedSegment) Len() int {
 952         return se.length
 953 }
 954
 955 func (se storedSegment) Slice(n, size int) segment {
 956         se.offset += n
 957         se.length -= n
 958         if size >= 0 && se.length > size {
 959                 se.length = size
 960         }
 961         return se
 962 }
 963
 964 func (se storedSegment) ReadAt(p []byte, off int64) (n int, err error) {
 965         if off > int64(se.length) {
 966                 return 0, io.EOF
 967         }
 968         maxlen := se.length - int(off)
 969         if len(p) > maxlen {
 970                 p = p[:maxlen]
 971                 n, err = se.kc.ReadAt(se.locator, p, int(off)+se.offset)
 972                 if err == nil {
 973                         err = io.EOF
 974                 }
 975                 return
 976         }
 977         return se.kc.ReadAt(se.locator, p, int(off)+se.offset)
 978 }
 979
 980 func canonicalName(name string) string {
 981         name = path.Clean("/" + name)
 982         if name == "/" || name == "./" {
 983                 name = "."
 984         } else if strings.HasPrefix(name, "/") {
 985                 name = "." + name
 986         }
 987         return name
 988 }
 989
 990 var manifestEscapeSeq = regexp.MustCompile(`\\([0-7]{3}|\\)`)
 991
 992 func manifestUnescapeFunc(seq string) string {
 993         if seq == `\\` {
 994                 return `\`
 995         }
 996         i, err := strconv.ParseUint(seq[1:], 8, 8)
 997         if err != nil {
 998                 // Invalid escape sequence: can't unescape.
 999                 return seq
1000         }
1001         return string([]byte{byte(i)})
1002 }
1003
1004 func manifestUnescape(s string) string {
1005         return manifestEscapeSeq.ReplaceAllStringFunc(s, manifestUnescapeFunc)
1006 }
1007
1008 var manifestEscapedChar = regexp.MustCompile(`[\000-\040:\s\\]`)
1009
1010 func manifestEscapeFunc(seq string) string {
1011         return fmt.Sprintf("\\%03o", byte(seq[0]))
1012 }
1013
1014 func manifestEscape(s string) string {
1015         return manifestEscapedChar.ReplaceAllStringFunc(s, manifestEscapeFunc)
1016 }