1 # Copyright (C) The Arvados Authors. All rights reserved.
3 # SPDX-License-Identifier: Apache-2.0
5 from __future__ import absolute_import
6 from __future__ import division
7 from future import standard_library
8 from future.utils import listitems, listvalues
9 standard_library.install_aliases()
10 from builtins import range
11 from builtins import object
28 from .errors import KeepWriteError, AssertionError, ArgumentError
29 from .keep import KeepLocator
30 from ._normalize_stream import normalize_stream
31 from ._ranges import locators_and_ranges, replace_range, Range, LocatorAndRange
32 from .retry import retry_method
37 _logger = logging.getLogger('arvados.arvfile')
40 """split(path) -> streamname, filename
42 Separate the stream name and file name in a /-separated stream path and
43 return a tuple (stream_name, file_name). If no stream name is available,
48 stream_name, file_name = path.rsplit('/', 1)
49 except ValueError: # No / in string
50 stream_name, file_name = '.', path
51 return stream_name, file_name
54 class UnownedBlockError(Exception):
55 """Raised when there's an writable block without an owner on the BlockManager."""
59 class _FileLikeObjectBase(object):
60 def __init__(self, name, mode):
66 def _before_close(orig_func):
67 @functools.wraps(orig_func)
68 def before_close_wrapper(self, *args, **kwargs):
70 raise ValueError("I/O operation on closed stream file")
71 return orig_func(self, *args, **kwargs)
72 return before_close_wrapper
77 def __exit__(self, exc_type, exc_value, traceback):
88 class ArvadosFileReaderBase(_FileLikeObjectBase):
89 def __init__(self, name, mode, num_retries=None):
90 super(ArvadosFileReaderBase, self).__init__(name, mode)
92 self.num_retries = num_retries
93 self._readline_cache = (None, None)
97 data = self.readline()
102 def decompressed_name(self):
103 return re.sub('\.(bz2|gz)$', '', self.name)
105 @_FileLikeObjectBase._before_close
106 def seek(self, pos, whence=os.SEEK_SET):
107 if whence == os.SEEK_CUR:
109 elif whence == os.SEEK_END:
112 raise IOError(errno.EINVAL, "Tried to seek to negative file offset.")
128 @_FileLikeObjectBase._before_close
130 def readall(self, size=2**20, num_retries=None):
132 data = self.read(size, num_retries=num_retries)
137 @_FileLikeObjectBase._before_close
139 def readline(self, size=float('inf'), num_retries=None):
140 cache_pos, cache_data = self._readline_cache
141 if self.tell() == cache_pos:
143 self._filepos += len(cache_data)
146 data_size = len(data[-1])
147 while (data_size < size) and (b'\n' not in data[-1]):
148 next_read = self.read(2 ** 20, num_retries=num_retries)
151 data.append(next_read)
152 data_size += len(next_read)
153 data = b''.join(data)
155 nextline_index = data.index(b'\n') + 1
157 nextline_index = len(data)
158 nextline_index = min(nextline_index, size)
159 self._filepos -= len(data) - nextline_index
160 self._readline_cache = (self.tell(), data[nextline_index:])
161 return data[:nextline_index].decode()
163 @_FileLikeObjectBase._before_close
165 def decompress(self, decompress, size, num_retries=None):
166 for segment in self.readall(size, num_retries=num_retries):
167 data = decompress(segment)
171 @_FileLikeObjectBase._before_close
173 def readall_decompressed(self, size=2**20, num_retries=None):
175 if self.name.endswith('.bz2'):
176 dc = bz2.BZ2Decompressor()
177 return self.decompress(dc.decompress, size,
178 num_retries=num_retries)
179 elif self.name.endswith('.gz'):
180 dc = zlib.decompressobj(16+zlib.MAX_WBITS)
181 return self.decompress(lambda segment: dc.decompress(dc.unconsumed_tail + segment),
182 size, num_retries=num_retries)
184 return self.readall(size, num_retries=num_retries)
186 @_FileLikeObjectBase._before_close
188 def readlines(self, sizehint=float('inf'), num_retries=None):
191 for s in self.readall(num_retries=num_retries):
194 if data_size >= sizehint:
196 return b''.join(data).decode().splitlines(True)
199 raise IOError(errno.ENOSYS, "Not implemented")
201 def read(self, size, num_retries=None):
202 raise IOError(errno.ENOSYS, "Not implemented")
204 def readfrom(self, start, size, num_retries=None):
205 raise IOError(errno.ENOSYS, "Not implemented")
208 class StreamFileReader(ArvadosFileReaderBase):
209 class _NameAttribute(str):
210 # The Python file API provides a plain .name attribute.
211 # Older SDK provided a name() method.
212 # This class provides both, for maximum compatibility.
216 def __init__(self, stream, segments, name):
217 super(StreamFileReader, self).__init__(self._NameAttribute(name), 'rb', num_retries=stream.num_retries)
218 self._stream = stream
219 self.segments = segments
221 def stream_name(self):
222 return self._stream.name()
225 n = self.segments[-1]
226 return n.range_start + n.range_size
228 @_FileLikeObjectBase._before_close
230 def read(self, size, num_retries=None):
231 """Read up to 'size' bytes from the stream, starting at the current file position"""
236 available_chunks = locators_and_ranges(self.segments, self._filepos, size)
238 lr = available_chunks[0]
239 data = self._stream.readfrom(lr.locator+lr.segment_offset,
241 num_retries=num_retries)
243 self._filepos += len(data)
246 @_FileLikeObjectBase._before_close
248 def readfrom(self, start, size, num_retries=None):
249 """Read up to 'size' bytes from the stream, starting at 'start'"""
254 for lr in locators_and_ranges(self.segments, start, size):
255 data.append(self._stream.readfrom(lr.locator+lr.segment_offset, lr.segment_size,
256 num_retries=num_retries))
257 return b''.join(data)
259 def as_manifest(self):
261 for r in self.segments:
262 segs.extend(self._stream.locators_and_ranges(r.locator, r.range_size))
263 return " ".join(normalize_stream(".", {self.name: segs})) + "\n"
266 def synchronized(orig_func):
267 @functools.wraps(orig_func)
268 def synchronized_wrapper(self, *args, **kwargs):
270 return orig_func(self, *args, **kwargs)
271 return synchronized_wrapper
274 class StateChangeError(Exception):
275 def __init__(self, message, state, nextstate):
276 super(StateChangeError, self).__init__(message)
278 self.nextstate = nextstate
280 class _BufferBlock(object):
281 """A stand-in for a Keep block that is in the process of being written.
283 Writers can append to it, get the size, and compute the Keep locator.
284 There are three valid states:
290 Block is in the process of being uploaded to Keep, append is an error.
293 The block has been written to Keep, its internal buffer has been
294 released, fetching the block will fetch it via keep client (since we
295 discarded the internal copy), and identifiers referring to the BufferBlock
296 can be replaced with the block locator.
306 def __init__(self, blockid, starting_capacity, owner):
309 the identifier for this block
312 the initial buffer capacity
315 ArvadosFile that owns this block
318 self.blockid = blockid
319 self.buffer_block = bytearray(starting_capacity)
320 self.buffer_view = memoryview(self.buffer_block)
321 self.write_pointer = 0
322 self._state = _BufferBlock.WRITABLE
325 self.lock = threading.Lock()
326 self.wait_for_commit = threading.Event()
330 def append(self, data):
331 """Append some data to the buffer.
333 Only valid if the block is in WRITABLE state. Implements an expanding
334 buffer, doubling capacity as needed to accomdate all the data.
337 if self._state == _BufferBlock.WRITABLE:
338 if not isinstance(data, bytes) and not isinstance(data, memoryview):
340 while (self.write_pointer+len(data)) > len(self.buffer_block):
341 new_buffer_block = bytearray(len(self.buffer_block) * 2)
342 new_buffer_block[0:self.write_pointer] = self.buffer_block[0:self.write_pointer]
343 self.buffer_block = new_buffer_block
344 self.buffer_view = memoryview(self.buffer_block)
345 self.buffer_view[self.write_pointer:self.write_pointer+len(data)] = data
346 self.write_pointer += len(data)
349 raise AssertionError("Buffer block is not writable")
351 STATE_TRANSITIONS = frozenset([
353 (PENDING, COMMITTED),
358 def set_state(self, nextstate, val=None):
359 if (self._state, nextstate) not in self.STATE_TRANSITIONS:
360 raise StateChangeError("Invalid state change from %s to %s" % (self._state, nextstate), self._state, nextstate)
361 self._state = nextstate
363 if self._state == _BufferBlock.PENDING:
364 self.wait_for_commit.clear()
366 if self._state == _BufferBlock.COMMITTED:
368 self.buffer_view = None
369 self.buffer_block = None
370 self.wait_for_commit.set()
372 if self._state == _BufferBlock.ERROR:
374 self.wait_for_commit.set()
381 """The amount of data written to the buffer."""
382 return self.write_pointer
386 """The Keep locator for this buffer's contents."""
387 if self._locator is None:
388 self._locator = "%s+%i" % (hashlib.md5(self.buffer_view[0:self.write_pointer]).hexdigest(), self.size())
392 def clone(self, new_blockid, owner):
393 if self._state == _BufferBlock.COMMITTED:
394 raise AssertionError("Cannot duplicate committed buffer block")
395 bufferblock = _BufferBlock(new_blockid, self.size(), owner)
396 bufferblock.append(self.buffer_view[0:self.size()])
401 self._state = _BufferBlock.DELETED
403 self.buffer_block = None
404 self.buffer_view = None
407 def repack_writes(self):
408 """Optimize buffer block by repacking segments in file sequence.
410 When the client makes random writes, they appear in the buffer block in
411 the sequence they were written rather than the sequence they appear in
412 the file. This makes for inefficient, fragmented manifests. Attempt
413 to optimize by repacking writes in file sequence.
416 if self._state != _BufferBlock.WRITABLE:
417 raise AssertionError("Cannot repack non-writable block")
419 segs = self.owner.segments()
421 # Collect the segments that reference the buffer block.
422 bufferblock_segs = [s for s in segs if s.locator == self.blockid]
424 # Collect total data referenced by segments (could be smaller than
425 # bufferblock size if a portion of the file was written and
427 write_total = sum([s.range_size for s in bufferblock_segs])
429 if write_total < self.size() or len(bufferblock_segs) > 1:
430 # If there's more than one segment referencing this block, it is
431 # due to out-of-order writes and will produce a fragmented
432 # manifest, so try to optimize by re-packing into a new buffer.
433 contents = self.buffer_view[0:self.write_pointer].tobytes()
434 new_bb = _BufferBlock(None, write_total, None)
435 for t in bufferblock_segs:
436 new_bb.append(contents[t.segment_offset:t.segment_offset+t.range_size])
437 t.segment_offset = new_bb.size() - t.range_size
439 self.buffer_block = new_bb.buffer_block
440 self.buffer_view = new_bb.buffer_view
441 self.write_pointer = new_bb.write_pointer
444 self.owner.set_segments(segs)
447 return "<BufferBlock %s>" % (self.blockid)
450 class NoopLock(object):
454 def __exit__(self, exc_type, exc_value, traceback):
457 def acquire(self, blocking=False):
464 def must_be_writable(orig_func):
465 @functools.wraps(orig_func)
466 def must_be_writable_wrapper(self, *args, **kwargs):
467 if not self.writable():
468 raise IOError(errno.EROFS, "Collection is read-only.")
469 return orig_func(self, *args, **kwargs)
470 return must_be_writable_wrapper
473 class _BlockManager(object):
474 """BlockManager handles buffer blocks.
476 Also handles background block uploads, and background block prefetch for a
477 Collection of ArvadosFiles.
481 DEFAULT_PUT_THREADS = 2
482 DEFAULT_GET_THREADS = 2
484 def __init__(self, keep, copies=None, put_threads=None, num_retries=None, storage_classes_func=None, get_threads=None):
485 """keep: KeepClient object to use"""
487 self._bufferblocks = collections.OrderedDict()
488 self._put_queue = None
489 self._put_threads = None
490 self._prefetch_queue = None
491 self._prefetch_threads = None
492 self.lock = threading.Lock()
493 self.prefetch_enabled = True
494 self.num_put_threads = put_threads or _BlockManager.DEFAULT_PUT_THREADS
495 self.num_get_threads = get_threads or _BlockManager.DEFAULT_GET_THREADS
497 self.storage_classes = storage_classes_func or (lambda: [])
498 self._pending_write_size = 0
499 self.threads_lock = threading.Lock()
500 self.padding_block = None
501 self.num_retries = num_retries
504 def alloc_bufferblock(self, blockid=None, starting_capacity=2**14, owner=None):
505 """Allocate a new, empty bufferblock in WRITABLE state and return it.
508 optional block identifier, otherwise one will be automatically assigned
511 optional capacity, otherwise will use default capacity
514 ArvadosFile that owns this block
517 return self._alloc_bufferblock(blockid, starting_capacity, owner)
519 def _alloc_bufferblock(self, blockid=None, starting_capacity=2**14, owner=None):
521 blockid = str(uuid.uuid4())
522 bufferblock = _BufferBlock(blockid, starting_capacity=starting_capacity, owner=owner)
523 self._bufferblocks[bufferblock.blockid] = bufferblock
527 def dup_block(self, block, owner):
528 """Create a new bufferblock initialized with the content of an existing bufferblock.
531 the buffer block to copy.
534 ArvadosFile that owns the new block
537 new_blockid = str(uuid.uuid4())
538 bufferblock = block.clone(new_blockid, owner)
539 self._bufferblocks[bufferblock.blockid] = bufferblock
543 def is_bufferblock(self, locator):
544 return locator in self._bufferblocks
546 def _commit_bufferblock_worker(self):
547 """Background uploader thread."""
551 bufferblock = self._put_queue.get()
552 if bufferblock is None:
555 if self.copies is None:
556 loc = self._keep.put(bufferblock.buffer_view[0:bufferblock.write_pointer].tobytes(), num_retries=self.num_retries, classes=self.storage_classes())
558 loc = self._keep.put(bufferblock.buffer_view[0:bufferblock.write_pointer].tobytes(), num_retries=self.num_retries, copies=self.copies, classes=self.storage_classes())
559 bufferblock.set_state(_BufferBlock.COMMITTED, loc)
560 except Exception as e:
561 bufferblock.set_state(_BufferBlock.ERROR, e)
563 if self._put_queue is not None:
564 self._put_queue.task_done()
566 def start_put_threads(self):
567 with self.threads_lock:
568 if self._put_threads is None:
569 # Start uploader threads.
571 # If we don't limit the Queue size, the upload queue can quickly
572 # grow to take up gigabytes of RAM if the writing process is
573 # generating data more quickly than it can be sent to the Keep
576 # With two upload threads and a queue size of 2, this means up to 4
577 # blocks pending. If they are full 64 MiB blocks, that means up to
578 # 256 MiB of internal buffering, which is the same size as the
579 # default download block cache in KeepClient.
580 self._put_queue = queue.Queue(maxsize=2)
582 self._put_threads = []
583 for i in range(0, self.num_put_threads):
584 thread = threading.Thread(target=self._commit_bufferblock_worker)
585 self._put_threads.append(thread)
589 def _block_prefetch_worker(self):
590 """The background downloader thread."""
593 b = self._prefetch_queue.get()
596 self._keep.get(b, prefetch=True)
598 _logger.exception("Exception doing block prefetch")
601 def start_get_threads(self):
602 if self._prefetch_threads is None:
603 self._prefetch_queue = queue.Queue()
604 self._prefetch_threads = []
605 for i in range(0, self.num_get_threads):
606 thread = threading.Thread(target=self._block_prefetch_worker)
607 self._prefetch_threads.append(thread)
613 def stop_threads(self):
614 """Shut down and wait for background upload and download threads to finish."""
616 if self._put_threads is not None:
617 for t in self._put_threads:
618 self._put_queue.put(None)
619 for t in self._put_threads:
621 self._put_threads = None
622 self._put_queue = None
624 if self._prefetch_threads is not None:
625 for t in self._prefetch_threads:
626 self._prefetch_queue.put(None)
627 for t in self._prefetch_threads:
629 self._prefetch_threads = None
630 self._prefetch_queue = None
635 def __exit__(self, exc_type, exc_value, traceback):
639 def repack_small_blocks(self, force=False, sync=False, closed_file_size=0):
640 """Packs small blocks together before uploading"""
642 self._pending_write_size += closed_file_size
644 # Check if there are enough small blocks for filling up one in full
645 if not (force or (self._pending_write_size >= config.KEEP_BLOCK_SIZE)):
648 # Search blocks ready for getting packed together before being
650 # A WRITABLE block always has an owner.
651 # A WRITABLE block with its owner.closed() implies that its
652 # size is <= KEEP_BLOCK_SIZE/2.
654 small_blocks = [b for b in listvalues(self._bufferblocks)
655 if b.state() == _BufferBlock.WRITABLE and b.owner.closed()]
656 except AttributeError:
657 # Writable blocks without owner shouldn't exist.
658 raise UnownedBlockError()
660 if len(small_blocks) <= 1:
661 # Not enough small blocks for repacking
664 for bb in small_blocks:
667 # Update the pending write size count with its true value, just in case
668 # some small file was opened, written and closed several times.
669 self._pending_write_size = sum([b.size() for b in small_blocks])
671 if self._pending_write_size < config.KEEP_BLOCK_SIZE and not force:
674 new_bb = self._alloc_bufferblock()
677 while len(small_blocks) > 0 and (new_bb.write_pointer + small_blocks[0].size()) <= config.KEEP_BLOCK_SIZE:
678 bb = small_blocks.pop(0)
679 new_bb.owner.append(bb.owner)
680 self._pending_write_size -= bb.size()
681 new_bb.append(bb.buffer_view[0:bb.write_pointer].tobytes())
682 files.append((bb, new_bb.write_pointer - bb.size()))
684 self.commit_bufferblock(new_bb, sync=sync)
686 for bb, new_bb_segment_offset in files:
687 newsegs = bb.owner.segments()
689 if s.locator == bb.blockid:
690 s.locator = new_bb.blockid
691 s.segment_offset = new_bb_segment_offset+s.segment_offset
692 bb.owner.set_segments(newsegs)
693 self._delete_bufferblock(bb.blockid)
695 def commit_bufferblock(self, block, sync):
696 """Initiate a background upload of a bufferblock.
699 The block object to upload
702 If `sync` is True, upload the block synchronously.
703 If `sync` is False, upload the block asynchronously. This will
704 return immediately unless the upload queue is at capacity, in
705 which case it will wait on an upload queue slot.
709 # Mark the block as PENDING so to disallow any more appends.
710 block.set_state(_BufferBlock.PENDING)
711 except StateChangeError as e:
712 if e.state == _BufferBlock.PENDING:
714 block.wait_for_commit.wait()
717 if block.state() == _BufferBlock.COMMITTED:
719 elif block.state() == _BufferBlock.ERROR:
726 if self.copies is None:
727 loc = self._keep.put(block.buffer_view[0:block.write_pointer].tobytes(), num_retries=self.num_retries, classes=self.storage_classes())
729 loc = self._keep.put(block.buffer_view[0:block.write_pointer].tobytes(), num_retries=self.num_retries, copies=self.copies, classes=self.storage_classes())
730 block.set_state(_BufferBlock.COMMITTED, loc)
731 except Exception as e:
732 block.set_state(_BufferBlock.ERROR, e)
735 self.start_put_threads()
736 self._put_queue.put(block)
739 def get_bufferblock(self, locator):
740 return self._bufferblocks.get(locator)
743 def get_padding_block(self):
744 """Get a bufferblock 64 MB in size consisting of all zeros, used as padding
745 when using truncate() to extend the size of a file.
747 For reference (and possible future optimization), the md5sum of the
748 padding block is: 7f614da9329cd3aebf59b91aadc30bf0+67108864
752 if self.padding_block is None:
753 self.padding_block = self._alloc_bufferblock(starting_capacity=config.KEEP_BLOCK_SIZE)
754 self.padding_block.write_pointer = config.KEEP_BLOCK_SIZE
755 self.commit_bufferblock(self.padding_block, False)
756 return self.padding_block
759 def delete_bufferblock(self, locator):
760 self._delete_bufferblock(locator)
762 def _delete_bufferblock(self, locator):
763 if locator in self._bufferblocks:
764 bb = self._bufferblocks[locator]
766 del self._bufferblocks[locator]
768 def get_block_contents(self, locator, num_retries, cache_only=False):
771 First checks to see if the locator is a BufferBlock and return that, if
772 not, passes the request through to KeepClient.get().
776 if locator in self._bufferblocks:
777 bufferblock = self._bufferblocks[locator]
778 if bufferblock.state() != _BufferBlock.COMMITTED:
779 return bufferblock.buffer_view[0:bufferblock.write_pointer].tobytes()
781 locator = bufferblock._locator
783 return self._keep.get_from_cache(locator)
785 return self._keep.get(locator, num_retries=num_retries)
787 def commit_all(self):
788 """Commit all outstanding buffer blocks.
790 This is a synchronous call, and will not return until all buffer blocks
791 are uploaded. Raises KeepWriteError() if any blocks failed to upload.
794 self.repack_small_blocks(force=True, sync=True)
797 items = listitems(self._bufferblocks)
800 if v.state() != _BufferBlock.COMMITTED and v.owner:
801 # Ignore blocks with a list of owners, as if they're not in COMMITTED
802 # state, they're already being committed asynchronously.
803 if isinstance(v.owner, ArvadosFile):
804 v.owner.flush(sync=False)
807 if self._put_queue is not None:
808 self._put_queue.join()
812 if v.state() == _BufferBlock.ERROR:
813 err.append((v.locator(), v.error))
815 raise KeepWriteError("Error writing some blocks", err, label="block")
818 # flush again with sync=True to remove committed bufferblocks from
821 if isinstance(v.owner, ArvadosFile):
822 v.owner.flush(sync=True)
823 elif isinstance(v.owner, list) and len(v.owner) > 0:
824 # This bufferblock is referenced by many files as a result
825 # of repacking small blocks, so don't delete it when flushing
826 # its owners, just do it after flushing them all.
827 for owner in v.owner:
828 owner.flush(sync=True)
829 self.delete_bufferblock(k)
831 def block_prefetch(self, locator):
832 """Initiate a background download of a block.
834 This assumes that the underlying KeepClient implements a block cache,
835 so repeated requests for the same block will not result in repeated
836 downloads (unless the block is evicted from the cache.) This method
841 if not self.prefetch_enabled:
845 if locator in self._bufferblocks:
848 self.start_get_threads()
849 self._prefetch_queue.put(locator)
852 class ArvadosFile(object):
853 """Represent a file in a Collection.
855 ArvadosFile manages the underlying representation of a file in Keep as a
856 sequence of segments spanning a set of blocks, and implements random
859 This object may be accessed from multiple threads.
863 __slots__ = ('parent', 'name', '_writers', '_committed',
864 '_segments', 'lock', '_current_bblock', 'fuse_entry')
866 def __init__(self, parent, name, stream=[], segments=[]):
868 ArvadosFile constructor.
871 a list of Range objects representing a block stream
874 a list of Range objects representing segments
878 self._writers = set()
879 self._committed = False
881 self.lock = parent.root_collection().lock
883 self._add_segment(stream, s.locator, s.range_size)
884 self._current_bblock = None
887 return self.parent.writable()
890 def permission_expired(self, as_of_dt=None):
891 """Returns True if any of the segment's locators is expired"""
892 for r in self._segments:
893 if KeepLocator(r.locator).permission_expired(as_of_dt):
898 def has_remote_blocks(self):
899 """Returns True if any of the segment's locators has a +R signature"""
901 for s in self._segments:
902 if '+R' in s.locator:
907 def _copy_remote_blocks(self, remote_blocks={}):
908 """Ask Keep to copy remote blocks and point to their local copies.
910 This is called from the parent Collection.
913 Shared cache of remote to local block mappings. This is used to avoid
914 doing extra work when blocks are shared by more than one file in
915 different subdirectories.
918 for s in self._segments:
919 if '+R' in s.locator:
921 loc = remote_blocks[s.locator]
923 loc = self.parent._my_keep().refresh_signature(s.locator)
924 remote_blocks[s.locator] = loc
926 self.parent.set_committed(False)
931 return copy.copy(self._segments)
934 def clone(self, new_parent, new_name):
935 """Make a copy of this file."""
936 cp = ArvadosFile(new_parent, new_name)
937 cp.replace_contents(self)
942 def replace_contents(self, other):
943 """Replace segments of this file with segments from another `ArvadosFile` object."""
947 for other_segment in other.segments():
948 new_loc = other_segment.locator
949 if other.parent._my_block_manager().is_bufferblock(other_segment.locator):
950 if other_segment.locator not in map_loc:
951 bufferblock = other.parent._my_block_manager().get_bufferblock(other_segment.locator)
952 if bufferblock.state() != _BufferBlock.WRITABLE:
953 map_loc[other_segment.locator] = bufferblock.locator()
955 map_loc[other_segment.locator] = self.parent._my_block_manager().dup_block(bufferblock, self).blockid
956 new_loc = map_loc[other_segment.locator]
958 self._segments.append(Range(new_loc, other_segment.range_start, other_segment.range_size, other_segment.segment_offset))
960 self.set_committed(False)
962 def __eq__(self, other):
965 if not isinstance(other, ArvadosFile):
968 othersegs = other.segments()
970 if len(self._segments) != len(othersegs):
972 for i in range(0, len(othersegs)):
973 seg1 = self._segments[i]
978 if self.parent._my_block_manager().is_bufferblock(loc1):
979 loc1 = self.parent._my_block_manager().get_bufferblock(loc1).locator()
981 if other.parent._my_block_manager().is_bufferblock(loc2):
982 loc2 = other.parent._my_block_manager().get_bufferblock(loc2).locator()
984 if (KeepLocator(loc1).stripped() != KeepLocator(loc2).stripped() or
985 seg1.range_start != seg2.range_start or
986 seg1.range_size != seg2.range_size or
987 seg1.segment_offset != seg2.segment_offset):
992 def __ne__(self, other):
993 return not self.__eq__(other)
996 def set_segments(self, segs):
997 self._segments = segs
1000 def set_committed(self, value=True):
1001 """Set committed flag.
1003 If value is True, set committed to be True.
1005 If value is False, set committed to be False for this and all parents.
1007 if value == self._committed:
1009 self._committed = value
1010 if self._committed is False and self.parent is not None:
1011 self.parent.set_committed(False)
1014 def committed(self):
1015 """Get whether this is committed or not."""
1016 return self._committed
1019 def add_writer(self, writer):
1020 """Add an ArvadosFileWriter reference to the list of writers"""
1021 if isinstance(writer, ArvadosFileWriter):
1022 self._writers.add(writer)
1025 def remove_writer(self, writer, flush):
1027 Called from ArvadosFileWriter.close(). Remove a writer reference from the list
1028 and do some block maintenance tasks.
1030 self._writers.remove(writer)
1032 if flush or self.size() > config.KEEP_BLOCK_SIZE // 2:
1033 # File writer closed, not small enough for repacking
1036 # All writers closed and size is adequate for repacking
1037 self.parent._my_block_manager().repack_small_blocks(closed_file_size=self.size())
1041 Get whether this is closed or not. When the writers list is empty, the file
1042 is supposed to be closed.
1044 return len(self._writers) == 0
1048 def truncate(self, size):
1049 """Shrink or expand the size of the file.
1051 If `size` is less than the size of the file, the file contents after
1052 `size` will be discarded. If `size` is greater than the current size
1053 of the file, it will be filled with zero bytes.
1056 if size < self.size():
1058 for r in self._segments:
1059 range_end = r.range_start+r.range_size
1060 if r.range_start >= size:
1061 # segment is past the trucate size, all done
1063 elif size < range_end:
1064 nr = Range(r.locator, r.range_start, size - r.range_start, 0)
1065 nr.segment_offset = r.segment_offset
1071 self._segments = new_segs
1072 self.set_committed(False)
1073 elif size > self.size():
1074 padding = self.parent._my_block_manager().get_padding_block()
1075 diff = size - self.size()
1076 while diff > config.KEEP_BLOCK_SIZE:
1077 self._segments.append(Range(padding.blockid, self.size(), config.KEEP_BLOCK_SIZE, 0))
1078 diff -= config.KEEP_BLOCK_SIZE
1080 self._segments.append(Range(padding.blockid, self.size(), diff, 0))
1081 self.set_committed(False)
1083 # size == self.size()
1086 def readfrom(self, offset, size, num_retries, exact=False):
1087 """Read up to `size` bytes from the file starting at `offset`.
1090 If False (default), return less data than requested if the read
1091 crosses a block boundary and the next block isn't cached. If True,
1092 only return less data than requested when hitting EOF.
1096 if size == 0 or offset >= self.size():
1098 readsegs = locators_and_ranges(self._segments, offset, size)
1099 prefetch = locators_and_ranges(self._segments, offset + size, config.KEEP_BLOCK_SIZE * self.parent._my_block_manager().num_get_threads, limit=32)
1104 block = self.parent._my_block_manager().get_block_contents(lr.locator, num_retries=num_retries, cache_only=(bool(data) and not exact))
1106 blockview = memoryview(block)
1107 data.append(blockview[lr.segment_offset:lr.segment_offset+lr.segment_size].tobytes())
1108 locs.add(lr.locator)
1113 if lr.locator not in locs:
1114 self.parent._my_block_manager().block_prefetch(lr.locator)
1115 locs.add(lr.locator)
1117 return b''.join(data)
1121 def writeto(self, offset, data, num_retries):
1122 """Write `data` to the file starting at `offset`.
1124 This will update existing bytes and/or extend the size of the file as
1128 if not isinstance(data, bytes) and not isinstance(data, memoryview):
1129 data = data.encode()
1133 if offset > self.size():
1134 self.truncate(offset)
1136 if len(data) > config.KEEP_BLOCK_SIZE:
1137 # Chunk it up into smaller writes
1139 dataview = memoryview(data)
1140 while n < len(data):
1141 self.writeto(offset+n, dataview[n:n + config.KEEP_BLOCK_SIZE].tobytes(), num_retries)
1142 n += config.KEEP_BLOCK_SIZE
1145 self.set_committed(False)
1147 if self._current_bblock is None or self._current_bblock.state() != _BufferBlock.WRITABLE:
1148 self._current_bblock = self.parent._my_block_manager().alloc_bufferblock(owner=self)
1150 if (self._current_bblock.size() + len(data)) > config.KEEP_BLOCK_SIZE:
1151 self._current_bblock.repack_writes()
1152 if (self._current_bblock.size() + len(data)) > config.KEEP_BLOCK_SIZE:
1153 self.parent._my_block_manager().commit_bufferblock(self._current_bblock, sync=False)
1154 self._current_bblock = self.parent._my_block_manager().alloc_bufferblock(owner=self)
1156 self._current_bblock.append(data)
1158 replace_range(self._segments, offset, len(data), self._current_bblock.blockid, self._current_bblock.write_pointer - len(data))
1160 self.parent.notify(WRITE, self.parent, self.name, (self, self))
1165 def flush(self, sync=True, num_retries=0):
1166 """Flush the current bufferblock to Keep.
1169 If True, commit block synchronously, wait until buffer block has been written.
1170 If False, commit block asynchronously, return immediately after putting block into
1173 if self.committed():
1176 if self._current_bblock and self._current_bblock.state() != _BufferBlock.COMMITTED:
1177 if self._current_bblock.state() == _BufferBlock.WRITABLE:
1178 self._current_bblock.repack_writes()
1179 if self._current_bblock.state() != _BufferBlock.DELETED:
1180 self.parent._my_block_manager().commit_bufferblock(self._current_bblock, sync=sync)
1184 for s in self._segments:
1185 bb = self.parent._my_block_manager().get_bufferblock(s.locator)
1187 if bb.state() != _BufferBlock.COMMITTED:
1188 self.parent._my_block_manager().commit_bufferblock(bb, sync=True)
1189 to_delete.add(s.locator)
1190 s.locator = bb.locator()
1192 # Don't delete the bufferblock if it's owned by many files. It'll be
1193 # deleted after all of its owners are flush()ed.
1194 if self.parent._my_block_manager().get_bufferblock(s).owner is self:
1195 self.parent._my_block_manager().delete_bufferblock(s)
1197 self.parent.notify(MOD, self.parent, self.name, (self, self))
1201 def add_segment(self, blocks, pos, size):
1202 """Add a segment to the end of the file.
1204 `pos` and `offset` reference a section of the stream described by
1205 `blocks` (a list of Range objects)
1208 self._add_segment(blocks, pos, size)
1210 def _add_segment(self, blocks, pos, size):
1211 """Internal implementation of add_segment."""
1212 self.set_committed(False)
1213 for lr in locators_and_ranges(blocks, pos, size):
1214 last = self._segments[-1] if self._segments else Range(0, 0, 0, 0)
1215 r = Range(lr.locator, last.range_start+last.range_size, lr.segment_size, lr.segment_offset)
1216 self._segments.append(r)
1220 """Get the file size."""
1222 n = self._segments[-1]
1223 return n.range_start + n.range_size
1228 def manifest_text(self, stream_name=".", portable_locators=False,
1229 normalize=False, only_committed=False):
1232 for segment in self._segments:
1233 loc = segment.locator
1234 if self.parent._my_block_manager().is_bufferblock(loc):
1237 loc = self.parent._my_block_manager().get_bufferblock(loc).locator()
1238 if portable_locators:
1239 loc = KeepLocator(loc).stripped()
1240 filestream.append(LocatorAndRange(loc, KeepLocator(loc).size,
1241 segment.segment_offset, segment.range_size))
1242 buf += ' '.join(normalize_stream(stream_name, {self.name: filestream}))
1248 def _reparent(self, newparent, newname):
1249 self.set_committed(False)
1250 self.flush(sync=True)
1251 self.parent.remove(self.name)
1252 self.parent = newparent
1254 self.lock = self.parent.root_collection().lock
1257 class ArvadosFileReader(ArvadosFileReaderBase):
1258 """Wraps ArvadosFile in a file-like object supporting reading only.
1260 Be aware that this class is NOT thread safe as there is no locking around
1261 updating file pointer.
1265 def __init__(self, arvadosfile, mode="r", num_retries=None):
1266 super(ArvadosFileReader, self).__init__(arvadosfile.name, mode=mode, num_retries=num_retries)
1267 self.arvadosfile = arvadosfile
1270 return self.arvadosfile.size()
1272 def stream_name(self):
1273 return self.arvadosfile.parent.stream_name()
1275 def readinto(self, b):
1276 data = self.read(len(b))
1277 b[:len(data)] = data
1280 @_FileLikeObjectBase._before_close
1282 def read(self, size=None, num_retries=None):
1283 """Read up to `size` bytes from the file and return the result.
1285 Starts at the current file position. If `size` is None, read the
1286 entire remainder of the file.
1290 rd = self.arvadosfile.readfrom(self._filepos, config.KEEP_BLOCK_SIZE, num_retries)
1293 self._filepos += len(rd)
1294 rd = self.arvadosfile.readfrom(self._filepos, config.KEEP_BLOCK_SIZE, num_retries)
1295 return b''.join(data)
1297 data = self.arvadosfile.readfrom(self._filepos, size, num_retries, exact=True)
1298 self._filepos += len(data)
1301 @_FileLikeObjectBase._before_close
1303 def readfrom(self, offset, size, num_retries=None):
1304 """Read up to `size` bytes from the stream, starting at the specified file offset.
1306 This method does not change the file position.
1308 return self.arvadosfile.readfrom(offset, size, num_retries)
1314 class ArvadosFileWriter(ArvadosFileReader):
1315 """Wraps ArvadosFile in a file-like object supporting both reading and writing.
1317 Be aware that this class is NOT thread safe as there is no locking around
1318 updating file pointer.
1322 def __init__(self, arvadosfile, mode, num_retries=None):
1323 super(ArvadosFileWriter, self).__init__(arvadosfile, mode=mode, num_retries=num_retries)
1324 self.arvadosfile.add_writer(self)
1329 @_FileLikeObjectBase._before_close
1331 def write(self, data, num_retries=None):
1332 if self.mode[0] == "a":
1333 self._filepos = self.size()
1334 self.arvadosfile.writeto(self._filepos, data, num_retries)
1335 self._filepos += len(data)
1338 @_FileLikeObjectBase._before_close
1340 def writelines(self, seq, num_retries=None):
1342 self.write(s, num_retries=num_retries)
1344 @_FileLikeObjectBase._before_close
1345 def truncate(self, size=None):
1347 size = self._filepos
1348 self.arvadosfile.truncate(size)
1350 @_FileLikeObjectBase._before_close
1352 self.arvadosfile.flush()
1354 def close(self, flush=True):
1356 self.arvadosfile.remove_writer(self, flush)
1357 super(ArvadosFileWriter, self).close()
1360 class WrappableFile(object):
1361 """An interface to an Arvados file that's compatible with io wrappers.
1364 def __init__(self, f):
1369 return self.f.close()
1371 return self.f.flush()
1372 def read(self, *args, **kwargs):
1373 return self.f.read(*args, **kwargs)
1375 return self.f.readable()
1376 def readinto(self, *args, **kwargs):
1377 return self.f.readinto(*args, **kwargs)
1378 def seek(self, *args, **kwargs):
1379 return self.f.seek(*args, **kwargs)
1381 return self.f.seekable()
1383 return self.f.tell()
1385 return self.f.writable()
1386 def write(self, *args, **kwargs):
1387 return self.f.write(*args, **kwargs)