X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/b5d352d6099b60db5dcdd9183dcab3e8e17d729e..ea6f25f0dde5c750eacea29662c19149c7800134:/sdk/python/arvados/arvfile.py diff --git a/sdk/python/arvados/arvfile.py b/sdk/python/arvados/arvfile.py index 2c653d7f09..6893b94bf7 100644 --- a/sdk/python/arvados/arvfile.py +++ b/sdk/python/arvados/arvfile.py @@ -1,3 +1,7 @@ +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + from __future__ import absolute_import from __future__ import division from future import standard_library @@ -84,9 +88,6 @@ class _FileLikeObjectBase(object): class ArvadosFileReaderBase(_FileLikeObjectBase): def __init__(self, name, mode, num_retries=None): super(ArvadosFileReaderBase, self).__init__(name, mode) - self._binary = 'b' in mode - if sys.version_info >= (3, 0) and not self._binary: - raise NotImplementedError("text mode {!r} is not implemented".format(mode)) self._filepos = 0 self.num_retries = num_retries self._readline_cache = (None, None) @@ -480,7 +481,7 @@ class _BlockManager(object): DEFAULT_PUT_THREADS = 2 DEFAULT_GET_THREADS = 2 - def __init__(self, keep, copies=None, put_threads=None): + def __init__(self, keep, copies=None, put_threads=None, num_retries=None): """keep: KeepClient object to use""" self._keep = keep self._bufferblocks = collections.OrderedDict() @@ -499,6 +500,7 @@ class _BlockManager(object): self._pending_write_size = 0 self.threads_lock = threading.Lock() self.padding_block = None + self.num_retries = num_retries @synchronized def alloc_bufferblock(self, blockid=None, starting_capacity=2**14, owner=None): @@ -553,9 +555,9 @@ class _BlockManager(object): return if self.copies is None: - loc = self._keep.put(bufferblock.buffer_view[0:bufferblock.write_pointer].tobytes()) + loc = self._keep.put(bufferblock.buffer_view[0:bufferblock.write_pointer].tobytes(), num_retries=self.num_retries) else: - loc = self._keep.put(bufferblock.buffer_view[0:bufferblock.write_pointer].tobytes(), copies=self.copies) + loc = self._keep.put(bufferblock.buffer_view[0:bufferblock.write_pointer].tobytes(), num_retries=self.num_retries, copies=self.copies) bufferblock.set_state(_BufferBlock.COMMITTED, loc) except Exception as e: bufferblock.set_state(_BufferBlock.ERROR, e) @@ -724,9 +726,9 @@ class _BlockManager(object): if sync: try: if self.copies is None: - loc = self._keep.put(block.buffer_view[0:block.write_pointer].tobytes()) + loc = self._keep.put(block.buffer_view[0:block.write_pointer].tobytes(), num_retries=self.num_retries) else: - loc = self._keep.put(block.buffer_view[0:block.write_pointer].tobytes(), copies=self.copies) + loc = self._keep.put(block.buffer_view[0:block.write_pointer].tobytes(), num_retries=self.num_retries, copies=self.copies) block.set_state(_BufferBlock.COMMITTED, loc) except Exception as e: block.set_state(_BufferBlock.ERROR, e) @@ -799,7 +801,7 @@ class _BlockManager(object): if v.state() != _BufferBlock.COMMITTED and v.owner: # Ignore blocks with a list of owners, as if they're not in COMMITTED # state, they're already being committed asynchronously. - if not isinstance(v.owner, list): + if isinstance(v.owner, ArvadosFile): v.owner.flush(sync=False) with self.lock: @@ -824,7 +826,7 @@ class _BlockManager(object): # of repacking small blocks, so don't delete it when flushing # its owners, just do it after flushing them all. for owner in v.owner: - owner.flush(sync=True, delete_bufferblock=False) + owner.flush(sync=True) self.delete_bufferblock(k) def block_prefetch(self, locator): @@ -862,6 +864,9 @@ class ArvadosFile(object): """ + __slots__ = ('parent', 'name', '_writers', '_committed', + '_segments', 'lock', '_current_bblock', 'fuse_entry') + def __init__(self, parent, name, stream=[], segments=[]): """ ArvadosFile constructor. @@ -893,6 +898,38 @@ class ArvadosFile(object): return True return False + @synchronized + def has_remote_blocks(self): + """Returns True if any of the segment's locators has a +R signature""" + + for s in self._segments: + if '+R' in s.locator: + return True + return False + + @synchronized + def _copy_remote_blocks(self, remote_blocks={}): + """Ask Keep to copy remote blocks and point to their local copies. + + This is called from the parent Collection. + + :remote_blocks: + Shared cache of remote to local block mappings. This is used to avoid + doing extra work when blocks are shared by more than one file in + different subdirectories. + """ + + for s in self._segments: + if '+R' in s.locator: + try: + loc = remote_blocks[s.locator] + except KeyError: + loc = self.parent._my_keep().refresh_signature(s.locator) + remote_blocks[s.locator] = loc + s.locator = loc + self.parent.set_committed(False) + return remote_blocks + @synchronized def segments(self): return copy.copy(self._segments) @@ -1129,7 +1166,7 @@ class ArvadosFile(object): return len(data) @synchronized - def flush(self, sync=True, num_retries=0, delete_bufferblock=True): + def flush(self, sync=True, num_retries=0): """Flush the current bufferblock to Keep. :sync: @@ -1155,8 +1192,10 @@ class ArvadosFile(object): self.parent._my_block_manager().commit_bufferblock(bb, sync=True) to_delete.add(s.locator) s.locator = bb.locator() - if delete_bufferblock: - for s in to_delete: + for s in to_delete: + # Don't delete the bufferblock if it's owned by many files. It'll be + # deleted after all of its owners are flush()ed. + if self.parent._my_block_manager().get_bufferblock(s).owner is self: self.parent._my_block_manager().delete_bufferblock(s) self.parent.notify(MOD, self.parent, self.name, (self, self)) @@ -1237,6 +1276,11 @@ class ArvadosFileReader(ArvadosFileReaderBase): def stream_name(self): return self.arvadosfile.parent.stream_name() + def readinto(self, b): + data = self.read(len(b)) + b[:len(data)] = data + return len(data) + @_FileLikeObjectBase._before_close @retry_method def read(self, size=None, num_retries=None): @@ -1315,3 +1359,33 @@ class ArvadosFileWriter(ArvadosFileReader): if not self.closed: self.arvadosfile.remove_writer(self, flush) super(ArvadosFileWriter, self).close() + + +class WrappableFile(object): + """An interface to an Arvados file that's compatible with io wrappers. + + """ + def __init__(self, f): + self.f = f + self.closed = False + def close(self): + self.closed = True + return self.f.close() + def flush(self): + return self.f.flush() + def read(self, *args, **kwargs): + return self.f.read(*args, **kwargs) + def readable(self): + return self.f.readable() + def readinto(self, *args, **kwargs): + return self.f.readinto(*args, **kwargs) + def seek(self, *args, **kwargs): + return self.f.seek(*args, **kwargs) + def seekable(self): + return self.f.seekable() + def tell(self): + return self.f.tell() + def writable(self): + return self.f.writable() + def write(self, *args, **kwargs): + return self.f.write(*args, **kwargs)