def packer():
yield "started %s" % datetime.now()
seen = IntSet()
items = self.gen_oid_record(start_oid=int8_to_str(0), seen=seen)
for step in Shelf.generate_shelf(file, items):
yield step
file.flush()
file.fsync()
shelf = Shelf(file)
yield "base written %s" % datetime.now()
# Invalidate oids that have been removed.
for hole in shelf.get_offset_map().gen_holes():
yield hole
oid = int8_to_str(hole)
if self.shelf.get_position(oid) is not None:
assert shelf.get_position(oid) is None
self.invalid.add(oid)
yield "invalidations identified %s" % datetime.now()
for oid in self.pack_extra:
seen.discard(str_to_int8(oid))
for oid in self.pack_extra:
shelf.store(self.gen_oid_record(start_oid=oid, seen=seen))
file.flush()
file.fsync()
if not self.shelf.get_file().is_temporary():
self.shelf.get_file().rename(file_path + '.prepack')
self.shelf.get_file().close()
shelf.get_file().rename(file_path)
self.shelf = shelf
self.pack_extra = None
yield "finished %s" % datetime.now()
def a(self):
int_set = IntSet(size=1000)
for x in range(1000):
assert x not in int_set
int_set.add(x)
assert x in int_set
if x > 0:
assert x - 1 in int_set
for x in range(1000):
int_set.discard(x)
assert x not in int_set
if x > 0:
assert x - 1 not in int_set
def gen_oid_record(self, start_oid=None, **other):
if start_oid is None:
for item in iteritems(self.shelf):
yield item
else:
todo = [start_oid]
seen = IntSet() # This eventually contains them all.
while todo:
oid = todo.pop()
if str_to_int8(oid) in seen:
continue
seen.add(str_to_int8(oid))
record = self.load(oid)
record_oid, data, refdata = unpack_record(record)
assert oid == record_oid
todo.extend(split_oids(refdata))
yield oid, record
def gen_oid_record(self, start_oid=None, **other):
if start_oid is None:
for item in iteritems(self.shelf):
yield item
else:
todo = [start_oid]
seen = IntSet() # This eventually contains them all.
while todo:
oid = todo.pop()
if str_to_int8(oid) in seen:
continue
seen.add(str_to_int8(oid))
record = self.load(oid)
record_oid, data, refdata = unpack_record(record)
assert oid == record_oid
todo.extend(split_oids(refdata))
yield oid, record
def gen_oid_record(self, start_oid=None, seen=None, **other):
if start_oid is None:
for item in iteritems(self.shelf):
yield item
else:
todo = [start_oid]
if seen is None:
seen = IntSet() # This eventually contains them all.
while todo:
oid = heapq.heappop(todo)
if str_to_int8(oid) in seen:
continue
seen.add(str_to_int8(oid))
record = self.load(oid)
record_oid, data, refdata = unpack_record(record)
assert oid == record_oid
for ref_oid in split_oids(refdata):
heapq.heappush(todo, ref_oid)
yield oid, record
