def exportFile(self, oid, f=None):
if f is None:
f = TemporaryFile()
elif isinstance(f, str):
f = open(f,'w+b')
f.write('ZEXP')
oids = [oid]
done_oids = {}
done=done_oids.has_key
load=self._storage.load
while oids:
oid = oids.pop(0)
if oid in done_oids:
continue
done_oids[oid] = True
try:
p, serial = load(oid, self._version)
except:
logger.debug("broken reference for oid %s", repr(oid),
exc_info=True)
else:
referencesf(p, oids)
f.writelines([oid, p64(len(p)), p])
f.write(export_end_marker)
return f
def exportFile(self, oid, f=None):
if f is None:
f = TemporaryFile()
elif isinstance(f, str):
f = open(f, 'w+b')
f.write('ZEXP')
oids = [oid]
done_oids = {}
done = done_oids.has_key
load = self._storage.load
while oids:
oid = oids.pop(0)
if oid in done_oids:
continue
done_oids[oid] = True
try:
p, serial = load(oid, self._version)
except:
logger.debug("broken reference for oid %s",
repr(oid),
exc_info=True)
else:
referencesf(p, oids)
f.writelines([oid, p64(len(p)), p])
f.write(export_end_marker)
return f
def referrers(storage):
result = {}
for transaction in storage.iterator():
for record in transaction:
for oid in referencesf(record.data):
result.setdefault(oid, []).append((record.oid, record.tid))
return result
def referrers(storage):
result = {}
for transaction in storage.iterator():
for record in transaction:
for oid in referencesf(record.data):
result.setdefault(oid, []).append((record.oid, record.tid))
return result
def buildRefmap(fs):
'''build a refmap from a filestorage. look in every record of every
transaction. build a dict of oid -> list((referenced oids, mod.klass))
'''
refmap = {}
fsi = fs.iterator()
size = os.stat(fs.__name__).st_size
start = time.time()
lastPercent = 0.0
interval = 0.005
print "[1/3] Computing the Reference Map"
for txn in fsi:
percent = float(fsi._file.tell())/float(size) * 100
if(percent - lastPercent > interval):
spentTime = time.time() - start
remainingTime = spentTime / float(fsi._file.tell()) * (float(size)) - spentTime
sys.stdout.write("\r%f%% complete, time spent %s, remaining time: %s" % (percent,GetInHMS(time.time() - start, True), GetInHMS(remainingTime, False)))
lastPercent = percent
for rec in txn:
if rec.data:
mod, klass = get_pickle_metadata(rec.data)
refs = referencesf(rec.data)
refmap[rec.oid] = (refs, mod+"."+klass)
print "\n",
return refmap
def get_references(state):
"""Return the set of OIDs the given state refers to."""
refs = set()
if state:
for oid in referencesf(str(state)):
refs.add(u64(oid))
return refs
def buildRefmap(fs):
'''build a refmap from a filestorage. look in every record of every
transaction. build a dict of oid -> list((referenced oids, mod.klass))
'''
refmap = {}
fsi = fs.iterator()
size = os.stat(fs.__name__).st_size
start = time.time()
lastPercent = 0.0
interval = 0.005
print "[1/3] Computing the Reference Map"
for txn in fsi:
percent = float(fsi._file.tell()) / float(size) * 100
if (percent - lastPercent > interval):
spentTime = time.time() - start
remainingTime = spentTime / float(
fsi._file.tell()) * (float(size)) - spentTime
sys.stdout.write(
"\r%f%% complete, time spent %s, remaining time: %s" %
(percent, GetInHMS(time.time() - start,
True), GetInHMS(remainingTime, False)))
lastPercent = percent
for rec in txn:
if rec.data:
mod, klass = get_pickle_metadata(rec.data)
refs = referencesf(rec.data)
refmap[rec.oid] = (refs, mod + "." + klass)
print "\n",
return refmap
def pack(self, t, referencesf):
with self._lock:
rindex = {}
rootl = ['\0\0\0\0\0\0\0\0']
# mark referenced objects
while rootl:
oid = rootl.pop()
if oid in rindex:
continue
p = self._opickle[oid]
referencesf(p, rootl)
rindex[oid] = None
# sweep unreferenced objects
for oid in self._index.keys():
if oid not in rindex:
self._takeOutGarbage(oid)
def pack(self, t, referencesf):
with self._lock:
rindex = {}
rootl = ['\0\0\0\0\0\0\0\0']
# mark referenced objects
while rootl:
oid = rootl.pop()
if oid in rindex:
continue
p = self._opickle[oid]
referencesf(p, rootl)
rindex[oid] = None
# sweep unreferenced objects
for oid in self._index.keys():
if oid not in rindex:
self._takeOutGarbage(oid)
def findrefs(self, pos):
"""Return a list of oids referenced as of packtime."""
dh = self._read_data_header(pos)
# Chase backpointers until we get to the record with the refs
while dh.back:
dh = self._read_data_header(dh.back)
if dh.plen:
return referencesf(self._file.read(dh.plen))
else:
return []
def findrefs(self, pos):
"""Return a list of oids referenced as of packtime."""
dh = self._read_data_header(pos)
# Chase backpointers until we get to the record with the refs
while dh.back:
dh = self._read_data_header(dh.back)
if dh.plen:
return referencesf(self._file.read(dh.plen))
else:
return []
def analyzeRecords(self, connection, records):
cursor = connection.cursor()
for record in records:
current_oid = u64(record.oid)
referred_oids = set(map(u64, referencesf(record.data)))
for referred_oid in referred_oids or [-1]:
cursor.execute("""
INSERT INTO links (source_oid, target_oid) VALUES
(?, ?)
""", (current_oid, referred_oid))
connection.commit()
def pack(self, t, referencesf):
self._lock_acquire()
try:
rindex={}
referenced=rindex.has_key
rootl=['\0\0\0\0\0\0\0\0']
# mark referenced objects
while rootl:
oid=rootl.pop()
if referenced(oid): continue
p = self._opickle[oid]
referencesf(p, rootl)
rindex[oid] = None
# sweep unreferenced objects
for oid in self._index.keys():
if not referenced(oid):
self._takeOutGarbage(oid)
finally:
self._lock_release()
def pack(self, t, referencesf):
self._lock_acquire()
try:
rindex={}
referenced=rindex.has_key
rootl=['\0\0\0\0\0\0\0\0']
# mark referenced objects
while rootl:
oid=rootl.pop()
if referenced(oid): continue
p = self._opickle[oid]
referencesf(p, rootl)
rindex[oid] = None
# sweep unreferenced objects
for oid in self._index.keys():
if not referenced(oid):
self._takeOutGarbage(oid)
finally:
self._lock_release()
def exportFile(self, oid, f=None):
if f is None:
f = TemporaryFile(prefix="EXP")
elif isinstance(f, six.string_types):
f = open(f, 'w+b')
f.write(b'ZEXP')
oids = [oid]
done_oids = {}
done = done_oids.__contains__
load = self._storage.load
supports_blobs = IBlobStorage.providedBy(self._storage)
while oids:
oid = oids.pop(0)
if oid in done_oids:
continue
done_oids[oid] = True
try:
p, serial = load(oid)
except:
logger.debug("broken reference for oid %s",
repr(oid),
exc_info=True)
else:
referencesf(p, oids)
f.writelines([oid, p64(len(p)), p])
if supports_blobs:
if not isinstance(self._reader.getGhost(p), Blob):
continue # not a blob
blobfilename = self._storage.loadBlob(oid, serial)
f.write(blob_begin_marker)
f.write(p64(os.stat(blobfilename).st_size))
blobdata = open(blobfilename, "rb")
cp(blobdata, f)
blobdata.close()
f.write(export_end_marker)
return f
def exportFile(self, oid, f=None):
if f is None:
f = TemporaryFile(prefix="EXP")
elif isinstance(f, six.string_types):
f = open(f,'w+b')
f.write(b'ZEXP')
oids = [oid]
done_oids = {}
done = done_oids.__contains__
load = self._storage.load
supports_blobs = IBlobStorage.providedBy(self._storage)
while oids:
oid = oids.pop(0)
if oid in done_oids:
continue
done_oids[oid] = True
try:
p, serial = load(oid)
except:
logger.debug("broken reference for oid %s", repr(oid),
exc_info=True)
else:
referencesf(p, oids)
f.writelines([oid, p64(len(p)), p])
if supports_blobs:
if not isinstance(self._reader.getGhost(p), Blob):
continue # not a blob
blobfilename = self._storage.loadBlob(oid, serial)
f.write(blob_begin_marker)
f.write(p64(os.stat(blobfilename).st_size))
blobdata = open(blobfilename, "rb")
cp(blobdata, f)
blobdata.close()
f.write(export_end_marker)
return f
def _total_size(oid, seen):
v = cache.get(oid)
if v is not None:
return v
data, serialno = load_current(fs, oid)
size = len(data)
for suboid in referencesf(data):
if suboid in seen:
continue
seen[suboid] = 1
size += _total_size(suboid, seen)
cache[oid] = size
if len(cache) == cache_size:
cache.popitem()
return size
def check_any_storage(storage):
checked = set()
to_check = deque([p64(0)])
while to_check:
check_oid = to_check.popleft()
checked.add(check_oid)
try:
state, tid = storage.load(check_oid)
except POSKeyError:
print "Bad reference found", u64(check_oid)
for reference in referencesf(state):
if reference not in checked:
to_check.append(reference)
def _total_size(oid, seen):
v = cache.get(oid)
if v is not None:
return v
data, serialno = load_current(fs, oid)
size = len(data)
for suboid in referencesf(data):
if suboid in seen:
continue
seen[suboid] = 1
size += _total_size(suboid, seen)
cache[oid] = size
if len(cache) == cache_size:
cache.popitem()
return size
def main(path):
fs = FileStorage(path, read_only=1)
if PACK:
fs.pack()
db = ZODB.DB(fs)
rt = db.open().root()
paths = find_paths(rt, 3)
def total_size(oid):
cache = {}
cache_size = 1000
def _total_size(oid, seen):
v = cache.get(oid)
if v is not None:
return v
data, serialno = load_current(fs, oid)
size = len(data)
for suboid in referencesf(data):
if suboid in seen:
continue
seen[suboid] = 1
size += _total_size(suboid, seen)
cache[oid] = size
if len(cache) == cache_size:
cache.popitem()
return size
return _total_size(oid, {})
keys = fs._index.keys()
keys.sort()
keys.reverse()
if not VERBOSE:
# If not running verbosely, don't print an entry for an object
# unless it has an entry in paths.
keys = filter(paths.has_key, keys)
fmt = "%8s %5d %8d %s %s.%s"
for oid in keys:
data, serialno = load_current(fs, oid)
mod, klass = get_pickle_metadata(data)
refs = referencesf(data)
path = paths.get(oid, '-')
print(fmt % (U64(oid), len(data), total_size(oid), path, mod, klass))
def check_relstorage(connection):
curr_objs = connection.cursor()
cursor = connection.cursor()
curr_objs.execute('select zoid, tid from current_object')
for zoid, tid in curr_objs:
cursor.execute('select state from object_state where zoid=%s and tid=%s',
(zoid, tid))
state = cursor.fetchone()[0]
if state is None:
continue # How would an object have a null state?
for ref_zoid in referencesf(state):
ref_zoid = u64(ref_zoid)
cursor.execute('select tid from current_object where zoid=%s',
(ref_zoid,))
if cursor.rowcount == 0:
print "Bad reference found", ref_zoid
def main(path):
fs = FileStorage(path, read_only=1)
if PACK:
fs.pack()
db = ZODB.DB(fs)
rt = db.open().root()
paths = find_paths(rt, 3)
def total_size(oid):
cache = {}
cache_size = 1000
def _total_size(oid, seen):
v = cache.get(oid)
if v is not None:
return v
data, serialno = load_current(fs, oid)
size = len(data)
for suboid in referencesf(data):
if suboid in seen:
continue
seen[suboid] = 1
size += _total_size(suboid, seen)
cache[oid] = size
if len(cache) == cache_size:
cache.popitem()
return size
return _total_size(oid, {})
keys = fs._index.keys()
keys.sort()
keys.reverse()
if not VERBOSE:
# If not running verbosely, don't print an entry for an object
# unless it has an entry in paths.
keys = filter(paths.has_key, keys)
fmt = "%8s %5d %8d %s %s.%s"
for oid in keys:
data, serialno = load_current(fs, oid)
mod, klass = get_pickle_metadata(data)
refs = referencesf(data)
path = paths.get(oid, '-')
print(fmt % (U64(oid), len(data), total_size(oid), path, mod, klass))
def exportXML(jar, oid, file=None):
# For performance reasons, exportXML does not use 'XMLrecord' anymore to map
# oids. This requires to initialize MinimalMapping.marked_reference before
# any string output, i.e. in ppml.Reference.__init__
# This also fixed random failures when DemoStorage is used, because oids
# can have values that have a shorter representation in 'repr' instead of
# 'base64' (see ppml.convert) and ppml.String does not support this.
load = jar._storage.load
pickle_dict = {oid: None}
max_cache = [1e7] # do not cache more than 10MB of pickle data
def getReorderedPickle(oid):
p = pickle_dict[oid]
if p is None:
# Versions are ignored, but some 'load()' implementations require them
# FIXME: remove "''" when TmpStore.load() on ZODB stops asking for it.
p = load(oid, "")[0]
p = reorderPickle(jar, p)[1]
if len(p) < max_cache[0]:
max_cache[0] -= len(p)
pickle_dict[oid] = p
return p
# Sort records and initialize id_mapping
id_mapping = ppml.MinimalMapping()
reordered_oid_list = [oid]
for oid in reordered_oid_list:
_mapOid(id_mapping, oid)
for oid in referencesf(getReorderedPickle(oid)):
if oid not in pickle_dict:
pickle_dict[oid] = None
reordered_oid_list.append(oid)
# Do real export
if file is None:
file = TemporaryFile()
elif isinstance(file, basestring):
file = open(file, "w+b")
write = file.write
write('<?xml version="1.0"?>\n<ZopeData>\n')
for oid in reordered_oid_list:
p = getReorderedPickle(oid)
write(XMLrecord(oid, len(p), p, id_mapping))
write("</ZopeData>\n")
return file
def exportXML(jar, oid, file=None):
# For performance reasons, exportXML does not use 'XMLrecord' anymore to map
# oids. This requires to initialize MinimalMapping.marked_reference before
# any string output, i.e. in ppml.Reference.__init__
# This also fixed random failures when DemoStorage is used, because oids
# can have values that have a shorter representation in 'repr' instead of
# 'base64' (see ppml.convert) and ppml.String does not support this.
load = jar._storage.load
pickle_dict = {oid: None}
max_cache = [1e7] # do not cache more than 10MB of pickle data
def getReorderedPickle(oid):
p = pickle_dict[oid]
if p is None:
# Versions are ignored, but some 'load()' implementations require them
# FIXME: remove "''" when TmpStore.load() on ZODB stops asking for it.
p = load(oid, '')[0]
p = reorderPickle(jar, p)[1]
if len(p) < max_cache[0]:
max_cache[0] -= len(p)
pickle_dict[oid] = p
return p
# Sort records and initialize id_mapping
id_mapping = ppml.MinimalMapping()
reordered_oid_list = [oid]
for oid in reordered_oid_list:
_mapOid(id_mapping, oid)
for oid in referencesf(getReorderedPickle(oid)):
if oid not in pickle_dict:
pickle_dict[oid] = None
reordered_oid_list.append(oid)
# Do real export
if file is None:
file = TemporaryFile()
elif isinstance(file, basestring):
file = open(file, 'w+b')
write = file.write
write('<?xml version="1.0"?>\n<ZopeData>\n')
for oid in reordered_oid_list:
p = getReorderedPickle(oid)
write(XMLrecord(oid, len(p), p, id_mapping))
write('</ZopeData>\n')
return file
def _finish(self, tid, u, d, e):
zeros = {}
referenceCount = self._referenceCount
referenceCount_get = referenceCount.get
oreferences = self._oreferences
serial = self._tid
index = self._index
opickle = self._opickle
self._ltid = tid
# iterate over all the objects touched by/created within this
# transaction
for entry in self._tmp:
oid, data = entry[:]
referencesl = []
referencesf(data, referencesl)
references = {}
for roid in referencesl:
references[roid] = 1
# Create a reference count for this object if one
# doesn't already exist
if referenceCount_get(oid) is None:
referenceCount[oid] = 0
# update references that are already associated with this
# object
roids = oreferences.get(oid, [])
for roid in roids:
if roid in references:
# still referenced, so no need to update
# remove it from the references dict so it doesn't
# get "added" in the next clause
del references[roid]
else:
# Delete the stored ref, since we no longer
# have it
oreferences[oid].remove(roid)
# decrement refcnt:
rc = referenceCount_get(roid, 1)
rc = rc - 1
if rc < 0:
# This should never happen
raise ReferenceCountError(
"%s (Oid %r had refcount %s)" %
(ReferenceCountError.__doc__, roid, rc))
referenceCount[roid] = rc
if rc == 0:
zeros[roid] = 1
# Create a reference list for this object if one
# doesn't already exist
if oreferences.get(oid) is None:
oreferences[oid] = []
# Now add any references that weren't already stored
for roid in references.keys():
oreferences[oid].append(roid)
# Create/update refcnt
rc = referenceCount_get(roid, 0)
if rc == 0 and zeros.get(roid) is not None:
del zeros[roid]
referenceCount[roid] = rc + 1
index[oid] = serial
opickle[oid] = data
now = time.time()
self._conflict_cache[(oid, serial)] = data, now
if zeros:
for oid in zeros.keys():
if oid == '\0\0\0\0\0\0\0\0':
continue
self._takeOutGarbage(oid)
self._tmp = []
def _finish(self, tid, u, d, e):
zeros={}
referenceCount=self._referenceCount
referenceCount_get=referenceCount.get
oreferences=self._oreferences
serial=self._tid
index=self._index
opickle=self._opickle
self._ltid = tid
# iterate over all the objects touched by/created within this
# transaction
for entry in self._tmp:
oid, data = entry[:]
referencesl=[]
referencesf(data, referencesl)
references={}
for roid in referencesl:
references[roid]=1
referenced=references.has_key
# Create a reference count for this object if one
# doesn't already exist
if referenceCount_get(oid) is None:
referenceCount[oid] = 0
#zeros[oid]=1
# update references that are already associated with this
# object
roids = oreferences.get(oid, [])
for roid in roids:
if referenced(roid):
# still referenced, so no need to update
# remove it from the references dict so it doesn't
# get "added" in the next clause
del references[roid]
else:
# Delete the stored ref, since we no longer
# have it
oreferences[oid].remove(roid)
# decrement refcnt:
rc = referenceCount_get(roid, 1)
rc=rc-1
if rc < 0:
# This should never happen
raise ReferenceCountError, (
"%s (Oid %s had refcount %s)" %
(ReferenceCountError.__doc__,`roid`,rc)
)
referenceCount[roid] = rc
if rc==0:
zeros[roid]=1
# Create a reference list for this object if one
# doesn't already exist
if oreferences.get(oid) is None:
oreferences[oid] = []
# Now add any references that weren't already stored
for roid in references.keys():
oreferences[oid].append(roid)
# Create/update refcnt
rc=referenceCount_get(roid, 0)
if rc==0 and zeros.get(roid) is not None:
del zeros[roid]
referenceCount[roid] = rc+1
index[oid] = serial
opickle[oid] = data
now = time.time()
self._conflict_cache[(oid, serial)] = data, now
if zeros:
for oid in zeros.keys():
if oid == '\0\0\0\0\0\0\0\0': continue
self._takeOutGarbage(oid)
self._tmp = []
