def snapshot(self) -> object:
"""Produce a comparable snapshot of this Option"""
# Under mypyc, we don't have a __dict__, so we need to do worse things.
d = dict(getattr(self, '__dict__', ()))
for k in get_class_descriptors(Options):
if hasattr(self, k):
d[k] = getattr(self, k)
del d['per_module_cache']
return d
def snapshot(self) -> object:
"""Produce a comparable snapshot of this Option"""
# Under mypyc, we don't have a __dict__, so we need to do worse things.
d = dict(getattr(self, '__dict__', ()))
for k in get_class_descriptors(Options):
if hasattr(self, k):
d[k] = getattr(self, k)
del d['per_module_cache']
return d
def snapshot(self) -> object:
"""Produce a comparable snapshot of this Option"""
# Under mypyc, we don't have a __dict__, so we need to do worse things.
d = dict(getattr(self, '__dict__', ()))
for k in get_class_descriptors(Options):
if hasattr(self, k) and k != "new_semantic_analyzer":
d[k] = getattr(self, k)
# Remove private attributes from snapshot
d = {k: v for k, v in d.items() if not k.startswith('_')}
return d
def collect_memory_stats() -> Tuple[Dict[str, int], Dict[str, int]]:
"""Return stats about memory use.
Return a tuple with these items:
- Dict from object kind to number of instances of that kind
- Dict from object kind to total bytes used by all instances of that kind
"""
objs = gc.get_objects()
find_recursive_objects(objs)
inferred = {}
for obj in objs:
if type(obj) is FakeInfo:
# Processing these would cause a crash.
continue
n = type(obj).__name__
if hasattr(obj, '__dict__'):
# Keep track of which class a particular __dict__ is associated with.
inferred[id(obj.__dict__)] = '%s (__dict__)' % n
if isinstance(obj, (Node, Type)): # type: ignore
if hasattr(obj, '__dict__'):
for x in obj.__dict__.values():
if isinstance(x, list):
# Keep track of which node a list is associated with.
inferred[id(x)] = '%s (list)' % n
if isinstance(x, tuple):
# Keep track of which node a list is associated with.
inferred[id(x)] = '%s (tuple)' % n
for k in get_class_descriptors(type(obj)):
x = getattr(obj, k, None)
if isinstance(x, list):
inferred[id(x)] = '%s (list)' % n
if isinstance(x, tuple):
inferred[id(x)] = '%s (tuple)' % n
freqs = {} # type: Dict[str, int]
memuse = {} # type: Dict[str, int]
for obj in objs:
if id(obj) in inferred:
name = inferred[id(obj)]
else:
name = type(obj).__name__
freqs[name] = freqs.get(name, 0) + 1
memuse[name] = memuse.get(name, 0) + sys.getsizeof(obj)
return freqs, memuse
def collect_memory_stats() -> Tuple[Dict[str, int],
Dict[str, int]]:
"""Return stats about memory use.
Return a tuple with these items:
- Dict from object kind to number of instances of that kind
- Dict from object kind to total bytes used by all instances of that kind
"""
objs = gc.get_objects()
find_recursive_objects(objs)
inferred = {}
for obj in objs:
if type(obj) is FakeInfo:
# Processing these would cause a crash.
continue
n = type(obj).__name__
if hasattr(obj, '__dict__'):
# Keep track of which class a particular __dict__ is associated with.
inferred[id(obj.__dict__)] = '%s (__dict__)' % n
if isinstance(obj, (Node, Type)):
if hasattr(obj, '__dict__'):
for x in obj.__dict__.values():
if isinstance(x, list):
# Keep track of which node a list is associated with.
inferred[id(x)] = '%s (list)' % n
for k in get_class_descriptors(type(obj)):
x = getattr(obj, k, None)
if isinstance(x, list):
inferred[id(x)] = '%s (list)' % n
freqs = {} # type: Dict[str, int]
memuse = {} # type: Dict[str, int]
for obj in objs:
if id(obj) in inferred:
name = inferred[id(obj)]
else:
name = type(obj).__name__
freqs[name] = freqs.get(name, 0) + 1
memuse[name] = memuse.get(name, 0) + sys.getsizeof(obj)
return freqs, memuse
