def __init__(self, execute, name=None, provides=None,
requires=None, auto_extract=True, rebind=None, revert=None,
version=None, inject=None):
if not six.callable(execute):
raise ValueError("Function to use for executing must be"
" callable")
if revert is not None:
if not six.callable(revert):
raise ValueError("Function to use for reverting must"
" be callable")
if name is None:
name = reflection.get_callable_name(execute)
super(FunctorTask, self).__init__(name, provides=provides,
inject=inject)
self._execute = execute
self._revert = revert
if version is not None:
self.version = version
mapping = self._build_arg_mapping(execute, requires, rebind,
auto_extract)
self.rebind, exec_requires, self.optional = mapping
if revert:
revert_mapping = self._build_arg_mapping(revert, requires, rebind,
auto_extract)
else:
revert_mapping = (self.rebind, exec_requires, self.optional)
(self.revert_rebind, revert_requires,
self.revert_optional) = revert_mapping
self.requires = exec_requires.union(revert_requires)
def _select_function(sort, typ):
if typ in ['F','D']:
if callable(sort):
#assume the user knows what they're doing
sfunction = sort
elif sort == 'lhp':
sfunction = lambda x,y: (np.real(x/y) < 0.0)
elif sort == 'rhp':
sfunction = lambda x,y: (np.real(x/y) >= 0.0)
elif sort == 'iuc':
sfunction = lambda x,y: (abs(x/y) <= 1.0)
elif sort == 'ouc':
sfunction = lambda x,y: (abs(x/y) > 1.0)
else:
raise ValueError("sort parameter must be None, a callable, or "
"one of ('lhp','rhp','iuc','ouc')")
elif typ in ['f','d']:
if callable(sort):
#assume the user knows what they're doing
sfunction = sort
elif sort == 'lhp':
sfunction = lambda x,y,z: (np.real((x+y*1j)/z) < 0.0)
elif sort == 'rhp':
sfunction = lambda x,y,z: (np.real((x+y*1j)/z) >= 0.0)
elif sort == 'iuc':
sfunction = lambda x,y,z: (abs((x+y*1j)/z) <= 1.0)
elif sort == 'ouc':
sfunction = lambda x,y,z: (abs((x+y*1j)/z) > 1.0)
else:
raise ValueError("sort parameter must be None, a callable, or "
"one of ('lhp','rhp','iuc','ouc')")
else: # to avoid an error later
raise ValueError("dtype %s not understood" % typ)
return sfunction
def add_state(self, state, terminal=False, on_enter=None, on_exit=None):
"""Adds a given state to the state machine.
:param on_enter: callback, if provided will be expected to take
two positional parameters, these being state being
entered and the second parameter is the event that is
being processed that caused the state transition
:param on_exit: callback, if provided will be expected to take
two positional parameters, these being state being
entered and the second parameter is the event that is
being processed that caused the state transition
:param state: state being entered or exited
:type state: string
"""
if state in self._states:
raise excp.Duplicate("State '%s' already defined" % state)
if on_enter is not None:
if not six.callable(on_enter):
raise ValueError("On enter callback must be callable")
if on_exit is not None:
if not six.callable(on_exit):
raise ValueError("On exit callback must be callable")
self._states[state] = {
'terminal': bool(terminal),
'reactions': {},
'on_enter': on_enter,
'on_exit': on_exit,
}
self._transitions[state] = OrderedDict()
def add_state(self, state, on_enter=None, on_exit=None,
target=None, terminal=None):
"""Adds a given state to the state machine.
The on_enter and on_exit callbacks, if provided will be expected to
take two positional parameters, these being the state being exited (for
on_exit) or the state being entered (for on_enter) and a second
parameter which is the event that is being processed that caused the
state transition.
"""
if state in self._states:
raise excp.Duplicate(_("State '%s' already defined") % state)
if on_enter is not None:
if not six.callable(on_enter):
raise ValueError(_("On enter callback must be callable"))
if on_exit is not None:
if not six.callable(on_exit):
raise ValueError(_("On exit callback must be callable"))
if target is not None and target not in self._states:
raise excp.InvalidState(_("Target state '%s' does not exist")
% target)
self._states[state] = {
'terminal': bool(terminal),
'reactions': {},
'on_enter': on_enter,
'on_exit': on_exit,
'target': target,
}
self._transitions[state] = OrderedDict()
def register(self, event_type, callback, args=None, kwargs=None, details_filter=None):
"""Register a callback to be called when event of a given type occurs.
Callback will be called with provided ``args`` and ``kwargs`` and
when event type occurs (or on any event if ``event_type`` equals to
:attr:`.ANY`). It will also get additional keyword argument,
``details``, that will hold event details provided to the
:meth:`.notify` method (if a details filter callback is provided then
the target callback will *only* be triggered if the details filter
callback returns a truthy value).
:param event_type: event type input
:param callback: function callback to be registered.
:param args: non-keyworded arguments
:type args: list
:param kwargs: key-value pair arguments
:type kwargs: dictionary
"""
if not six.callable(callback):
raise ValueError("Event callback must be callable")
if details_filter is not None:
if not six.callable(details_filter):
raise ValueError("Details filter must be callable")
if not self.can_be_registered(event_type):
raise ValueError("Disallowed event type '%s' can not have a" " callback registered" % event_type)
if self.is_registered(event_type, callback, details_filter=details_filter):
raise ValueError("Event callback already registered with" " equivalent details filter")
if kwargs:
for k in self.RESERVED_KEYS:
if k in kwargs:
raise KeyError("Reserved key '%s' not allowed in " "kwargs" % k)
self._topics[event_type].append(Listener(callback, args=args, kwargs=kwargs, details_filter=details_filter))
def add_state(self, state, terminal=False, on_enter=None, on_exit=None):
"""Adds a given state to the state machine.
The on_enter and on_exit callbacks, if provided will be expected to
take two positional parameters, these being the state being exited (for
on_exit) or the state being entered (for on_enter) and a second
parameter which is the event that is being processed that caused the
state transition.
"""
if self.frozen:
raise excp.FrozenMachine()
if state in self._states:
raise excp.Duplicate("State '%s' already defined" % state)
if on_enter is not None:
if not six.callable(on_enter):
raise ValueError("On enter callback must be callable")
if on_exit is not None:
if not six.callable(on_exit):
raise ValueError("On exit callback must be callable")
self._states[state] = {
'terminal': bool(terminal),
'reactions': {},
'on_enter': on_enter,
'on_exit': on_exit,
}
self._transitions[state] = collections.OrderedDict()
def _prepare(cls, elem, attr):
if not isinstance(elem, six.binary_type):
serialize = getattr(elem, 'serialize', None)
if six.callable(serialize):
return serialize()
attr_class = cls._get_attr_class(attr)
serialize = getattr(attr_class, 'serialize', None)
if six.callable(serialize):
return serialize(elem)
return elem
def _set_obj_extra_data_key(obj, eng):
my_value = value
my_key = key
if callable(my_value):
while callable(my_value):
my_value = my_value(obj, eng)
if callable(my_key):
while callable(my_key):
my_key = my_key(obj, eng)
obj.extra_data[str(my_key)] = my_value
def _foreach(obj, eng):
my_list_to_process = []
step = str(eng.getCurrTaskId())
try:
if "_Iterators" not in eng.extra_data:
eng.extra_data["_Iterators"] = {}
except KeyError:
eng.extra_data["_Iterators"] = {}
if step not in eng.extra_data["_Iterators"]:
eng.extra_data["_Iterators"][step] = {}
if cache_data:
if callable(get_list_function):
eng.extra_data["_Iterators"][step]["cache"] = get_list_function(obj, eng)
elif isinstance(get_list_function, list):
eng.extra_data["_Iterators"][step]["cache"] = get_list_function
else:
eng.extra_data["_Iterators"][step]["cache"] = []
my_list_to_process = eng.extra_data["_Iterators"][step]["cache"]
if order == "ASC":
eng.extra_data["_Iterators"][step].update({"value": 0})
elif order == "DSC":
eng.extra_data["_Iterators"][step].update({"value": len(my_list_to_process) - 1})
eng.extra_data["_Iterators"][step]["previous_data"] = obj.data
if callable(get_list_function):
if cache_data:
my_list_to_process = eng.extra_data["_Iterators"][step]["cache"]
else:
my_list_to_process = get_list_function(obj, eng)
elif isinstance(get_list_function, list):
my_list_to_process = get_list_function
else:
my_list_to_process = []
if order == "ASC" and eng.extra_data["_Iterators"][step]["value"] < len(my_list_to_process):
obj.data = my_list_to_process[eng.extra_data["_Iterators"][step]["value"]]
if savename is not None:
obj.extra_data[savename] = obj.data
eng.extra_data["_Iterators"][step]["value"] += 1
elif order == "DSC" and eng.extra_data["_Iterators"][step]["value"] > -1:
obj.data = my_list_to_process[eng.extra_data["_Iterators"][step]["value"]]
if savename is not None:
obj.extra_data[savename] = obj.data
eng.extra_data["_Iterators"][step]["value"] -= 1
else:
obj.data = eng.extra_data["_Iterators"][step]["previous_data"]
del eng.extra_data["_Iterators"][step]
coordonatex = len(eng.getCurrTaskId()) - 1
coordonatey = eng.getCurrTaskId()[coordonatex]
new_vector = eng.getCurrTaskId()
new_vector[coordonatex] = coordonatey + 2
eng.setPosition(eng.getCurrObjId(), new_vector)
def add_listener(self, rule, view_func, **options):
"""Adds a listener to the listeners container; verifies that
`rule` and `view_func` are callable.
:raises TypeError: if rule is not callable.
:raises TypeError: if view_func is not callable
"""
if not six.callable(rule):
raise TypeError('rule should be callable')
if not six.callable(view_func):
raise TypeError('view_func should be callable')
self.listeners.append(Listener(rule, view_func, options))
def _get_files_list(obj, eng):
if callable(parameter):
unknown = parameter
while callable(unknown):
unknown = unknown(obj, eng)
else:
unknown = parameter
result = glob.glob1(path, unknown)
for i in range(0, len(result)):
result[i] = path + os.sep + result[i]
return result
def test_namespace_dereferencing(self):
service = self.object.nested
assert service is self.object.nested
assert six.callable(service)
assert hasattr(service, 'method')
assert six.callable(service.method)
service = self.object.nested.deeper
assert service is self.object.nested.deeper
assert six.callable(service)
assert hasattr(service, 'method')
assert six.callable(service.method)
def register(self, event_type, callback,
args=None, kwargs=None, details_filter=None,
weak=False):
"""Register a callback to be called when event of a given type occurs.
Callback will be called with provided ``args`` and ``kwargs`` and
when event type occurs (or on any event if ``event_type`` equals to
:attr:`.ANY`). It will also get additional keyword argument,
``details``, that will hold event details provided to the
:meth:`.notify` method (if a details filter callback is provided then
the target callback will *only* be triggered if the details filter
callback returns a truthy value).
:param event_type: event type to get triggered on
:param callback: function callback to be registered.
:param args: non-keyworded arguments
:type args: list
:param kwargs: key-value pair arguments
:type kwargs: dictionary
:param weak: if the callback retained should be referenced via
a weak reference or a strong reference (defaults to
holding a strong reference)
:type weak: bool
:returns: the listener that was registered
:rtype: :py:class:`~.Listener`
"""
if not six.callable(callback):
raise ValueError("Event callback must be callable")
if details_filter is not None:
if not six.callable(details_filter):
raise ValueError("Details filter must be callable")
if not self.can_be_registered(event_type):
raise ValueError("Disallowed event type '%s' can not have a"
" callback registered" % event_type)
if kwargs:
for k in self.RESERVED_KEYS:
if k in kwargs:
raise KeyError("Reserved key '%s' not allowed in "
"kwargs" % k)
with self._lock:
if self.is_registered(event_type, callback,
details_filter=details_filter):
raise ValueError("Event callback already registered with"
" equivalent details filter")
listener = Listener(_make_ref(callback, weak=weak),
args=args, kwargs=kwargs,
details_filter=details_filter,
weak=weak)
listeners = self._topics.setdefault(event_type, [])
listeners.append(listener)
return listener
def _update_last_update(obj, eng):
if "_should_last_run_be_update" in obj.extra_data:
if obj.extra_data["_should_last_run_be_update"]:
repository_list_to_process = repository_list
if not isinstance(repository_list_to_process, list):
if callable(repository_list_to_process):
while callable(repository_list_to_process):
repository_list_to_process = repository_list_to_process(
obj, eng)
else:
repository_list_to_process = [
repository_list_to_process]
for repository in repository_list_to_process:
update_lastrun(repository["id"])
def __init__(self, root, default_renderer='mako',
template_path='templates', hooks=lambda: [],
custom_renderers={}, extra_template_vars={},
force_canonical=True, guess_content_type_from_ext=True,
context_local_factory=None, **kw):
self.init_context_local(context_local_factory)
if isinstance(root, six.string_types):
root = self.__translate_root__(root)
self.root = root
self.renderers = RendererFactory(custom_renderers, extra_template_vars)
self.default_renderer = default_renderer
# pre-sort these so we don't have to do it per-request
if six.callable(hooks):
hooks = hooks()
self.hooks = list(sorted(
hooks,
key=operator.attrgetter('priority')
))
self.template_path = template_path
self.force_canonical = force_canonical
self.guess_content_type_from_ext = guess_content_type_from_ext
def url_encode(obj, charset="utf8", encode_keys=False):
items = []
if isinstance(obj, dict):
for k, v in list(obj.items()):
items.append((k, v))
else:
items = list(items)
tmp = []
for k, v in items:
if encode_keys:
k = encode(k, charset)
if not isinstance(v, (tuple, list)):
v = [v]
for v1 in v:
if v1 is None:
v1 = ''
elif six.callable(v1):
v1 = encode(v1(), charset)
else:
v1 = encode(v1, charset)
tmp.append('%s=%s' % (quote(k), quote_plus(v1)))
return '&'.join(tmp)
def power(*args):
'''Returns the "power" composition of several functions.
Examples::
>>> import operator
>>> f = power(partial(operator.mul, 3), 3)
>>> f(23) == 3*(3*(3*23))
True
>>> power(operator.neg)
Traceback (most recent call last):
...
TypeError: Function `power` requires at least two arguments
'''
try:
funcs, times = args[:-1], args[-1]
except IndexError:
msg = "Function `power` requires at least two arguments"
raise TypeError(msg)
if not funcs:
raise TypeError('Function `power` requires at least two arguments')
if any(not callable(func) for func in funcs):
raise TypeError('First arguments of `power` must be callables')
if not isinstance(times, int):
raise TypeError('Last argument of `power` must be int')
if len(funcs) > 1:
base = (compose(funcs), )
else:
base = (funcs[0], )
return compose(*(base * times))
def __getattr__(self, name):
attr = getattr(self.obj, name)
if not six.callable(attr):
return attr
return functools.partial(self.proxy_callable, name)
def _func(request, *args, **kwargs):
user = getattr(request, 'user', None)
is_authenticated = getattr(user, 'is_authenticated', lambda: False)
if ((user is not None
and six.callable(is_authenticated) and not is_authenticated())
or user is None):
user = None
try:
creds = args[:len(authentication_arguments)]
if len(creds) == 0:
raise IndexError
# Django's authenticate() method takes arguments as dict
user = _authenticate(username=creds[0], password=creds[1], *creds[2:])
if user is not None:
args = args[len(authentication_arguments):]
except IndexError:
auth_kwargs = {}
try:
for auth_kwarg in authentication_arguments:
auth_kwargs[auth_kwarg] = kwargs[auth_kwarg]
except KeyError:
raise InvalidParamsError(
'Authenticated methods require at least '
'[%s] or {%s} arguments', authentication_arguments)
user = _authenticate(**auth_kwargs)
if user is not None:
for auth_kwarg in authentication_arguments:
kwargs.pop(auth_kwarg)
if user is None:
raise InvalidCredentialsError
request.user = user
return func(request, *args, **kwargs)
def merge_graphs(graph, *graphs, **kwargs):
"""Merges a bunch of graphs into a new graph.
If no additional graphs are provided the first graph is
returned unmodified otherwise the merged graph is returned.
"""
tmp_graph = graph
allow_overlaps = kwargs.get('allow_overlaps', False)
overlap_detector = kwargs.get('overlap_detector')
if overlap_detector is not None and not six.callable(overlap_detector):
raise ValueError("Overlap detection callback expected to be callable")
elif overlap_detector is None:
overlap_detector = (lambda to_graph, from_graph:
len(to_graph.subgraph(from_graph.nodes_iter())))
for g in graphs:
# This should ensure that the nodes to be merged do not already exist
# in the graph that is to be merged into. This could be problematic if
# there are duplicates.
if not allow_overlaps:
# Attempt to induce a subgraph using the to be merged graphs nodes
# and see if any graph results.
overlaps = overlap_detector(graph, g)
if overlaps:
raise ValueError("Can not merge graph %s into %s since there "
"are %s overlapping nodes (and we do not "
"support merging nodes)" % (g, graph,
overlaps))
graph = nx.algorithms.compose(graph, g)
# Keep the first graphs name.
if graphs:
graph.name = tmp_graph.name
return graph
def process_response(self, request, response):
"""Sets the cache, if needed."""
if (
not hasattr(request, '_cache_update_cache') or
not request._cache_update_cache
):
# We don't need to update the cache, just return.
return response
if request.method != 'GET':
# This is a stronger requirement than above. It is needed
# because of interactions between this middleware and the
# HTTPMiddleware, which throws the body of a HEAD-request
# away before this middleware gets a chance to cache it.
return response
if not response.status_code == 200:
return response
# Try to get the timeout from the "max-age" section of the "Cache-
# Control" header before reverting to using the default cache_timeout
# length.
timeout = get_max_age(response)
if timeout is None:
timeout = self.cache_timeout
elif timeout == 0:
# max-age was set to 0, don't bother caching.
return response
if self.patch_headers:
patch_response_headers(response, timeout)
if timeout:
if callable(self.key_prefix):
key_prefix = self.key_prefix(request)
else:
key_prefix = self.key_prefix
if self.post_process_response:
response = self.post_process_response(
response,
request
)
with RequestPath(request, self.only_get_keys, self.forget_get_keys):
cache_key = learn_cache_key(
request,
response,
timeout,
key_prefix
)
if self.remember_all_urls:
self.remember_url(request, cache_key, timeout)
self.cache.set(cache_key, response, timeout)
if self.post_process_response_always:
response = self.post_process_response_always(
response,
request
)
return response
def handle_key(self, key, check_parent=True):
'''Handle a key press in this window.'''
# First try the first responder if this window has one, but don't allow
# it to check with its parent (False second parameter) so we don't recurse
# and get a stack overflow
for first_responder in reversed(self.first_responders):
if first_responder.handle_key(key, False):
return True
# Check our key map to see if we have any actions. Actions don't take
# any arguments, they must be callable
if key in self.key_actions:
key_action = self.key_actions[key]
key_action['callback']()
return True
# Check if there is a wildcard key for any key
if -1 in self.key_actions:
key_action = self.key_actions[-1]
key_action['callback']()
return True
# Check if the window delegate wants to handle this key press
if self.delegate:
if six.callable(getattr(self.delegate, "handle_key", None)):
if self.delegate.handle_key(self, key):
return True
if self.delegate(self, key):
return True
# Check if we have a parent window and if so, let the parent
# window handle the key press
if check_parent and self.parent:
return self.parent.handle_key(key, True)
else:
return False # Key not handled
def convert(self, value, receiver, context, function_spec, engine,
*convert_args, **convert_kwargs):
if six.callable(value) and hasattr(value, '__unwrapped__'):
value = value.__unwrapped__
def func(*args, **kwargs):
function_context = self._find_function_context(
function_spec, context)
parent_function_context = function_context.parent
if parent_function_context is None:
raise exceptions.NoFunctionRegisteredException(
function_spec.name)
new_name = function_spec.name
if self.with_name:
new_name = args[0]
args = args[1:]
new_receiver = receiver
if self.method is True:
new_receiver = args[0]
args = args[1:]
elif self.method is False:
new_receiver = utils.NO_VALUE
if self.with_context:
new_context = args[0]
args = args[1:]
else:
new_context = context.create_child_context()
return parent_function_context(
new_name, engine, new_receiver, new_context)(*args, **kwargs)
func.__unwrapped__ = value
return func
def convert(self, value, receiver, context, function_spec, engine,
*convert_args, **convert_kwargs):
if six.callable(value) and hasattr(value, '__unwrapped__'):
value = value.__unwrapped__
def func(*args, **kwargs):
name = self.name
if not name:
name = args[0]
args = args[1:]
new_receiver = utils.NO_VALUE
if self.method:
new_receiver = args[0]
args = args[1:]
if self.with_context:
new_context = args[0]
args = args[1:]
else:
new_context = context.create_child_context()
return new_context(
name, engine, new_receiver,
use_convention=self.use_convention)(*args, **kwargs)
func.__unwrapped__ = value
return func
def set_bandwidth(self, bw_method=None):
"""Compute the estimator bandwidth with given method.
The new bandwidth calculated after a call to `set_bandwidth` is used
for subsequent evaluations of the estimated density.
Parameters
----------
bw_method : str, scalar or callable, optional
The method used to calculate the estimator bandwidth. This can be
'scott', 'silverman', a scalar constant or a callable. If a
scalar, this will be used directly as `kde.factor`. If a callable,
it should take a `gaussian_kde` instance as only parameter and
return a scalar. If None (default), nothing happens; the current
`kde.covariance_factor` method is kept.
Notes
-----
.. versionadded:: 0.11
Examples
--------
>>> x1 = np.array([-7, -5, 1, 4, 5.])
>>> kde = stats.gaussian_kde(x1)
>>> xs = np.linspace(-10, 10, num=50)
>>> y1 = kde(xs)
>>> kde.set_bandwidth(bw_method='silverman')
>>> y2 = kde(xs)
>>> kde.set_bandwidth(bw_method=kde.factor / 3.)
>>> y3 = kde(xs)
>>> fig = plt.figure()
>>> ax = fig.add_subplot(111)
>>> ax.plot(x1, np.ones(x1.shape) / (4. * x1.size), 'bo',
... label='Data points (rescaled)')
>>> ax.plot(xs, y1, label='Scott (default)')
>>> ax.plot(xs, y2, label='Silverman')
>>> ax.plot(xs, y3, label='Const (1/3 * Silverman)')
>>> ax.legend()
>>> plt.show()
"""
if bw_method is None:
pass
elif bw_method == 'scott':
self.covariance_factor = self.scotts_factor
elif bw_method == 'silverman':
self.covariance_factor = self.silverman_factor
elif np.isscalar(bw_method) and not isinstance(bw_method, string_types):
self._bw_method = 'use constant'
self.covariance_factor = lambda: bw_method
elif callable(bw_method):
self._bw_method = bw_method
self.covariance_factor = lambda: self._bw_method(self)
else:
msg = "`bw_method` should be 'scott', 'silverman', a scalar " \
"or a callable."
raise ValueError(msg)
self._compute_covariance()
def convert(self, value, receiver, context, function_spec, engine,
*convert_args, **convert_kwargs):
super(Lambda, self).convert(
value, receiver, context, function_spec, engine,
*convert_args, **convert_kwargs)
if value is None:
return None
elif six.callable(value) and hasattr(value, '__unwrapped__'):
value = value.__unwrapped__
def func(*args, **kwargs):
if self.method and self.with_context:
new_receiver, new_context = args[:2]
args = args[2:]
elif self.method and not self.with_context:
new_receiver, new_context = \
args[0], context.create_child_context()
args = args[1:]
elif not self.method and self.with_context:
new_receiver, new_context = utils.NO_VALUE, args[0]
args = args[1:]
else:
new_receiver, new_context = \
utils.NO_VALUE, context.create_child_context()
return self._call(value, new_receiver, new_context,
engine, args, kwargs)
func.__unwrapped__ = value
return func
def add_reaction(self, state, event, reaction, *args, **kwargs):
"""Adds a reaction that may get triggered by the given event & state.
Reaction callbacks may (depending on how the state machine is ran) be
used after an event is processed (and a transition occurs) to cause the
machine to react to the newly arrived at stable state.
These callbacks are expected to accept three default positional
parameters (although more can be passed in via *args and **kwargs,
these will automatically get provided to the callback when it is
activated *ontop* of the three default). The three default parameters
are the last stable state, the new stable state and the event that
caused the transition to this new stable state to be arrived at.
The expected result of a callback is expected to be a new event that
the callback wants the state machine to react to. This new event
may (depending on how the state machine is ran) get processed (and
this process typically repeats) until the state machine reaches a
terminal state.
"""
if self.frozen:
raise excp.FrozenMachine()
if state not in self._states:
raise excp.NotFound("Can not add a reaction to event '%s' for an"
" undefined state '%s'" % (event, state))
if not six.callable(reaction):
raise ValueError("Reaction callback must be callable")
if event not in self._states[state]['reactions']:
self._states[state]['reactions'][event] = (reaction, args, kwargs)
else:
raise excp.Duplicate("State '%s' reaction to event '%s'"
" already defined" % (state, event))
def valid_int(x, cast=None):
'''Ensure that an input value is integer-typed.
This is primarily useful for ensuring integrable-valued
array indices.
Parameters
----------
x : number
A scalar value to be cast to int
cast : function [optional]
A function to modify `x` before casting.
Default: `np.floor`
Returns
-------
x_int : int
`x_int = int(cast(x))`
Raises
------
TypeError
If `cast` is provided and is not callable.
'''
if cast is None:
cast = np.floor
if not six.callable(cast):
raise TypeError('cast parameter must be callable.')
return int(cast(x))
def default(self, obj):
'''
Converts an object and returns a ``JSON``-friendly structure.
:param obj: object or structure to be converted into a
``JSON``-ifiable structure
Considers the following special cases in order:
* object has a callable __json__() attribute defined
returns the result of the call to __json__()
* date and datetime objects
returns the object cast to str
* Decimal objects
returns the object cast to float
* SQLAlchemy objects
returns a copy of the object.__dict__ with internal SQLAlchemy
parameters removed
* SQLAlchemy ResultProxy objects
Casts the iterable ResultProxy into a list of tuples containing
the entire resultset data, returns the list in a dictionary
along with the resultset "row" count.
.. note:: {'count': 5, 'rows': [('Ed Jones',), ('Pete Jones',),
('Wendy Williams',), ('Mary Contrary',), ('Fred Smith',)]}
* SQLAlchemy RowProxy objects
Casts the RowProxy cursor object into a dictionary, probably
losing its ordered dictionary behavior in the process but
making it JSON-friendly.
* webob_dicts objects
returns webob_dicts.mixed() dictionary, which is guaranteed
to be JSON-friendly.
'''
if hasattr(obj, '__json__') and six.callable(obj.__json__):
return obj.__json__()
elif isinstance(obj, (date, datetime)):
return str(obj)
elif isinstance(obj, Decimal):
# XXX What to do about JSONEncoder crappy handling of Decimals?
# SimpleJSON has better Decimal encoding than the std lib
# but only in recent versions
return float(obj)
elif is_saobject(obj):
props = {}
for key in obj.__dict__:
if not key.startswith('_sa_'):
props[key] = getattr(obj, key)
return props
elif isinstance(obj, ResultProxy):
props = dict(rows=list(obj), count=obj.rowcount)
if props['count'] < 0:
props['count'] = len(props['rows'])
return props
elif isinstance(obj, RowProxy):
return dict(obj)
elif isinstance(obj, webob_dicts):
return obj.mixed()
else:
return JSONEncoder.default(self, obj)
def expectedFailure(expected_fn, bugnumber=None):
def expectedFailure_impl(func):
if isinstance(func, type) and issubclass(func, unittest2.TestCase):
raise Exception("Decorator can only be used to decorate a test method")
@wraps(func)
def wrapper(*args, **kwargs):
self = args[0]
if funcutils.requires_self(expected_fn):
xfail_reason = expected_fn(self)
else:
xfail_reason = expected_fn()
if xfail_reason is not None:
if configuration.results_formatter_object is not None:
# Mark this test as expected to fail.
configuration.results_formatter_object.handle_event(
EventBuilder.event_for_mark_test_expected_failure(self))
xfail_func = unittest2.expectedFailure(func)
xfail_func(*args, **kwargs)
else:
func(*args, **kwargs)
return wrapper
# Some decorators can be called both with no arguments (e.g. @expectedFailureWindows)
# or with arguments (e.g. @expectedFailureWindows(compilers=['gcc'])). When called
# the first way, the first argument will be the actual function because decorators are
# weird like that. So this is basically a check that says "which syntax was the original
# function decorated with?"
if six.callable(bugnumber):
return expectedFailure_impl(bugnumber)
else:
return expectedFailure_impl
def frame(est):
"""
Arguments:
est: either an estimator class or an estimator object. The class (or class of the
object) should subclass :py:class:`ibex.sklearn.base.BaseEstimator`.
Returns:
If ``est`` is a class, returns a class; if ``est`` is an object,
returns an object. Note that the result will subclass ``est``
and :py:class:`ibex.FrameMixin`
Example:
>>> from sklearn import linear_model
>>> from ibex import frame
We can use ``frame`` to adapt an object:
>>> prd = frame(linear_model.LinearRegression())
>>> prd
Adapter[LinearRegression](copy_X=True, fit_intercept=True, n_jobs=1, normalize=False)
We can use ``frame`` to adapt a class:
>>> PDLinearRegression = frame(linear_model.LinearRegression)
>>> PDLinearRegression()
Adapter[LinearRegression](copy_X=True, fit_intercept=True, n_jobs=1, normalize=False)
>>> PDLinearRegression(fit_intercept=False)
Adapter[LinearRegression](copy_X=True, fit_intercept=False, n_jobs=1, normalize=False)
"""
from ._base import FrameMixin
if isinstance(est, FrameMixin):
return est
# Tmp Ami - what about feature union?
if isinstance(est, pipeline.Pipeline):
return frame(pipeline.Pipeline)(ests=est.ests)
if not inspect.isclass(est):
params = est.get_params()
f = frame(type(est))(**params)
return f
_Adapter = make_adapter(est)
update_class_wrapper(_Adapter, est)
_Adapter.__name__ = est.__name__
for name, func in vars(_Adapter).items():
if name.startswith('_'):
continue
parfunc = getattr(est, name, None)
if parfunc and getattr(parfunc, '__doc__', None):
func.__doc__ = parfunc.__doc__
wrapped = [
'fit_transform',
'predict_proba',
'sample_y',
'score_samples',
'score',
'staged_predict_proba',
'apply',
'bic',
'perplexity',
'fit',
'decision_function',
'aic',
'partial_fit',
'predict',
'radius_neighbors',
'staged_decision_function',
'staged_predict',
'inverse_transform',
'fit_predict',
'kneighbors',
'predict_log_proba',
'transform',
]
for wrap in wrapped:
if not hasattr(est, wrap) and hasattr(_Adapter, wrap):
delattr(_Adapter, wrap)
elif six.callable(getattr(_Adapter, wrap)):
try:
update_method_wrapper(_Adapter, est, wrap)
except AttributeError:
pass
return _Adapter
def istft(stft_matrix,
hop_length=None,
win_length=None,
window=None,
center=True,
dtype=np.float32):
"""
Inverse short-time Fourier transform (ISTFT).
Converts a complex-valued spectrogram `stft_matrix` to time-series `y`
by minimizing the mean squared error between `stft_matrix` and STFT of
`y` as described in [1]_.
In general, window function, hop length and other parameters should be same
as in stft, which mostly leads to perfect reconstruction of a signal from
unmodified `stft_matrix`.
.. [1] D. W. Griffin and J. S. Lim,
"Signal estimation from modified short-time Fourier transform,"
IEEE Trans. ASSP, vol.32, no.2, pp.236–243, Apr. 1984.
Parameters
----------
stft_matrix : np.ndarray [shape=(1 + n_fft/2, t)]
STFT matrix from `stft`
hop_length : int > 0 [scalar]
Number of frames between STFT columns.
If unspecified, defaults to `win_length / 4`.
win_length : int <= n_fft = 2 * (stft_matrix.shape[0] - 1)
When reconstructing the time series, each frame is windowed
and each sample is normalized by the sum of squared window
according to the `window` function (see below).
If unspecified, defaults to `n_fft`.
window : None, function, np.ndarray [shape=(n_fft,)]
- None (default): use an asymmetric Hann window
- a window function, such as `scipy.signal.hanning`
- a user-specified window vector of length `n_fft`
center : boolean
- If `True`, `D` is assumed to have centered frames.
- If `False`, `D` is assumed to have left-aligned frames.
dtype : numeric type
Real numeric type for `y`. Default is 32-bit float.
Returns
-------
y : np.ndarray [shape=(n,)]
time domain signal reconstructed from `stft_matrix`
Raises
------
ParameterError
If `window` is supplied as a vector of length `n_fft`
See Also
--------
stft : Short-time Fourier Transform
Examples
--------
>>> y, sr = librosa.load(librosa.util.example_audio_file())
>>> D = librosa.stft(y)
>>> y_hat = librosa.istft(D)
>>> y_hat
array([ -4.812e-06, -4.267e-06, ..., 6.271e-06, 2.827e-07], dtype=float32)
"""
n_fft = 2 * (stft_matrix.shape[0] - 1)
# By default, use the entire frame
if win_length is None:
win_length = n_fft
# Set the default hop, if it's not already specified
if hop_length is None:
hop_length = int(win_length / 4)
if window is None:
# Default is an asymmetric Hann window.
ifft_window = scipy.signal.hann(win_length, sym=False)
elif six.callable(window):
# User supplied a windowing function
ifft_window = window(win_length)
else:
# User supplied a window vector.
# Make it into an array
ifft_window = np.asarray(window)
# Verify that the shape matches
if ifft_window.size != n_fft:
raise ParameterError('Size mismatch between n_fft and window size')
# Pad out to match n_fft
ifft_window = util.pad_center(ifft_window, n_fft)
n_frames = stft_matrix.shape[1]
expected_signal_len = n_fft + hop_length * (n_frames - 1)
y = np.zeros(expected_signal_len, dtype=dtype)
ifft_window_sum = np.zeros(expected_signal_len, dtype=dtype)
ifft_window_square = ifft_window * ifft_window
for i in range(n_frames):
sample = i * hop_length
spec = stft_matrix[:, i].flatten()
spec = np.concatenate((spec.conj(), spec[-2:0:-1]), 0)
ytmp = ifft_window * fft.ifft(spec).real
y[sample:(sample + n_fft)] = y[sample:(sample + n_fft)] + ytmp
ifft_window_sum[sample:(sample + n_fft)] += ifft_window_square
# Normalize by sum of squared window
approx_nonzero_indices = ifft_window_sum > util.SMALL_FLOAT
y[approx_nonzero_indices] /= ifft_window_sum[approx_nonzero_indices]
if center:
y = y[int(n_fft // 2):-int(n_fft // 2)]
return y
def logamplitude(S, ref_power=1.0, amin=1e-10, top_db=80.0):
"""Log-scale the amplitude of a spectrogram.
Parameters
----------
S : np.ndarray [shape=(d, t)]
input spectrogram
ref_power : scalar or function
If scalar, `log(abs(S))` is compared to `log(ref_power)`.
If a function, `log(abs(S))` is compared to `log(ref_power(abs(S)))`.
This is primarily useful for comparing to the maximum value of `S`.
amin : float > 0[scalar]
minimum amplitude threshold for `abs(S)` and `ref_power`
top_db : float >= 0 [scalar]
threshold log amplitude at top_db below the peak:
``max(log(S)) - top_db``
Returns
-------
log_S : np.ndarray [shape=(d, t)]
``log_S ~= 10 * log10(S) - 10 * log10(abs(ref_power))``
See Also
--------
perceptual_weighting
Examples
--------
Get a power spectrogram from a waveform ``y``
>>> y, sr = librosa.load(librosa.util.example_audio_file())
>>> S = np.abs(librosa.stft(y))
>>> librosa.logamplitude(S**2)
array([[-33.293, -27.32 , ..., -33.293, -33.293],
[-33.293, -25.723, ..., -33.293, -33.293],
...,
[-33.293, -33.293, ..., -33.293, -33.293],
[-33.293, -33.293, ..., -33.293, -33.293]], dtype=float32)
Compute dB relative to peak power
>>> librosa.logamplitude(S**2, ref_power=np.max)
array([[-80. , -74.027, ..., -80. , -80. ],
[-80. , -72.431, ..., -80. , -80. ],
...,
[-80. , -80. , ..., -80. , -80. ],
[-80. , -80. , ..., -80. , -80. ]], dtype=float32)
Or compare to median power
>>> librosa.logamplitude(S**2, ref_power=np.median)
array([[-0.189, 5.784, ..., -0.189, -0.189],
[-0.189, 7.381, ..., -0.189, -0.189],
...,
[-0.189, -0.189, ..., -0.189, -0.189],
[-0.189, -0.189, ..., -0.189, -0.189]], dtype=float32)
And plot the results
>>> import matplotlib.pyplot as plt
>>> plt.figure()
>>> plt.subplot(2, 1, 1)
>>> librosa.display.specshow(S**2, sr=sr, y_axis='log', x_axis='time')
>>> plt.colorbar()
>>> plt.title('Power spectrogram')
>>> plt.subplot(2, 1, 2)
>>> librosa.display.specshow(librosa.logamplitude(S**2, ref_power=np.max),
... sr=sr, y_axis='log', x_axis='time')
>>> plt.colorbar(format='%+2.0f dB')
>>> plt.title('Log-Power spectrogram')
>>> plt.tight_layout()
"""
if amin <= 0:
raise ParameterError('amin must be strictly positive')
magnitude = np.abs(S)
if six.callable(ref_power):
# User supplied a function to calculate reference power
__ref = ref_power(magnitude)
else:
__ref = np.abs(ref_power)
log_spec = 10.0 * np.log10(np.maximum(amin, magnitude))
log_spec -= 10.0 * np.log10(np.maximum(amin, __ref))
if top_db is not None:
if top_db < 0:
raise ParameterError('top_db must be non-negative positive')
log_spec = np.maximum(log_spec, log_spec.max() - top_db)
return log_spec
def sinc_window(num_zeros=64, precision=9, window=None, rolloff=0.945):
'''Construct a windowed sinc interpolation filter
Parameters
----------
num_zeros : int > 0
The number of zero-crossings to retain in the sinc filter
precision : int > 0
The number of filter coefficients to retain for each zero-crossing
window : callable
The window function. By default, uses Blackman-Harris.
rolloff : float > 0
The roll-off frequency (as a fraction of nyquist)
Returns
-------
interp_window: np.ndarray [shape=(num_zeros * num_table + 1)]
The interpolation window (right-hand side)
num_bits: int
The number of bits of precision to use in the filter table
rolloff : float > 0
The roll-off frequency of the filter, as a fraction of Nyquist
Raises
------
TypeError
if `window` is not callable or `None`
ValueError
if `num_zeros < 1`, `precision < 1`,
or `rolloff` is outside the range `(0, 1]`.
Examples
--------
>>> # A filter with 10 zero-crossings, 32 samples per crossing, and a
... # Hann window for tapering.
>>> halfwin, prec, rolloff = resampy.filters.sinc_window(num_zeros=10, precision=5,
... window=scipy.signal.hann)
>>> halfwin
array([ 9.450e-01, 9.436e-01, ..., -7.455e-07, -0.000e+00])
>>> prec
32
>>> rolloff
0.945
>>> # Or using sinc-window filter construction directly in resample
>>> y = resampy.resample(x, sr_orig, sr_new, filter='sinc_window',
... num_zeros=10, precision=5,
... window=scipy.signal.hann)
'''
if window is None:
window = _blackman_harris_window
elif not six.callable(window):
raise TypeError('window must be callable, not type(window)={}'.format(
type(window)))
if not 0 < rolloff <= 1:
raise ValueError('Invalid roll-off: rolloff={}'.format(rolloff))
if num_zeros < 1:
raise ValueError('Invalid num_zeros: num_zeros={}'.format(num_zeros))
if precision < 0:
raise ValueError('Invalid precision: precision={}'.format(precision))
# Generate the right-wing of the sinc
num_bits = 2**precision
n = num_bits * num_zeros
sinc_win = rolloff * np.sinc(
rolloff * np.linspace(0, num_zeros, num=n + 1, endpoint=True))
# Build the window function and cut off the left half
taper = window(2 * n + 1)[n:]
interp_win = (taper * sinc_win)
return interp_win, num_bits, rolloff
def get_simple_name(component):
if six.callable(component):
return component.__name__
return str(component)
def default(self, obj): # pylint: disable=method-hidden
if hasattr(obj, '__json__') and six.callable(obj.__json__):
return obj.__json__()
else:
return JSONEncoder.default(self, obj)
def on(self, name, callback):
assert six.callable(callback), 'callback is not callable.'
if callback in self.__listeners[name]:
raise DuplicateListenerError()
self.__listeners[name].append(callback)
def get_name(component):
if six.callable(component):
name = getattr(component, "__qualname__", component.__name__)
return '.'.join([component.__module__, name])
return str(component)
def fn(self):
skip_for_os = _match_decorator_property(
lldbplatform.translate(oslist), self.getPlatform())
skip_for_hostos = _match_decorator_property(
lldbplatform.translate(hostoslist),
lldbplatformutil.getHostPlatform())
skip_for_compiler = _match_decorator_property(
compiler, self.getCompiler()) and self.expectedCompilerVersion(compiler_version)
skip_for_arch = _match_decorator_property(
archs, self.getArchitecture())
skip_for_debug_info = _match_decorator_property(
debug_info, self.getDebugInfo())
skip_for_triple = _match_decorator_property(
triple, lldb.DBG.GetSelectedPlatform().GetTriple())
skip_for_remote = _match_decorator_property(
remote, lldb.remote_platform is not None)
skip_for_swig_version = (
swig_version is None) or (
not hasattr(
lldb,
'swig_version')) or (
_check_expected_version(
swig_version[0],
swig_version[1],
lldb.swig_version))
skip_for_py_version = (
py_version is None) or _check_expected_version(
py_version[0], py_version[1], sys.version_info)
skip_for_macos_version = (macos_version is None) or (
_check_expected_version(
macos_version[0],
macos_version[1],
platform.mac_ver()[0]))
# For the test to be skipped, all specified (e.g. not None) parameters must be True.
# An unspecified parameter means "any", so those are marked skip by default. And we skip
# the final test if all conditions are True.
conditions = [(oslist, skip_for_os, "target o/s"),
(hostoslist, skip_for_hostos, "host o/s"),
(compiler, skip_for_compiler, "compiler or version"),
(archs, skip_for_arch, "architecture"),
(debug_info, skip_for_debug_info, "debug info format"),
(triple, skip_for_triple, "target triple"),
(swig_version, skip_for_swig_version, "swig version"),
(py_version, skip_for_py_version, "python version"),
(macos_version, skip_for_macos_version, "macOS version"),
(remote, skip_for_remote, "platform locality (remote/local)")]
reasons = []
final_skip_result = True
for this_condition in conditions:
final_skip_result = final_skip_result and this_condition[1]
if this_condition[0] is not None and this_condition[1]:
reasons.append(this_condition[2])
reason_str = None
if final_skip_result:
mode_str = {
DecorateMode.Skip: "skipping",
DecorateMode.Xfail: "xfailing"}[mode]
if len(reasons) > 0:
reason_str = ",".join(reasons)
reason_str = "{} due to the following parameter(s): {}".format(
mode_str, reason_str)
else:
reason_str = "{} unconditionally"
if bugnumber is not None and not six.callable(bugnumber):
reason_str = reason_str + " [" + str(bugnumber) + "]"
return reason_str
def _log_info(obj, eng):
message_buffer = message
while callable(message_buffer):
message_buffer = message_buffer(obj, eng)
eng.log.info(message_buffer)
def ifgram(y,
sr=22050,
n_fft=2048,
hop_length=None,
win_length=None,
window='hann',
norm=False,
center=True,
ref_power=1e-6,
clip=True,
dtype=np.complex64,
pad_mode='reflect'):
'''Compute the instantaneous frequency (as a proportion of the sampling rate)
obtained as the time-derivative of the phase of the complex spectrum as
described by [1]_.
Calculates regular STFT as a side effect.
.. [1] Abe, Toshihiko, Takao Kobayashi, and Satoshi Imai.
"Harmonics tracking and pitch extraction based on instantaneous
frequency."
International Conference on Acoustics, Speech, and Signal Processing,
ICASSP-95., Vol. 1. IEEE, 1995.
Parameters
----------
y : np.ndarray [shape=(n,)]
audio time series
sr : number > 0 [scalar]
sampling rate of `y`
n_fft : int > 0 [scalar]
FFT window size
hop_length : int > 0 [scalar]
hop length, number samples between subsequent frames.
If not supplied, defaults to `win_length / 4`.
win_length : int > 0, <= n_fft
Window length. Defaults to `n_fft`.
See `stft` for details.
window : string, tuple, number, function, or np.ndarray [shape=(n_fft,)]
- a window specification (string, tuple, number);
see `scipy.signal.get_window`
- a window function, such as `scipy.signal.hanning`
- a user-specified window vector of length `n_fft`
See `stft` for details.
.. see also:: `filters.get_window`
norm : bool
Normalize the STFT.
center : boolean
- If `True`, the signal `y` is padded so that frame
`D[:, t]` (and `if_gram`) is centered at `y[t * hop_length]`.
- If `False`, then `D[:, t]` at `y[t * hop_length]`
ref_power : float >= 0 or callable
Minimum power threshold for estimating instantaneous frequency.
Any bin with `np.abs(D[f, t])**2 < ref_power` will receive the
default frequency estimate.
If callable, the threshold is set to `ref_power(np.abs(D)**2)`.
clip : boolean
- If `True`, clip estimated frequencies to the range `[0, 0.5 * sr]`.
- If `False`, estimated frequencies can be negative or exceed
`0.5 * sr`.
dtype : numeric type
Complex numeric type for `D`. Default is 64-bit complex.
pad_mode : string
If `center=True`, the padding mode to use at the edges of the signal.
By default, STFT uses reflection padding.
Returns
-------
if_gram : np.ndarray [shape=(1 + n_fft/2, t), dtype=real]
Instantaneous frequency spectrogram:
`if_gram[f, t]` is the frequency at bin `f`, time `t`
D : np.ndarray [shape=(1 + n_fft/2, t), dtype=complex]
Short-time Fourier transform
See Also
--------
stft : Short-time Fourier Transform
Examples
--------
>>> y, sr = librosa.load(librosa.util.example_audio_file())
>>> frequencies, D = librosa.ifgram(y, sr=sr)
>>> frequencies
array([[ 0.000e+00, 0.000e+00, ..., 0.000e+00, 0.000e+00],
[ 3.150e+01, 3.070e+01, ..., 1.077e+01, 1.077e+01],
...,
[ 1.101e+04, 1.101e+04, ..., 1.101e+04, 1.101e+04],
[ 1.102e+04, 1.102e+04, ..., 1.102e+04, 1.102e+04]])
'''
if win_length is None:
win_length = n_fft
if hop_length is None:
hop_length = int(win_length // 4)
# Construct a padded hann window
fft_window = util.pad_center(get_window(window, win_length, fftbins=True),
n_fft)
# Window for discrete differentiation
freq_angular = np.linspace(0, 2 * np.pi, n_fft, endpoint=False)
d_window = np.sin(-freq_angular) * np.pi / n_fft
stft_matrix = stft(y,
n_fft=n_fft,
hop_length=hop_length,
win_length=win_length,
window=window,
center=center,
dtype=dtype,
pad_mode=pad_mode)
diff_stft = stft(y,
n_fft=n_fft,
hop_length=hop_length,
window=d_window,
center=center,
dtype=dtype,
pad_mode=pad_mode).conj()
# Compute power normalization. Suppress zeros.
mag, phase = magphase(stft_matrix)
if six.callable(ref_power):
ref_power = ref_power(mag**2)
elif ref_power < 0:
raise ParameterError('ref_power must be non-negative or callable.')
# Pylint does not correctly infer the type here, but it's correct.
# pylint: disable=maybe-no-member
freq_angular = freq_angular.reshape((-1, 1))
bin_offset = (-phase * diff_stft).imag / mag
bin_offset[mag < ref_power**0.5] = 0
if_gram = freq_angular[:n_fft // 2 + 1] + bin_offset
if norm:
stft_matrix = stft_matrix * 2.0 / fft_window.sum()
if clip:
np.clip(if_gram, 0, np.pi, out=if_gram)
if_gram *= float(sr) * 0.5 / np.pi
return if_gram, stft_matrix
def istft(stft_matrix,
hop_length=None,
win_length=None,
window=None,
center=True,
dtype=np.float32):
"""
Inverse short-time Fourier transform.
Converts a complex-valued spectrogram `stft_matrix` to time-series `y`.
Parameters
----------
stft_matrix : np.ndarray [shape=(1 + n_fft/2, t)]
STFT matrix from `stft`
hop_length : int > 0 [scalar]
Number of frames between STFT columns.
If unspecified, defaults to `win_length / 4`.
win_length : int <= n_fft = 2 * (stft_matrix.shape[0] - 1)
When reconstructing the time series, each frame is windowed
according to the `window` function (see below).
If unspecified, defaults to `n_fft`.
window : None, function, np.ndarray [shape=(n_fft,)]
- None (default): use an asymmetric Hann window * 2/3
- a window function, such as `scipy.signal.hanning`
- a user-specified window vector of length `n_fft`
center : boolean
- If `True`, `D` is assumed to have centered frames.
- If `False`, `D` is assumed to have left-aligned frames.
dtype : numeric type
Real numeric type for `y`. Default is 32-bit float.
Returns
-------
y : np.ndarray [shape=(n,)]
time domain signal reconstructed from `stft_matrix`
Raises
------
ParameterError
If `window` is supplied as a vector of length `n_fft`
See Also
--------
stft : Short-time Fourier Transform
Examples
--------
>>> y, sr = librosa.load(librosa.util.example_audio_file())
>>> D = librosa.stft(y)
>>> y_hat = librosa.istft(D)
>>> y_hat
array([ -4.812e-06, -4.267e-06, ..., 6.271e-06, 2.827e-07], dtype=float32)
"""
n_fft = 2 * (stft_matrix.shape[0] - 1)
# By default, use the entire frame
if win_length is None:
win_length = n_fft
# Set the default hop, if it's not already specified
if hop_length is None:
hop_length = int(win_length / 4)
if window is None:
# Default is an asymmetric Hann window.
# 2/3 scaling is to make stft(istft(.)) identity for 25% hop
ifft_window = scipy.signal.hann(win_length, sym=False) * (2.0 / 3)
elif six.callable(window):
# User supplied a windowing function
ifft_window = window(win_length)
else:
# User supplied a window vector.
# Make it into an array
ifft_window = np.asarray(window)
# Verify that the shape matches
if ifft_window.size != n_fft:
raise ParameterError('Size mismatch between n_fft and window size')
# Pad out to match n_fft
ifft_window = pad_center(ifft_window, n_fft)
n_frames = stft_matrix.shape[1]
y = np.zeros(n_fft + hop_length * (n_frames - 1), dtype=dtype)
for i in range(n_frames):
sample = i * hop_length
spec = stft_matrix[:, i].flatten()
spec = np.concatenate((spec.conj(), spec[-2:0:-1]), 0)
ytmp = ifft_window * fft.ifft(spec).real
y[sample:(sample + n_fft)] = y[sample:(sample + n_fft)] + ytmp
if center:
y = y[int(n_fft // 2):-int(n_fft // 2)]
return y
def _actionformat(self, action):
if six.callable(action):
return self.sublist_template.format(action.__name__, "function")
else:
return self.sublist_template.format(action, "command")
def stft(y,
n_fft=2048,
hop_length=None,
win_length=None,
window=None,
center=True,
dtype=np.complex64):
"""Short-time Fourier transform (STFT)
Returns a complex-valued matrix D such that
`np.abs(D[f, t])` is the magnitude of frequency bin `f`
at frame `t`
`np.angle(D[f, t])` is the phase of frequency bin `f`
at frame `t`
Parameters
----------
y : np.ndarray [shape=(n,)], real-valued
the input signal (audio time series)
n_fft : int > 0 [scalar]
FFT window size
hop_length : int > 0 [scalar]
number audio of frames between STFT columns.
If unspecified, defaults `win_length / 4`.
win_length : int <= n_fft [scalar]
Each frame of audio is windowed by `window()`.
The window will be of length `win_length` and then padded
with zeros to match `n_fft`.
If unspecified, defaults to ``win_length = n_fft``.
window : None, function, np.ndarray [shape=(n_fft,)]
- None (default): use an asymmetric Hann window
- a window function, such as `scipy.signal.hanning`
- a vector or array of length `n_fft`
center : boolean
- If `True`, the signal `y` is padded so that frame
`D[:, t]` is centered at `y[t * hop_length]`.
- If `False`, then `D[:, t]` begins at `y[t * hop_length]`
dtype : numeric type
Complex numeric type for `D`. Default is 64-bit complex.
Returns
-------
D : np.ndarray [shape=(1 + n_fft/2, t), dtype=dtype]
STFT matrix
Raises
------
ParameterError
If `window` is supplied as a vector of length `n_fft`.
See Also
--------
istft : Inverse STFT
ifgram : Instantaneous frequency spectrogram
Examples
--------
>>> y, sr = librosa.load(librosa.util.example_audio_file())
>>> D = librosa.stft(y)
>>> D
array([[ 2.576e-03 -0.000e+00j, 4.327e-02 -0.000e+00j, ...,
3.189e-04 -0.000e+00j, -5.961e-06 -0.000e+00j],
[ 2.441e-03 +2.884e-19j, 5.145e-02 -5.076e-03j, ...,
-3.885e-04 -7.253e-05j, 7.334e-05 +3.868e-04j],
...,
[ -7.120e-06 -1.029e-19j, -1.951e-09 -3.568e-06j, ...,
-4.912e-07 -1.487e-07j, 4.438e-06 -1.448e-05j],
[ 7.136e-06 -0.000e+00j, 3.561e-06 -0.000e+00j, ...,
-5.144e-07 -0.000e+00j, -1.514e-05 -0.000e+00j]], dtype=complex64)
Use left-aligned frames, instead of centered frames
>>> D_left = librosa.stft(y, center=False)
Use a shorter hop length
>>> D_short = librosa.stft(y, hop_length=64)
Display a spectrogram
>>> import matplotlib.pyplot as plt
>>> librosa.display.specshow(librosa.logamplitude(np.abs(D)**2,
... ref_power=np.max),
... y_axis='log', x_axis='time')
>>> plt.title('Power spectrogram')
>>> plt.colorbar(format='%+2.0f dB')
>>> plt.tight_layout()
"""
# By default, use the entire frame
if win_length is None:
win_length = n_fft
# Set the default hop, if it's not already specified
if hop_length is None:
hop_length = int(win_length / 4)
if window is None:
# Default is an asymmetric Hann window
fft_window = scipy.signal.hann(win_length, sym=False)
elif six.callable(window):
# User supplied a window function
fft_window = window(win_length)
else:
# User supplied a window vector.
# Make sure it's an array:
fft_window = np.asarray(window)
# validate length compatibility
if fft_window.size != n_fft:
raise ParameterError('Size mismatch between n_fft and len(window)')
# Pad the window out to n_fft size
fft_window = pad_center(fft_window, n_fft)
# Reshape so that the window can be broadcast
fft_window = fft_window.reshape((-1, 1))
# Pad the time series so that frames are centered
if center:
valid_audio(y)
y = np.pad(y, int(n_fft // 2), mode='reflect')
# Window the time series.
y_frames = frame(y, frame_length=n_fft, hop_length=hop_length)
# Pre-allocate the STFT matrix
stft_matrix = np.empty((int(1 + n_fft // 2), y_frames.shape[1]),
dtype=dtype,
order='F')
# how many columns can we fit within MAX_MEM_BLOCK?
n_columns = int(MAX_MEM_BLOCK /
(stft_matrix.shape[0] * stft_matrix.itemsize))
for bl_s in range(0, stft_matrix.shape[1], n_columns):
bl_t = min(bl_s + n_columns, stft_matrix.shape[1])
# RFFT and Conjugate here to match phase from DPWE code
stft_matrix[:,
bl_s:bl_t] = fft.fft(fft_window * y_frames[:, bl_s:bl_t],
axis=0)[:stft_matrix.shape[0]].conj()
return stft_matrix
def istft_noDiv(stft_matrix,
hop_length=None,
win_length=None,
window=None,
center=True,
dtype=np.float32):
"""
#Copied from librosa's spectrum.py file, removing division by squared
window, which shouldn't be necessary and can cause problems in recon.
Inverse short-time Fourier transform (ISTFT).
Converts a complex-valued spectrogram `stft_matrix` to time-series `y`
by minimizing the mean squared error between `stft_matrix` and STFT of
`y` as described in [1]_.
In general, window function, hop length and other parameters should be same
as in stft, which mostly leads to perfect reconstruction of a signal from
unmodified `stft_matrix`.
Parameters
----------
stft_matrix : np.ndarray [shape=(1 + n_fft/2, t)]
STFT matrix from `stft`
hop_length : int > 0 [scalar]
Number of frames between STFT columns.
If unspecified, defaults to `win_length / 4`.
win_length : int <= n_fft = 2 * (stft_matrix.shape[0] - 1)
When reconstructing the time series, each frame is windowed
and each sample is normalized by the sum of squared window
according to the `window` function (see below).
If unspecified, defaults to `n_fft`.
window : None, function, np.ndarray [shape=(n_fft,)]
- None (default): use an asymmetric Hann window
- a window function, such as `scipy.signal.hanning`
- a user-specified window vector of length `n_fft`
center : boolean
- If `True`, `D` is assumed to have centered frames.
- If `False`, `D` is assumed to have left-aligned frames.
dtype : numeric type
Real numeric type for `y`. Default is 32-bit float.
Returns
-------
y : np.ndarray [shape=(n,)]
time domain signal reconstructed from `stft_matrix`
Raises
------
ParameterError
If `window` is supplied as a vector of length `n_fft`
See Also
--------
stft : Short-time Fourier Transform
Examples
--------
>>> y, sr = librosa.load(librosa.util.example_audio_file())
>>> D = librosa.stft(y)
>>> y_hat = librosa.istft(D)
>>> y_hat
array([ -4.812e-06, -4.267e-06, ..., 6.271e-06, 2.827e-07], dtype=float32)
Exactly preserving length of the input signal requires explicit padding.
Otherwise, a partial frame at the end of `y` will not be represented.
>>> n = len(y)
>>> n_fft = 2048
>>> y_pad = librosa.util.fix_length(y, n + n_fft // 2)
>>> D = librosa.stft(y_pad, n_fft=n_fft)
>>> y_out = librosa.util.fix_length(librosa.istft(D), n)
>>> np.max(np.abs(y - y_out))
1.4901161e-07
"""
n_fft = 2 * (stft_matrix.shape[0] - 1)
# By default, use the entire frame
if win_length is None:
win_length = n_fft
# Set the default hop, if it's not already specified
if hop_length is None:
hop_length = int(win_length / 4)
if window is None:
# Default is an asymmetric Hann window.
ifft_window = scipy.signal.hann(win_length, sym=False)
elif six.callable(window):
# User supplied a windowing function
ifft_window = window(win_length)
else:
# User supplied a window vector.
# Make it into an array
ifft_window = np.asarray(window)
# Verify that the shape matches
if ifft_window.size != n_fft:
raise ParameterError('Size mismatch between n_fft and window size')
# Pad out to match n_fft
ifft_window = util.pad_center(ifft_window, n_fft)
# scale the window
ifft_window = ifft_window * (2.0 / (win_length / hop_length))
n_frames = stft_matrix.shape[1]
expected_signal_len = n_fft + hop_length * (n_frames - 1)
y = np.zeros(expected_signal_len, dtype=dtype)
ifft_window_sum = np.zeros(expected_signal_len, dtype=dtype)
ifft_window_square = ifft_window * ifft_window
for i in range(n_frames):
sample = i * hop_length
spec = stft_matrix[:, i].flatten()
spec = np.concatenate((spec.conj(), spec[-2:0:-1]), 0)
ytmp = ifft_window * fft.ifft(spec).real
y[sample:(sample + n_fft)] = y[sample:(sample + n_fft)] + ytmp
# shouldn't need to do this sum of the squared window:
#ifft_window_sum[sample:(sample + n_fft)] += ifft_window_square
# don't do this:
## Normalize by sum of squared window
#approx_nonzero_indices = ifft_window_sum > util.SMALL_FLOAT
#y[approx_nonzero_indices] /= ifft_window_sum[approx_nonzero_indices]
if center:
y = y[int(n_fft // 2):-int(n_fft // 2)]
return y
def skip_if_callable(test, mycallable, reason):
if six.callable(mycallable):
if mycallable(test):
test.skipTest(reason)
return True
return False
def zero_crossings(y,
threshold=1e-10,
ref_magnitude=None,
pad=True,
zero_pos=True,
axis=-1):
'''Find the zero-crossings of a signal `y`: indices `i` such that
`sign(y[i]) != sign(y[j])`.
If `y` is multi-dimensional, then zero-crossings are computed along
the specified `axis`.
Parameters
----------
y : np.ndarray
The input array
threshold : float > 0 or None
If specified, values where `-threshold <= y <= threshold` are
clipped to 0.
ref_magnitude : float > 0 or callable
If numeric, the threshold is scaled relative to `ref_magnitude`.
If callable, the threshold is scaled relative to
`ref_magnitude(np.abs(y))`.
pad : boolean
If `True`, then `y[0]` is considered a valid zero-crossing.
zero_pos : boolean
If `True` then the value 0 is interpreted as having positive sign.
If `False`, then 0, -1, and +1 all have distinct signs.
axis : int
Axis along which to compute zero-crossings.
Returns
-------
zero_crossings : np.ndarray [shape=y.shape, dtype=boolean]
Indicator array of zero-crossings in `y` along the selected axis.
Notes
-----
This function caches at level 20.
Examples
--------
>>> # Generate a time-series
>>> y = np.sin(np.linspace(0, 4 * 2 * np.pi, 20))
>>> y
array([ 0.000e+00, 9.694e-01, 4.759e-01, -7.357e-01,
-8.372e-01, 3.247e-01, 9.966e-01, 1.646e-01,
-9.158e-01, -6.142e-01, 6.142e-01, 9.158e-01,
-1.646e-01, -9.966e-01, -3.247e-01, 8.372e-01,
7.357e-01, -4.759e-01, -9.694e-01, -9.797e-16])
>>> # Compute zero-crossings
>>> z = librosa.zero_crossings(y)
>>> z
array([ True, False, False, True, False, True, False, False,
True, False, True, False, True, False, False, True,
False, True, False, True], dtype=bool)
>>> # Stack y against the zero-crossing indicator
>>> np.vstack([y, z]).T
array([[ 0.000e+00, 1.000e+00],
[ 9.694e-01, 0.000e+00],
[ 4.759e-01, 0.000e+00],
[ -7.357e-01, 1.000e+00],
[ -8.372e-01, 0.000e+00],
[ 3.247e-01, 1.000e+00],
[ 9.966e-01, 0.000e+00],
[ 1.646e-01, 0.000e+00],
[ -9.158e-01, 1.000e+00],
[ -6.142e-01, 0.000e+00],
[ 6.142e-01, 1.000e+00],
[ 9.158e-01, 0.000e+00],
[ -1.646e-01, 1.000e+00],
[ -9.966e-01, 0.000e+00],
[ -3.247e-01, 0.000e+00],
[ 8.372e-01, 1.000e+00],
[ 7.357e-01, 0.000e+00],
[ -4.759e-01, 1.000e+00],
[ -9.694e-01, 0.000e+00],
[ -9.797e-16, 1.000e+00]])
>>> # Find the indices of zero-crossings
>>> np.nonzero(z)
(array([ 0, 3, 5, 8, 10, 12, 15, 17, 19]),)
'''
# Clip within the threshold
if threshold is None:
threshold = 0.0
if six.callable(ref_magnitude):
threshold = threshold * ref_magnitude(np.abs(y))
elif ref_magnitude is not None:
threshold = threshold * ref_magnitude
if threshold > 0:
y = y.copy()
y[np.abs(y) <= threshold] = 0
# Extract the sign bit
if zero_pos:
y_sign = np.signbit(y)
else:
y_sign = np.sign(y)
# Find the change-points by slicing
slice_pre = [slice(None)] * y.ndim
slice_pre[axis] = slice(1, None)
slice_post = [slice(None)] * y.ndim
slice_post[axis] = slice(-1)
# Since we've offset the input by one, pad back onto the front
padding = [(0, 0)] * y.ndim
padding[axis] = (1, 0)
return np.pad((y_sign[tuple(slice_post)] != y_sign[tuple(slice_pre)]),
padding,
mode='constant',
constant_values=pad)
def _crawl(self, url, **kwargs):
"""
real crawl API
checking kwargs, and repack them to each sub-dict
"""
task = {}
if kwargs.get('callback'):
callback = kwargs['callback']
if isinstance(callback, six.string_types) and hasattr(
self, callback):
func = getattr(self, callback)
elif six.callable(
callback) and six.get_method_self(callback) is self:
func = callback
kwargs['callback'] = func.__name__
else:
raise NotImplementedError("self.%s() not implemented!" %
callback)
if hasattr(func, '_config'):
for k, v in iteritems(func._config):
kwargs.setdefault(k, v)
for k, v in iteritems(self.crawl_config):
kwargs.setdefault(k, v)
url = quote_chinese(_build_url(url.strip(), kwargs.get('params')))
if kwargs.get('files'):
assert isinstance(
kwargs.get('data', {}),
dict), "data must be a dict when using with files!"
content_type, data = _encode_multipart_formdata(
kwargs.get('data', {}), kwargs.get('files', {}))
kwargs.setdefault('headers', {})
kwargs['headers']['Content-Type'] = content_type
kwargs['data'] = data
if kwargs.get('data'):
kwargs['data'] = _encode_params(kwargs['data'])
if kwargs.get('data'):
kwargs.setdefault('method', 'POST')
schedule = {}
for key in ('priority', 'retries', 'exetime', 'age', 'itag',
'force_update'):
if key in kwargs and kwargs[key] is not None:
schedule[key] = kwargs[key]
task['schedule'] = schedule
fetch = {}
for key in ('method', 'headers', 'data', 'timeout', 'allow_redirects',
'cookies', 'proxy', 'etag', 'last_modifed', 'save',
'js_run_at', 'js_script', 'load_images', 'fetch_type'):
if key in kwargs and kwargs[key] is not None:
fetch[key] = kwargs[key]
task['fetch'] = fetch
process = {}
for key in ('callback', ):
if key in kwargs and kwargs[key] is not None:
process[key] = kwargs[key]
task['process'] = process
task['project'] = self.project_name
task['url'] = url
task['taskid'] = task.get('taskid') or self.get_taskid(task)
cache_key = "%(project)s:%(taskid)s" % task
if cache_key not in self._follows_keys:
self._follows_keys.add(cache_key)
self._follows.append(task)
return task
def combine_pipeline(source, pipeline, debugger=None):
"""Build the "pipeline" generator.
Optionally a debugging object may be passed in.
>>> def step(iterator):
... for entry in iterator:
... yield entry
>>> def remover(iterator):
... for count, entry in enumerate(iterator):
... if count % 2 == 0:
... yield entry
>>> class Identity(PipelineStep):
... def process(self, iterator):
... for entry in iterator:
... yield entry
>>> assert hasattr(Identity, '__call__')
It's possible for iterator-factories to return lists of iterators. These
will then be flattened into the pipeline in the order of occurrence.
>>> pipeline = [step, [remover, Identity()], step]
Example: We construct 1000 elements and pass it through our pipeline. As
the remover removes every second element we can expect to have only half
as many elements out than we sent in.
>>> gen = combine_pipeline(range(1000), pipeline)
>>> assert len(list(gen)) == 500
The `source` can also be a callable. If it is, `combine_pipeline` will also
return a callable, which - when called - will pass the supplied arguments
on to the source callable.
>>> gen = combine_pipeline(range, pipeline)
>>> assert len(list(gen(1000))) == 500
"""
def identity(x):
return x
if debugger is not None:
head = debugger.head
tail = debugger.tail
track = debugger.track
report = debugger.report
else:
head = tail = track = report = identity
gen = source
if _six.callable(gen):
# Source is (hopefully) an iterator-factory, so we need
# to convert all our steps into iterator-factories
# as well, to enable passing on the arguments passed
# to our collapsed pipeline.
defer = defer_call
else:
defer = identity
track = defer(track)
head = defer(head)
tail = defer(tail)
report = defer(report)
gen = head(gen)
for step in recursive_flatten(pipeline):
gen = defer(step)(track(gen))
return report(tail(track(gen)))
def contains(self, mouseevent):
"""
Test whether the mouse event occurred on the line. The pick
radius determines the precision of the location test (usually
within five points of the value). Use
:meth:`~matplotlib.lines.Line2D.get_pickradius` or
:meth:`~matplotlib.lines.Line2D.set_pickradius` to view or
modify it.
Returns *True* if any values are within the radius along with
``{'ind': pointlist}``, where *pointlist* is the set of points
within the radius.
TODO: sort returned indices by distance
"""
if six.callable(self._contains):
return self._contains(self, mouseevent)
if not is_numlike(self.pickradius):
raise ValueError("pick radius should be a distance")
# Make sure we have data to plot
if self._invalidy or self._invalidx:
self.recache()
if len(self._xy) == 0:
return False, {}
# Convert points to pixels
transformed_path = self._get_transformed_path()
path, affine = transformed_path.get_transformed_path_and_affine()
path = affine.transform_path(path)
xy = path.vertices
xt = xy[:, 0]
yt = xy[:, 1]
# Convert pick radius from points to pixels
if self.figure is None:
warnings.warn('no figure set when check if mouse is on line')
pixels = self.pickradius
else:
pixels = self.figure.dpi / 72. * self.pickradius
# the math involved in checking for containment (here and inside of
# segment_hits) assumes that it is OK to overflow. In case the
# application has set the error flags such that an exception is raised
# on overflow, we temporarily set the appropriate error flags here and
# set them back when we are finished.
olderrflags = np.seterr(all='ignore')
try:
# Check for collision
if self._linestyle in ['None', None]:
# If no line, return the nearby point(s)
d = (xt - mouseevent.x)**2 + (yt - mouseevent.y)**2
ind, = np.nonzero(np.less_equal(d, pixels**2))
else:
# If line, return the nearby segment(s)
ind = segment_hits(mouseevent.x, mouseevent.y, xt, yt, pixels)
finally:
np.seterr(**olderrflags)
ind += self.ind_offset
# Debugging message
if False and self._label != '':
print("Checking line", self._label, "at", mouseevent.x,
mouseevent.y)
print('xt', xt)
print('yt', yt)
#print 'dx,dy', (xt-mouseevent.x)**2., (yt-mouseevent.y)**2.
print('ind', ind)
# Return the point(s) within radius
return len(ind) > 0, dict(ind=ind)
def _crawl(self, url, **kwargs):
"""
real crawl API
checking kwargs, and repack them to each sub-dict
"""
task = {}
assert len(url) < 1024, "Maximum (1024) URL length error."
if kwargs.get('callback'):
callback = kwargs['callback']
if isinstance(callback, six.string_types) and hasattr(
self, callback):
func = getattr(self, callback)
elif six.callable(
callback) and six.get_method_self(callback) is self:
func = callback
kwargs['callback'] = func.__name__
else:
raise NotImplementedError("self.%s() not implemented!" %
callback)
if hasattr(func, '_config'):
for k, v in iteritems(func._config):
if isinstance(v, dict) and isinstance(kwargs.get(k), dict):
kwargs[k].update(v)
else:
kwargs.setdefault(k, v)
url = quote_chinese(_build_url(url.strip(), kwargs.pop('params',
None)))
if kwargs.get('files'):
assert isinstance(
kwargs.get('data', {}),
dict), "data must be a dict when using with files!"
content_type, data = _encode_multipart_formdata(
kwargs.pop('data', {}), kwargs.pop('files', {}))
kwargs.setdefault('headers', {})
kwargs['headers']['Content-Type'] = content_type
kwargs['data'] = data
if kwargs.get('data'):
kwargs['data'] = _encode_params(kwargs['data'])
if kwargs.get('data'):
kwargs.setdefault('method', 'POST')
if kwargs.get('user_agent'):
kwargs.setdefault('headers', {})
kwargs['headers']['User-Agent'] = kwargs.get('user_agent')
schedule = {}
for key in self.schedule_fields:
if key in kwargs:
schedule[key] = kwargs.pop(key)
elif key in self.crawl_config:
schedule[key] = self.crawl_config[key]
task['schedule'] = schedule
fetch = {}
for key in self.fetch_fields:
if key in kwargs:
fetch[key] = kwargs.pop(key)
task['fetch'] = fetch
process = {}
for key in self.process_fields:
if key in kwargs:
process[key] = kwargs.pop(key)
task['process'] = process
task['project'] = self.project_name
task['url'] = url
if 'taskid' in kwargs:
task['taskid'] = kwargs.pop('taskid')
else:
task['taskid'] = self.get_taskid(task)
if kwargs:
raise TypeError('crawl() got unexpected keyword argument: %s' %
kwargs.keys())
if self.is_debugger():
task = self.task_join_crawl_config(task, self.crawl_config)
cache_key = "%(project)s:%(taskid)s" % task
if cache_key not in self._follows_keys:
self._follows_keys.add(cache_key)
self._follows.append(task)
return task
def wraps(callable_thing):
if isinstance(callable_thing, _types.FunctionType):
return _functools.wraps(callable_thing)
else:
assert _six.callable(callable_thing)
return identity_step
def replica_device_setter(ps_tasks=0,
ps_device="/job:ps",
worker_device="/job:worker",
merge_devices=True,
cluster=None,
ps_ops=None,
ps_strategy=None):
"""Return a `device function` to use when building a Graph for replicas.
Device Functions are used in `with tf.device(device_function):` statement to
automatically assign devices to `Operation` objects as they are constructed,
Device constraints are added from the inner-most context first, working
outwards. The merging behavior adds constraints to fields that are yet unset
by a more inner context. Currently the fields are (job, task, cpu/gpu).
If `cluster` is `None`, and `ps_tasks` is 0, the returned function is a no-op.
Otherwise, the value of `ps_tasks` is derived from `cluster`.
By default, only Variable ops are placed on ps tasks, and the placement
strategy is round-robin over all ps tasks. A custom `ps_strategy` may be used
to do more intelligent placement, such as
`tf.contrib.training.GreedyLoadBalancingStrategy`.
For example,
```python
# To build a cluster with two ps jobs on hosts ps0 and ps1, and 3 worker
# jobs on hosts worker0, worker1 and worker2.
cluster_spec = {
"ps": ["ps0:2222", "ps1:2222"],
"worker": ["worker0:2222", "worker1:2222", "worker2:2222"]}
with
tf.device(tf.compat.v1.train.replica_device_setter(cluster=cluster_spec)):
# Build your graph
v1 = tf.Variable(...) # assigned to /job:ps/task:0
v2 = tf.Variable(...) # assigned to /job:ps/task:1
v3 = tf.Variable(...) # assigned to /job:ps/task:0
# Run compute
```
Args:
ps_tasks: Number of tasks in the `ps` job. Ignored if `cluster` is
provided.
ps_device: String. Device of the `ps` job. If empty no `ps` job is used.
Defaults to `ps`.
worker_device: String. Device of the `worker` job. If empty no `worker`
job is used.
merge_devices: `Boolean`. If `True`, merges or only sets a device if the
device constraint is completely unset. merges device specification rather
than overriding them.
cluster: `ClusterDef` proto or `ClusterSpec`.
ps_ops: List of strings representing `Operation` types that need to be
placed on `ps` devices. If `None`, defaults to `STANDARD_PS_OPS`.
ps_strategy: A callable invoked for every ps `Operation` (i.e. matched by
`ps_ops`), that takes the `Operation` and returns the ps task index to
use. If `None`, defaults to a round-robin strategy across all `ps`
devices.
Returns:
A function to pass to `tf.device()`.
Raises:
TypeError if `cluster` is not a dictionary or `ClusterDef` protocol buffer,
or if `ps_strategy` is provided but not a callable.
"""
if cluster is not None:
if isinstance(cluster, server_lib.ClusterSpec):
cluster_spec = cluster.as_dict()
else:
cluster_spec = server_lib.ClusterSpec(cluster).as_dict()
# Get ps_job_name from ps_device by stripping "/job:".
ps_job_name = pydev.DeviceSpec.from_string(ps_device).job
if ps_job_name not in cluster_spec or cluster_spec[ps_job_name] is None:
return None
ps_tasks = len(cluster_spec[ps_job_name])
if ps_tasks == 0:
return None
if ps_ops is None:
# TODO(sherrym): Variables in the LOCAL_VARIABLES collection should not be
# placed in the parameter server.
ps_ops = list(STANDARD_PS_OPS)
if not merge_devices:
logging.warning(
"DEPRECATION: It is recommended to set merge_devices=true in "
"replica_device_setter")
if ps_strategy is None:
ps_strategy = _RoundRobinStrategy(ps_tasks)
if not six.callable(ps_strategy):
raise TypeError("ps_strategy must be callable")
chooser = _ReplicaDeviceChooser(ps_tasks, ps_device, worker_device,
merge_devices, ps_ops, ps_strategy)
return chooser.device_function
def _crawl(self, url, **kwargs):
"""
real crawl API
checking kwargs, and repack them to each sub-dict
"""
task = {}
assert len(url) < 1024, "Maximum (1024) URL length error."
if kwargs.get('callback'):
callback = kwargs['callback']
if isinstance(callback, six.string_types) and hasattr(
self, callback):
func = getattr(self, callback)
elif six.callable(
callback) and six.get_method_self(callback) is self:
func = callback
kwargs['callback'] = func.__name__
else:
raise NotImplementedError("self.%s() not implemented!" %
callback)
if hasattr(func, '_config'):
for k, v in iteritems(func._config):
if isinstance(v, dict) and isinstance(kwargs.get(k), dict):
kwargs[k].update(v)
else:
kwargs.setdefault(k, v)
for k, v in iteritems(self.crawl_config):
if isinstance(v, dict) and isinstance(kwargs.get(k), dict):
kwargs[k].update(v)
else:
kwargs.setdefault(k, v)
url = quote_chinese(_build_url(url.strip(), kwargs.pop('params',
None)))
if kwargs.get('files'):
assert isinstance(
kwargs.get('data', {}),
dict), "data must be a dict when using with files!"
content_type, data = _encode_multipart_formdata(
kwargs.pop('data', {}), kwargs.pop('files', {}))
kwargs.setdefault('headers', {})
kwargs['headers']['Content-Type'] = content_type
kwargs['data'] = data
if kwargs.get('data'):
kwargs['data'] = _encode_params(kwargs['data'])
if kwargs.get('data'):
kwargs.setdefault('method', 'POST')
schedule = {}
for key in ('priority', 'retries', 'exetime', 'age', 'itag',
'force_update', 'auto_recrawl', 'cancel'):
if key in kwargs:
schedule[key] = kwargs.pop(key)
task['schedule'] = schedule
fetch = {}
for key in ('method', 'headers', 'data', 'connect_timeout', 'timeout',
'allow_redirects', 'cookies', 'proxy', 'etag',
'last_modifed', 'last_modified', 'save', 'js_run_at',
'js_script', 'js_viewport_width', 'js_viewport_height',
'load_images', 'fetch_type', 'use_gzip', 'validate_cert',
'max_redirects', 'robots_txt'):
if key in kwargs:
fetch[key] = kwargs.pop(key)
task['fetch'] = fetch
process = {}
for key in ('callback', ):
if key in kwargs:
process[key] = kwargs.pop(key)
task['process'] = process
task['project'] = self.project_name
task['url'] = url
if 'taskid' in kwargs:
task['taskid'] = kwargs.pop('taskid')
else:
task['taskid'] = self.get_taskid(task)
if kwargs:
raise TypeError('crawl() got unexpected keyword argument: %s' %
kwargs.keys())
cache_key = "%(project)s:%(taskid)s" % task
if cache_key not in self._follows_keys:
self._follows_keys.add(cache_key)
self._follows.append(task)
return task
def css_raw(self):
if six.callable(self._css_raw):
self._css_raw = self._css_raw()
return self._css_raw
def fBmnd(shape,
power_spectrum,
unit_length=1,
seed=None,
statistic=np.random.normal,
fft=np.fft,
fft_args=dict()):
"""
Generates a field given a stastitic and a power_spectrum.
author: A. Marchal based on FieldGenerator (C. Cadiou) and MAMDLIB
Parameters
----------
statistic: callable
A function that takes returns a random array of a given signature,
with signature (s) -> (B) with s == B.shape
shape: tuple
The shape of the output field
unit_length: float
How much physical length represent 1pixel. For example a value of 10
mean that each pixel stands for 10 physical units. It has the
dimension of a physical_unit/pixel.
fft: a numpy-like fft API
fft_args: array
a dictionary of kwargs to pass to the FFT calls
Returns:
--------
field: a real array of shape `shape` following the statistic
with the given power_spectrum
"""
if seed is not None: np.random.seed(seed)
if not six.callable(statistic):
raise Exception('`statistic` should be callable')
# Draw a random sample
normal = statistic(size=shape)
# Compute the FFT of the field and take the phase
phase = np.angle(fft.fftn(normal, **fft_args))
try:
fftfreq = fft.fftfreq
except:
# Fallback on numpy for the frequencies
fftfreq = np.fft.fftfreq
# Compute the k grid
ks = [np.fft.fftshift(np.fft.fftfreq(s, d=unit_length)) for s in shape]
if len(ks) == 3:
kgrid = np.meshgrid(ks[1], ks[0], ks[2])
else:
kgrid = np.meshgrid(*ks)
knorm = np.sqrt(np.sum(np.power(kgrid, 2), axis=0))
power_k = np.where(knorm == 0, 0, np.sqrt(power_spectrum(knorm)))
imfft = np.zeros(shape, dtype=complex)
imfft.real = power_k * np.cos(phase)
imfft.imag = power_k * np.sin(phase)
return fft.ifftn(np.fft.ifftshift(imfft)).real
def get_window(window, Nx, fftbins=True):
'''Compute a window function.
This is a wrapper for `scipy.signal.get_window` that additionally
supports callable or pre-computed windows.
Parameters
----------
window : string, tuple, number, callable, or list-like
The window specification:
- If string, it's the name of the window function (e.g., `'hann'`)
- If tuple, it's the name of the window function and any parameters
(e.g., `('kaiser', 4.0)`)
- If numeric, it is treated as the beta parameter of the `'kaiser'`
window, as in `scipy.signal.get_window`.
- If callable, it's a function that accepts one integer argument
(the window length)
- If list-like, it's a pre-computed window of the correct length `Nx`
Nx : int > 0
The length of the window
fftbins : bool, optional
If True (default), create a periodic window for use with FFT
If False, create a symmetric window for filter design applications.
Returns
-------
get_window : np.ndarray
A window of length `Nx` and type `window`
See Also
--------
scipy.signal.get_window
Notes
-----
This function caches at level 10.
Raises
------
ParameterError
If `window` is supplied as a vector of length != `n_fft`,
or is otherwise mis-specified.
'''
if six.callable(window):
return window(Nx)
elif (isinstance(window, (six.string_types, tuple)) or
np.isscalar(window)):
# TODO: if we add custom window functions in librosa, call them here
return scipy.signal.get_window(window, Nx, fftbins=fftbins)
elif isinstance(window, (np.ndarray, list)):
if len(window) == Nx:
return np.asarray(window)
raise ParameterError('Window size mismatch: '
'{:d} != {:d}'.format(len(window), Nx))
else:
raise ParameterError('Invalid window specification: {}'.format(window))
def _maybe_model_repr(obj):
if hasattr(obj, '_repr_model_') and six.callable(obj._repr_model_):
return obj._repr_model_()
return obj
def power_to_db(S, ref=1.0, amin=1e-10, top_db=80.0):
"""Convert a power spectrogram (amplitude squared) to decibel (dB) units
This computes the scaling ``10 * log10(S / ref)`` in a numerically
stable way.
Parameters
----------
S : np.ndarray
input power
ref : scalar or callable
If scalar, the amplitude `abs(S)` is scaled relative to `ref`:
`10 * log10(S / ref)`.
Zeros in the output correspond to positions where `S == ref`.
If callable, the reference value is computed as `ref(S)`.
amin : float > 0 [scalar]
minimum threshold for `abs(S)` and `ref`
top_db : float >= 0 [scalar]
threshold the output at `top_db` below the peak:
``max(10 * log10(S)) - top_db``
Returns
-------
S_db : np.ndarray
``S_db ~= 10 * log10(S) - 10 * log10(ref)``
See Also
--------
perceptual_weighting
db_to_power
amplitude_to_db
db_to_amplitude
Notes
-----
This function caches at level 30.
Examples
--------
Get a power spectrogram from a waveform ``y``
>>> y, sr = librosa.load(librosa.util.example_audio_file())
>>> S = np.abs(librosa.stft(y))
>>> librosa.power_to_db(S**2)
array([[-33.293, -27.32 , ..., -33.293, -33.293],
[-33.293, -25.723, ..., -33.293, -33.293],
...,
[-33.293, -33.293, ..., -33.293, -33.293],
[-33.293, -33.293, ..., -33.293, -33.293]], dtype=float32)
Compute dB relative to peak power
>>> librosa.power_to_db(S**2, ref=np.max)
array([[-80. , -74.027, ..., -80. , -80. ],
[-80. , -72.431, ..., -80. , -80. ],
...,
[-80. , -80. , ..., -80. , -80. ],
[-80. , -80. , ..., -80. , -80. ]], dtype=float32)
Or compare to median power
>>> librosa.power_to_db(S**2, ref=np.median)
array([[-0.189, 5.784, ..., -0.189, -0.189],
[-0.189, 7.381, ..., -0.189, -0.189],
...,
[-0.189, -0.189, ..., -0.189, -0.189],
[-0.189, -0.189, ..., -0.189, -0.189]], dtype=float32)
And plot the results
>>> import matplotlib.pyplot as plt
>>> plt.figure()
>>> plt.subplot(2, 1, 1)
>>> librosa.display.specshow(S**2, sr=sr, y_axis='log')
>>> plt.colorbar()
>>> plt.title('Power spectrogram')
>>> plt.subplot(2, 1, 2)
>>> librosa.display.specshow(librosa.power_to_db(S**2, ref=np.max),
... sr=sr, y_axis='log', x_axis='time')
>>> plt.colorbar(format='%+2.0f dB')
>>> plt.title('Log-Power spectrogram')
>>> plt.tight_layout()
"""
S = np.asarray(S)
if amin <= 0:
raise ParameterError('amin must be strictly positive')
if np.issubdtype(S.dtype, np.complexfloating):
warnings.warn(
'power_to_db was called on complex input so phase '
'information will be discarded. To suppress this warning, '
'call power_to_db(magphase(D, power=2)[0]) instead.')
magnitude = np.abs(S)
else:
magnitude = S
if six.callable(ref):
# User supplied a function to calculate reference power
ref_value = ref(magnitude)
else:
ref_value = np.abs(ref)
log_spec = 10.0 * np.log10(np.maximum(amin, magnitude))
log_spec -= 10.0 * np.log10(np.maximum(amin, ref_value))
if top_db is not None:
if top_db < 0:
raise ParameterError('top_db must be non-negative')
log_spec = np.maximum(log_spec, log_spec.max() - top_db)
return log_spec
def getConsumer(self, consumerId, username=None, password=None):
if hasattr(self, 'consumer') and self.consumer:
return self.consumer
if six.callable(self.registered_consumer_info):
return self.registered_consumer_info()
return self.registered_consumer_info
